mhealth and health information technology tools for...

Journal of Diabetes Research

mHealth and Health Information Technology Tools for Diverse Patients with Diabetes

Guest Editors: Courtney Lyles, Neda Ratanawongsa, Shari Bolen, and Lipika Samal

mHealth and Health Information TechnologyTools for Diverse Patients with Diabetes

Journal of Diabetes Research

mHealth and Health Information TechnologyTools for Diverse Patients with Diabetes

Guest Editors: Courtney Lyles, Neda Ratanawongsa,Shari Bolen, and Lipika Samal

Copyright © 2017 Hindawi Publishing Corporation. All rights reserved.

This is a special issue published in “Journal of Diabetes Research.” All articles are open access articles distributed under the CreativeCommons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the originalwork is properly cited.

Editorial Board

Steven F. Abcouwer, USAReza Abdi, USAAbdelaziz Amrani, CanadaFabrizio Barbetti, ItalyIrene D. Blackberry, AustraliaSimona Bo, ItalySihem Boudina, USAMonica Bullo, SpainStefania Camastra, ItalyNorman Cameron, UKIlaria Campesi, ItalyRiccardo Candido, ItalyBrunella Capaldo, ItalyDanila Capoccia, ItalySergiu Catrina, SwedenSubrata Chakrabarti, CanadaMunmun Chattopadhyay, USAEusebio Chiefari, ItalySecundino Cigarran, SpainKim Connelly, CanadaLaurent Crenier, BelgiumChristophe De Block, BelgiumDevon A. Dobrosielski, USAKhalid M. Elased, USAUlf J. Eriksson, SwedenPaolo Fiorina, USAAndrea Flex, ItalyDaniela Foti, ItalyGeorgia Fousteri, ItalyMaria Pia Francescato, ItalyPedro M. Geraldes, CanadaMargalit D. Goldfracht, Israel

Thomas Haak, GermanyThomas J. Hawke, CanadaOle Kristian Hejlesen, DenmarkDario Iafusco, ItalyKonstantinos Kantartzis, GermanyDaisuke Koya, JapanFrida Leonetti, ItalySandra MacRury, UKAfshan Malik, UKRoberto Mallone, FranceRaffaele Marfella, ItalyCarlos Martinez Salgado, SpainLucy Marzban, CanadaRaffaella Mastrocola, ItalyDavid Meyre, CanadaMaria G. Montez, USAStephan Morbach, GermanyJiro Nakamura, JapanPratibha V. Nerurkar, USAMonika A. Niewczas, USAFrancisco Javier Nóvoa, SpainCraig S. Nunemaker, USAHiroshi Okamoto, JapanIke S. Okosun, USAFernando Ovalle, USAJun Panee, USACesare Patrone, SwedenSubramaniam Pennathur, USAAndreas Pfützner, GermanyBernard Portha, FranceEd Randell, CanadaJordi Lluis Reverter, Spain

Ute Christine Rogner, FranceUlrike Rothe, GermanyToralph Ruge, SwedenChristoph H. Saely, AustriaPonnusamy Saravanan, UKToshiyasu Sasaoka, JapanAndrea Scaramuzza, ItalyYael Segev, IsraelSuat Simsek, NetherlandsMarco Songini, ItalyJanet H. Southerland, USADavid Strain, UKKiyoshi Suzuma, JapanGiovanni Targher, ItalyPatrizio Tatti, ItalyFarook Thameem, USAMichael J. Theodorakis, UKPeter Thule, USAAndrea Tura, ItalyRuben Varela-Calvino, SpainChristian Wadsack, AustriaMatthias Weck, GermanyPer Westermark, SwedenJennifer L. Wilkinson-Berka, AustraliaDane K. Wukich, USAKazuya Yamagata, JapanShi Fang Yan, USAMark A. Yorek, USALiping Yu, USADavid Zangen, IsraelThomas J. Zgonis, USADan Ziegler, Germany

Contents

mHealth and Health Information Technology Tools for Diverse Patients with DiabetesCourtney R. Lyles, Neda Ratanawongsa, Shari D. Bolen, and Lipika SamalVolume 2017, Article ID 1704917, 3 pages

TheNext Frontier in Communication and the ECLIPPSE Study: Bridging the Linguistic Divide inSecure MessagingDean Schillinger, Danielle McNamara, Scott Crossley, Courtney Lyles, Howard H. Moffet, Urmimala Sarkar,Nicholas Duran, Jill Allen, Jennifer Liu, Danielle Oryn, Neda Ratanawongsa, and Andrew J. KarterVolume 2017, Article ID 1348242, 9 pages

TheChallenges of Electronic Health Records and Diabetes Electronic Prescribing: Implications forSafety Net Care for Diverse PopulationsNeda Ratanawongsa, Lenny L. S. Chan, Michelle M. Fouts, and Elizabeth J. MurphyVolume 2017, Article ID 8983237, 7 pages

TheDesign, Usability, and Feasibility of a Family-Focused Diabetes Self-Care Support mHealthIntervention for Diverse, Low-Income Adults with Type 2 DiabetesLindsay Satterwhite Mayberry, Cynthia A. Berg, Kryseana J. Harper, and Chandra Y. OsbornVolume 2016, Article ID 7586385, 13 pages

Adaptation and Feasibility Study of a Digital Health Program to Prevent Diabetes among Low-IncomePatients: Results from a Partnership between a Digital Health Company and an Academic ResearchTeamValy Fontil, Kelly McDermott, Lina Tieu, Christina Rios,Eliza Gibson, Cynthia Castro Sweet, Mike Payne, and Courtney R. LylesVolume 2016, Article ID 8472391, 10 pages

Linking High Risk PostpartumWomen with a Technology Enabled Health Coaching Program toReduce Diabetes Risk and ImproveWellbeing: ProgramDescription, Case Studies, andRecommendations for Community Health Coaching ProgramsPriyanka Athavale, Melanie Thomas, Adriana T. Delgadillo-Duenas, Karen Leong, Adriana Najmabadi,Elizabeth Harleman, Christina Rios, Judy Quan, Catalina Soria, and Margaret A. HandleyVolume 2016, Article ID 4353956, 16 pages

Pilot Study of a Web-Delivered Multicomponent Intervention for Rural Teens with Poorly ControlledType 1 DiabetesAmy Hughes Lansing, Catherine Stanger, Alan Budney, Ann S. Christiano, and Samuel J. CasellaVolume 2016, Article ID 7485613, 8 pages

EditorialmHealth and Health Information Technology Tools forDiverse Patients with Diabetes

Courtney R. Lyles,1 Neda Ratanawongsa,1 Shari D. Bolen,2 and Lipika Samal3

1University of California, San Francisco, CA, USA2Case Western Reserve University, Cleveland, OH, USA3Harvard Medical School, Boston, MA, USA

Correspondence should be addressed to Courtney R. Lyles; [email protected]

Received 5 February 2017; Accepted 5 February 2017; Published 23 February 2017

Copyright © 2017 Courtney R. Lyles et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Mobile health (mHealth) and health information technology(HIT) tools to enhance diabetes health and healthcare man-agement have proliferated rapidly, including websites, mobilephone applications, texting or interactive voice responsephone calls, remote monitoring devices/sensors, and per-sonal health records (PHRs) linked to electronic healthrecords [1, 2]. Many studies and systematic reviews havedemonstrated that the additional communication and sup-port provided by such technologies can improve outcomeslike patient confidence, self-management, quality of life, andeven health outcomes like glycemic control [3–10].

However, emerging evidence reveals a digital divide inhealth technology use, with lower use of widely dissemi-nated technologies among racial/ethnic minority groups orthose who have limited health literacy [11–14]. Althoughoverall ownership and use of devices are increasing amongracial/ethnic minorities, lower income individuals, and othersubgroups [15, 16], there remain access, skills, and interestbarriers that influence this overall digital divide [17–19].Furthermore, research in mHealth or HIT has not oftendirectly engaged diverse end users, as evidenced by fewpublished studies which report that the usability of diabetestechnologies among participants represents the spectrum oftechnological proficiency or income and educational attain-ment [20].

This special issue therefore provides crucial evidenceabout the design, testing, and implementation of mHealth orhealth information technology platforms for diverse patients

with diabetes. The included studies cover a range of relevantresearch on these topics.

Two studies describe HIT-facilitated interventions toenhance diabetes self-management support by engaging bothpatients and their families. L. S. Mayberry et al. describe anapproach to user-centered design and iterative usability test-ing with low-income patients to develop an mHealth inter-vention to promote family engagement in self-managementsupport. Although nontechnology facilitators may be neededto engage social support networks in diabetes self-care forthis population, their findings suggest preliminary feasibilityfor low-income patients to engage in this text messagingself-management support. Meanwhile, A. H. Lansing et al.describe the feasibility of engaging rural teens with poorlycontrolled type 1 diabetes and their families through anInternet-delivered intervention to improve blood glucoseself-monitoring and glycemic control. Their findings promptfuture research directions about sustainability and scalability,particularly to teens and families who may not be as easilyincentivized in similar web-delivered self-management pro-grams.

Two studies describe novel approaches to tailoringmHealth interventions for culture and language. V. Fontilet al. describe an approach to leveraging academic-industrypartnership in developing a culturally and linguisticallyappropriate diabetes prevention program tailored to safetynet patients with limited health literacy. While their findingsare limited to a small sample drawn from an academically

Hindawi Publishing CorporationJournal of Diabetes ResearchVolume 2017, Article ID 1704917, 3 pageshttp://dx.doi.org/10.1155/2017/1704917

http://dx.doi.org/10.1155/2017/1704917

2 Journal of Diabetes Research

affiliated clinic, the approach suggests a model for engaginghealth technology companies in designing products that willdecrease the digital divide. P. Athavale et al. describe a healthcoaching intervention facilitated by automated telemedicineoutreach for reducing diabetes among postpartum Latinawomen.They address the advantages and limitations of usingHIT to improve the scalability of health coaching throughcommunity organizations like local Women, Infants, andChildren (WIC) Programs in an effort to reach vulnerablewomen at high risk of loss to follow-up.

Finally, two papers describe future directions for researchin the design and implementation of online patient portalsand electronic health record systems (EHRs). D. Schillinger etal. describe a research protocol to partner with computationallinguistics experts to study the linguistic complexity of securemessages between diverse patients with diabetes and theirhealthcare teams. They propose using this novel approachto quantify and study health literacy at a population level,while also developing tools to help care teams tailor theirsecure message content. N. Ratanawongsa et al. describevulnerabilities in the ways EHR electronic prescribing hasaffected diabetes care for diverse patients and advocate forspecific changes in EHR design, implementation, policy, andresearch.

It is well known that many existing technological inter-ventions have not seen wide uptake among heterogeneoussettings and patient populations [21] and multifaceted, real-world research strategies that can create insights formeaning-ful change in the near future. Overall, we believe this specialissue offers innovative approaches for including diversepopulations in mHealth and health technology researchand inspires future informatics, implementation, and policyresearchers to build on this important work. Notably, nointerventions focused on using aggregated data from mobiletechnology or social media to design interventions. Forinstance, use of aggregated data on opportunities for healthyfood and safe places to be active could be used to design futurepublic health interventions to improve the built environment.Moving forward, we must continue these multiple strandsof research to truly advance the field: from discovery ofnew technology programs that impact health behaviors, toadaptation of existing technologies for diverse user needs, tocareful consideration of implementation strategies thatmightdifferentially impact patient subgroups. Findings from thesestudies also indicate that policy work around increasing andsustaining low-cost broadband access will also be critical tothe future success of interventions to improve care and reducedisparities using HIT.

Courtney R. LylesNeda Ratanawongsa

Shari D. BolenLipika Samal

References

[1] S. R. Steinhubl, E. D. Muse, and E. J. Topol, “Can mobile healthtechnologies transform health care?” JAMA, vol. 310, no. 22, pp.2395–2396, 2013.

[2] R. Hillestad, J. Bigelow, A. Bower et al., “Can electronic medicalrecord systems transform health care? Potential health benefits,savings, and costs,” Health Affairs, vol. 24, no. 5, pp. 1103–1117,2005.

[3] P. C. Tang and D. Lansky, “The missing link: bridging thepatient-provider health information gap,”Health Affairs, vol. 24,no. 5, pp. 1290–1295, 2005.

[4] K. Pal, S. V. Eastwood, S. Michie et al., “Computer-based inter-ventions to improve self-management in adults with type 2diabetes: a systematic review andmeta-analysis,”Diabetes Care,vol. 37, no. 6, pp. 1759–1766, 2014.

[5] J. D. Ralston, I. B. Hirsch, J. Hoath, M. Mullen, A. Cheadle,and H. I. Goldberg, “Web-based collaborative care for type 2diabetes: a pilot randomized trial,” Diabetes Care, vol. 32, no. 2,pp. 234–239, 2009.

[6] D. Schillinger, F.Wang,M.Handley, andH.Hammer, “Effects ofself-management support on structure, process, and outcomesamong vulnerable patients with diabetes: a three-arm practicalclinical trial,” Diabetes Care, vol. 32, no. 4, pp. 559–566, 2009.

[7] A. K. Hall, H. Cole-Lewis, and J. M. Bernhardt, “Mobile textmessaging for health: a systematic review of reviews,” AnnualReview of Public Health, vol. 36, pp. 393–415, 2015.

[8] S. Hamine, E. Gerth-Guyette, D. Faulx, B. B. Green, and A.S. Ginsburg, “Impact of mHealth chronic disease managementon treatment adherence and patient outcomes: a systematicreview,” Journal of Medical Internet Research, vol. 17, no. 2, 2015.

[9] B. M. Costa, K. J. Fitzgerald, K. M. Jones, and T. Dunning Am,“Effectiveness of IT-based diabetes management interventions:a review of the literature,” BMC Family Practice, vol. 10, articleno. 72, 2009.

[10] C. Y. Osborn, L. S. Mayberry, S. A. Mulvaney, and R. Hess,“Patient web portals to improve diabetes outcomes: a systematicreview,” Current Diabetes Reports, vol. 10, no. 6, pp. 422–435,2010.

[11] U. Sarkar, A. J. Karter, J. Y. Liu et al., “The literacy divide: healthliteracy and the use of an internet-based patient portal in anintegrated health system-results from the diabetes study ofNorthern California (DISTANCE),” Journal of Health Commu-nication, vol. 15, no. 2, pp. 183–196, 2010.

[12] U. Sarkar, A. J. Karter, J. Y. Liu et al., “Social disparities ininternet patient portal use in diabetes: evidence that the digitaldivide extends beyond access,” Journal of the American MedicalInformatics Association, vol. 18, no. 3, pp. 318–321, 2011.

[13] C. K. Yamin, S. Emani, D. H. Williams et al., “The digital dividein adoption and use of a personal health record,” Archives ofInternal Medicine, vol. 171, no. 6, pp. 568–574, 2011.

[14] T. Isaacs, D. Hunt, D.Ward, L. Rooshenas, and L. Edwards, “Theinclusion of ethnic minority patients and the role of language intelehealth trials for type 2 diabetes: a systematic review,” Journalof Medical Internet Research, vol. 18, no. 9, Article ID e256, 2016.

[15] Pew Internet & American Life Project,Digital Differences, 2012,http://pewinternet.org/Reports/2012/Digital-differences.aspx.

[16] Pew Research Internet Project, Internet User Demographics:January 2014, 2014, http://www.pewinternet.org/data-trend/internet-use/latest-stats/.

[17] K.Mossberger, C. J. Tolbert, D. Bowen, and B. Jimenez, “Unrav-eling different barriers to internet use: urban residents andneighborhood effects,” Urban Affairs Review, vol. 48, no. 6, pp.771–810, 2012.

[18] R. F. McCloud, C. A. Okechukwu, G. Sorensen, and K. Viswa-nath, “Beyond access: barriers to internet health information

http://pewinternet.org/Reports/2012/Digital-differences.aspx

http://www.pewinternet.org/data-trend/internet-use/latest-stats/

http://www.pewinternet.org/data-trend/internet-use/latest-stats/

Journal of Diabetes Research 3

seeking among the urban poor,” JAMA, vol. 23, no. 6, pp. 1053–1059, 2016.

[19] Pew Research Internet Project, Technology Adoption by LowerIncome Populations, 2013, http://www.pewinternet.org/2013/10/08/technology-adoption-by-lower-income-populations/.

[20] C. R. Lyles, U. Sarkar, and C. Y. Osborn, “Getting a technology-based diabetes intervention ready for prime time: a review ofusability testing studies,” Current diabetes reports, vol. 14, no. 10,article no. 534, 2014.

[21] S. Kumar, W. J. Nilsen, A. Abernethy et al., “Mobile health tech-nology evaluation: the mHealth evidence workshop,” AmericanJournal of Preventive Medicine, vol. 45, no. 2, pp. 228–236, 2013.

http://www.pewinternet.org/2013/10/08/technology-adoption-by-lower-income-populations/

http://www.pewinternet.org/2013/10/08/technology-adoption-by-lower-income-populations/

Review ArticleThe Next Frontier in Communication and the ECLIPPSE Study:Bridging the Linguistic Divide in Secure Messaging

Dean Schillinger,1 Danielle McNamara,2 Scott Crossley,3 Courtney Lyles,1

Howard H. Moffet,4 Urmimala Sarkar,1 Nicholas Duran,2 Jill Allen,4 Jennifer Liu,4

Danielle Oryn,5 Neda Ratanawongsa,1 and Andrew J. Karter4

1University of California, San Francisco, CA, USA2Arizona State University, Tempe, AZ, USA3Georgia State University, Atlanta, GA, USA4Kaiser Permanente, Oakland, CA, USA5Redwood Community Health Coalition, Petaluma, CA, USA

Correspondence should be addressed to Dean Schillinger; [email protected]

Received 4 August 2016; Accepted 12 December 2016; Published 7 February 2017

Academic Editor: Raffaele Marfella

Copyright © 2017 Dean Schillinger et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Health systems are heavily promoting patient portals. However, limited health literacy (HL) can restrict online communicationvia secure messaging (SM) because patients’ literacy skills must be sufficient to convey and comprehend content while cliniciansmust encourage and elicit communication from patients and match patients’ literacy level. This paper describes the EmployingComputational Linguistics to Improve Patient-Provider Secure Email (ECLIPPSE) study, an interdisciplinary effort bringingtogether scientists in communication, computational linguistics, and health services to employ computational linguisticmethods to(1) create a novel Linguistic Complexity Profile (LCP) to characterize communications of patients and clinicians and demonstrate itsvalidity and (2) examine whether providers accommodate communication needs of patients with limited HL by tailoring their SMresponses.Wewill study>5million SMs generated by>150,000 ethnically diverse type 2 diabetes patients and>9000 clinicians fromtwo settings: an integrated delivery system and a public (safety net) system. Finally, we will then create an LCP-based automatedaid that delivers real-time feedback to clinicians to reduce the linguistic complexity of their SMs. This research will support healthsystems’ journeys to become health literate healthcare organizations and reduce HL-related disparities in diabetes care.

1. Introduction

“The single biggest problem in communicationis the illusion that it has taken place.” GeorgeBernard Shaw.

Health literacy (HL) refers to a patient’s capacity to obtain,process, communicate, and understand basic health informa-tion and services needed to make appropriate health deci-sions [1]. Limited HL can place individuals at greater risk oftype 2 diabetes (DM2) and its complications, making limitedHL a critical clinical and public health problem [2, 3]. Sub-optimal communication exchange, and resultant problems

with medication adherence, is a mediator between limitedHL and DM2 outcomes. However, effective communicationbetween patients and providers can mitigate the impactof limited HL. Online patient portals that support asyn-chronous, between-visit electronic communications (securemessages [SM]) are heavily promoted by health systemsand patient uptake is high [4, 5]. Secure messaging (SM)represents a “disruptive” innovation that is rapidly expand-ing (∼5–10%/year) in systems such as Kaiser Permanente(KP), thereby complementing, and at times, supplanting, orstimulating in person/visit-based communication. Enablingpatients with DM2 to easily access their medical informationis a novel approach to facilitate patient engagement and


http://dx.doi.org/10.1155/2017/1348242


activation; empowering patients in this fashion improves dis-ease knowledge, enhances patient-provider communication,and increases adherence to treatment [6]. While researchdemonstrates that patients who access portals are more likelyto have favorable healthcare utilization patterns, adhere toprescribed regimens, and achieve better outcomes, minoritystatus, low levels of income, limited HL, and older age areall associated with lower portal usage. In addition, patientsspecifically require a certain degree of linguistic facility totake advantage of SM as a means of communication, andpatients with limited HL may have difficulty messagingtheir provider or understanding the provider’s replies orinstructions. Providers too must engage with patients in amanner that providesmeaningful and actionable informationand support in an easily comprehended style that promotesshared meaning.

Secure messaging has been shown to be particularlyrelevant for patients with DM2, given their need for relativelyfrequent communication between outpatient encounters. Nostudies have examined how DM2 patients with limited HLand their providers interact via SM. In the Employing Com-putational Linguistics to Improve Patient-Provider SecureEmail (ECLIPPSE) study, funded by National Library ofMedicine, we examine to what extent clinicians accommo-date the communication needs of their DM2 patients withlimited HL. The IOM recently advocated that health systemsand clinicians must accommodate the communication needsof patients with limited HL to overcome the challenges facingthis clinical vulnerable population. The degree to whichproviders adjust linguistic complexity in their SM exchangesto “match” the level observed in their patients will serve asone indicator of the extent to which providers are, or are not,making such accommodations.

This study integrates 3 conceptual frameworks and theo-ries to assess the effect of LCP discordance in SM on a rangeof DM2 outcomes.TheConceptual Framework (Figure 1) inte-grates 3 complementary research and operational paradigms:(1) communication in Chronic Illness Care Framework ofSchillinger [7], (2) Information-Motivation-Behavior Model(IMB) of DM2 medication adherence [2], and (3) IOMattributes of “health literate healthcare organizations.” [8]Our framework revolves around achieving provider-patientconcordance and sharedmeaning across 4 domains involvingelicitation-type communication and explanatory-type commu-nication. Communication concordance can improve chronicdisease outcomes [9, 10]. We will examine whether patientLCP is associated with medication adherence, HbA1c, hypo-glycemia, or long-term outcomes. We augmented this frame-work with the IMB Model of DM2 medication adherence[11]. IMB elements explain >40% of variance in adherenceand predict glycemic control, even in low SES populations[2]. Interventions that target IMB have been successful inprevious studies [12]. In IMB, adherence is determined bythe extent individuals are informed about their regimen,motivated to adhere, and possess or are provided with skills.All 3 domains of IMB are sensitive to interactive patient-provider communication, making them appropriate targetsfor discussions of barriers to adherence (Figure 1, Box 2) ortreatment (Figure 1, Box 4). In 2012, the Institute of Medicine

commissioned a White Paper to define attributes of “healthliterate healthcare organizations” [8]. This paper served asa national call to shift focus from HL skills of individualstoward HL-promoting actions of organizations, includingproviders. Five of 10 attributes bear relevance to this proposal:(1)meeting needs of populations with a range ofHL; (2) usingHL strategies in interpersonal communications and confirm-ing understanding; (3) providing easy access to health infor-mation/services; (4) designing and distributing content thatis easy to understand and act on; and (5) preparing workforceto be health literate [8]. Many providers are unprepared forcommunicating with patients with limited HL [4] and lacktools to improve communication-sensitive outcomes. A goalof this proposal is to create an automated communication aidprototype, based on provider LCPs, that delivers feedback toproviders to reduce linguistic complexity of SMs. Implemen-tation of this tool could advance an organizations’ journey tobecome more health literate.

2. Materials and Methods

Our study involves 2 settings. The first, Kaiser PermanenteNorthern California (KPNC), is a nonprofit, fully integratedhealthcare delivery system, providing services through 37outpatient centers and ∼3,300 providers to 3.3 million planmembers, in a 14-county region of Northern California.Except for extremes of income, sociodemographic char-acteristics of KP members are representative of the localpopulation [13]. KP provides care to a population insuredthrough employer-based plans, Medicare, Medicaid, and newhealth insurance exchanges. Thus, our study findings shouldbe widely generalizable outside of KP. KP has a very well-developed, mature patient portal, kp.org, used by KP mem-bers to SM their healthcare team. KP maintains integratedadministrative and clinical databases (pharmacy, lab, diag-noses and procedures, clinical notes, SMs, utilization, andcost) linked to individualmembers. Part of our sample will bedrawn from the KP DM2 Registry (𝑛 = 229,027 between 1/06and 6/14). Descriptions of the Registry have been publishedpreviously [14–17]. Nested within KP DM2 Registry are ∼11,000 DM2 patients from the well-characterized DISTANCEcohort [18], for whom we have measures of self-reportedHL [19], patient reports of provider’s communication quality,and a broad array of sociobehavioral and psychologicalmeasures obtained in 2006 [20]. Nearly 40% of DISTANCErespondents had limited HL, and the cohort is diverse (19%Hispanic, 17% African American, 23% Asian/Pacific Islander,23% White/non-Hispanic, and 18% Multiracial). To achievesome of the aims of ECLIPPSE, we will use data from KP’spatient portal, which includes SM capabilities. The secondsite, The San Francisco Health Network (SFHN), is a publicdelivery system that includes the following: 13 primary carecenters and specialty and inpatient services at SF GeneralHospital and provides >1.5 million visits and 24-hour careto ∼120,000 low income patients annually. SFHN patients areinsured byMedicaid (39%),Medicare (28%), and commercialhealth insurance (2%) or are of no insurance (32%, a num-ber falling due to ACA). The population is 30% Hispanic,20% African American, 32% Asian/Pacific Islander, 13%

https://healthy.kaiserpermanente.org/


Elicitation

Provider-patientconcordance

Attributes of health literate health care systems

Clinicaldecision-making Health

outcomes

Treatmentadherence

Sharedmeaning

(3) Diagnosis

(1) Disease state

Patient: advocacy skills, healthliteracy, English proficiency, gender,

Communicationcharacteristics

linguistic facility

Relationship: trust/therapeuticalliance, continuity

Provider: collaborativecommunication style, linguistic

complexity, experience

Explanation

(2) Barriers(i) Adherence information(ii) Adherence motivation(iii) Adherence behavioral

skills

(4) Treatment plan(i) Adherence information(ii) Adherence motivation(iii) Adherence behavioral

skills

Figure 1: Conceptual Framework for concordant health communication in DM2 care in the Patient-Centered Medical Home.

White/non-Hispanic, 3% Native American, and 2% Multira-cial. SFHN cares for 8,105 DM2 patients who have seen theirprimary provider (𝑛 = 270) >once in the prior year; ∼50%have limitedHL [21–26]. In 2013, 42% of SFHNDM2 patientshad HbA1c >8% (versus 32% in KP). The SF Departmentof Public Health created The Hospital Record ElectronicData Set (THREDS) that contains demographics, phar-macy, lab, diagnoses, and utilization. In 10/14, SFHN’s EHRadded a self-reported HL measure: “how confident are youfilling out medical forms by yourself?” [27]. SFHN launchedmySFHealth patient portal [28] in 10/14, which includesSM functionality in 2015. We anticipate that >10% of DM2patients from SFHN will engage in SM by 2016, 20% in 2017,30% by 2018, and 35% by 2019 (𝑛 = 2,837,225 providers) [29].

Our study includes 3 aims. The first aim is to developand validate a novel, automated Linguistic Complexity Profile(LCP) to assess securemessage content generated by English-speaking DM2 patients and their providers. We will employnatural language processing (NLP) indices to develop andvalidate the LCP, based on >5 million SM and covariate datafrom >150,000 ethnically diverse, DM2 patients receivingcare in 2 integrated health systems: a large, integrated, health-care delivery system with a mature patient portal (KaiserPermanente Northern California, KPNC) and a county-run,integrated, public (safety net) delivery system with a newlylaunched patient portal (San Francisco Hospital Network,SFHN). We will first aggregate our selected automated lin-guistic indices into larger components using a principal com-ponent analysis [30, 31]. A PCA examines cooccurrence pat-terns among the selected linguistic indices and, using these

cooccurrence patterns, we will develop larger componentscores related to linguistic complexity. We will use thesecomponent scores inmachine learning algorithms to patients’self-reported HL when applied to SMs from ∼12,000 patientSMs. We will also use the component scores to predict 9,535DM2 patients’ ratings of physicians’ communication from950 providers. These analyses will provide construct validityfor the component scores and be used as the foundation ofthe LCP. We will then use the LCP models derived from thepatient’s self-reported HL and ratings of physicians’ com-munications to determine, among >150,000 DM2 patients,whether patient LCP is associated with HbA1c, adherencefor cardiometabolic medications, DM2 complications, hypo-glycemia, and utilization of services.

We hypothesize that (1) patients’ and their providers’LCP will demonstrate construct validity based on patients’HL level and reports of provider communication (using theAHRQ Consumer Assessment of Healthcare Providers &Systems score, CAHPS), respectively; and (2) patient LCPwillcorrelate with clinically relevant outcomes, for example, medadherence, HbA1c, hypoglycemia, and utilization of services.Patients’ HL level will be based on a previously validatedinstrument [19, 32]This itemhas test characteristics similar tothe cumulative 3-item scale and is validated against in person,interview-based tools, such as REALM and TOFHLA (areasunder ROC: 0.70–0,82) [19, 27].We have shown its predictivevalue for DM2 outcomes in DISTANCE [27, 33]. We will usethe CAHPS instrument to assess provider communicationskills. Provider LCP will be assessed for construct validity


against items from the patient-reported, provider commu-nication subscale (over 12 months, how often their providerlistens carefully to them; explains things in a way they couldunderstand; spends enough timewith them; and involves theminmaking decisions about care) from the CAHPS survey [34]in DISTANCE. We will calculate a summary CAHPS score[35] (range, 0–100) by linearly transforming and averagingresponses (Cronbach 𝛼 in DISTANCE = 0.80) [20]. We willonly measure provider LCP during the period the patientreported CAHPS. We hypothesize that higher CAHPS willcorrelate negatively with LCP of providers’ SMs. CAHPS datawere captured by a previous survey conducted by theDiabetesStudy of Northern California (DISTANCE) [18].

The second aim is to examine whether concordancebetween providers’ and patient’s LCP is associated with betteradherence among DM2 patients newly prescribed insulin oran antidepressant, excluding patientswith pharmacy dispens-ing in prior 2 years. We have previously shown that limitedHL predicts poor adherence to communication-sensitivemedications, for example, insulin and antidepressants inDM2 [33, 36]. Based on prior analyses, we estimate ∼52,000patients in KP Registry will have started insulin and ∼28,000antidepressants (selective serotonin reuptake inhibitor orserotonin norepinephrine reuptake inhibitor, mirtazapine orbupropion) during our study period [33]. We hypothesizethat greater degrees of patient-provider concordance LCP inthe period surrounding the start of insulin or antidepressantwill be associated with better adherence to these newlyprescribed medications. Our outcome measure will combineprimary nonadherence (never filling Rx) and early nonper-sistence (never refilling Rx). These standard adherence mea-sures are obtained via pharmacy claims 6 months after initia-tion [20, 33, 37].The third aim is to create an automated, LCP-based prototype to deliver automated feedback to providersto reduce their SM linguistic complexity and enhance sharedmeaning. After developing the prototype, we will enlist 42providers to evaluate whether the prototype improves LCPconcordance using a series of in vitro experiments with sim-ulated, standardized clinical scenarios. We evaluate whetherit improves LCP concordance for DM2 providers via 3-armrandomized controlled trial (RCT) design. Formative feed-back must be timely, impersonal, suggestive, digestible, andnot interrupting workflow [38]. Providers will be givenautomated feedback on SMs as they draft them. Issues flaggedwill be related to the SM (not the individual provider) and topotential (not definitive) problems. Feedback content will bedeveloped andpilotedwith the team’s primary care physiciansto ensure it is impersonal, suggestive, digestible, and action-able. Arm 1 (active control feedback) will receive automatedlinguistic feedback based onNLP algorithm related to socioe-motional tone, rapport building, and degree of empathy andsupport. Support was selected as an active control because (a)it has face validity with providers, (b) we expect a high pro-portion to receive feedback [39], and (c) there is no evidencethat increasing socioemotional content of SMs affects lin-guistic complexity [40, 41]. Arm 2 (Flesch-Kincaid feedback)will receive automated linguistic feedback based on Flesch-Kincaid’s algorithm [42], selected as a comparison because(a) it is ubiquitous and recognizable, (b) it generates grade

level of text, which has face validity with providers, and (c)our pilot work suggests providers rarely generate SMs at <6thgrade level; therefore, we expect a high proportionwill receivefeedback. In Arm 3 (LCP-based feedback), the LCP developedinAims 1 and 2will provide algorithms that assess complexityof patient and provider SMs. Algorithms will be translatedto linguistic feedback to guide providers to generate moreconcordant SMs. The nature of feedback will allow for moregranular, specific, and tailored feedback than Flesch-Kincaid,which we hypothesize will more likely lead to linguisticallyconcordant SMs. We hypothesize that this automated com-munication aid deployed in clinical simulations can reduceprovider’s linguistic complexity to better accommodate DM2patients’ linguistic skills and HL.

3. Preliminary Results

Our preliminary research suggests that patients who accessportals, albeit not necessary using SM, are more likely tohave better (a) healthcare utilization [43], (b) prescriptionadherence [37], and (c) glycemic control [44, 45]. AmongDM2 patients, better ratings of physician communication areassociated with greater SM usage [46]. While we found thatlimitedHL posed a barrier to portal and SMuse [29], dispari-ties have rapidly narrowed. In 2014, 68% versus 84% of KaiserPermanente DM2 patients with limited versus adequate HL,respectively, accessed the portal. Overall, 46% used SM in2014, compared to 30% in 2009. Those with limited HL arerapidly gaining ground, with a 65% increase in 5 years, com-pared to a 41% increase for adequate HL (20%–>33% for lowHL versus 39%–>55%, with greatest gains among Latinos andAfrican Americans) (unpublished data, 2014).

Our research further shows that SMs serve as a criticalmode of communication of clinically relevantmatter inDM2.Among a sample of DM2 subjects (𝑛 = 9,535), a mean of19 SMs involving a mean of 8 SM threads (closed commu-nication loops) were generated annually. While prevalence ofSM use in this sample increased 53% (30 to 46%) from 2009to 2014, the number of outpatient visits in SM users wentdown 4% (from 13.2 to 12.7 total visits per year) during thesame timeframe. A SFHN study just before launch of theirportal revealed 60% of safety net patients used email, 71%were interested in SM, and 19% currently use email withproviders [47]. For SFHN, pilot work with 22 patients whoengaged in email with providers in 2013 revealed a mean of 5SMs, yielding a projection of ∼15,000 SMs by 2019, assuming35% uptake of SM among DM2 patients [48]. The KPMCDM2Registry cohort who sent ≥1 SM toDM2 providers from2006 to 2014 (𝑛 = 151,804) generated >1.5 million SMs in2013 alone. We also calculated the number of patients fromthe KPNC Registry who initiated insulin or antidepressantsfrom 2006 to 2010 and the proportion with SMs 3 monthsbefore and 3 months after initiation. We found that 37,628patients (∼25%) had insulin initiated in this timeframe: 7,264(∼20%) had SMs in the 3 months before, 9,231 (∼25%) after,and 5,720 (∼15%) before and after. For antidepressants, 20,440DM2 patients (∼13%) initiated in that timeframe. Of these,16% had SM exchanges 3 months before, 20% after, and ∼12%both before and after.


To further understand SMcommunication,we conducteda pilot examination of SM content among 50 ethnicallydiverse DISTANCE respondents. Major themes includedthe following: provision/explanation of lab/diagnostic tests;requests for/discussion of medications and side effects;requests for follow-up appointments and discussions aboutspecialty visits; reports of symptoms/self-care; and preventivecare and DM2 guideline-related reminders. In parallel, wecarried out a study of email exchanges between 22 SFHNpatients and their primary care physicians in 2013. Themost common patient requests were the following: actionregardingmedication or treatment; lab tests, x-rays, and otherstudies; referral requests; information regarding symptoms,tests, or procedures; and information regarding medications,side effects, or treatments. Patients requested action in 77% ofthreads. The most common requests were for a prescription(22%), appointment (21%), clarification (16%), medical guid-ance (14%), and paperwork (13%), resulting in 62 physicianactions. We found that SM content was highly relevant toDM2 patients’ clinical care in both health systems [48]. Wealso identified high degrees of clinical jargon use in clinicians’SMs [49].

Qualitative analyses such as those reported above aretime and resource intensive. To address these issues, wehave shown that NLP tools involving automatic extractionof linguistic features using computer programming can sup-plant human ratings to a degree. NLP provides informationabout language at multiple levels and dimensions [50, 51]and affords the ability to glean just about any aspect of text,language, or discourse.ManyNLP techniques are specialized,providing information about different aspects of language.A distinctive aspect of this study is that we will incorporateindices from a variety of tools to paint a rich picture oflanguage, discourse, and communication patterns. We willalso develop new NLP techniques for our data. We will useCoh-Metrix [50, 52], NLP tool developed by our team, whichintegrates various NLP indices, including pattern classifiers,part-of-speech taggers [53], syntactic parsers [54], and LatentSemantic Analysis [55]. This is the most commonly usedtool to measure linguistic complexity on a broad profile oflanguage [41, 56]. Other NLP tools developed by our teamwill augment our analyses. The Writing Assessment Tool(WAT) [57, 58] provides computational indices related towriting quality: global cohesion, contextual cohesion, rhetor-ical strategies, and 𝑛-gram (contiguous sequence of 𝑛 itemsfrom a sequence of text or speech) accuracy. The Tool for theAutomatic Analysis of Lexical Sophistication (TAALES) [59]examines text features specific to sophistication of word use:lexical frequency and range, academic words, concreteness,and meaningfulness. The Tool for Automatic Assessment ofCohesion (TAACO) [60] provides additional indices relatedto text cohesion including word, lemma (mental abstractionof a word to be uttered or written), argument, and synonymoverlap between sentences and paragraphs. The SentimentAnalysis and Cognition Engine (SEANCE) provides anoverview of a text’s affective, cognitive, and social features.

We have used NLP tools to assess linguistic complexity toestimate text comprehensibility.Cohesion, a construct centralto Coh-Metrix and TAACO, is the degree that relations

between concepts are explicit in text by using cues such asword overlap and connectives. High cohesion text enhancesreading comprehension, particularly for less skilled readerswith less knowledge. Coh-Metrix indices related to textcohesion, lexical sophistication, and syntactic complexityhave been used to develop readability measures [50, 61–63].In contrast, a popular index of readability, Flesch-Kincaid’sformula, estimates ease of readability of text by deriving ratiosamong only 3 linguistic units: numbers of syllables, words,and sentences. NLP indices found in tools such as Coh-Metrix demonstrate significant improvements in predictingreadability compared to indices such as Flesch-Kincaid [64]and provide better theoretical overlap with cognitive modelsof text processing and comprehension [62]. Previous studieshave reported that Coh-Metrix indices also perform betterthan traditional readability formulas in distinguishing amongtexts simplified to beginning, intermediate, and advancedreading levels [63]. Coh-Metrix,WAT, TAALES, and TAACOalso predict quality of writing using linguistic indices relatedto lexical sophistication, syntactic complexity, and cohesion[57–59, 63, 65–67]. Measures of linguistic complexity andindividual differences can also be used to examine linksbetween writing skills and reading comprehension [68]. Wehave investigated relationships between latent factors under-lying writing development and found correlations as high as𝑟 = .54 between writing quality and reading comprehension[69]. Collectively, these studies demonstrate that linguisticfeatures can be used to examine text complexity, readability,comprehensibility, and links between reading comprehensionand writing skills.

In a pilot study, we assessed feasibility of using automatedNLP queries to examine SM content in a stratified randomsample of DM2 subjects. We were able to efficiently and reli-ably capture and distinguish provider SMs, patient SMs, andsystem-generated messages. These preliminary analyses usedindices from TAALES and TAACO to examine the potentialof developing LCPs from SMs. We examined 402 SMs gener-ated by 13 providers and 51 English-speaking DM2 patients,stratified into 25 low and 26 high HL. We were able to assessprovider-patient communication differences in SMs usinga number of linguistic features. Preliminary findings werepromising. Indices related to word frequency and entropy(i.e., number of documents or contexts in which a wordoccurs) distinguished SMs by low versus high HL patients,with high HL patients using more infrequent (less familiar)and more specific words that occur in fewer contexts (𝑝 <.05). Providers used more complex linguistic features thanpatients, producing more rare words (e.g., jargon), specificwords, words with fewer associations to other concepts,and less semantic overlap between paragraphs (𝑝 < .05).Providers judged by patients to be less communicative onCAHPS used more infrequent (unfamiliar) words and wordsequences.This preliminary work suggests that (a) providers’use of complex linguistic features better correlated to highthan low HL patients (i.e., providers and low HL patientsdemonstrated discordance in language use, 𝑝 < .05) and (b)providersmodify linguistic output based on patientHL, usingmore familiar words with low HL patients than high HLpatients (𝑝 < .05), although the gap between providers and


adequate HL patients remained large. The findings suggestedthe feasibility of developing patient LCPs to assess constructand predictive validity with respect to DM2 outcomes,analyzing complexity of providers’ SMs, and using LCP toprovide feedback to providers.

4. Discussion

Most clinicians are untrained in communicating with DM2patients with limited HL, and health systems have no feasiblemeans to identify patients with limited HL or those clinicianswho need support in communicating effectively [4]. TheLinguistic Complexity Profile (LCP) presented in this projectrepresents methodological and measurement research focu-sed on HL of individual providers and health systems thatcan facilitate comparisons across ethnicity and health sys-tem settings. Studying whether LCP gaps between providerand patient influence medication adherence reflects basicresearch into how HL impacts health processes and out-comes. Our communication aid based on LCP representsapplied research addressing issues pertinent to HL.

LimitedHL places individuals at risk ofDM2 and its com-plications, making it a critical public health problem [2, 3]. Inmany settings, limitedHL, aswell as numeracy, has been asso-ciated with higher prevalence of DM2, poor glycemic control[21, 70], DM2 complications [21, 71], hypoglycemia [29], andpoorer medication adherence [33, 72]. Inadequate healthcommunication is a mediator of the relationship betweenlimited HL and DM2 outcomes [7]. Provider communica-tion shortcomings limit the effectiveness of DM2 self-man-agement interventions, especially for patients with limitedHL [9, 23]. DM2 providers often are poorly prepared tocommunicate with patients with limited HL [4, 5]. Securemessaging (SM) represents a “disruptive” innovation that israpidly expanding (∼5–10%/year) in systems such as KPNC,thereby complementing, and at times, supplanting, or stim-ulating [73–75] in person/visit-based communication. SM isespecially important for patients with DM2, who use SM forcritical self-management functions, use portals more oftenthan healthier patients [76], and often have competing prior-ities during in person visits, creating greater communicationneeds between visits [75]. CMS has promotedmeaningful useof electronic health records (EHRs) via substantial incentives(>$25 billion to date) for clinicians and systems that imple-ment EHRs, encouraging patients to engage in EHR [77]. Toreceive incentives, systems must demonstrate that significantproportions of patients register for the EHR and exchangeSMs with providers. While this program has stimulateduptake of SM in the private sector and has established SM asa new standard of care in the US, SM is now enjoying uptakein safety net and public healthcare systems which dispropor-tionately care for patients with limited HL. However, there isvery limited understanding of the impact of SM use acrossHL levels. Additionally we are unaware of intervention workto enhance the effectiveness of clinician SM use with DM2patients with limited HL. We do know that ratings of physi-cians’ communication are associated with DM2 medicationadherence [20] and that medication adherence is an impor-tant target for health communications and HL research. Poor

adherence in DM2 is common and associated with highercosts and worse outcomes [2]. Limited HL predicts pooradherence to communication-sensitive medications, forexample, insulin and antidepressants in DM2 [33, 36, 78]. Totake full advantage of patient portals, patients must have thelinguistic competencies to convey and comprehend clinicalcontent from providers. Providers’ must encourage and elicitpatient communication and must match or approximatepatients’ linguistic complexity levels to enhance SM effective-ness (“linguistic complexity concordance”). These problemsunderlie the need to develop and test the LCP described inthis paper. The ECLIPPSE Study will be the first systematicstudy of SM between DM2 patients and their clinicians and,to our knowledge, is the first study to employ computationallinguistics to analyze and improve digital patient-providercommunications.

5. Conclusions

This research has the potential to achieve important gainsin the effort to translate both diabetes and HL researchinto meaningful action. First, measuring individual HL inhealthcare populations is widely recognized to be a dauntingand time-consuming undertaking, making it infeasible andcost-prohibitive to carry out. If the patient LCP proves to bea valid indicator of individual HL and is predictive of healthoutcomes in DM2, then health systems will have an efficient,automatic method to harness “big data” in order to identifypatients with limited HL and that may benefit from outreachand additional self-management support. In addition, ifpatient-clinician discordance in LCP is found to be prevalentand is associated with suboptimal communication-sensitiveoutcomes, such as medication nonadherence or hypogly-cemia, then health systems can target communication train-ing efforts for those clinicians. Finally, if ECLIPPSE’s LCP-based prototype communication aids to provide real-timefeedback to clinicians is found to be both feasible for clini-cians and effective in increasing the degree of linguisticmatching, then clinicians will be much better equipped topromote shared meaning in their communications with vul-nerable patients. Insofar as electronic patient portals arebeing heavily promoted by health systems and becoming thestandard of care, and insofar as patients who access themappear more likely to achieve better outcomes, the researchproducts of ECLIPPSE will undoubtedly support healthsystems’ journeys to reduce HL-related disparities in diabetescare.

Competing Interests

The authors declare that they have no competing interests.

Acknowledgments

Support for this manuscript came from Grants NLM R01,LM012355, and NIH UL1TR000004. Drs. Schillinger andKarter are also supported by P30 Grant NIDDK092924.


References

[1] E. G. Grossman and Office of the Legislative Counsel, PatientProtection and Affordable Care Act, Edited by U.D.o.H.H. Ser-vices, Department of Health & Human Services, Washington,DC, USA, 2010.

[2] S. C. Bailey, A. G. Brega, T. M. Crutchfield et al., “Update onhealth literacy and diabetes,” Diabetes Educator, vol. 40, no. 5,pp. 581–604, 2014.

[3] Institute of Medicine, Health Literacy: A Prescription to EndConfusion, Edited by L. Nielsen-Bohlman, A. Panzer, and D. A.Kindig, The National Academies Press, Washington, DC, USA,2004.

[4] H. K. Seligman, F. F. Wang, J. L. Palacios et al., “Physician noti-fication of their diabetes patients’ limited health literacy: a ran-domized, controlled trial,” Journal of General Internal Medicine,vol. 20, no. 11, pp. 1001–1007, 2005.

[5] N. K. Ali, R. P. Ferguson, S. Mitha, and A. Hanlon, “Domedicaltrainees feel confident communicating with low health literacypatients?” Journal of Community Hospital Internal MedicinePerspectives, vol. 4, no. 2, 2014.

[6] M. Jhamb, K. L. Cavanaugh, A. Bian et al., “Disparities in elec-tronic health record patient portal use in nephrology clinics,”Clinical Journal of the American Society of Nephrology, vol. 10,no. 11, pp. 2013–2022, 2015.

[7] D. Schillinger, “Literacy and health communication: reversingthe “inverse care law”,” American Journal of Bioethics, vol. 7, no.11, pp. 15–18, 2007.

[8] C. Brach, D. Keller, L. M. Hernandez et al., Ten Attributes ofHealth Literate Health Care Organizations, Institute ofMedicineof the National Academies, Washington, DC, USA, 2012.

[9] E. L.Machtinger, F.Wang, L.-L. Chen,M. Rodriguez, S.Wu, andD. Schillinger, “A visual medication schedule to improve anti-coagulation control: a randomized, controlled trial,” Joint Com-mission Journal on Quality and Patient Safety, vol. 33, no. 10, pp.625–635, 2007.

[10] J. L. Barton, J. Imboden, J. Graf, D. Glidden, E. H. Yelin, andD. Schillinger, “Patient-physician discordance in assessments ofglobal disease severity in rheumatoid arthritis,” Arthritis Careand Research, vol. 62, no. 6, pp. 857–864, 2010.

[11] C. Y. Osborn and L. E. Egede, “Validation of an information-motivation-behavioral skills model of diabetes self-care (IMB-DSC),” Patient Education and Counseling, vol. 79, no. 1, pp. 49–54, 2010.

[12] J. D. Fisher, W. A. Fisher, K. R. Amico, and J. J. Harman, “Aninformation-motivation-behavioral skills model of adherenceto antiretroviral therapy,” Health Psychology, vol. 25, no. 4, pp.462–473, 2006.

[13] N. Krieger, “Overcoming the absence of socioeconomic data inmedical records: validation and application of a census-basedmethodology,” American Journal of Public Health, vol. 82, no. 5,pp. 703–710, 1992.

[14] A. J. Karter, A. Ferrara, J. Y. Liu, H. H. Moffet, L. M. Ackerson,and J. V. Selby, “Ethnic disparities in diabetic complications inan insured population,” JAMA, vol. 287, no. 19, pp. 2519–2527,2002.

[15] A. J. Karter, A. Ferrara, J. A. Darbinian, L.M. Ackerson, and J. V.Selby, “Self-monitoring of blood glucose: language and financialbarriers in a managed care population with diabetes,” DiabetesCare, vol. 23, no. 4, pp. 477–483, 2000.

[16] A. J. Karter, L. M. Ackerson, J. A. Darbinian et al., “Self-monitoring of blood glucose levels and glycemic control: the

Northern California Kaiser Permanente Diabetes registry,”TheAmerican Journal of Medicine, vol. 111, no. 1, pp. 1–9, 2001.

[17] C. Iribarren, A. J. Karter, A. S. Go et al., “Glycemic control andheart failure among adult patients with diabetes,” Circulation,vol. 103, no. 22, pp. 2668–2673, 2001.

[18] H. H. Moffet, N. Adler, D. Schillinger et al., “Cohort profile: thediabetes study of Northern California (DISTANCE)—objec-tives and design of a survey follow-up study of social healthdisparities in a managed care population,” International Journalof Epidemiology, vol. 38, no. 1, pp. 38–47, 2009.

[19] L. D. Chew, J. M. Griffin, M. R. Partin et al., “Validation ofscreening questions for limited health literacy in a large VAoutpatient population,” Journal of General Internal Medicine,vol. 23, no. 5, pp. 561–566, 2008.

[20] N. Ratanawongsa, A. J. Karter, M. M. Parker et al., “Commu-nication and medication refill adherence the diabetes study ofNorthern California,” JAMA Internal Medicine, vol. 173, no. 3,pp. 210–218, 2013.

[21] D. Schillinger, K. Grumbach, J. Piette et al., “Association ofhealth literacy with diabetes outcomes,” JAMA, vol. 288, no. 4,pp. 475–482, 2002.

[22] D. Schillinger, J. Piette, K. Grumbach et al., “Closing the loop:physician communication with diabetic patients who have lowhealth literacy,” Archives of Internal Medicine, vol. 163, no. 1, pp.83–90, 2003.

[23] D. Schillinger,M.Handley, F.Wang, andH.Hammer, “Effects ofself-management support on structure, process, and outcomesamong vulnerable patients with diabetes,”Diabetes Care, vol. 32,no. 4, pp. 559–566, 2009.

[24] D. Schillinger, H. Hammer, F. Wang et al., “Seeing in 3-D: examining the reach of diabetes self-management supportstrategies in a public health care system,” Health Education andBehavior, vol. 35, no. 5, pp. 664–682, 2008.

[25] N. Ratanawongsa, M. A. Handley, U. Sarkar et al., “Diabeteshealth information technology innovation to improve qualityof life for health plan members in urban safety net,” Journal ofAmbulatory Care Management, vol. 37, no. 2, pp. 127–137, 2014.

[26] N. Ratanawongsa, V. K. Bhandari, M. Handley, T. Rundall,H. Hammer, and D. Schillinger, “Primary care provider per-ceptions of the effectiveness of two self-management supportprograms for vulnerable patients with diabetes,” Journal ofDiabetes Science and Technology, vol. 6, no. 1, pp. 116–124, 2012.

[27] U. Sarkar, D. Schillinger, A. Lopez, and R. Sudore, “Validationof self-reported health literacy questions among diverse englishand spanish-speaking populations,” Journal of General InternalMedicine, vol. 26, no. 3, pp. 265–271, 2011.

[28] eClinicalWorks, http://www.eclinicalworks.com/.[29] U. Sarkar, A. J. Karter, J. Y. Liu, H. H. Moffet, N. E. Adler, and D.

Schillinger, “Hypoglycemia is more common among type 2 dia-betes patients with limited health literacy: the Diabetes Study ofNorthern California (DISTANCE),” Journal of General InternalMedicine, vol. 25, no. 9, pp. 962–968, 2010.

[30] Y. Attali and D. Powers, A Developmental Writing Scale, ETSResearch Report Series, 2008.

[31] S. A. Crossley, K. Kyle, L. Varner, L. Gou, and D. S. McNamara,“Linguistic microfeatures to predict L2 writing proficiency: acase study in automated writing evaluation,” Journal of WritingAssessment, vol. 7, no. 1, 2014.

[32] L. D. Chew, K. A. Bradley, and E. J. Boyko, “Brief questionsto identify patients with inadequate health literacy,” FamilyMedicine, vol. 36, no. 8, pp. 588–594, 2004.

http://www.eclinicalworks.com/


[33] A. M. Bauer, D. Schillinger, M. M. Parker et al., “Health literacyand antidepressant medication adherence among adults withdiabetes: the diabetes study of Northern California (DIS-TANCE),” Journal of General Internal Medicine, vol. 28, no. 9,pp. 1181–1187, 2013.

[34] Agency for Healthcare Research and Quality, CAHPS Clinician& Group Survey, 2011.

[35] L. S.Morales,M.N. Elliott, R.Weech-Maldonado, K. L. Spritzer,andR.D.Hays, “Differences inCAHPS adult survey reports andratings by race and ethnicity: an analysis of theNational CAHPSBenchmarking Data 1.0,”Health Services Research, vol. 36, no. 3,pp. 595–617, 2001.

[36] A. J. Karter, U. Subramanian, C. Saha et al., “Barriers to insulininitiation: the translating research into action for diabetesinsulin starts project,” Diabetes Care, vol. 33, no. 4, pp. 733–735,2010.

[37] U. Sarkar, C. R. Lyles, M. M. Parker et al., “Use of the refillfunction through an online patient portal is associated withimproved adherence to statins in an integrated health system,”Medical Care, vol. 52, no. 3, pp. 194–201, 2014.

[38] R. Roscoe, A. Varner, E. L. Snow, andD. S.McNamara, “Design-ing usable automated formative feedback for intelligent tutor-ing,” in Proceedings of the 11th International Conference of theLearning Sciences (ICLS ’14), pp. 1423–1425, Boulder, Colo, USA,2014.

[39] D. L. Roter, S. Larson, D. Z. Sands, D. E. Ford, and T. Houston,“Can E-mail messages between patients and physicians bepatient-centered?”Health Communication, vol. 23, no. 1, pp. 80–86, 2008.

[40] N. D. Duran, S. Crossley, C. Hall, P. McCarthy, and D. McNa-mara, “Expanding a catalogue of deceptive linguistic featureswith NLP technologies,” in Proceedings of the 22nd Interna-tional Florida Artificial Intelligence Research Society Conference(FLAIRS-22 ’09), pp. 243–248, Sanibel, Fla, USA, March 2009.

[41] N. D. Duran, C. Hall, P. M. McCarthy, and D. S. McNa-mara, “The linguistic correlates of conversational deception:comparing natural language processing technologies,” AppliedPsycholinguistics, vol. 31, no. 3, pp. 439–462, 2010.

[42] J. P. Kincaid, R. P. Fishburne Jr., R. L. Rogers, and B. S. Chissom,“Derivation of new readability formulas (automated readabilityindex, fog count and flesch reading ease formula) for navyenlisted personnel,” DTIC Document, 1975.

[43] M. Reed, J. Huang, R. Brand et al., “Implementation of anoutpatient electronic health record and emergency departmentvisits, hospitalizations, and office visits among patients withdiabetes,” JAMA, vol. 310, no. 10, pp. 1060–1065, 2013.

[44] M. Reed, J. Huang, I. Graetz et al., “Outpatient electronic healthrecords and the clinical care and outcomes of patients withdiabetes mellitus,” Annals of Internal Medicine, vol. 157, no. 7,pp. 482–489, 2012.

[45] L. T. Harris, T. D. Koepsell, S. J. Haneuse, D. P. Martin, and J.D. Ralston, “Glycemic control associated with secure patient-provider messaging within a shared electronic medical record:a longitudinal analysis,” Diabetes Care, vol. 36, no. 9, pp. 2726–2733, 2013.

[46] C. R. Lyles, U. Sarkar, J. D. Ralston et al., “Patient-providercommunication and trust in relation to use of an online patientportal among diabetes patients: the Diabetes and Aging Study,”Journal of the AmericanMedical Informatics Association, vol. 20,no. 6, pp. 1128–1131, 2013.

[47] A. Schickedanz, D. Huang, A. Lopez et al., “Access, interest,and attitudes toward electronic communication for health care

among patients in the medical safety net,” Journal of GeneralInternal Medicine, vol. 28, no. 7, pp. 914–920, 2013.

[48] J. B. Mirsky, L. Tieu, C. Lyles, and U. Sarkar, “A mixed-methodsstudy of patient–provider e-mail content in a safety-net setting,”Journal of Health Communication, vol. 21, no. 1, pp. 85–91, 2016.

[49] C. M. Castro, C. Wilson, F. Wang, and D. Schillinger, “Babelbabble: physicians’ use of unclarified medical jargon withpatients,” American Journal of Health Behavior, vol. 31, supple-ment 1, pp. s85–s95, 2007.

[50] A.C.Graesser andD. S.McNamara, “Computational analyses ofmultilevel discourse comprehension,” Topics in Cognitive Sci-ence, vol. 3, no. 2, pp. 371–398, 2011.

[51] S. A. Crossley, “Advancing research in second language writingthrough computational tools and machine learning techniques:a research agenda,” Language Teaching, vol. 46, no. 2, pp. 256–271, 2013.

[52] D. S. McNamara, A. C. Graesser, P. M. McCarthy, and Z. Cai,Automated Evaluation of Text and Discourse with Coh-Metrix,Cambridge University Press, 2014.

[53] E. Brill, “Transformation-based error-driven learning and natu-ral language processing: a case study in part-of-speech tagging,”Computational Linguistics, vol. 21, no. 4, 1995.

[54] E. Charniak, “Amaximum-entropy-inspired parser,” inProceed-ings of the 1st North American Chapter of the Association forComputational Linguistics Conference, pp. 132–139, Associationfor Computational Linguistics, Seattle, Wash, USA, 2000.

[55] T. K. Landauer, D. S. McNamara, S. Dennis, and W. Kintsch,Handbook of Latent Semantic Analysis, Psychology Press, 2013.

[56] P. M. McCarthy, R. H. Guess, and D. S. McNamara, “The com-ponents of paraphrase evaluations,” Behavior ResearchMethods,vol. 41, no. 3, pp. 682–690, 2009.

[57] D. S. McNamara, S. A. Crossley, and R. Roscoe, “Natural lang-uage processing in an intelligent writing strategy tutoring sys-tem,” Behavior Research Methods, vol. 45, no. 2, pp. 499–515,2013.

[58] S. A. Crossley, R. Roscoe, and D. S. McNamara, “Using auto-matic scoring models to detect changes in student writing inan intelligent tutoring system,” in Proceedings of the 26th Inter-national FLAIRS Conference, May 2013.

[59] K. Kyle and S. A. Crossley, “Automatically assessing lexicalsophistication: indices, tools, findings, and application,” TESOLQuarterly, vol. 49, no. 4, pp. 757–786, 2015.

[60] S. A. Crossley, K. Kyle, and D. S. McNamara, “The tool for theautomatic analysis of text cohesion (TAACO): automatic assess-ment of local, global, and text cohesion,” Behavior ResearchMethods, vol. 48, no. 4, pp. 1227–1237, 2015.

[61] N. Duran, C. Bellissens, R. Taylor, andD. S.McNamara, “Quali-fying text difficulty with automated indices of cohesion andsemantics,” in Proceedings of the 29th Annual Meeting of theCognitive Science Society, Austin, Tex, USA, 2007.

[62] S. A. Crossley, J. Greenfield, and D. S. McNamara, “Assessingtext readability using cognitively based indices,” TESOL Quar-terly, vol. 42, no. 3, pp. 475–493, 2008.

[63] S. A. Crossley, R. Roscoe, and D. S. McNamara, “Predictinghuman scores of essay quality using computational indices oflinguistic and textual features,” inArtificial Intelligence in Educa-tion, Springer, Berlin, Germany, 2011.

[64] R. Flesch, “A new readability yardstick,” Journal of Applied Psy-chology, vol. 32, no. 3, pp. 221–233, 1948.

[65] S. A. Crossley, Z. Cai, and D. S. McNamara, “Syntagmatic, para-digmatic, and automatic N-gram approaches to assessing essay


quality,” inProceedings of the 25th International FloridaArtificialIntelligence Research Society Conference (FLAIRS-25 ’12), pp.214–219, Marco Island, Fla, USA, May 2012.

[66] L. Guo, S. A. Crossley, and D. S. McNamara, “Predicting humanjudgments of essay quality in both integrated and independentsecond language writing samples: a comparison study,” Assess-ing Writing, vol. 18, no. 3, pp. 218–238, 2013.

[67] D. S. McNamara, M. M. Louwerse, P. M. McCarthy, and A. C.Graesser, “Coh-Metrix: capturing linguistic features of cohe-sion,” Discourse Processes, vol. 47, no. 4, pp. 292–330, 2010.

[68] C. Juel, P. Griffith, and P. Gough, “A longitudinal study of thechanging relationships of word recognition, spelling, readingcomprehension, andwriting from first to second grade,” Journalof Educational Psychology, vol. 78, no. 4, pp. 243–255, 1986.

[69] R. D. Abbott and V. W. Berninger, “Structural equation mod-eling of relationships among developmental skills and writingskills in primary-and intermediate-grade writers,” Journal ofEducational Psychology, vol. 85, no. 3, pp. 478–508, 1993.

[70] A. G. Brega, A. Ang, W. Vega et al., “Mechanisms underlyingthe relationship between health literacy and glycemic control inAmerican Indians and Alaska Natives,” Patient Education andCounseling, vol. 88, no. 1, pp. 61–68, 2012.

[71] A. S. Laramee, N. Morris, and B. Littenberg, “Relationship ofliteracy and heart failure in adults with diabetes,” BMC HealthServices Research, vol. 7, article 98, 2007.

[72] A. J. Karter, M. M. Parker, H. H. Moffet, A. T. Ahmed, J. A.Schmittdiel, and J. V. Selby, “New prescriptionmedication gaps:a comprehensive measure of adherence to new prescriptions,”Health Services Research, vol. 44, no. 5, pp. 1640–1661, 2009.

[73] R. Pearl, “Kaiser permanente Northern California: current exp-eriences with internet, mobile, and video technologies,” HealthAffairs, vol. 33, no. 2, pp. 251–257, 2014.

[74] L. T. Harris, S. J. Haneuse, D. P. Martin, and J. D. Ralston, “Dia-betes quality of care and outpatient utilization associated withelectronic patient-provider messaging: a cross-sectional analy-sis,” Diabetes Care, vol. 32, no. 7, pp. 1182–1187, 2009.

[75] R. W. Grant, A. S. Adams, E. A. Bayliss, and M. Heisler, “Estab-lishing visit priorities for complex patients: a summary of theliterature and conceptual model to guide innovative interven-tions,” Healthcare, vol. 1, no. 3-4, pp. 117–122, 2013.

[76] J. D. Ralston, C. M. Rutter, D. Carrell, J. Hecht, D. Rubanowice,and G. E. Simon, “Patient use of secure electronic messagingwithin a sharedmedical record: a cross-sectional study,” Journalof General Internal Medicine, vol. 24, no. 3, pp. 349–355, 2009.

[77] D. Blumenthal and M. Tavenner, “The ‘meaningful use’ regula-tion for electronic health records,” The New England Journal ofMedicine, vol. 363, no. 6, pp. 501–504, 2010.

[78] N. Ratanawongsa, J. C. Crosson, D. Schillinger, A. J. Karter, C.K. Saha, and D. G. Marrero, “Getting under the skin of clinicalinertia in insulin initiation: the translating research into actionfor diabetes (TRIAD) insulin starts project,” Diabetes Educator,vol. 38, no. 1, pp. 94–100, 2012.

Review ArticleThe Challenges of Electronic Health Recordsand Diabetes Electronic Prescribing: Implications forSafety Net Care for Diverse Populations

Neda Ratanawongsa,1 Lenny L. S. Chan,2 Michelle M. Fouts,3 and Elizabeth J. Murphy4

1Division of General Internal Medicine, Department of Medicine, UCSF Center for Vulnerable Populations, University of California,San Francisco, 1001 Potrero Avenue, Box 1364, San Francisco, CA 94143, USA2San Francisco Department of Public Health, 1001 Potrero Avenue, San Francisco, CA 94110, USA3Laguna Honda Hospital and Rehabilitation Center, 375 Laguna Honda Blvd, San Francisco, CA 94116, USA4Division of Endocrinology, Department of Medicine, University of California, San Francisco, 1001 Potrero Avenue,Box 0862, San Francisco, CA 94143, USA

Correspondence should be addressed to Neda Ratanawongsa; [email protected]

Received 2 August 2016; Accepted 4 January 2017; Published 18 January 2017

Academic Editor: Andrea Flex

Copyright © 2017 Neda Ratanawongsa et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Widespread electronic health record (EHR) implementation creates new challenges in the diabetes care of complex and diversepopulations, including safe medication prescribing for patients with limited health literacy and limited English proficiency. Thisreview highlights how the EHR electronic prescribing transformation has affected diabetes care for vulnerable patients and offersrecommendations for improving patient safety through EHR electronic prescribing design, implementation, policy, and research.Specifically, we present evidence for (1) the adoption of RxNorm; (2) standardized naming and picklist options for high alertmedications such as insulin; (3) the widespread implementation of universal medication schedule and language-concordant labels,with the expansion of electronic prescription 140-character limit; (4) enhanced bidirectional communication with pharmacypartners; and (5) informatics and implementation research in safety net healthcare systems to examine howEHR tools and practicesaffect diverse vulnerable populations.

1. Introduction

Mrs. D, a 67-year-old Latina woman, suffers fromthirst and frequent urination, with a hemoglobinA1c that jumped from 7.5% to 9.7%. Dr. P notesin the pharmacy claims section of the electronichealth record (EHR) that Mrs. D has not filledher pioglitazone in 6 months. Mrs. D reluctantlyadmits she did not like the medication because herface and legs felt swollen. So, Dr. P counseled Mrs.D about adding a short-acting mealtime insulinto her metformin and glargine regimen. Using theEHR electronic prescribing feature, Dr. P triedto submit a prescription to her local pharmacyfor Humulin R insulin but was warned by theformulary check that this was not covered. So, Dr.

P prescribed Novolin R insulin, typing into the“Take” field “5 units with breakfast and dinner”while leaving the defaulted frequency as “twicedaily.”

One week later, Mrs. D returns to Dr. P after threeepisodes of feeling “shaky” and “sweaty” promptedher to stop this new insulin. She shows Dr. Pan insulin vial with Novolog, a fast-acting aspartinsulin, rather than the regular insulin Novolin Rinsulin. In addition, Mrs. D has been taking hernew insulin when she wakes up and at bedtimebecause the bottle said “twice daily.” Finally, Mrs.D showed a newly dispensed bottle of pioglitazoneand said she was confused about whether she wassupposed to restart that medication.


http://dx.doi.org/10.1155/2017/8983237


Rapid deployment of electronic health record (EHR)systems across US safety net clinics has transformed thecare delivery system for vulnerable patients. Fueled by 2009Health Information Technology for Economic and ClinicalHealth Act, the percentage of outpatient clinics with any typeof EHR doubled from 42% in 2008 to 83% in 2014 [1]. Mean-ingful use requires clinicians to document and place ordersfor medical care in structured and reportable ways, includingthe electronic recording and prescribing of medications [2–4]. Federally qualified health centers (FQHC)—previouslylacking resources and support to implement EHRs—are nowparticipating in this transformation, with high rates of adop-tion [5].

EHRs are a necessary and valuable tool for healthcaredelivery, with extensive research investigating EHRs’ impacton the quality of care delivery, including medication safety[4, 6–11]. However, studies have also raised concerns that thedesign and implementation of EHR computerized providerorder entry (CPOE) may either fail to mitigate or introducenew medication errors [12–17]. This may pose particularrisks in safety net clinics, which serve a disproportionatelyhigh number of patients with limited health literacy (LHL)and limited English proficiency (LEP), who are shown toexperience disparities in communication and care [18–27].

In this article, we highlight how EHR electronic prescrib-ing has affected diabetes care for vulnerable patients and sug-gest recommendations for EHR design, implementation, pol-icy, and research. Specifically, we present evidence for (1) theadoption of RxNorm; (2) standardized naming and picklistoptions for high alert medications such as insulin; (3) thewidespread implementation of universal medication sched-ule and language-concordant labels, with the expansion ofelectronic prescription 140-character limit; (4) enhancedbidirectional communication with pharmacy partners; and(5) informatics and implementation research in safety nethealthcare systems to examine how EHR tools and practicesaffect diverse vulnerable populations.

2. Potential EHR Benefits in Medication Safetyand Adherence for Outpatient Diabetes Care

2.1. Adverse Drug Events. In the US, approximately 4.5million ambulatory visits relate to adverse drug events (ADE)each year, with the majority of these occurring in outpatientoffice practices [28]. Patientwith diabetesmay be at particularrisk, with cardiovascular and hypoglycemic medicationscomprising the majority of preventable ADE [29].

Two decades ago, computerized prescribing was heraldedas a potentially powerful tool to improve medication safetyand promotemore evidence-based prescribing in ambulatorymedical care [30, 31]. However, studies have raised concernsabout the potential errors resulting from electronic prescrib-ing platforms in ambulatory care. Research shows electronicprescription error rates ranging from 5 to 38% [16], whichis similar to rates reported in the era before computerizedprescribing [32]. In addition, studies have shown that 16–19%of electronic prescriptions convey contradictory informationwithin a single prescription, with 2.7% of all prescriptionshaving the potential for a severe ADE [15, 33]. Meanwhile,

although electronic prescribing platforms technically sup-port the ability to discontinue medications, these electronicmessages incur charges to pharmacies and thus may notbe accepted, leaving providers to embed discontinuationinstructions in free-text fields [33–35]. Patients with diabetesmay be disproportionately affected, as a study showed that,among the top nine medications dispensed after EHR dis-continuation, seven were commonly used in diabetes, hyper-tension, and hyperlipidemia [36]. Overall, whether EHRplat-formsmitigate errormay depend on the specific implementa-tion within individual institutions, as illustrated by a studyshowing that the same medication-related decision supportknowledge base was associated with ADE decreases in onemajor academicmedical center and increases in another [37].

2.2. Patient-Centered Prescribing. Two features of EHRs offerthe potential to enhance the patient-centeredness of treat-ment decision-making: formulary support and pharmacy filldata. Cost remains a major andmodifiable barrier to medica-tion adherence, particularly for low-income patients [38, 39].By integrating formulary checks into the clinical decisionsupport of electronic prescribing, EHRs can make cost andformulary information readily available to the prescribingprovider. This could reduce delays in patients receiving theirmedications and reduce patient’s out-of-pocket costs therebyimproving adherence [11]. In addition, EHRsmay also permitclinicians to view claims data from pharmacies, providing asurrogate measure of medication adherence and informationabout what medications are dispensed outside of the pre-scribers’ healthcare system [40–42]. Both of these tools canfacilitate patient-centered discussions about patient’s beliefs,concerns, and behaviors around medications, promotingmore treatment decision-making tailored to the individualpatient and enhancing future medication adherence.

Despite the promise of these electronic prescribing fea-tures, it is not clear how much these tools lead to morepatient-centered prescribing. For example, a study of inter-ruptive formulary decision support (in which EHRs stop pre-scribers from moving forward) shifted prescriptions towardspreferred tier medications, but patients’ out-of-pocket costswere only slightly lower ($10.60 versus $11.81 for angiotensinreceptor blockers) and adherence rates did not improve [43].

3. Specific Limitations and Recommendationsfor Current EHRs in Caring for VulnerablePatients with Diabetes

EHRs and electronic prescribing provide opportunities forimproved safety, but they also provide opportunities for newtypes of errors.

3.1. Requiring Brand Name Prescriptions. Both providers andpatients are at risk for confusion by databases that force theuse of specific brand names based on the National DrugCode (NDC) Directory, which identifies each medicationuniquely based on its manufacturer, product formulation,and package size [44]. By forcing the prescriber to overspec-ify the medication choice, electronic prescribing platformsremove the pharmacist’s ability to select a medication that


meets the prescriber’s true intentions, taking into accountavailable formulations, patients’ prescription drug history,and cost and formulary considerations [12, 17]. For example,when EHRs force providers to prescribe regular insulin byselecting a brand name, pharmacists cannot offer patientsthe formulary or lower cost brand without contacting theprovider for a change. This can lead to delays in dispensingor increased cost-related medication nonadherence in low-income patients.

In the case above, Dr. P selected Novolog from the EHRscreen in a hurry when shemeant to prescribe Novolin R. Shelikely would have realized her mistake if she was presentedwith both the brand and generic names.

Thus, some have recommended using RxNorm, a listusing standard terminology that contains all medicationsavailable onUSmarket that ismaintained byNational Libraryof Medicine, as the standardized identifier for choosing andprescribingmedications [12].This labelingwould also complywith Joint Commission best practices of including both thegeneric and the brand name of a medication on the labeland EHR, to minimize confusion and error when switchingformulations is required [45, 46].

3.2. Errors Associated with Insulin Electronic Prescribing. TheInstitute for Safe Medication Practices classifies insulin as ahigh alert medication, “drugs that bear a heightened risk ofcausing significant patient harm when used in error” [47].EHRs pose specific risks around insulin prescribing andmedication reconciliation.

EHR medication searches may list similar medicationnames in close proximity on the screen, which increases therisk of selecting the wrong medication [48–50]This is partic-ularly problematic for insulin, since many insulin productshave similar brand or generic names but differ in onset andduration of action [51]. The use of “tall man letters,” forexample, distinguishing “NovoLOG” versus “NovoLIN,” canreduce such risk [52]. In addition, insulin formulations maybe up to 5 times more concentrated than the standard U-100(100 units per mL) insulin: for example, U-200 (200 units permL) insulin lispro, U-300 (300 units permL) insulin glargine,and U-500 (500 units per mL) regular insulin. Althoughproviders may not know that this concentrated insulin exists,EHR databases may display these selections side by sidewith similarly named standard formulations, introducingnew risks of confusion and ADE.

Moreover, the vendor-approved terminology for a partic-ular medication may be unfamiliar to prescribing providers.For example, a standardized e-Prescribing terminology forNPH insulin, “insulin isophane,” is not used in educationalor clinical literature, and after unsuccessfully searching for“NPH,” providers may erroneously select an insulin formu-lation with a different onset or concentration.

Provider training is not sufficient to prevent these kinds oferrors; thus software vendors should commit to adopting safemedication practices for high alert medications like insulin.

3.3. Sig Confusion and Vulnerable Patients. Safety net pop-ulations include large proportions of patients with limitedhealth literacy (LHL), limited English proficiency (LEP),

and polypharmacy—all risk factors formisunderstanding theinstructions or “sig” on prescription drug labels [53–56].Thispresents particular risks with high alert medications suchas insulin, where the risk of severe ADE (including hypo-glycemia and death) with improper timing of insulin admin-istration ismuch greater than formostmedications.However,the risk of sig confusion also extends to other commonlyprescribed oral hypoglycemic, antihypertensive, and lipid-lowering medications that are required for cardiometaboliccontrol in diabetes. In examining a common diabetes “sig”instruction “take two tablets by mouth twice daily,” only36% of patients could correctly demonstrate this instruction,with higher odds of incorrect demonstration among thosewith limited health literacy [53]. In addition, patients withlimited English proficiency have significantly increased oddsof reporting difficulty in understanding prescription druglabels [56].

EHRs offer an important opportunity to increase pre-scribing safety by eliminating these confusing sigs. Universalmedication schedule (UMS) is a “plain-language” approachto standardizing and simplifying medication instructions tosupport safe and effective prescription drug use, highlightedby the Institute of Medicine as best practice for caringfor patients across the health literacy spectrum [57]. Allpatients—particularly those with LHL and LEP—are morelikely to interpret accurately and demonstrate comprehen-sion of UMS instructions compared with current standardinstructions [58–60]. A recent trial of patient-centered UMSprescription labels showed that LHL patients had higher oddsof adhering to medications, compared with those receivingstandard labels [60].

Unfortunately, despite strong support for the use of UMSlabeling from stakeholders such as the National Councilon Prescription Drug Programs (NCPDP) and the NationalAssociation of Boards of Pharmacy [61], EHR vendors havenot incorporated UMS as their standard instruction format,instead prepopulating medication instructions using moreconfusing older jargon [62]. Moreover, for more complexmedication instructions, such as regimens involving insulin,the default limit of 140 characters of the direction field pre-sents a barrier to prescribers or health systems adding theirown UMS instructions.

In the case ofMrs.D, the use ofUMS languagewould haveprevented Mrs. D from using a mealtime insulin at bedtime.Ideally, Dr. P should have been offered a menu of commonUMS instructions when she prescribed Novolin R insulin.

Given the wealth of evidence suggesting that UMS canincrease comprehension, medication safety, and adherencefor safety net patients [57–60], combinedwith the clear direc-tion in which labeling requirements are headed, electronicprescription vendors and EHRs should convert their stan-dardized drug frequencies to universal medication schedule(UMS) language.

Finally, language-concordant prescriptions in several dif-ferent languages have been shown to result in increasedcomprehension of instructions by patients [63, 64]. EHR ven-dors and policy makers should partner with implementation


researchers and innovative health information technologyplatforms to incorporate language-concordant instructionsfor limited English proficiency patients [63–65].

3.4. Missed Opportunity to Facilitate Clinically MeaningfulCollaboration with Pharmacists. Community pharmacistshave a unique perspective on electronic prescribing due totheir significant responsibilities with the downstream resultof the electronic prescriptions [17, 66]. Pharmacists are taskedwith clarifying the often conflicting instructions within theprescription, and successful implementation of EHRs canimprove both the quality of electronic prescriptions and thequality of communication with pharmacists facilitated byEHRs [66].

First, as described above, EHR electronic prescribingvendors should adopt RxNorm to facilitate best practicesimplementation of both generic and brand names within theEHR and the medication label, to minimize confusion anderror when switching formulations is required [12, 45, 46].

Second, EHRs and electronic prescribing vendors shouldfocus on enhancing bidirectional communication betweenprescribers and pharmacists, integrating pharmacists elec-tronically into the healthcare team supporting patients [16].Providers currently use free-text electronic prescribing fieldsto communicate a diverse set of information to pharmacists,including requests to discontinue refills because the per-message feedisincentivizes pharmacies from accepting struc-tured discontinuation messages [33–35]. Policies to removethis financial barrier will improve safety due to adverse drugevents from failed discontinuation and allow providers tofocus on conveying clinically meaningful information notcaptured through structured fields. Researchers have alsouncovered several other categories of communication fromprescribers to pharmacies, using text mining to codify thediverse wording of free-text prescriptions instructions [67].Further research is needed about how to codify and conveythese instructions through optimal design of EHR computerorder entry and EHR-pharmacy communication interfaces toreduce omission and dosing errors and to help pharmacistsdiscover potential prescription errors.

In addition, despite pharmacists’ role in clarifying poten-tially erroneous prescriptions, pharmacists still lack con-venient and efficient ways to reach prescribing providers.Pharmacists play significant roles uncovering barriers tomedication safety and adherence for diverse and under-served LHL populations, through personalized education,and phone outreach [68–72]. Pharmacists and pharmacytechnicians need platforms beyond faxing to convey thisvaluable information and the pharmacist team’s interventionsto prescriber care teams, thus expanding the patient-centeredmedical home into the patient’s neighborhood. While policyleaders have lamented EHRs’ overall failure to acknowledgethe important contribution of the entire patient care teamin providing care [73], improved EHR-pharmacy integrationshould also facilitate collaboration with the pharmacy teamswho work closely with patients in their communities [17].

4. Conclusion

EHRs have yet to realize their promise to improve the quality,safety, and patient-centeredness of diabetes medication pre-scribing. In fact, the literature has prompted enough concernthat informatics experts are calling for changes in EHR plat-forms, including better provider-centered design and usabil-ity testing of prescribing fields, tools for minimizing internalprescription discrepancies or omitted data, and enhancingprovider training on the optimal use of EHRs [12, 16, 74].TheOffice of the National Coordinator for Health InformationTechnology (ONC) is leading efforts to bring together EHRvendors, informatics leaders, and healthcare systems towardsimproving medication safety [75].

To improve the medication safety and health of safetynet patients with diabetes—including LHL and LEP popula-tions—we recommend policy changes to facilitate the fol-lowing: (1) the adoption of RxNorm to reduce patient andprescriber confusion about medication names; (2) focusedstakeholder engagement to standardize the naming andpicklist options for high alert medications such as insulin; (3)the widespread implementation of the universal medicationschedule and language-concordant labels, with the expansionof electronic prescription 140-character limit, to improvepatient comprehension of how and when to take their medi-cations; and (4) enhanced bidirectional communication withpharmacy partners, enabling pharmacies to receive “discon-tinuation” messages without cost and improving interoper-ability to allow pharmacist communication back to providers.Finally, informatics and implementation researchers shouldinclude safety net healthcare systems in examining the spe-cific positive and negative consequences of EHR tools such aselectronic prescribing in the care of diverse vulnerable popu-lations.

EHR-Facilitated Support of Mrs. D

Mrs. D, a 67-year-old Latina woman, suffers fromthirst and frequent urination, with a hemoglobinA1c that jumped from 7.5% to 9.7%. Her commu-nity pharmacist sends a message to Dr. P statingthatMrs. D was asking her questions about piogli-tazone side effects, reporting swollen face and legs.So, Dr. P brought Mrs. D in for a visit and coun-seled Mrs. D about adding a short-acting meal-time insulin to her metformin and glargine regi-men. Dr. P searches the EHR RxNorm databasefor regular insulin and chooses a version coveredbyMrs. D’s insurance, but indicating flexibility forpharmacists to fill the least expensive formulationof regular insulin. After choosing the medication,the default instructions read “Take — units withbreakfast and take— units with dinner,” allowingDr. P to fill in the correct number of units foreach dose. Dr. P also selects a “STOP” message forpioglitazone, which also allows her to record thisadverse reaction automatically in Mrs. D’s chartsin both Dr. P’s EHR and the pharmacy’s electronicsystem. Mrs. D’s after-visit summary includes


an easy-to-read Spanish version of her medica-tion schedule, with her medications automaticallysorted by therapeutic indication. Similarly, thecommunity pharmacist technician knows Mrs. Dprefers instructions in Spanish and prints the pre-scription label in Spanish with the UMS instruct-ions. Mrs. D is able to teach back both to Dr. Pand to the community pharmacist how she willstop her old medication and how she will take hernewmedication, expressing thanks for the supportfrom her entire care team.

Disclosure

No funders had any role in the design and conduct of thestudy; collection, management, analysis, and interpretationof the data; writing of the manuscript; preparation, review,or approval of the manuscript; or decision to submit themanuscript for publication.

Competing Interests

None of the authors had conflict of interests.

Authors’ Contributions

All authors contributed to the conception and design anddrafting and critical revision of the manuscript, includingfinal approval of the version to be published.

Acknowledgments

This research was supported by Agency for HealthcareResearch and Quality Grants 1K08HS022561 andP30HS023558; Health Delivery Systems Center for DiabetesTranslational Research (CDTR) funded through NIDDKGrant 1P30-DK092924; the National Center for AdvancingTranslational Sciences of the NIH under Award no.KL2TR000143; and the UCSF Open Access PublishingFund.

References

[1] Office of the National Coordinator for Health InformationTechnology, Office-based Physician Electronic Health RecordAdoption: 2004–2014, 2015, http://dashboard.healthit.gov/quickstats/pages/physician-ehr-adoption-trends.php.

[2] Office of the National Coordinator for Health InformationTechnology, Office-based Health Care ProfessionalsParticipating in the CMS EHR Incentive Programs, Health ITQuick-Stat #44, 2016, http://dashboard.healthit.gov/quickstats/pages/FIG-Health-Care-Professionals-EHR-Incentive-Programs.php.

[3] A. K. Jha, “Meaningful use of electronic health records: the roadahead,” JAMA, vol. 304, no. 15, pp. 1709–1710, 2010.

[4] R. D. Cebul, T. E. Love, A. K. Jain, and C. J. Hebert, “Electronichealth records and quality of diabetes care,” New EnglandJournal of Medicine, vol. 365, no. 9, pp. 825–833, 2011.

[5] D. Heisey-Grove, K. Hawkins, E. Jones, K. Shanks, and K.Lynch, “Supporting Health Information Technology Adoptionin Federally Qualified Health Centers,” ONC Data Brief 2013.

[6] L.M.Kern andR. Kaushal, “Measuring the impact of “meaning-ful use” on quality of care,” JAMA InternalMedicine, vol. 174, no.6, pp. 998–999, 2014.

[7] L. Samal, A. Wright, M. J. Healey, J. A. Linder, and D. W. Bates,“Meaningful use and quality of care,” JAMA Internal Medicine,vol. 174, no. 6, pp. 997–998, 2014.

[8] L. M. Kern, Y. Barron, R. V. Dhopeshwarkar, A. Edwards, andR. Kaushal, “Electronic health records and ambulatory qualityof care,” Journal of General Internal Medicine, vol. 28, no. 4, pp.496–503, 2013.

[9] E. G. Poon, A.Wright, S. R. Simon et al., “Relationship betweenuse of electronic health record features and health care quality:results of a statewide survey,” Medical Care, vol. 48, no. 3, pp.203–209, 2010.

[10] B. Chaudhry, J.Wang, S.Wu et al., “Systematic review: impact ofhealth information technology on quality, efficiency, and costsof medical care,”Annals of InternalMedicine, vol. 144, no. 10, pp.742–752, 2006.

[11] M.A. Fischer, C.Vogeli,M. Stedman, T. Ferris,M.A. Brookhart,and J. S.Weissman, “Effect of electronic prescribingwith formu-lary decision support on medication use and cost,” Archives ofInternal Medicine, vol. 168, no. 22, pp. 2433–2439, 2008.

[12] E.W.Ashton, E-prescribing: E for error?, 2012, https://psnet.ahrq.gov/webmm/case/260.

[13] S. P. Slight, T. Eguale, M. G. Amato et al., “The vulnerabilitiesof computerized physician order entry systems: a qualitativestudy,” Journal of the American Medical Informatics Association,vol. 23, no. 2, pp. 311–316, 2016.

[14] G. Schiff, M. G. Amato, T. Eguale et al., “Computerised physi-cian order entry-related medication errors: analysis of reportederrors and vulnerability testing of current systems,”BMJQualityand Safety, vol. 24, no. 4, pp. 264–271, 2015.

[15] M. B. Palchuk, E. A. Fang, J. M. Cygielnik et al., “An unintendedconsequence of electronic prescriptions: prevalence and impactof internal discrepancies,” Journal of the American MedicalInformatics Association, vol. 17, no. 4, pp. 472–476, 2010.

[16] K. C. Nanji, J. M. Rothschild, C. Salzberg et al., “Errorsassociated with outpatient computerized prescribing systems,”Journal of the AmericanMedical Informatics Association, vol. 18,no. 6, pp. 767–773, 2011.

[17] K. C. Nanji, J. M. Rothschild, J. J. Boehne, C. A. Keohane,J. S. Ash, and E. G. Poon, “Unrealized potential and residualconsequences of electronic prescribing on pharmacy workflowin the outpatient pharmacy,” Journal of the American MedicalInformatics Association, vol. 21, no. 3, pp. 481–486, 2014.

[18] S. Shortell, “Safety Net Challenges In Delivering AccountableCare,” Health Affairs Blog, 2012, http://healthaffairs.org/blog/2012/11/01/safety-net-challenges-in-delivering-accountable-care/.

[19] T. Sentell, “Implications for reform: survey of California adultssuggests low health literacy predicts likelihood of being unin-sured,” Health Affairs, vol. 31, no. 5, pp. 1039–1048, 2012.

[20] R. Rivadeneyra, V. Elderkin-Thompson, R. C. Silver, and H.Waitzkin, “Patient centeredness in medical encounters requir-ing an interpreter,” The American Journal of Medicine, vol. 108,no. 6, pp. 470–474, 2000.

[21] N.D. Berkman, S. L. Sheridan,K. E.Donahue,D. J.Halpern, andK. Crotty, “Low health literacy and health outcomes: an updated

http://dashboard.healthit.gov/quickstats/pages/physician-ehr-adoption-trends.php

http://dashboard.healthit.gov/quickstats/pages/physician-ehr-adoption-trends.php

http://dashboard.healthit.gov/quickstats/pages/FIG-Health-Care-Professionals-EHR-Incentive-Programs.php



https://psnet.ahrq.gov/webmm/case/260

https://psnet.ahrq.gov/webmm/case/260

http://healthaffairs.org/blog/2012/11/01/safety-net-challenges-in-delivering-accountable-care/




systematic review,” Annals of Internal Medicine, vol. 155, no. 2,pp. 97–107, 2011.

[22] E. Wilson, A. H. Chen, K. Grumbach, F. Wang, and A.Fernandez, “Effects of limited English proficiency and physicianlanguage on health care comprehension,” Journal of GeneralInternal Medicine, vol. 20, no. 9, pp. 800–806, 2005.

[23] C. A. Dubard and Z. Gizlice, “Language spoken and differencesin health status, access to care, and receipt of preventive servicesamongUShispanics,”American Journal of PublicHealth, vol. 98,no. 11, pp. 2021–2028, 2008.

[24] U. Sarkar, A. J. Karter, J. Y. Liu, H. H. Moffet, N. E. Adler, andD. Schillinger, “Hypoglycemia is more common among type 2diabetes patients with limited health literacy: the diabetes studyof northern California (distance),” Journal of General InternalMedicine, vol. 25, no. 9, pp. 962–968, 2010.

[25] D. Schillinger, K. Grumbach, J. Piette et al., “Association ofhealth literacy with diabetes outcomes,” The Journal of theAmerican Medical Association, vol. 288, no. 4, pp. 475–482,2002.

[26] D. A. DeWalt, N. D. Berkman, S. Sheridan, K. N. Lohr, andM. P.Pignone, “Literacy and health outcomes: a systematic review ofthe literature,” Journal of General Internal Medicine, vol. 19, no.12, pp. 1228–1239, 2004.

[27] A. Fernandez, D. Schillinger, E. M. Warton et al., “Languagebarriers, physician-patient language concordance, and glycemiccontrol among insured Latinos with diabetes: the DiabetesStudy of Northern California (DISTANCE),” Journal of GeneralInternal Medicine, vol. 26, no. 2, pp. 170–176, 2011.

[28] U. Sarkar, A. Lopez, J. H. Maselli, and R. Gonzales, “Adversedrug events in U.S. adult ambulatory medical care,” HealthServices Research, vol. 46, no. 5, pp. 1517–1533, 2011.

[29] L. A. Thomsen, A. G. Winterstein, B. Søndergaard, L. S.Haugbølle, and A. Melander, “Systematic review of the inci-dence and characteristics of preventable adverse drug events inambulatory care,” Annals of Pharmacotherapy, vol. 41, no. 9, pp.1411–1426, 2007.

[30] G. D. Schiff and T. D. Rucker, “Computerized prescribing:building the electronic infrastructure for better medicationusage,” The Journal of the American Medical Association, vol.279, no. 13, pp. 1024–1029, 1998.

[31] D. W. Bates, M. Cohen, L. L. Leape, J. M. Overhage, M. M.Shabot, and T. Sheridan, “Reducing the frequency of errorsin medicine using information technology,” Journal of theAmericanMedical Informatics Association, vol. 8, no. 4, pp. 299–308, 2001.

[32] T. K. Gandhi, S. N. Weingart, A. C. Seger et al., “Outpatientprescribing errors and the impact of computerized prescribing,”Journal of General Internal Medicine, vol. 20, no. 9, pp. 837–841,2005.

[33] A. A. Dhavle, Y. Yang, M. T. Rupp, H. Singh, S. Ward-Charlerie,and J. Ruiz, “Analysis of prescribers’ notes in electronic prescrip-tions in ambulatory practice,” JAMA Internal Medicine, vol. 176,no. 4, pp. 463–470, 2016.

[34] Surescripts. What is considered a “billable e-prescribing trans-action”?, http://surescripts.com/support/faqs/pharmacists/detail/what-is-considered-a-billable-transaction.

[35] A. S. Allen and T. D. Sequist, “Pharmacy dispensing of electron-ically discontinued medications,” Annals of Internal Medicine,vol. 157, no. 10, pp. 700–705, 2012.

[36] P. J. Baranowski, K. L. Peterson, J. L. Statz-Paynter, and J. A.Zorek, “Incidence and cost of medications dispensed despite

electronicmedical record discontinuation,” Journal of the Amer-ican Pharmacists Association, vol. 55, no. 3, pp. 313–319, 2015.

[37] J. M. Overhage, T. K. Gandhi, C. Hope et al., “Ambulatorycomputerized prescribing and preventable adverse drug events,”Journal of Patient Safety, vol. 12, no. 2, pp. 69–74, 2016.

[38] M. Salas, D. Hughes, A. Zuluaga, K. Vardeva, and M. Lebmeier,“Costs of medication nonadherence in patients with diabetesmellitus: a systematic review and critical analysis of the liter-ature,” Value in Health, vol. 12, no. 6, pp. 915–922, 2009.

[39] J. E. Kurlander, E. A. Kerr, S. Krein, M. Heisler, and J. D. Piette,“Cost-related nonadherence to medications among patientswith diabetes and chronic pain: factors beyond finances,”Diabetes Care, vol. 32, no. 12, pp. 2143–2148, 2009.

[40] D. Comer, J. Couto, R. Aguiar, P. Wu, and D. Elliott, “Usingaggregated pharmacy claims to identify primary nonadher-ence,” American Journal of Managed Care, vol. 21, pp. e655–e660, 2015.

[41] D. Comer, J. Couto, R. Aguiar, P. Wu, and D. J. Elliott,“Usefulness of pharmacy claims for medication reconciliationin primary care,”American Journal of Managed Care, vol. 21, no.7, pp. 486–493, 2015.

[42] M. Heisler, T. P. Hofer, J. A. Schmittdiel et al., “Improvingblood pressure control through a clinical pharmacist out-reach program in patients with diabetes mellitus in 2 high-performing health systems: the adherence and intensificationof medications cluster randomized, controlled pragmatic trial,”Circulation, vol. 125, no. 23, pp. 2863–2872, 2012.

[43] J. M. Pevnick, N. Li, S. M. Asch, C. A. Jackevicius, and D. S.Bell, “Effect of electronic prescribing with formulary decisionsupport on medication tier, copayments, and adherence,” BMCMedical Informatics and Decision Making, vol. 14, no. 1, article79, 2014.

[44] US Food and Drug Administration, “National drug code direc-tory,” 2016, http://www.fda.gov/Drugs/InformationOnDrugs/ucm142438.htm.

[45] The Joint Commission, “What Did the Doctor Say?”: ImprovingHealth Literacy to Protect Patient Safety, The Joint Commission,Oakbrook Terrace, Ill, USA, 2007, https://www.jointcommis-sion.org/assets/1/18/improving health literacy.pdf.

[46] S. J. Nelson, K. Zeng, J. Kilbourne, T. Powell, and R. Moore,“Normalized names for clinical drugs: RxNorm at 6 years,”Journal of the AmericanMedical Informatics Association, vol. 18,no. 4, pp. 441–448, 2011.

[47] Institute for SafeMedicationPractices, “ISMPList ofHigh-AlertMedications in Community/Ambulatory Healthcare,” 2011,https://www.ismp.org/communityRx/tools/highAlert-commu-nity.pdf.

[48] J. S. Ash, D. F. Sittig, R. H. Dykstra, K. Guappone, J. D.Carpenter, and V. Seshadri, “Categorizing the unintendedsociotechnical consequences of computerized provider orderentry,” International Journal of Medical Informatics, vol. 76,supplement 1, pp. 21–27, 2007.

[49] C. Zhan, R. W. Hicks, C. M. Blanchette, M. A. Keyes, and D. D.Cousins, “Potential benefits and problems with computerizedprescriber order entry: analysis of a voluntary medicationerror-reporting database,” American Journal of Health-SystemPharmacy, vol. 63, no. 4, pp. 353–358, 2006.

[50] M. Smith, D. Dang, and J. Lee, “E-prescribing: clinical implica-tions for patients with diabetes,” Journal of Diabetes Science andTechnology, vol. 3, no. 5, pp. 1215–1218, 2009.

http://surescripts.com/support/faqs/pharmacists/detail/what-is-considered-a-billable-transaction

http://surescripts.com/support/faqs/pharmacists/detail/what-is-considered-a-billable-transaction

http://www.fda.gov/Drugs/InformationOnDrugs/ucm142438.htm

http://www.fda.gov/Drugs/InformationOnDrugs/ucm142438.htm

https://www.jointcommission.org/assets/1/18/improving_health_literacy.pdf

https://www.jointcommission.org/assets/1/18/improving_health_literacy.pdf

https://www.ismp.org/communityRx/tools/highAlert-community.pdf

https://www.ismp.org/communityRx/tools/highAlert-community.pdf


[51] L. Meneghini, “New insulin preparations: a primer for theclinician,” Cleveland Clinic Journal of Medicine, vol. 83, pp. S27–S33, 2016.

[52] Institute for Safe Medication Practices, “FDA and ISMP Lists ofLook-Alike DrugNames with Recommended TallMan Letters,”2016, https://www.ismp.org/tools/tallmanletters.pdf.

[53] T. C. Davis, M. S. Wolf, P. F. Bass III et al., “Literacy andmisunderstanding prescription drug labels,” Annals of InternalMedicine, vol. 145, no. 12, pp. 887–894, 2006.

[54] M. S. Wolf, T. C. Davis, H. H. Tilson, P. F. Bass III, and R. M.Parker, “Misunderstanding of prescription drug warning labelsamong patients with low literacy,” American Journal of Health-System Pharmacy, vol. 63, no. 11, pp. 1048–1055, 2006.

[55] T. C. Davis, M. S. Wolf, P. F. Bass III et al., “Low literacy impairscomprehension of prescription drug warning labels,” Journal ofGeneral Internal Medicine, vol. 21, no. 8, pp. 847–851, 2006.

[56] M. C. Masland, S. H. Kang, and Y. Ma, “Association betweenlimited English proficiency and understanding prescriptionlabels among five ethnic groups in California,” Ethnicity andHealth, vol. 16, no. 2, pp. 125–144, 2011.

[57] IOM (Institute of Medicine), Standardizing MedicationLabels: Confusing Patients Less, Workshop Summary, NationalAcademies Press, Washington, DC, USA, 2008.

[58] M. S. Wolf, T. C. Davis, L. M. Curtis et al., “Effect ofstandardized, patient-centered label instructions to improvecomprehension of prescription drug use,”Medical Care, vol. 49,no. 1, pp. 96–100, 2011.

[59] L. J. Sahm, M. S. Wolf, L. M. Curtis et al., “What’s in a label?An exploratory study of patient-centered drug instructions,”European Journal of Clinical Pharmacology, vol. 68, no. 5, pp.777–782, 2012.

[60] M. S. Wolf, T. C. Davis, L. M. Curtis et al., “A patient-centeredprescription drug label to promote appropriate medication useand adherence,” Journal of General InternalMedicine, vol. 31, no.12, pp. 1482–1489, 2016.

[61] National Council for Prescription Drug Programs, UniversalMedication Schedule White Paper, National Council for Pre-scription Drug Programs, Scottsdale, Ariz, USA, 2013.

[62] Surescripts, “E-Prescribing Quality Guidelines,” 2016, http://surescripts.com/docs/default-source/Resources/quality e-prescription guidelines surescripts.pdf.

[63] S. C. Bailey, U. Sarkar, A. H. Chen, D. Schillinger, and M.S. Wolf, “Evaluation of language concordant, patient-centereddrug label instructions,” Journal of General Internal Medicine,vol. 27, no. 12, pp. 1707–1713, 2012.

[64] S. C. Bailey, R. Hasnain-Wynia, A. H. Chen et al., “Developingmultilingual prescription instructions for patients with limitedEnglish proficiency,” Journal of Health Care for the Poor andUnderserved, vol. 23, no. 1, pp. 81–87, 2012.

[65] A. Mohan, M. B. Riley, D. Boyington et al., “Development ofa patient-centered bilingual prescription drug label,” Journal ofHealth Communication, vol. 18, no. 1, pp. 49–61, 2013.

[66] A. Singer and R. D. Fernandez, “The effect of electronic medicalrecord system use on communication between pharmacists andprescribers,” BMCFamily Practice, vol. 16, no. 1, article 155, 2015.

[67] G. Karystianis, T. Sheppard, W. G. Dixon, and G. Nenadic,“Modelling and extraction of variability in free-text medicationprescriptions from an anonymised primary care electronicmedical record research database Healthcare Information Sys-tems,” BMC Medical Informatics and Decision Making, vol. 16,no. 1, article no. 18, 2016.

[68] J. D. Hirsch, M. Gonzales, A. Rosenquist, T. A. Miller, T. P.Gilmer, and B.M. Best, “Antiretroviral therapy adherence, med-ication use, and health care costs during 3 years of a communitypharmacy medication therapy management program for Medi-Cal beneficiaries with HIV/AIDS,” Journal of Managed CarePharmacy, vol. 17, no. 3, pp. 213–223, 2011.

[69] J. K. Lee, K. A. Grace, and A. J. Taylor, “Effect of a pharmacycare program on medication adherence and persistence, bloodpressure, and low-density lipoprotein cholesterol: a randomizedcontrolled trial,” Journal of the American Medical Association,vol. 296, no. 21, pp. 2563–2571, 2006.

[70] H. B. Congdon, T. C. Dowling, I. Cheng, and H.-A. Truong,“Impact of medication therapy management on underserved,primarily Hispanic patients with diabetes,” Annals of Pharma-cotherapy, vol. 47, no. 5, pp. 665–670, 2013.

[71] M.Noureldin, K. S. Plake,D.G.Morrow,W.Tu, J.Wu, andM.D.Murray, “Effect of health literacy on drug adherence in patientswith heart failure,” Pharmacotherapy, vol. 32, no. 9, pp. 819–826,2012.

[72] J. S. Skinner, B. Poe, R. Hopper, A. Boyer, and C. H. Wilkins,“Assessing the effectiveness of pharmacist-directed medica-tion therapy management in improving diabetes outcomes inpatients with poorly controlled diabetes,” Diabetes Educator,vol. 41, no. 4, pp. 459–465, 2015.

[73] S. A. Martin and C. A. Sinsky, “The map is not the territory:medical records and 21st century practice,”The Lancet, 2016.

[74] J. L. Schnipper, “Free-text notes as a marker of needed improve-ments in electronic prescribing,” JAMA Internal Medicine, vol.176, no. 4, pp. 471–472, 2016.

[75] RTI International on behalf of Office of the National Coor-dinator for Health Information Technology, Report on theSafe Use of Pick Lists in Ambulatory Care Settings: Issuesand Recommended Solutions for Improved Usability in PatientSelection and Medication Ordering, Research Triangle Park,NC, 2016, https://www.healthit.gov/sites/default/files/report-on-the-safe-use-of-pick-lists-in-ambulatory-care-settings.pdf.

https://www.ismp.org/tools/tallmanletters.pdf

http://surescripts.com/docs/default-source/Resources/quality_e-prescription_guidelines_surescripts.pdf



https://www.healthit.gov/sites/default/files/report-on-the-safe-use-of-pick-lists-in-ambulatory-care-settings.pdf

https://www.healthit.gov/sites/default/files/report-on-the-safe-use-of-pick-lists-in-ambulatory-care-settings.pdf

Research ArticleThe Design, Usability, and Feasibility of a Family-FocusedDiabetes Self-Care Support mHealth Intervention for Diverse,Low-Income Adults with Type 2 Diabetes

Lindsay Satterwhite Mayberry,1,2 Cynthia A. Berg,3

Kryseana J. Harper,1,2 and Chandra Y. Osborn1,2,4

1Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA2Center for Health Behavior and Health Education, Vanderbilt University Medical Center, Nashville, TN, USA3Department of Psychology, University of Utah, Salt Lake City, UT, USA4Department of Biomedical Informatics, Vanderbilt University Medical Center, Nashville, TN, USA

Correspondence should be addressed to Lindsay Satterwhite Mayberry; [email protected]

Received 21 July 2016; Accepted 4 September 2016

Academic Editor: Shari Bolen

Copyright © 2016 Lindsay Satterwhite Mayberry et al. This is an open access article distributed under the Creative CommonsAttribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work isproperly cited.

Family members’ helpful and harmful actions affect adherence to self-care and glycemic control among adults with type 2diabetes (T2D) and low socioeconomic status. Few family interventions for adults with T2D address harmful actions or usetext messages to reach family members. Through user-centered design and iterative usability/feasibility testing, we developed amHealth intervention for disadvantaged adults with T2D called FAMS. FAMS delivers phone coaching to set self-care goals andimprove patient participant’s (PP) ability to identify and address family actions that support/impede self-care. PPs receive textmessage support and can choose to invite a support person (SP) to receive text messages. We recruited 19 adults with T2D fromthree Federally Qualified Health Centers to use FAMS for two weeks and complete a feedback interview. Coach-reported datacaptured coaching success, technical data captured user engagement, and PP/SP interviews captured the FAMS experience. PPswere predominantly AfricanAmerican, 83%had incomes<$35,000, and 26%weremarried.Most SPs (𝑛 = 7) were spouses/partnersor adult children. PPs reported FAMS increased self-care and both PPs and SPs reported FAMS improved support for andcommunication about diabetes. FAMS is usable and feasible and appears to help patients manage self-care support, although somePPs may not have a SP.

1. Introduction

Family members and other close loved ones participate inpatients’ daily routines and are often present for self-careactivities (e.g., food selection, meal preparation, and disease-related problem solving and coping) [1–5]. For adults withtype 2 diabetes (T2D), the receipt of helpful actions (i.e.,instrumental support) ismore predictive of adherence to self-care than other types of support (e.g., emotional support) [2,6, 7]. According to both social control theory [8, 9] and familysystems theory [10, 11], family members are well-positionedto provide instrumental support for diabetes self-care and tocreate contexts valuing and supporting self-care adherence.

However, patients who experience more harmful actions(e.g., nagging, arguing, and sabotaging self-care efforts) areless adherent to self-care [2, 12, 13] and have worse glycemiccontrol [12]. Moreover, helpful and harmful actions are eachstrongly associated with patients’ being more or less adherentto diet and exercise recommendations, respectively [12].

According to social control theory, certain actions areharmful because they are misaligned with the types of sup-port patients need [8, 14] and/or infringe on their autonomyand create resentment [15] (e.g., nagging, arguing about self-care). According to family systems theory, harmful actionsundermine patients’ efforts to initiate and sustain self-careefforts (i.e., undermining or sabotaging self-care efforts, such


http://dx.doi.org/10.1155/2016/7586385


as bringing tempting unhealthy foods into patients’ homes)[10, 11]. Families who become involved in adults’ self-careperform both helpful and harmful actions [2, 12, 16–18],but family interventions for adults with T2D have not ade-quately addressed the harmful aspects of family involvement[19].

Family interventions targeting both helpful and harmfulactionsmay be particularly useful for racial/ethnicminoritiesand persons with low socioeconomic status (SES) with T2D.These groups have high rates of limited health literacy [20],more life stressors [21–23], and depression [24, 25], whichmaymake themmore vulnerable to the detrimental effects ofothers’ harmful actions on their diabetes self-care and control[12, 26]. Family interventions may also be challenging forpatients with T2D who (a) have diverse living and familysituations [1, 27, 28], (b) live apart from the person(s) pro-viding the most support [27, 28], (c) have difficulty attendingand/or bringing family members to in-person interventionsdue to competing priorities and limited resources [4, 27, 29],and (d) are concerned family involvement will underminetheir autonomy and self-efficacy [27]. On the contrary, familymembers want to be more involved in adults’ diabetesmanagement and often feel frustratedwhen they do not knowhow to best help [30]. Thus, patients may appreciate andbenefit from one-on-one guidance on how to identify andcommunicate their desires and needs for specific supportiveactions from family members [30].

In other chronic disease contexts (e.g., cardiovasculardiseases, cancer, and arthritis), family interventions havesignificant and stable effects on health over and above patient-only interventions [31] but have been less successful in T2D[19, 32]. Therefore, we developed FAMS (Family-FocusedAdd-On for Motivating Self-Care) to help patients identifya diet or exercise goal, get desired support from familymembers or close loved ones, and redirect or cope withundesired or harmful actions without compromising theirown health goals. FAMS seeks to improve patients’ adherenceto diet and exercise recommendations via increased self-efficacy, increased receipt of helpful actions, and reducedreceipt of harmful actions.

FAMS is delivered via basic mobile phone technology(i.e., phone calls and text messages), which is ubiquitouslyavailable in the USA [33], even among adults with the lowestSES and racial/ethnic minority groups [34, 35]. A robust,multistep approach is recommended to develop effectivemHealth interventions for patients with diabetes, particularlyfor underserved or vulnerable patients [36].Therefore, FAMSwas developed from (a) front-end qualitative and quantitativeformative research with adults with T2D [2] and low SES [12,27], alongside (b) early feedback from potential users (adultswith T2D and low SES) and (c)members of our research team(who used and critiqued FAMS during internal beta testing),followed by (d) iterative usability and feasibility testing withpotential users. Our objectives were to develop a family-focused intervention acceptable to patients receiving carefrom Federally Qualified Health Centers (FQHCs), obtainfeedback and data to improve the intervention, and ensureour research processes were sound prior to an evaluativetrial.

2. User-Centered Intervention Design

For mHealth interventions in diabetes [36, 37], user-centereddesign entails formative research prior to and during inter-vention development to understand the needs, values, andabilities of users, as well as iteratively assessing the designto improve users’ perceptions of and interactions with thetechnology and content. Following these principles, wedeveloped FAMS to harness universal text messaging (i.e.,not requiring a smartphone or the Internet). To accom-modate the diversity of family types among adults withT2D [1, 27, 28, 38], we designed FAMS to acknowledgethat patients’ “families” include whomever the patient con-siders included, regardless of legal/biological relationships.FAMS content is sensitive and applicable to various livingarrangements (e.g., living alone, with children, or withfriends/roommates). FAMS text messages were designed tobe≤6th-grade reading level and testedwith the Flesh-KincaidGrade Level and Automated Readability Index (tested withhttps://readability-score.com/). To ensure plain language andaccommodate those with literacy limitations, we ensuredeach sentence expressed one thought, simplified sentencestructure, and used active voice [39]. We then replaced orplainly defined multisyllabic (≥three syllables) and uncom-mon words and medical jargon [39]. Messages avoid refer-ences to potentially unavailable resources (e.g., gym mem-berships). We worked with a digital content developer toshortenmessages to ≤160 characters (a common limit for textmessages) while maintaining their meaning.

2.1. FAMS Intervention Components. Each FAMS componentis briefly described below and in Table 1 with example mes-sages. Phone coaching seeks to improve the patient’s abilityto identify family members’ actions that support or impedehis/her self-care goals and his/her skills and self-efficacyto ask for needed support and manage harmful actions tomeet these goals. Authors LSM and KJH, who have graduatedegrees in counseling, developed FAMS phone coachingto be deliverable by counselors/counselors-in-training orhealth coaches/health coaches-in-training (i.e., persons withtraining in helping skills). FAMS coaching occurs with thePP alone and combines family therapy with basic healthcoaching. Among adults with T2D, health coaching has suc-cessfully improved adherence to exercise recommendations[40], psychological functioning [41], and illness-related cop-ing [41]. FAMS coaching employs evidence-based techniquesfromgoal setting theory [42] (assessing current behaviors andproblem solving and assessing confidence to achieve a goaland revising goals with a confidence rating <7 on a 10-pointscale), cognitive behavioral therapy [43, 44] (role playing,homework), and best practices in health communication[45] (teach back). FAMS sends the PP one-way and two-way/interactive text message support tailored to his/her self-care goal selected during coaching and preferred time of day(Table 1).

Patients can choose to invite a support person (SP) regard-less of relationship type or living arrangement to receive textmessages (Table 1). PPs were encouraged to select someonewho is part of their daily life and routine and not someone

https://readability-score.com/


Table1:FA

MSinterventio

ncompo

nentsa

nddescrip

tion.

Com

ponent

Descriptio

n

Patie

ntparticipants

Phon

ecoaching

Each

sessioninclu

desthe

following:

(i)Goalsettin

g :identifyaS

MART

(specific,m

easurable,attainable,

relevant,tim

e-bo

und)

dietor

exercise

goal

(ii)D

iscussin

ghelpful&

harm

fulfam

ilyactio

ns:identify

anddifferentiatebetweenfamily

mem

bers’help

fuland

harm

fulactions

related

totheS

MART

goal

(iii)Skill

build

ing:completee

xercise

sinrelationto

aspecific

wantedhelpfulactionor

unwantedharm

fulactionandconfi

rmcomprehensio

nviar

olep

laying

orteaching

back

(iv)V

erbalcon

tract:makea

plan

toem

ploy

thes

killwith

aspecific

family

mem

bera

ndassessparticipant’s

self-effi

cacy

todo

so

Patie

ntparticipants

Text

messages

Goalsup

portmessages :on

e-way

textsw

ithtip

sand

motivationalcon

tent

tailo

redto

theS

MART

goalsetd

uringph

onec

oaching:

(i)2textse

achweekaren

ottailo

redto

goaltype

Changing

your

lifestyleisn

’teasy,but

smallsteps

matter!Reaching

your

SMAR

Tgoaltodayc

anaddup

tobigimprovem

ents!

(ii)1

text

each

weekistailo

redto

thep

artic

ipant’s

goaltype

(dieto

rexercise

)Manyp

eopleforgeta

bout

their

dietgoalsa

roun

dholid

ayso

rvacations.P

lanaheadandpreparefor

thesechangesin

your

daily

routine.

Workyour

SMAR

Texerciseg

oalintoyour

day.Ex

ercisew

hiledoingo

therthingsliketalking

onthep

hone

orwa

tchingT

V.Th

eone-w

aytext

issent

with

inusers’preferredwindo

wof

time(thee

xacttim

evariesw

ithin

thiswindo

w)

Goalassessm

entm

essages:Ea

chSaturday,partic

ipantsreceivea

two-way/in

teractivetexttailoredto

theirg

oal

Your

goalwa

stowa

lk15

minuteseach

day.How

manyd

aysd

idyoumeetthisgoallastw

eek(Sun

-Sat)?Replyw

iththen

umbero

fdays,0–7.

(i)Participant’s

respon

seisfollo

wed

byafeedb

acktext,tailoredto

ther

espo

nse

Excellent

progress!

Considerwh

atmotivated

youtodo

bette

rthisw

eekthan

inwe

eksb

efore.K

eepup

your

good

work!

Supp

ortp

ersons

Invitatio

n

Participantscanchoo

seto

invitean

adultfam

ilymem

bero

rfrie

ndto

receivetextm

essages

(i)Participantswho

inviteas

uppo

rtperson

provideh

is/herfi

rstn

amea

ndph

onen

umber

(ii)S

uppo

rtperson

mustb

eEng

lish-speaking

andow

nac

ellpho

ne(iii)Stud

ysta

ffcalls

thes

uppo

rtperson

toinvitehim/her

toparticipate,completeinformed

consent,enrollviap

hone

Supp

ortp

ersons

Text

messages

Each

week,enrolledsupp

ortp

ersons

receive3

one-way

textsw

ithcontenttailoredto

thep

atient

participant’s

selected

goal,n

ame,gend

erBo

bhassetagoaltoim

proveh

isdiet.A

skhim

aboutitthe

next

timey

ouspeaktohim.

(i)2textse

achweekaren

ottailo

redto

goaltype

Oftenwe

want

tohelpourloved

oneswith

their

health

butd

on’tknow

how.

AskBo

bwh

atyoucando

tomakeiteasier

forh

imtoreachhisgoal.

(ii)1

text

each

weekistailo

redto

thep

atient

participant’s

goaltype

(dieto

rexercise

)Be

sensitive

abouteatingtreatsinfro

ntofBo

bas

itmay

tempt

ormakeh

imfee

lleft

out.Inste

ad,choosetreatsyou

canenjoytogether.

Dancin

gisthe

mostfun

exercisey

oucando

athome.Help

Bobmeeth

isexerciseg

oalbyp

uttin

gonhisfavorite

jam

andsta

rtinga

2person

party!

Theo

ne-w

aytextsa

resent

atthes

uppo

rtperson’spreferredtim

eofd

ay


with whom they “have a lot of conflict.”The SP does not haveto be the identified family member in the coaching session.An enrolled SP receives messages tailored to the PP’s name,gender, and goal type (Table 1). These aim to help SPs bethoughtful about the support they provide and to increasecommunication about diabetes and the PP’s goal. The SPtext messages do not provide information on the PP’s goalachievement to avoid encouraging nagging/arguing.

2.2. Community Engaged Research Studio. Before developingFAMS functionality, we shared FAMS design and contentwith adults with T2D through a Community EngagedResearch Studio (CES) [46] and made improvementsbased on their recommendations. The CES, conductedthrough the Meharry-Vanderbilt Community EngagedResearch Core, employs direct feedback from communitymembers (“experts”) who share similar demographiccharacteristics of a researcher’s desired sample to identifyand address concerns (e.g., cultural appropriateness ofstudy materials, fair compensation, and intervention/surveydesign) [46]. Our CES experts (∼12) were diagnosed withT2D, predominantly African American (AA), and withlow SES - no family members were included. The CESfacilitator read each text message and participants sharedfeedback ranging from “good, like it” to discussions withmultiple view-points expressed. Authors LSM and KJH askedfollow-up or clarifying questions as needed during the CES;afterwards, the facilitators compiled and provided notes.

CES experts provided helpful feedback. They did not likemessages presuming patients were struggling to meet theirgoal or that evoked strong negative language. For example,in one text message they advised the word “dangerous” wastoo negative: “Traveling is dangerous for diet goals.” Theyrecommended it be changed to the following: “Traveling cancause you not to meet your diet goals.” CES experts alsorecommended defining “self-care” to avoid confusion, so wereplaced this phrase with a specific self-care behavior (e.g.,diet or exercise) or “taking care of your diabetes.”Communityexperts said they would feel comfortable inviting a SP andprovided feedback on text frequency (recommended daily)and coaching frequency (most recommended every twoweeks or monthly) and duration (most recommended 20–30minutes). Finally, experts were enthusiastic about communityclinic recruitment.

3. Iterative Feasibility and Usability Testing

FAMS functionality was developed and tested for usabilityalongside REACH (Rapid Education/Encouragement AndCommunications for Health), a text messaging interventionsupporting diabetes self-care [47]. REACH (a) sends tailoredmessages addressing users’ barriers to diabetes medicationadherence, (b) sends nontailoredmessages addressing adher-ence to diet, exercise, and self-monitoring of blood glucose,(c) sends daily adherence assessment messages with weeklyfeedback and support, (d) sends messages with HIPAA-compliant access to A1c test results, and (e) provides accessto a helpline tethered to a clinical pharmacist. In addition tothe REACH elements, FAMS users received phone coaching

and had the option to invite a support person (SP) to receivemessages. FAMS also replaced REACH’s general diet andexercise messages with messages tailored to users’ coachinggoal.

Text messages were tailored and sent by MEMOTEXT�,an algorithmic communications and data management plat-form supporting personalized user outputs and inputs viatext messaging. Interventions using this platform have >90%retention rate [48]. We worked with MEMOTEXT to designFAMS functionality and conduct three rounds of internalbeta testing during which research team members served aspseudousers to identify technical bugs and improve the userexperience before usability and feasibility testing. Costs owedto MEMOTEXT included the initial development, a monthlymaintenance fee of $200, and ∼$0.07 per message sent (per-message costs decrease when more messages are sent).

3.1. Study Design. We conducted three iterative rounds oftesting, eachwith a new sample of users. Each round includeda single phone coaching session with the PP followed by twoweeks of text message support for the PP and (if enrolled) theSP. PPs were adults with T2D recruited from three FQHCsin Nashville, TN, via flyers, interest cards, and referrals fromclinic staff. Eligible PPs were receiving care at one of theclinic sites and taking diabetes medications (not caregiveradministered). Both PPs and SPs had to be adults (≥18years old), speak and read English, have a cell phone withtext messaging, and provide a social security number (toprocess compensation). PPs and SPs were excluded if theyhad an existing diagnosis of dementia, were unable to orallycommunicate, or had an auditory limitation (for interviews).SPs verbally confirmed they could receive text messages.During enrollment, PPs were screened to confirm they couldreceive/read/send a message [47]. We aimed to enroll 𝑛 = 6PPs per round to satisfy the recommendation of three smallstudies with five participants in each round to identify allusability problems [49].

3.1.1. Procedures. Study procedures were approved by theVanderbilt University IRB. Trained research assistants (RAs)conducted eligibility screening and administered informedconsent and survey measures in a private room at the PP’sclinic. After enrollment, coaches (authors LSM and KJH)called the PP to complete phone coaching. Coaches thencalled and invited the SP to participate, asking if they wouldlike to “receive text messages related to being a continuedsupport” for the PP and complete an interview telling theresearch team what they thought about FAMS and how toimprove it. Interested SPs gave their consent via phone andprovided a preferred time of day to receive text messages.

Data fromPPs’ enrollment survey, coaching, and enrolledSPs were entered into REDCap� and transferred auto-matically to MEMOTEXT via an application-programminginterface for the tailoring and delivery of textmessages. Com-pensation was $54 for PPs and $40 for SPs. The study did notpay for or supply cell phones or plans. Authors LSM and KJHmet between rounds to discuss successes and shortcomings ofthe coaching protocols and SP enrollment process. Followingeach round of testing, we compiled PP- and SP-user feedback


to improve FAMS and research processes. We received IRBapproval before recruiting participants for the next round oftesting.

3.1.2. Measures. During the enrollment survey, PPs self-re-ported demographic and diabetes information and com-pleted validated survey instruments to characterize thesample (health literacy [50], family involvement [16], andadherence to self-care behaviors [51, 52]). We also asked PPsif they were comfortable using their cell phone and textmessaging. Glycemic control (A1c) was assessed via lab testwith a blood sample drawn at enrollment.

Immediately after each coaching session, coaches record-ed the PP’s goal and type of family action—helpful orharmful—identified and addressed with psychoeducationand assessed the success of each phase of the coaching proto-col using a “Coaching Assessment” we developed. Technicaldata collected by MEMOTEXT captured user engagement:we assessed PPs engagement as the number of text responsesreceived out of the number of two-way texts sent. The FAMSuser experience was assessed via semistructured interviewswith the PP and, separately, the SP. Interviewswere conductedby phone and included open-ended questions and closed-and open-ended question pairs (e.g., yes/no followed by“Why/Why not?” and 10-point scales followed by “Can youtell me why you chose that number?”). We asked participantsto rate different aspects of the intervention (e.g., how easythe text messages were to understand, how helpful they were,and how much each FAMS component motivated them toreach their goal) on a scale from 1 (least favorable) to 10 (mostfavorable).

3.1.3. Analysis. Interviews with PPs and SPs were audiore-corded with transcribed verbatim by an independent profes-sional transcriber. Interview questions and responses wereentered into REDCap. We exported survey and quantitativeinterview data from REDCap and conducted all descriptiveanalyses using Stata v. 13.1. Qualitative interview data wereexported to Excel. Author LSM categorized and summarizedfeedback by FAMS intervention component.

3.2. Findings and Changes between Testing Rounds. We en-rolled 19 PPs (mean per round 𝑛 = 6.3). PP characteristicsare shown in Table 2. Nine PPs invited a SP and seven SPsenrolled, including four boyfriends/girlfriends/fiances, onespouse, one ex-spouse, and one son. There were no dif-ferences (significant or meaningful nonsignificant) betweenparticipants who chose to invite a SP and those who did notin any variable in Table 2. Of the 19 enrolled PPs, 17 (89%)completed an exit interview. Of the seven enrolled SPs, six(86%) completed an exit interview. Findings are describedbelow with corresponding participant quotes in text andadditional quotes in Table 3. Changes made between roundsand prior to a longer trial are described below and in Table 4.

3.2.1. Phone Coaching. Table 5 shows Coaching Assessmentdata. Most participants focused on a helpful family actionduring skill building (examples in Table 5). All but one of thePPs who completed skill building were very confident (≥7 on

Table 2: Participant characteristics.

M± SD or 𝑛 (%) Total𝑁 = 19

Support PersonInvited

Yes𝑛 = 9

No𝑛 = 10

DemographicsAge, years 51.7± 10.2 52.0± 9.4 51.3± 11.6Gender, female 10 (53) 4 (44) 6 (60)RaceaCaucasian/white 7 (39) 4 (44) 3 (33)African American/black 8 (44) 4 (44) 4 (44)Othersb 3 (17) 1 (11) 2 (22)

Education, years 12.8± 2.5 13.4± 1.9 12.2± 2.9Annual household income,US$a

<10,000 8 (44) 3 (33) 5 (50)10,000–34,999 7 (39) 3 (33) 4 (40)≥35,000 3 (17) 3 (33) 0 (0)

Limited health literacy(BHLS) 2 (11) 1 (11) 1 (10)

Family characteristicsMarried/partnered 5 (26) 3 (33) 2 (20)Helpful actions (DFBC-II) 1.6± 0.7 1.5± 0.8 1.6± 0.7Harmful actions (DFBC-II) 1.4± 0.7 1.2± 0.7 1.6± 0.8Diabetes characteristicsDiabetes duration, years 6.9± 5.9 5.1± 5.1 8.3± 6.4Insulin status, takinginsulin 8 (42) 4 (44) 4 (40)

Previous cell phone useComfortable using cellphone 18 (95) 9 (100) 9 (90)

Used text messages 17 (90) 8 (89) 9 (90)Self-care adherenceMedication adherence(ARMS-D) 25.8± 2.8 25.9± 3.9 25.7± 1.4

General diet (SDSCA) 3.6± 2.5 3.7± 2.8 3.5± 2.4Specific diet (SDSCA) 3.5± 1.7 3.3± 1.6 3.6± 2.0Exercise (SDSCA) 2.0± 1.5 2.2± 1.6 2.1± 1.5SMBG (SDSCA) 3.9± 3.1 3.0± 3.2 4.7± 2.9Glycemic control (A1c, %) 7.4± 1.6 7.4± 1.6 7.5± 1.7aOne participant refused/did not know.bTwo Hispanic people and one Native American.Note. Mann–Whitney 𝑈 and Fisher’s exact tests identified no associationbetween any variable and inviting a support person. A1c, hemoglobinA1c; BHLS, Brief Health Literacy Screen (limited if score ≤ 9); ARMS-D, Adherence to Refills and Medications Scale for Diabetes medicationtaking subscale (possible range 7–28; higher scores indicatemore adherence);DFBC-II, Diabetes Family Behavior Checklist-II (possible range 1 = never to5 = once a day); SDSCA, Summary of Diabetes Self-Care Activities (possiblerange 0–7; it indicates number of days adherent per week); and SMBG, self-monitoring of blood glucose.

a 10-point scale) they could apply the skill with an identifiedfamily member and, separately, doing so would result in adesired change (Table 5). In interviews, PPs said coachinghelped them set a goal that was important to them (8.8 ± 1.4,range 6–10, example goals in Table 5). Several explained thecoach helped them learn how to set a realistic and attainablegoal:


Table 3: Participant feedback on FAMS.

Phone coachingI would have never set a goal [without the coaching]. I would just, you know - just took it day-by-day, not really set a goal. (PP, 57-year-oldAA female)Well, it gave me incentive to try to cut down - my goal was to cut down on my colas. And you gave me incentive to do that, but you also leftme some leeway, you know, where I could still maybe have one a day, you know. But you gave me some wiggle room instead of just saying,“Stop it.” You know, it was almost like a detox, a gradual detox from cola. (PP, 60-year-old Caucasian male)It made me be more honest about my health issues to my fiance. (PP, 46-year-old Caucasian female)Well, it made me want to tell [my family] more about [my diabetes]. And then it makes them want to be more interested, you know, in whatI have to say. If they come to realize how important, you know, diabetes is. (PP, 55-year-old AA female)

Patient participant text message supportI liked reading them. They were short, so it was not like I spent a lot of time reading them. . .It felt like I had a little bit of support thatnormally you just don’t have. You had backup that you just normally - like I say, I have nobody, and it felt like I’d come to have somebodythere for a while. (PP, 36-year-old Hispanic female)Well, when I get a text from the program - then people are like, “Who’s that?” I’m like, “Oh, it’s the study I’m doing for my diabetes.” Andthey’re like, “Oh. Well, what does it do?” You know, so, it opens a conversation for some people, you know, that I wouldn’t have told probably.(PP, 50-year-old male, race unknown)And it wasn’t real intrusive. Just a gentle reminder, “Hey,” you know, a text popped up. And just addressed some different things I mightpossibly be going through. And, you know, it was perfect. (PP 58-year-old Caucasian male)

Support person text messagesI think it was fun. I mean, I kind of looked forward to getting to see what it was going to be the next time, you know? [I would think], “Well,it’s going to come through here in a minute.” You know, because it gave me an idea of what to - kind of get it in my head, you know, what Ineed to ask him for that evening, you know, just to throw it out there. I usually try to call him on my way home, and we discuss things. Andthat just kind of gives me an idea of something to actually throw out there, you know, and get his input on it. And he was always very openand honest and told me. . . I enjoyed the text messages. (SP, ex-wife)Interviewer: Did you discuss any of the text messages with [the PP]?Every last one of them. (SP, husband)It made us talk when I got a text and she got a text. . .She gets her text. I get my text, we sit and discuss them when we both get them. BecauseI get mine early in the morning, and she gets hers at night. It reminds [family members] of what they’re not doing, or it brings it up so me andher can talk about it. (SP, boyfriend)

Overall opinions and experiencesIt reminds her that somebody actually cares about what she’s doing, what’s going on with her and her health. (SP; fiance)It was a really good goal for me. And to do it, it’s been wonderful. [My husband] and I have been doing stuff more together. And I’ve beengetting more exercise because of it. And, you know, it’s had the benefit of we’re more tired by the end of the day. (PP, 61-year-old Caucasianfemale)

AA, African American; PP, patient participant; and SP, support person.

Oh, I thought [coaching] was helpful because it pin-pointed what I could do and then helped me keep itdoable or obtainable. It helped me focus on what I wantto do, mademe vocalize what I want to do, and then thecoach part helped me make it - choose something thatwas obtainable. Also with the help, it was explained tome that it didn’t need to be 15 solidminutes [of exercise].I could do 5 minutes 3 times a day. And so, that reallyhelped. So, it made me take a break at work, go walk inthemall for my 15 today. And tomorrow, take a break atwork, or I could walk an entire floor, which would takeabout 5 minutes. You know, and then I’d do that 3 timesand I’d have my 15 minutes. So, I could come home andwatch a show and stand up and move around all thetime it was on. Hey, it was movement and it’s a start.(PP, 50-year-old male, race unknown)

Two-thirds (67%) said coaching improved communicationabout diabetes with family members:

I talk to my husband, and I explain about the goal, andthen he helped me more. Like, now he probably is goingto understand when I tell him, if we go out, what kindof food to order instead [of our regular order], becauseI already talked to him about what I talked about [incoaching]. (PP, 36-year-old Hispanic female)

Participants said they would want coaching once (33%) ortwice (33%) per month in a longer intervention, and 65% saideach session should last 15–30 minutes.

However, more than one in three participants (37%) didnot “buy into” the coaching session (Table 5), and these sameparticipants rated the coaching as less helpful in interviews.The most common reason (𝑛 = 4; 21%) was that theparticipant did not accept a connection between his/hergoal and the actions of close loved ones. Coaches reportedthey had an insufficient amount of time to discuss familyactions before asking PPs to think of how they would wantsuch actions to change. Two PPs did not “buy into” thecoaching session because the skill in the coaching protocol


Table4:FA

MSinterventio

nchanges.

Com

ponent

Usability&feasibilitytesting

Rand

omized

controlledtrial

Patie

ntparticipants

Phon

ecoaching

(i)Skill

build

ingexercise

andtype

offamily

actio

nassig

nedby

roun

dof

usabilitytesting

(a)A

ssertiv

ecom

mun

icationto

askfora

needed

helpfulaction

(Rou

nd1)

(b)A

ssertiv

ecom

mun

icationto

stop/redirectan

unwanted

harm

fulaction(Rou

nd2)

(c)P

roblem

solvingto

identifyho

wfamily

mem

berscanhelp

overcomea

barriertogoalcompletion(Rou

nd3)

(i)Session1focuses

oncollabo

rativ

egoalsettin

gandbriefskill

build

ingexercise

(Fam

ilyBe

havior

Observatio

n)with

homew

orkto

observefam

ilyactio

ns(ii)C

oach

andparticipantcollabo

rativ

elychosew

hether

tofocuso

na

helpfulorh

armfulfam

ilyactio

nforskillbu

ilding

(iii)Coach

choo

sesskillbu

ildingexercise

basedon

participantn

eeds

durin

gsessions

2–5

(a)A

ssertiv

ecom

mun

ication

(b)P

roblem

solving

(c)C

ognitiv

ebehavioralcop

ing

Patie

ntparticipants

Text

messages

Content

(i)Goalassessm

entm

essage

andfeedback

assumea

daily

goal

Your

goalwa

stowa

lk15

minuteseach

day .How

manyd

aysd

idyou

meetthisgoallastw

eek(Sun

-Sat)?Replyw

iththen

umbero

fdays,0–7.

(a)F

eedb

ackbasedon

numbero

fdaysg

oalw

ascompleted

outo

f7days;successwas

identifi

edas

meetin

ggoal7days

(i)Goalassessm

entm

essage

andfeedback

accommod

atea

goalfor≥

4days

perw

eek

Your

goalwa

stowa

lk15

minutes4days.H

owmanyd

aysd

idyoumeet

thisgoallastwe

ek(Sun

-Sat)?Replyw

iththen

umbero

fdays,0–7.

(a)F

eedb

ackcong

ratulated

respon

dent

form

eetin

gtheg

oal,4–

6days,and

challenged

him/her

tostr

ivefor

7days

next

week

Functio

nality

(i)Participantshave

2ho

urstorespon

dto

goalassessmenttext

(a)F

eedb

ackistailo

redto

participants’

respon

sefro

mthew

eekin

questio

nandthep

riorw

eek

(b)N

onrespon

seaffectsfeedback

forthe

weekin

questio

nandthe

next

week

(i)Participantshave

4ho

urstorespon

dto

goalassessmenttext

(a)F

eedb

ackistailo

redto

participants’

respon

sefro

mthew

eekin

questio

nandifthep

artic

ipantrespo

nded

thep

riorw

eek,whether

goalattainmentimproved/declin

ed/w

asconsistent

(b)N

onrespon

seon

lyaffectsfeedback

forthe

weekin

questio

n

Supp

ortp

ersons

Invitatio

n&enrollm

ent(i)

Participantg

iven

theo

ptionto

invitea“

family

mem

ber”up

onenrollm

ent(by

enrolling

research

assistant),on

lyon

ce

(i)Participantg

iven

optio

nto

invitea“

supp

ortp

erson”

(SP)

atendof

each

coaching

session(bycoach),m

ultip

letim

es(a)E

mph

asized

toparticipantthatallSP

involvem

ento

ccursv

iaph

one

(ii)T

oldSP

alltextm

essagesa

resent

toallSPs

inthes

tudy

andare

notsele

cted

basedon

anything

theP

Prepo

rted

abou

tthe

SPor

their

relatio

nship

Supp

ortp

ersons

Text

messages

Content

(i)Sometextsused

lang

uage

like“

nagging”

and“diabetesc

op”

Doyouexam

ineB

eth’s

platea

ndnagthem

durin

gmeals?

Thinkofmore

helpfulw

aystoprom

oteb

ettereatingb

esidesbeinga

diabetescop.

Doyouarguew

ithBeth

abouth

erdiabetes?T

akea

stepback.N

exttim

easkwh

atyoucando

tohelpBeth

makeb

etterh

ealth

choices.

Your

roleas

alovedoneistosupportB

ethin

meetin

gherhealth

goalby

asking

howyoucanbe

helpful.Remem

bern

ottobe

thed

iabetesp

olice

!

(i)Av

oidedlang

uage

thatcanbe

perceivedas

accusatory

oroff

ensiv

eInste

adofasking

Beth

what

sheisd

oing

totake

careofherd

iabetes,ask

what

youcando

tohelp.

Even

ifyoudisagree

with

Beth

abouth

erhealth,beg

entle

andfin

da

waytohelpherreach

herg

oals.

SupportB

ethby

brainstorm

ingtogetherw

aysy

oucanhelpherb

esuccessfu

linmeetin

gherhealth

goals.

Functio

nality

(i)Dailytexts

(i)Th

reetextsperw

eek


Table 5: Coaching Assessment data evaluating FAMS protocol success.

Goal settingWas the patient able to set a SMART goal?47%: yes, independently32%: yes, with help from the coach21%: no, needed the coach to set a goal for them

What type of goal was set?53%: diet (e.g., decrease to one 12-ounce soda, eat 3 servings of vegetables)47%: exercise (e.g., walk 15min or until feet hurt, lift canned goods as weights for 10min)

Discussing helpful & harmful family actionsCould the patient identify helpful family actions he/she had experienced?74%: yes, independently16%: yes, with help from the coach10%: no

Could the patient identify harmful family actions he/she had experienced?42%: yes, independently16%: yes, with help from the coach42%: no

Skill building84% engaged in the skill building exercise (percentages below reflect the 16 participants who completed skill building)

Desired change used for skill building:86%: wanted helpful action (e.g., choose healthy places to eat out, cook meals with me, exercise with me, do accountabilitycalls or texts)13%: unwanted harmful action (e.g., stop bringing unhealthy food to my house, stop bringing food over after dinner time)

68% were able to role play or teach back the skills learnedWas there any portion of the coaching protocol the patient did not “buy into”?

21% (𝑛 = 4) did not accept a connection between health goal and loved ones’ actions5% (𝑛 = 1) tried assertive communication with his wife with no results—the assertive communication skill was not a good fit5% (𝑛 = 1) has good support and lacks personal motivation5% (𝑛 = 1) didn’t accept idea that family could be harmful to diabetes self-care

Verbal contract79% made a verbal contract to implement the skill with an identified family memberParticipants’ confidence he/she can complete the verbal contract on scale 1–10 (How confident are you that you can use assertivecommunication to ask your sister to walk with you?)1 – 6% (not at all confident) 8 – 18% 10 – 53% (totally confident)7 – 12% 9 – 12%Participants’ confidence in success on scale 1–10 (How confident are you that doing so will result in your sister walking with you?)1 – 11% (not at all confident) 8 – 17% 10 – 50% (totally confident)7 – 11% 9 – 11%

was not a good fit for the identified family actions or his/herfamily relationships (Table 5). For example, one participanthad unsuccessfully tried assertive communication with hiswife about the type of food she cooked and felt doing soagain would not be productive (in his interview, this PPexplained, “Because my family members, they ain’t goingto do that for me.” 68-year-old AA male). Finally, coachesreported spending a disproportionate amount of time on goalsetting and rushing through the family-focused portions ofthe coaching protocol.

3.2.2. Patient Participant Text Message Support. PPs rated thehelpfulness of the goal supportmessages 8.4±2.7 (range 1–10)

and 80% said these messages motivated them to meet theirgoal. PPs liked that messages were short and easy to readand provided encouragement and support. One PP savedmessages she wanted to memorize or apply to review in herfree time. Three described using receipt of message as a cueto action to meet their goal and three others reported sharingmessages with loved ones:

You know, when I got the texts, I took a few minutesand did some arm rows and leg lifts and stretched myleg out, bring it to my stomach as much as I could. (PP,46-year-old Caucasian female)


I mean, a lot of stuff I can explain to my wife that -she might understand it but seeing it for herself [in themessages], she understand it different. (PP, 37-year-oldAA male)

Two PPs and one SP wanted messages to be more interactive.One PP said texts that were not interactive make “people feellike they’re being preached at.” Often, responses to one-waymessages were received from PPs (e.g., “That’s right!” and “Iwill, thank you”) and SPs (e.g., “We can work on it together”).Others appreciated not having to respond and said they wereused to receiving text messages without responding.

PPs responded to 65% of the goal assessment messagesand rated the helpfulness of these messages in keeping themon trackwith their goal 8.7±2.1 (range 4–10). PPs appreciatedgetting feedback on their progress, and said the encouragingtone motivated them to keep trying to meet their goal:

I would get a message: “Did you do it?” And then I said,“Oh, only 3 times.” And it said, “Ooh, You’re halfwayto your goal. What could - think about what wouldhelp you, you know, maybe get closer?” So, I didn’t feelashamed or I like “Ugh, I need to think of why didn’t Ido that?” You know? So, that part was really good aboutreaching my goal. It was very helpful - it reinforced thecoaching that we did. It helped me not - I wanted to dobetter the second week. So, it’s like, “Oh, dang it. I didn’tdo it the first week. I’ve got to do it this week.” (PP, 50-year-old male, race unknown)

We made three changes to improve the goal assessmentmessages during testing (Table 4). First, we created functionalflexibility to allow PPs to set a goal for ≥4 days. We alteredfeedback messages and functionality so a PP’s response of0–3 implied the participant had not met their goal, 4–6implied success, and 7 received especially congratulatoryfeedback. Second, we widened the window during whichMEMOTEXT could record a response. Third, we initiallydeveloped feedback to reference progress relative to theprior week (i.e., improvement, decline, or consistency) whichresulted in nonresponse affecting feedback for the week inquestion and the subsequent week.Therefore, we changed thefunctionality of the feedback text to reference progress onlywhen two weeks’ responses were provided and to default tomore basic feedback if not (e.g., “Keep trying to reach yourgoal each day. Use the tips you get in your texts. You can dobetter next week!”).

3.2.3. Support Persons. Of the 19 PPs, nine (47%) invited aSP to enroll in the study and seven SPs were enrolled (wecould not contact one; one declined). Six more PPs wantedto have a SP receive text messages but did not provide theSP’s contact information to study staff. Reasons included thefollowing: SP was ineligible (having no cell phone or beingnon-English-speaking), PP could not identify a SP, and SPexpressed disinterest to the PP. Only four PPs (21%) did notwant to invite a SP:

Well, I guess I do have family members. But I just feltlike this was my deal and I just wanted to own it myself

because that really drives me more. (PP, 50-year-oldmale, race unknown)My family members are really busy. I didn’t want toput more weight on their shoulders. (PP, 56-year-oldCaucasian female)

Enrolled SPs reported the messages motivated them totalk with the PP about diabetes (9.3 ± 1.2, range 7–10) andimproved their support of the PP (9.3 ± 1.2, range 7–10).When asked how many messages per week they would wantto receive in a six-month intervention, SPs said three timesper week (𝑛 = 3), daily (𝑛 = 2), or every other day (𝑛 = 1).Of the six SPs who completed an interview, four said FAMSincreased their knowledge about diabetes and all reporteddiscussing message content with PPs.

I think it was a great experience for me because I hadnever known that much about diabetes and I was ableto learn a great deal about it, you know, through thosetexts and then talking with [the PP]. I learned quite abit. . .and in ways that I can actually help someone else.You know, if they’re going through the same things. . .Ialso tried to, you know, watch what I eat as well. And,in turn, you know, we talk to each other about alwaystrying to eat healthy and try to make the right choicesand, you know, just try not to overdo things that wouldhurt her health-wise. (SP, boyfriend)I think it was fun. I mean, I kind of looked forward togetting to see what it was going to be the next time, youknow? [I would think], “Well, it’s going to come throughhere in aminute.” You know, because it gaveme an ideaof what to - kind of get it in my head, you know, whatI need to ask him for that evening, you know, just tothrow it out there. I usually try to call him on my wayhome, and we discuss things. And that just kind of givesme an idea of something to actually throw out there,you know, and get his input on it. And he was alwaysvery open and honest and told me. . . I enjoyed the textmessages. (SP, ex-wife)

In round one, a SP-PP dyad reported the SP’s textmessages led to interpersonal conflict. The SP received thistext message: “Do you argue with George about his diabetes?Take a step back. Next time ask what you can do to helpGeorge make better health choices” (name changed). The SPthought this message was sent to her because of somethingthe PP told us:

But there was one that kind of got me a littlebit. . .it was kind of like I was being aggressivewith him or something and that’d never been thecase. And I was like, “I don’t understand this onebecause I don’t do that to him - pressure him aboutanything.” But anyway, that was the only one thatwas kind of strange. . .That one really needs to beworked on because that was disturbing (laughs). Imean, it just made me feel like I’d done somethingthat maybe he had told you or something that Idid.


The PP was very upset about this:

Y’all sent her text messages asking her why shewant to argue withmewhen she’s not arguing withme. She has never been negative about this; she’sthe one who wanted me to do this [study]. You allmake her feel like I’m telling you bad shit abouther and I’m not. . .That thoroughly pissed me off,and I texted back some pretty hot shit to them. I’mnot sure if the computer understood it or not, butit made me feel better.

We reported this unexpected adverse event to the IRB andfixed it for subsequent rounds. With family interventionexperts, we reviewed and edited all SP messages to avoidinsinuating SPs were performing a harmful action or experi-encing conflictwith the PP (Table 5).We also added this to theinformed consent process for both the PP (i.e., conflict withfriends/family members as a risk) and the SP (i.e., explainedall of the SP text messages were sent randomly to all SPs inthe study and not based on any information from PPs).

4. Lessons to Improve FAMS

In addition to described changes made between rounds oftesting, we made several decisions and changes to FAMS foran upcoming evaluative trial based on these results (Table 4).Coaching sessions will occur monthly and are designed tolast 20–30 minutes. The first session will allocate half of thesession to assessing current diet or exercise activities andcollaboratively setting a SMART goal, and the other half willbe allocated to discussing the role of family actions/supportinmeeting and sustaining health goals.This session endswitha brief “Family Behavior Observation” skill building exercise,requesting that the PP observe the role of close loved ones ashe/she tries to meet the goal for the first month.We hope thischange will allow sufficient time for goal setting and increasethe PP’s acceptance of family members’ role prior to askingPPs to identify desired family changes in subsequent sessions.We also created flexibility in the coaching protocol so coachescan tailor the skill to the PP in vivo. Each session involves thesame elements (Table 1), but coaches choose the skill mostapplicable to PPs during sessions 2–5 (Table 4). We addedthe skill building exercise “Cognitive Behavioral Coping” forPPs who report not having supportive family members orexperiencing persistent harmful family actions. We created a“wrap-up” session 6.We alsomade changes to our process forenrolling and inviting SPs to receive text messages (Table 4).With these changes, we hope PPs will understand the SP doesnot need to be a biological/legal family member and will haveopportunity to ask the SP’s permission to give study staffhis/her contact information.

5. Conclusions

We designed FAMS to improve T2D adults’ (a) adherenceto diet and exercise recommendations and (b) self-efficacyand skills to engage loved ones in self-care goals in ways thatfacilitate behavior change.Through an iterative user-centered

design process followed by rigorous andmultimethod usabil-ity/feasibility testing [36], we designed an intervention thatappeals to end users, is easy to use, and is applicable to a vari-ety of patient and family situations. PPs said FAMS improvedtheir diet/physical activity, communicationwith familymem-bers, and confidence in soliciting helpful actions/support. SPsreported FAMS increased the amount they communicatedwith PPs about diabetes and made it possible for them to bemore helpful. Comments from users indicate families findit difficult to know how to communicate about diabetes orsupport diabetes self-care efforts and appreciated concretesuggestions provided during coaching or in text messages.Over half of the PPs identified a harmful family action, butfewer (13%) focused on the harmful action in skill building.Patients who discussed harmful actions often preferred eitherasking for a helpful action to replace or offset the harmfulaction or asking for a helpful action from another familymember instead of dealing with the family member perform-ing the harmful action.These are productive ways to manageharmful actions, as helpful actions have been found to protectagainst the detrimental effects of harmful actions on patients’A1c [12].

We discovered limitations of FAMS. For instance, only37% of PPs successfully invited a SP and a few PPs did notwant to discuss family issues at all.Wehope to have developedappropriate and effective solutions to these problems, but thesuccess of these changes remains unknown. We do know,however, that our iterative process dramatically reducedtechnical bugs and problems with content, functionality, andstudy processes, thereby maximizing our ability to evaluateFAMS. FAMS targets patients’ adherence to diet and exercise,but other self-care behaviors (e.g., medication adherence,self-monitoring of blood glucose) are impacted by familyactions [2, 12] and affect diabetes control [2, 12].

5.1. Future Research. Future work should assess patients’expectations of others’ involvement in diabetes self-care.Such expectations may moderate the relationship betweenfamily involvement and patients’ outcomes. For instance, inother disease contexts, helpful [53] and harmful [54] familyinvolvement have been shown to affect dietary adherence[54], self-efficacy [53], and depressive symptoms [53] differ-ently based on perceptions of the role of family in adults’ self-management. Thus, future research should examine whetherindividuals with T2D who view their illness as an individualconcern benefit from family involvement or interventionslike FAMS. Additionally, we found SPs may react negativelyto the idea that their actions might be less than helpful orthey might nag or argue with the patient. Future interventionefforts may benefit from normalizing nagging and arguingand inadvertently making self-care more challenging. Thismay make it less emotionally fraught for families to identifyand replace these actions with more helpful ones.

Although mHealth interventions have improved adher-ence and glycemic control among adults with T2D [55, 56],engaging family members in these interventions remainsunderstudied. To our knowledge, the only other mHealthintervention involving a patient-selected family member/support person for adults with T2D ismHealth +CarePartner


[57, 58]. Both mHealth + CarePartner and FAMS weredesigned to be inclusive of patients who live alone, recruitfrom community clinics, and provide tailored mobile com-munications to an adult family member/friend. However,there are key differences between Piette et al.’s [59] approachto engaging a support person and our own. CarePartnersreceive weekly information about the patients’ health statusalong with tailored advice about how to help. In contrast,FAMS does not provide information about patients’ progressto SPs but rather aims to empower the patient to identify andcommunicate support needs. Lessons learned from FAMSand mHealth + CarePartner will improve future efforts toinvolve meaningful others in adults’ T2D self-management.

5.2. Strengths and Limitations of Our Approach. We enrolleda small convenience sample with no control group andusers experienced the intervention for a brief time, limitingour understanding to participants’ anecdotal accounts andshort-term engagement data. “Down time” was required fortroubleshooting technical bugs, making fixes/improvements,and adding new functionality. As a result, we had to waitbetween each development phase and paused recruitmentbetween rounds of usability/feasibility testing. These nec-essary and inevitable occurrences—rarely accounted for instudy planning or funding schedules—nearly doubled thelength of FAMS development and testing. However, sim-ilar “down time” during an evaluative trial would havebeen much more problematic and costly. Because identi-fying and fixing bugs are inevitable with mHealth inter-ventions, we stress the importance of starting rigorousmultiround testing early in the design/development processrather than waiting until the intervention is developed andfunctioning properly to allow users to experience it. Moretime and energy in these phases likely pay off with fasterevaluative trials with more engagement, less attrition, andfewer study limitations associated with functional prob-lems [36]. Such work is especially critical when designingmHealth interventions for underserved and disadvantagedpatients.

Disclosure

Thecontent is solely the responsibility of the authors and doesnot necessarily represent the official views of the NIH.

Competing Interests

The authors declare that they have no competing interests.

Authors’ Contributions

Lindsay Satterwhite Mayberry designed the study, devel-oped research protocols, directed analyses, and wrote themanuscript. Lindsay Satterwhite Mayberry and KryseanaJ. Harper codeveloped FAMS coaching protocols and textmessage content. Kryseana J. Harper managed the study,conductedmost coaching, assisted with analyses, and draftedsubsections and edited the manuscript. Cynthia A. Berg and

Chandra Y. Osborn guided study design and wrote portionsof and edited the manuscript.

Acknowledgments

The authors would like to thank the Federally QualifiedHealth Centers—University Community Health Services,Faith Family Medical Center, and the Clinic at MercuryCourts—and Drs. Shari Barkin, Jeffrey Gonzalez, KerriCavanaugh, Tom Elasy, and Russell Rothman for assist-ing with text content and informed consent revisions inresponse to the adverse event. This study was funded bythe National Institute of Diabetes and Digestive and Kid-ney Diseases (NIDDK) through Dr. Mayberry’s Pilot andFeasibility Award from the Vanderbilt Center for DiabetesTranslational Research [P30DK092986, PI: Elasy] and Dr.Mayberry’s Career Development Award [K01DK106306]. Dr.Mayberry was also supported by the Agency for HealthcareResearch & Quality [K12HS022990, PI: Penson]. The Van-derbilt Institute for Clinical Translational Research [NationalCenter for Research Resources, Grant UL1 RR024975-01, nowat the National Center for Advancing Translational Sciences,Grant 2 UL1 TR000445-06] supported this project via theCommunity Engagement Studio and REDCap. Dr. Berg wasfunded by the NIDDK (R01DK092939, DP3DK103999) whilepreparing this article. Dr. Osborn was funded by the NIDDK(K01DK087894, R01DK100694) while preparing this article.

References

[1] L. Fisher, C. A. Chesla, R. J. Bartz et al., “The family and type 2diabetes: a framework for intervention,” Diabetes Educator, vol.24, no. 5, pp. 599–607, 1998.

[2] L. S. Mayberry and C. Y. Osborn, “Family support, medicationadherence, and glycemic control among adults with type 2diabetes,” Diabetes Care, vol. 35, no. 6, pp. 1239–1245, 2012.

[3] C. Crespo, S. Santos, M. C. Canavarro, M. Kielpikowski, J.Pryor, and T. Feres-Carneiro, “Family routines and rituals in thecontext of chronic conditions: a review,” International Journal ofPsychology, vol. 48, no. 5, pp. 729–746, 2013.

[4] A.-M. Rosland and J. D. Piette, “Emerging models for mobi-lizing family support for chronic disease management: a struc-tured review,” Chronic Illness, vol. 6, no. 1, pp. 7–21, 2010.

[5] L. Fisher, C.A.Chesla,M.M. Skaff et al., “The family and diseasemanagement in hispanic and European-American patients withtype 2 diabetes,”Diabetes Care, vol. 23, no. 3, pp. 267–272, 2000.

[6] M. R. DiMatteo, “Social support and patient adherence tomedical treatment: a meta-analysis,” Health Psychology, vol. 23,no. 2, pp. 207–218, 2004.

[7] A.-M. Rosland, M. Heisler, and J. D. Piette, “The impactof family behaviors and communication patterns on chronicillness outcomes: a systematic review,” Journal of BehavioralMedicine, vol. 35, no. 2, pp. 221–239, 2012.

[8] B. N.Uchino, Social Support and Physical Health: Understandingthe Health Consequences of Relationships, Yale University Press,2004.

[9] D. Umberson, “Family status and health behaviors: socialcontrol as a dimension of social integration,” Journal of Healthand Social Behavior, vol. 28, no. 3, pp. 306–319, 1987.


[10] G. G. Whitchurch and L. L. Constantine, “Systems theory,”in Sourcebook of Family Theories and Methods, pp. 325–355,Springer, Berlin, Germany, 2009.

[11] P. Ryan and K. J. Sawin, “The individual and family self-management theory: background and perspectives on context,process, and outcomes,”Nursing Outlook, vol. 57, no. 4, pp. 217–225.e6, 2009.

[12] L. S.Mayberry andC. Y.Osborn, “Family involvement is helpfuland harmful to patients’ self-care and glycemic control,” PatientEducation and Counseling, vol. 97, no. 3, pp. 418–425, 2014.

[13] A.-M. Rosland, M. Heisler, H.-J. Choi, M. J. Silveira, andJ. D. Piette, “Family influences on self-management amongfunctionally independent adults with diabetes or heart failure:do family members hinder as much as they help?” ChronicIllness, vol. 6, no. 1, pp. 22–33, 2010.

[14] M. A. Lewis and K. S. Rook, “Social control in personalrelationships: impact on health behaviors and psychologicaldistress,” Health Psychology, vol. 18, no. 1, pp. 63–71, 1999.

[15] M. A. Lewis, K. S. Rook, and R. Schwarzer, “Social support,social control, and health among the elderly,” Health Psychol: ALifespan Perspective, pp. 191–211, 1994.

[16] R. E. Glashow and D. J. Toobert, “Social environment andregimen adherence among type II diabetic patients,” DiabetesCare, vol. 11, no. 5, pp. 377–386, 1988.

[17] M. A. P. Stephens, M. M. Franks, K. S. Rook, M. Iida, R. C.Hemphill, and J. K. Salem, “Spouses’ attempts to regulate day-to-day dietary adherence among patients with type 2 diabetes,”Health Psychology, vol. 32, no. 10, pp. 1029–1037, 2013.

[18] S. L. Henry, K. S. Rook, M. A. P. Stephens, and M. M. Franks,“Spousal undermining of older diabetic patients’ disease man-agement,” Journal of Health Psychology, vol. 18, no. 12, pp. 1550–1561, 2013.

[19] A. A. Baig, A. Benitez, M. T. Quinn, and D. L. Burnet, “Familyinterventions to improve diabetes outcomes for adults,” Annalsof the New York Academy of Sciences, vol. 1353, no. 1, pp. 89–112,2015.

[20] C. L. Arnold, A. Rademaker, S. C. Bailey et al., “Literacy barriersto colorectal cancer screening in community clinics,” Journal ofHealth Communication, vol. 17, no. 3, pp. 252–264, 2012.

[21] K. M. Neckerman, G. S. Lovasi, S. Davies et al., “Disparities inurban neighborhood conditions: evidence from GIS measuresand field observation in New York city,” Journal of Public HealthPolicy, vol. 30, no. 1, pp. S264–S285, 2009.

[22] Y.-H. Jeon, B. Essue, S. Jan, R. Wells, and J. A. Whitworth,“Economic hardship associated with managing chronic illness:a qualitative inquiry,” BMC Health Services Research, vol. 9,article 182, 2009.

[23] L. Jiang, J. Beals, N. R. Whitesell, Y. Roubideaux, and S. M.Manson, “Stress burden and diabetes in two American Indianreservation communities,”Diabetes Care, vol. 31, no. 3, pp. 427–429, 2008.

[24] L. Fisher, C. A. Chesla, J. T. Mullan, M. M. Skaff, and R. A.Kanter, “Contributors to depression in latino and European-American patients with type 2 diabetes,” Diabetes Care, vol. 24,no. 10, pp. 1751–1757, 2001.

[25] M. S. Spencer, E. C. Kieffer, B. R. Sinco et al., “Diabetes-specificemotional distress among African Americans and Hispanicswith type 2 diabetes,” Journal of Health Care for the Poor andUnderserved, vol. 17, no. 2, pp. 88–105, 2006.

[26] L. S. Mayberry, L. E. Egede, J. A. Wagner, and C. Y. Osborn,“Stress, depression and medication nonadherence in diabetes:

test of the exacerbating and buffering effects of family support,”Journal of Behavioral Medicine, vol. 38, no. 2, pp. 363–371, 2015.

[27] L. S. Mayberry, K. J. Harper, and C. Y. Osborn, “Familybehaviors and type 2 diabetes: what to target and how to addressin interventions for adults with low socioeconomic status,”Chronic Illness, vol. 12, no. 3, pp. 199–215, 2016.

[28] A. Nicolucci, K. Kovacs Burns, R. I. G. Holt et al., “Diabetesattitudes, wishes and needs second study (DAWN2�): cross-national benchmarking of diabetes-related psychosocial out-comes for people with diabetes,” Diabetic Medicine, vol. 30, no.7, pp. 767–777, 2013.

[29] N. Wayne and P. Ritvo, “Smartphone-enabled health coachintervention for people with diabetes from a modest socioe-conomic strata community: single-arm longitudinal feasibilitystudy,” Journal of Medical Internet Research, vol. 16, no. 6, articlee149, 2014.

[30] K.K. Burns, A.Nicolucci, R. I. G.Holt et al., “DiabetesAttitudes,Wishes and Needs second study (DAWN2�): cross-nationalbenchmarking indicators for familymembers livingwith peoplewith diabetes,” Diabetic Medicine, vol. 30, no. 7, pp. 778–788,2013.

[31] M. Hartmann, E. Bazner, B. Wild, I. Eisler, and W. Herzog,“Effects of interventions involving the family in the treatment ofadult patients with chronic physical diseases: a meta-analysis,”Psychotherapy and Psychosomatics, vol. 79, no. 3, pp. 136–148,2010.

[32] T. A. Armour, S. L. Norris, L. Jack Jr., X. Zhang, and L.Fisher, “The effectiveness of family interventions in people withdiabetes mellitus: a systematic review,” Diabetic Medicine, vol.22, no. 10, pp. 1295–1305, 2005.

[33] L. Raine and K. Zickuhr, Americans’ Views on Mobile Etiquette,Pew Research Center, 2015.

[34] M. Duggan, Cell Phone Activities, Pew Research Center, Wash-ington, DC, USA, 2013.

[35] R. J. Chakkalakal, S. Kripalani, D. G. Schlundt, T. A. Elasy, andC. Y. Osborn, “Disparities in using technology to access healthinformation: race versus health literacy,” Diabetes Care, vol. 37,no. 3, pp. e53–e54, 2014.

[36] C. R. Lyles, U. Sarkar, and C. Y. Osborn, “Getting a technology-based diabetes intervention ready for prime time: a review ofusability testing studies,” Current Diabetes Reports, vol. 14, no.10, pp. 1–12, 2014.

[37] C. LeRouge andN.Wickramasinghe, “A review of user-centereddesign for diabetes-related consumer health informatics tech-nologies,” Journal of Diabetes Science and Technology, vol. 7, no.4, pp. 1039–1056, 2013.

[38] L. Fisher, “Family relationships and diabetes care during theadult years,” Diabetes Spectrum, vol. 19, no. 2, pp. 71–74, 2006.

[39] F. Hill-Briggs, K. P. Schumann, andO.Dike, “Five-stepmethod-ology for evaluation and adaptation of print patient healthinformation to meet the <5th grade readability criterion,”Medical Care, vol. 50, no. 4, pp. 294–301, 2012.

[40] R. Q.Wolever, M. Dreusicke, J. Fikkan et al., “Integrative healthcoaching for patients with type 2 diabetes: a randomized clinicaltrial,” Diabetes Educator, vol. 36, no. 4, pp. 629–639, 2010.

[41] M. J. Kreitzer, V. S. Sierpina, and K. Lawson, “Health coaching:innovative education and clinical programs emerging,” Explore:the Journal of Science and Healing, vol. 4, no. 2, pp. 154–155,2008.

[42] C. K. Miller and J. Bauman, “Goal setting: an integral compo-nent of effective diabetes care,”Current Diabetes Reports, vol. 14,no. 8, article 509, 8 pages, 2014.


[43] B. Shedon, Cognitive-Behavioural Therapy: Research and Prac-tice in Health and Social Care, Rutledge, New York, NY, USA,2011.

[44] J. S. Beck, Cognitive Behavior Therapy: Basics and Beyond,Guilford Press, New York, NY, USA, 2nd edition, 2011.

[45] D. Schillinger, J. Piette, K. Grumbach et al., “Closing the loop:physician communication with diabetic patients who have lowhealth literacy,” Archives of Internal Medicine, vol. 163, no. 1, pp.83–90, 2003.

[46] Y. A. Joosten, T. L. Israel, N. A. Williams et al., “Communityengagement studios: a structured approach to obtaining mean-ingful input from stakeholders to inform research,” AcademicMedicine, vol. 90, no. 12, pp. 1646–1650, 2015.

[47] L. A. Nelson, L. S. Mayberry, K. A. Wallston, S. Kripalani, E.M. Bergner, and C. Y. Osborn, “Development and usability ofREACH: a tailored theory-based text messaging interventionfor disadvantaged adults with type 2 diabetes,” JMIR HumanFactors, vol. 3, no. 2, article e23, 2016.

[48] M. V. Boland, D. S. Chang, T. Frazier, R. Plyler, J. L. Jefferys,and D. S. Friedman, “Automated telecommunication-basedreminders and adherencewith once-daily glaucomamedicationdosing: The Automated Dosing Reminder Study,” JAMA Oph-thalmology, vol. 132, no. 7, pp. 845–850, 2014.

[49] R. A. Virzi, “Refining the test phase of usability evaluation: howmany subjects is enough?” Human Factors, vol. 34, no. 4, pp.457–468, 1992.

[50] K. A.Wallston, C. Cawthon, C. D.McNaughton, R. L. Rothman,C. Y. Osborn, and S. Kripalani, “Psychometric properties ofthe brief health literacy screen in clinical practice,” Journal ofGeneral Internal Medicine, vol. 29, no. 1, pp. 119–126, 2014.

[51] L. S. Mayberry, J. S. Gonzalez, K. A. Wallston, S. Kripalani, andC. Y. Osborn, “The ARMS-D out performs the SDSCA, butboth are reliable, valid, and predict glycemic control,” DiabetesResearch and Clinical Practice, vol. 102, no. 2, pp. 96–104, 2013.

[52] D. J. Toobert, S. E. Hampson, and R. E. Glasgow, “The summaryof diabetes self-care activities measure: results from 7 studiesand a revised scale,” Diabetes Care, vol. 23, no. 7, pp. 943–950,2000.

[53] L. M. Martire, M. A. P. Stephens, and R. Schulz, “Independencecentrality as a moderator of the effects of spousal support onpatient well-being and physical functioning,”Health Psychology,vol. 30, no. 5, pp. 651–655, 2011.

[54] A. J. Seidel, M. M. Franks, M. A. P. Stephens, and K. S. Rook,“Spouse control and type 2 diabetes management: moderatingeffects of dyadic expectations for spouse involvement,” FamilyRelations, vol. 61, no. 4, pp. 698–709, 2012.

[55] S. Krishna and S. A. Boren, “Diabetes self-management care viacell phone: a systematic review,” Journal of Diabetes Science andTechnology, vol. 2, no. 3, pp. 509–517, 2008.

[56] X. Liang, Q. Wang, X. Yang et al., “Effect of mobile phoneintervention for diabetes on glycaemic control: ameta-analysis,”Diabetic Medicine, vol. 28, no. 4, pp. 455–463, 2011.

[57] J. D. Piette, D. Striplin, N. Marinec et al., “A mobile healthintervention supporting heart failure patients and their infor-mal caregivers: a randomized comparative effectiveness trial,”Journal of Medical Internet Research, vol. 17, no. 6, article e142,2015.

[58] J. D. Piette, N. Marinec, K. Janda et al., “Structured caregiverfeedback enhances engagement and impact of mobile healthsupport: a randomized trial in a lower-middle-income country,”Telemedicine Journal and e-health, vol. 22, no. 4, pp. 261–268,2016.

[59] J. D. Piette, A. Sen, and J. E. Aikens, “Randomized controlledtrial of mhealth telemonitoring with enhanced caregiver sup-port for diabetes self-management,” Journal of Clinical Trials,vol. 4, article 194, 2014.

Research ArticleAdaptation and Feasibility Study of a DigitalHealth Program to Prevent Diabetes among Low-IncomePatients: Results from a Partnership between a DigitalHealth Company and an Academic Research Team

Valy Fontil,1,2 Kelly McDermott,3 Lina Tieu,1,2 Christina Rios,1,2

Eliza Gibson,3 Cynthia Castro Sweet,3 Mike Payne,3 and Courtney R. Lyles1,2

1Division of General Internal Medicine, University of California, San Francisco (UCSF) at Zuckerberg San FranciscoGeneral Hospital and Trauma Center, P.O. Box 1364, San Francisco, CA 94143, USA2Center for Vulnerable Populations, University of California, San Francisco (UCSF) at Zuckerberg San FranciscoGeneral Hospital and Trauma Center, P.O. Box 1364, San Francisco, CA 94143, USA3Omada Health, 500 Sansome St., Suite 200, San Francisco, CA 94111, USA

Correspondence should be addressed to Courtney R. Lyles; [email protected]

Received 6 May 2016; Revised 3 August 2016; Accepted 19 September 2016

Academic Editor: Andrea Scaramuzza

Copyright © 2016 Valy Fontil et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Background. The feasibility of digital health programs to prevent and manage diabetes in low-income patients has not beenadequately explored. Methods. Researchers collaborated with a digital health company to adapt a diabetes prevention programfor low-income prediabetes patients at a large safety net clinic. We conducted focus groups to assess patient perspectives, revisedlessons for improved readability and cultural relevance to low-income and Hispanic patients, conducted a feasibility study of theadapted program in English and Spanish speaking cohorts, and implemented real-time adaptations to the program for commercialuse and for a larger trial of in multiple safety net clinics. Results. The majority of focus group participants were receptive to theprogram.Wemodified the curriculum to a 5th-grade reading level and adapted content based on patient feedback. In the feasibilitystudy, 54% of eligible contacted patients expressed interest in enrolling (𝑛 = 23). Although some participants’ computer accessand literacy made registration challenging, they were highly satisfied and engaged (80% logged in at least once/week). Conclusions.Underserved prediabetic patients displayed high engagement and satisfaction with a digital diabetes prevention program despitelower digital literacy skills. The collaboration between researchers and a digital health company enabled iterative improvements intechnology implementation to address challenges in low-income populations.

1. Introduction

Nearly half of Americans will develop a chronic disease suchas diabetes during their lifetime [1]. Optimal management ofchronic diseases can be very complex and requires activat-ing patients to proactively engage in self-management thatincludes behavioral changes and execution of complex med-ical treatment regimens needed to achieve optimal control ofthe disease. To this end, interventions that provide supportfor self-management have become a cornerstone for health

system innovations toward preventing and treating chronicdiseases such as diabetes [2–4]. The diabetes preventionprogram (DPP) was a landmark trial of an intensive lifestyleintervention that reduced risk for development of type 2diabetes by 58% after 3 years. Subsequent practice-basedinterventions have validated the effectiveness of DPP inweight reduction and preventing diabetes in real-world clin-ical settings [5, 6]. As a result, the Center for Disease Controlhas established the National Diabetes Prevention Program todisseminate DPP programs across the country [7].


http://dx.doi.org/10.1155/2016/8472391


Advancements in information technology (IT) haveexpanded the ability to engage patients in the healthcareprocess, motivate health behavior change, and offer thepotential to disseminate lifestyle self-management programslike DPP on a large scale [8]. Digital health tools (Internet-and/or mobile-phone-based) to enhance self-management ofdiabetes have proliferated rapidly [8, 9] with inconsistent butoften positive results in increasing healthy lifestyle practices(e.g., diet and exercise) and affecting clinical outcomes likeglycemic control [10, 11]. While there is great promise forthese digital health interventions, our study focuses on twomajor gaps that need to be addressed in order to see moreconsistent and widespread effectiveness. First, digital healthinterventions for self-management need to utilize contentand curricula that are based on validated evidence for behav-ioral change. Second, very few feasibility studies of digitalhealth interventions have focused on real-world clinicalsettings that care for underserved populations at highest riskof developing diabetes and poor health outcomes associatedwith the disease [10, 12]. This is a very important gap in theliterature because high risk populations with greater diseaseburden, such as older adults, low-income individuals, andethnic minorities [13], are often the same groups associatedwith lower computer literacy and encounter greater chal-lenges in accessing and using digital health technologies [14,15].

The Omada Health Program� (formerly known as Pre-vent) is an Internet- and mobile-phone-based educationalprogram modeled after the DPP lifestyle intervention, whichincludes small group support, personalized health coaching,a weekly curriculum, and digital tracking tools, including awireless scale delivered to each participant’s home. Partici-pants are placed into a private online social network wherethey can discuss their progress toward their goals and provideeach other with social support. Participants are encouragedto read and post weekly comments in this forum. Theprogram starts with a 16-week intensive curriculum focusingon weight loss and continues with a 36-week curriculumfocusing on weight maintenance. The online platform allowsparticipants to asynchronously complete weekly lessons,privately message, text message, and call a health coachfor individual counseling, track weight loss and physicalactivity using a wireless weight scale and pedometer, andmonitor their engagement and weight loss progress. In aprevious quasi-experimental pre- and postintervention study,the Omada Health Program was associated with significantreductions in body weight and A1C that were maintainedafter 2 years [16]. In 2014, Omada Health sought to adapt itsprogram for vulnerable populations at high risk of diabetes.Healthcare institutions that serve predominantly low-incomeand uninsured patients in the USA are often referred to as thehealthcare safety net [17, 18].

In this paper, we describe a collaboration between healthservices researchers at University of California, San Franciscoand Omada Health to adapt the program through a real-world, user-centered process and test its feasibility in predi-abetic patients at a large, urban, county-operated safety netclinic.

2. Methods

2.1. Objectives. The work began at Omada in 2014 with thegoal of adapting the program for more vulnerable patientpopulations. This included a literacy and content overhaul ofthe existing program, the first (beta) version of a Spanish-language version, and then a usability test of the modifiedcontent within a real-world clinic setting. To this end, Omadapartneredwith researchers at theUCSFCenter forVulnerablePopulations (CVP) at the Zuckerberg San Francisco GeneralHospital and Trauma Center (ZSFG) to collaborate on thisprocess.

More specifically, the partnership between Omada andUCSF had two primary objectives: (1) adapting the literacylevel and cultural relevance of the online program con-tent for low-income, underserved populations, using bothfocus groups (phase 1) and in-depth editing of the entireweekly curriculum (phase 2); (2) testing the feasibility andacceptability of the modified program in a small sampleof patients at a large safety net clinic (using observationsof in-person registration and follow-up phone interviews,phase 3), with the overall goal of using the results from thiswork to inform further improvement of the program (phase4).

2.2. Research Setting. From June to November 2015, we re-cruited patients receiving primary care at the Richard H.Fine People’s Clinic, an adult primary care clinic based inZSFG. This clinic serves 6,000 low-income patients in thecity and county of San Francisco, among whom about three-fourths have public insurance, three-fourths are nonwhite,40% prefer to speak a language other than English, and 15%are monolingual Spanish speakers. This same clinic was thestudy setting for all phases of this work.

2.3. Research Approach. We used a user-centered approachto design a prototype. User-centered design involves incor-porating the perspectives and experiences of end-users inplanning, designing, and finalizing a technology or toolwith the ultimate goal of improving usability, acceptability,and value to potential users [19]. User-centered design isincreasingly being used to inform the design of healthtechnologies, including those geared toward facilitatinglifestyle management or the self-management of chronicconditions, including diabetes [19–21]. We employed user-centered design in a 4-phase approach to inform and iteratethe design of a prototype adaptation of the program forpatients receiving medical care in a safety net healthcaresetting.

In phase 1, we conducted focus groups to understandthe needs and perspectives of potential end-users. In phase2 based on this feedback, we adapted the program’s sign-upprocess and online curriculum. In phase 3, we conducteda feasibility study to test the modified program with safetynet patients. In the final phase, we provided and adoptedrecommendations for the next iteration that will inform alarger controlled trial in safety net clinics throughout theregion. The Institutional Review Board of the University ofCalifornia, San Francisco approved the study.


2.3.1. Phase 1: Assessing Perspectives and Preferences for Life-style Management. In June 2015, we recruited English speak-ing and Spanish speaking participants to participate in twolanguage-specific focus groups. We recruited participantsby contacting diabetic and prediabetic patients who wereenrolled in upcoming in-person Spanish andEnglish diabeteseducation classes, which were held on a biweekly or monthlybasis at ZSFG. All patients were referred to these classes bytheir primary care provider team, and all were eligible forrecruitment to participate in phase 1 focus groups if they werefluent in English or Spanish and able to give written informedconsent.

The purpose of the focus groups was to assess the overallacceptability of the program (i.e., initial reaction, attitude,and baseline receptiveness to health tips for behavioralchange) and inform content modifications that would betteralign the program’s curricular content with socioeconomicconditions and sociocultural preferences in this population.Participants discussed their overall perspectives on the useof technology for lifestyle management. After reviewingdirect excerpts from the curriculum, participants providedfeedback about the clarity and relevance of the content. Todocument participant feedback during this phase, we had tworesearch analysts taking detailed notes with verbatim quotesto capture the discussion in these group sessions.

The authors used a team-based approach to conductdescriptive qualitative content analysis of focus group discus-sions, directly informed from the semistructured interviewquestions that captured perceptions of the acceptability of theonline weight loss platform and feedback on specific excerptsfrom the program’s curricular content. The three leads ofthe focus groups (LT, CR, and CRL) independently reviewedthe written notes from the focus group sessions to achievea consensus on the original list of codes, which were thenreviewed by the entire research team. Any discrepancies wereresolved by consensus.

2.3.2. Phase 2: Adapting the ProgramCurriculum for Readabil-ity and Relevance. Applying the feedback we received fromthe focus groups, we created an adaptation of the existingcurriculum to improve the readability and relevance of thecontent for a safety net population. To assess the readability ofthe existing curriculum and guide the adaptation of the con-tent, we used the SMOG readability formula, an index usedto determine the grade level required to understand a writtenpassage [22]. In adapting the readability of the curriculum,wefollowed recommendations from the US National Institutesof Health to aim for 6th-grade reading level or belowwhen developing easy-to-read health materials [23], makingfurther simplifications to address the high prevalence oflimitations in literacy, health literacy, and English proficiencyamong our clinic population. We modified the content ofthe program lessons to achieve SMOG readability indices of5th-grade (on average) reading levels, while preserving coreconcepts of the original lessons. We also adapted the contentto address lifestyle preferences and limitations reflected byparticipants in phase 1 focus groups.

Once a lower literacy version of the program content wascompleted in English, it was then translated into Spanish.

A bilingual, bicultural native Spanish speaker completed theSpanish translations. A second Spanish speaker reviewedand revised the final versions to ensure that the translatedlanguage was relevant to Latino patients and not just a directliteral translation of the English text. Because this was thefirst Spanish version of the program, we considered it a betaversion with which to gain very early feedback about thecontent, rather than a complete adaptation into Spanish.

2.3.3. Phase 3: Assessing the Feasibility of theModified Programin a Safety Net Population. Next, our team moved fromcontent adaptation to testing of the implementation of theprogram within a clinical setting. From August to November2015, we recruited English speaking and Spanish speakingpatients to participate in the phase 3 feasibility study, whichfollowed two small prospective cohorts of English and Span-ish speakers enrolling into the program at the same time.Thisfeasibility study covered enrollment through the first 4 weeksof the core program.

We designed the recruitment protocol to be consistentwith existing workflows for panel management in the clinic.To this end, we queried the electronic health record toidentify patients whomet eligibility criteria for language, age,Hemoglobin A1c (HbA1c) test result, and body mass index(BMI). We then sent the list to primary care providers toscreen out individuals who were not suitable for the study(based on exclusion criteria) and refer additional patients thatour electronic querymayhavemissed.We also posted flyers atthe clinic to allow individuals to self-refer (i.e., volunteer) forthe study. Research staff called individuals who were referredby their providers or self-referred in response to posted flyers,verified eligibility via chart review and telephone screening,explained the purpose and procedures of the study, andscheduled in-person sessions to obtain written informedconsent for phase 3 of the study.

Participants were eligible for phase 3 if they met all ofthe following eligibility criteria: (1) being fluent in English orSpanish; (2) age 18–75 at screening; (3) having had an HbA1ctest result of 5.7–6.4% or fasting glucose test of 110–125mg/dLin the past 6 months or as their most recent result; (4) BMI≥ 24 kg/m2 (or ≥22 kg/m2 if Asian American); (5) using theInternet at least weekly; and (6) being able to give informedconsent. We excluded participants who were already diag-nosed with diabetes, taking any hypoglycemic medications,and had serious unmanaged mental health conditions (e.g.,untreated bipolar disorder, severe untreated depression) orany other conditions that would preclude or make it difficultto participate in physical activity involving walking (e.g.,severe arthritis and limited lower limb mobility).

Sign-Up Process. We asked individuals interested in partic-ipating to attend an orientation session with a member ofthe study staff. During this session, participants receivedinstructions adapted for low literacy on how to completethe sign-up process for the program. For participants ofthe English speaking cohort who expressed moderate toadvanced comfort with computers, we asked participants tocomplete the sign-up process on their own time. For thosewith limited computer literacy, we allowed participants to


complete the sign-up process during the orientation session,providing one-on-one assistance if needed. Since the Spanishversion of the program was still in beta form, we asked allparticipants in the Spanish speaking cohort to complete thesign-up process in person to offer assistance if needed. Werecorded the types of barriers experienced by participantswho completed the sign-up process in person. Participantswho successfully enrolled were placed in a small cohort witha common program start date.

Follow-Up Interviews. Two and four weeks after the cohortsstarted the full 16-week program, we conducted semistruc-tured phone interviews with study participants to explorebarriers and facilitators to feasibility. At the two-week follow-up, we called all participants who had attended an orientationsession. Interviews focused on (1) experiences of participantsin completing the sign-up process, (2) getting oriented to theprogram, (3) using the online platform, (4) communicatingwith the health coach, (5) engaging with cohort membersonline, and (6) other barriers or facilitators to participatingin the program. At the four-week follow-up, we called allparticipants who had been placed into a program cohort.Interviews focused on overall satisfaction and components ofongoing engagement with the program, including communi-cating with cohort members and the health coach, trackingchanges in diet and exercise, and the feasibility and desire toapply newhealth knowledge and sustain engagementwith theprogram.

2.3.4. Phase 4: Applying Recommendations to Finalize a Proto-type Program. Based on observations and perspectives fromparticipants in signing up, getting oriented to, and engag-ing with the program, we provided recommendations formodifications to iterate the program in preparation for acontrolled clinical trial in safety net clinics and for com-mercial deployments with safety net providers. We outlinedoverall barriers and potential strategies to address thesechallenges and recommended modifications to the programwhere feasible.

3. Results

3.1. Phase 1: Assessing Perspectives and Preferences for LifestyleManagement. We enrolled four English speaking and sixSpanish speaking participants in phase 1 focus groups.Among the English speaking group, three participants weremale, three had been diagnosed with diabetes, and one hadbeen diagnosed with prediabetes. Among the Spanish speak-ing group, three were male and all six had been diagnosedwith diabetes.

The majority of informants in both English and Spanishspeaking focus groups were very receptive to the program’seducational content, expressed a high level of interest inparticipating, and conveyed a willingness to change theirbehavior as a result of the program.

I’m not just interested, I’m fascinated.I need to wisen up a bit and stop being silly. These[health] tips would be helpful.

However, our focus group discussions highlighted twolimitations to the program’s educational lessons that in-formed adaptation of the program to underserved popula-tions who tend to receive care in safety net clinics. First, afew informants reported that the educational content wastoo complex. Participants suggestedmodifying the content toexplain concepts that were difficult to understand in simplerterms.

The word [placebo] looks familiar, but I haven’theard this before.

How do I calculate a serving? How can wecalculate those daily percentages?

Second, more than half of focus group informants sug-gested that the health tips contained in the lessons neededto be more practical. Many of the health tips did notresonate with participants because they did not align withthe socioeconomic and sociocultural realities or preferencesin this low-income population. For example, health tipson physical exercise requiring gym equipment were oftenimpractical in this setting because access to fitness centerswaslimited. One participant noted the following:

Affordability [of the gym] is something I findfrustrating.

In addition, the majority of informants expressed thatfinding motivation to exercise was often a challenge. Ratherthanmore intense exercise regimens or going to the gym, par-ticipants expressed interest in walking or activities involvingmusic as a motivator, such as dancing.

Exercise can be about pleasure and not obligation.

The Spanish speaking focus group additionally empha-sized the importance of nutrition labels and false healthclaims. Participants had many questions about food labelssuch as “whole grain,” “organic,” and genetically modifiedorganisms (GMOs). They suggested a need for Spanishlanguage nutrition labels or having a glossary of simplifieddefinitions for nutritional terms included in nutrition labels(e.g., sodium = salt).

It’s important to explain to people what are car-bohydrates and other nutrients on the food labels.[Give] tips to know how to select foods that arewhole grain.

3.2. Phase 2: Adapting the ProgramCurriculum for Readabilityand Relevance. Changes made to the program curriculumto befit the literacy and cultural preferences of patients areshown in Table 1.

Specifically, to address concerns about the complexityof the curriculum, we adapted the readability level of eachlesson (originally 9th grade or higher) to mostly a 5th-gradelevel or below. In addition to simplifying overall language,we simplified explanations of scientific concepts, preservingcore concepts while improving understandability. To addressthe feasibility of and preferences for lifestyle changes, we


Table1:Ex

amples

ofcontentand

lang

uage

mod

ificatio

nto

adapto

nlinelessons

intheO

madap

rogram

tofit

literacylevels

andcultu

ralrelevance

associated

with

low-SES

patie

ntsinalarge

safetynetclin

ic.

Orig

inalcontent

Adaptedcontent

Adaptatio

nQuo

teRe

adability

index∗

Quo

teRe

adabilityindex∗

Protein

saretheb

uildingb

lockso

fmuscle

and

tissue,makingthem

essentialtogrow

thand

healing.Th

eysupportimmun

efun

ction,andare

used

tomakeh

ormonesandenzymes.P

rotein

isalso

abackup

source

ofenergy

when

carbohydrates

aren’tavailable.Whenitcomestoeatin

gprotein,

leansourcesa

regreatfor

your

health

6.0

Protein

build

smuscle

sand

tissue,and

helpsy

ougrow

andheal.Yourb

odya

lsousesprotein

asfuelwh

enitruns

outof

carbs.Low-

fatp

roteinsa

regreatfor

you

1.8Im

proveu

ndersta

ndabilityby

usingfewer

scientificterms

Liste

ning

tomusicwh

ileyouwa

lk,w

aitin

gtoread

afavoritem

agazineo

rwatch

aguilty-pleasureTV

show

until

you’r

eonthestatio

nary

bike,orinviting

your

funn

iestfrie

ndtojoin

youcanturn

exercise

intosomething

tolook

forw

ardto

18.2

Liste

ntoyour

favoritem

usicwh

ileyou

walk,orinvite

afriendtojoin

you.Th

iscanmakee

xercise

something

tolook

forw

ardto

7.2Em

phasizee

xercise

sthatd

ono

trequire

significantfi

nancialinvestm

ents

“Finally,

paya

ttentiontothec

ircum

stancesand

environm

entsthat

inspire

youtosta

yhealth

y,and

soak

upas

muchofthat

inspira

tionas

youcan.

Playingsports,perfe

ctinga

newyoga

pose,

spending

timeinnature,bein

garoun

dkids,reading

aboutp

eoplew

hoovercomeilln

essa

gainstall

odds...Whateverinspiresyou,embracethata

ctivity

andplan

todo

itmoreo

ften

10.6

Find

thethingsthatinspireyoutosta

yhealthya

ndusethem

tokeep

youon

track.

Maybe

it’splayingsportso

rbein

garoun

dkids.M

aybe

youenjoyspend

ingtim

ein

natureor

learninga

newdance.Find

what

drivesyou,useittosta

yinspired,andplan

todo

itmoreo

ften

1.8Em

phasizes

ources

ofmotivationand

activ

ities

thatarea

ccessib

leregardlessof

incomelevel

There’s

onem

oreimpressiv

estata

bout

theD

PPthat

youshouldknow

:75%

ofthep

eoplew

hoparticipatedin

thelifesty

leprogram

andmettheir

firstwe

ight-lo

ssgoal,keptthatw

eight

offforthree

years.In

comparison,only5

%ofpeoplewh

olose

weight

ontra

ditio

naldietsd

othesam

e

9.2

One

moreg

reat

thinga

bout

thep

rogram

:peopledidn’tjustlosewe

ight,buttheyk

ept

thew

eight

off.3

outof4

peoplekept

the

weight

offfor3

years.With

aregulard

iet,

only1o

utof20

peoplearea

bletokeep

the

weight

off

3.8

Improveu

ndersta

ndabilityby

simplify

ing

numericalconcepts

∗Whiletheentirelesson

swereon

averageat5th-gradereadinglevel,afewexcerptswith

inalesson

couldbe

atahigh

erSM

OGindexforreadabilitybasedon

theleng

thof

asentence

orthenu

mbero

f3-syllable

words.


Potential participants via chart review and provider referral (39)

Provider-referred (30)

Excluded by provider orEHR review (9)

Potential participants via flyer (12)

Excluded by EHR reviewor upon contact (7)

Self-referred (5)

Total referred (35)

Consent scheduled (14)

Consented to in-person (5)and at home (7) sign-up

Grouped in program (9)

Remained in program for4 weeks (8)

Unable to reach (6)

No-shows (2)

Unable to completePrevent sign-up at home (3)

Dropped out (1)

Contacted (29)Declined (9)

Ineligible: no Internet or tech use (6)

Completed 4-weekfollow-up interview (5)


Figure 1: Flowchart of recruitment for phase 3 English-based feasibility study.

adapted curriculum examples to fit perspectives expressedby phase 1 participants. In particular, we emphasized sourcesof motivation and exercises that did not require significantfinancial investments. For example, we replaced activitieswith potential financial burden (e.g., gymmemberships, yogaclasses, and races) with no or low-cost options (e.g., dancing,sports, and classes at community recreation centers). SeeTable 1 for examples of specific content changes. Additionally,to support healthy food choices that are accessible andculturally relevant, we added low-budget and ethnic recipesuggestions.

3.3. Phase 3: Assessing the Feasibility of the Modified Programin a Safety Net Population. During the feasibility study, wecontacted 64 potentially eligible patients: 29 English speakersand 35 Spanish speakers [Figures 1 and 2]. A total of 6English speakers (21%) and 17 Spanish speakers (49%) werenot eligible because they did not use the Internet regularly.Another 9 English speakers (31%) and 9 Spanish speakers(26%) declined to participate, citing reasons of disinterest inresearch, current participation in another health program, orother personal or health issues.

Overall, 23 out of 41 eligible patients (54%) expressedinterest in the program, and 18 of these eligible participantsbegan the enrollment process (12 English speakers and 6Spanish speakers). Ten of the 12 (83%) English speakingparticipants completed the two-week follow-up interview,and five of the nine actively enrolled in the program (55%)completed the four-week follow-up interview. Four of thesix (67%) Spanish-speaking participants completed the 2-week follow-up interview, and all five (100%) of the activelyenrolled Spanish speakers completed the 4-week follow-upinterview.

3.3.1. Sign-Up. The final feasibility sample was evenly splitby gender (53% female), and the mean age was 53. Of the12 participants of the English speaking cohort, 5 participants(42%) noted that English was not their first language.

All six Spanish speaking participants completed the sign-up process with guidance from a member of the researchteam. Seven English speaking participants wanted to com-plete the online sign-up process on their own at home; weasked five participants to complete the process in persondue to limited experience/confidence using computers orthe Internet. Of those 7 participants who were asked tocomplete sign-up independently at home, 4 (57%) finishedthe enrollment process. Among those who did not completethe sign-up process, it wasmostly due to technical issues withtheir computer or Internet not working. One participant evenstated, “Just real guidance (would have helped me sign up). . .Ijust got really frustrated.”

Even among those who completed the sign-up processwith one-on-one assistance, we observed several computerliteracy challenges for a majority of participants: (1) manyparticipants had difficulties using uniform resource locators(URLs). For example, some participants did not know whereto type in theURL, while others did not know to press “enter”after typing in the URL. Participants made frequent errors(i.e., typos) in entering the URL or entered the URL intosearch bars leading to Google search results instead of accessto the Omada website. This left participants confused andbewildered as they often could not recognize and correcttheir mistakes. (2) Navigating the online sign-up form wasalso challenging for some participants because they did notunderstand skip patterns that required users to click on anicon in order to move on to the next screen. (3) Participantsalso experienced challenges with the parts of the sign-up pro-cess that required using email (e.g., confirmation codes and


Potential participants via chartreview and provider referral (61)

Provider-referred (51)

Excluded by provider orEHR review (10)

Potential participants via flyer (5)

Excluded by EHR review, uponcontact, or unable to reach (4)

Self-referred (1)

Total referred (52)

Consent scheduled (9)

Consented to in-personsign-up (6)

Grouped in program (5)

Remained in program for4 weeks (5)

Unable to reach (17)

No-shows (3)

Only partially completed sign-up anddid not continue (1)

Contacted (35)Declined (9)

Ineligible: no Internet or tech use (17)



Figure 2: Flowchart of recruitment for phase 3 Spanish-based feasibility study.

links). Although they reported frequent Internet use, manyparticipants rarely checked email and some participants didnot have email accounts, requiring help to set up new ones.

3.3.2. Engagement with the Weekly Lessons

Weekly Logins. There was a high level of engagement amongfeasibility study participants, with 80% signing in at leasttwice at the two-week follow-up, and an average of 4.7 loginsper week for English speakers and 5.5 logins per week forSpanish speakers at the four-week follow-up. Of the programactivities, participants engaged most frequently in groupdiscussions, weigh-ins, and weekly lesson completion.

However, while logins overall were high, a few partic-ipants expressed technology accessibility barriers that mir-rored their challenges in completing the sign-up process, suchas not being able to remember their passwords to get backinto the online content and having inconsistent access to thecomputer.

Perceptions of the Curricular Content. The majority of pro-gram participants who completed either the two-week orfour-week interview reported satisfaction with the programand intentions to keep upwith the healthy behaviors. Overall,participants also expressed that they thought the lessons wereclear and useful.

It’s a lot of information that I never [heard] about.It’s great for me, I tell my family, you have to goand read those lessons.

Balance, control, and food portions—that is thepart that has helped me.

Aside from a few participants who noted a lack of timeto make lifestyle changes, participants noted that it was not

difficult to complete the program’s tracking requirements viafood diaries and exercise/pedometer logs.

I log my food every day. . .This program helps youhave awareness of what you’re doing.

Support from the Health Coach and Online Peer Group.Among those who successfully completed the sign-up pro-cess and participated in a follow-up interview, all but oneparticipant engaged with a health coach by telephone and50% contacted the health coach by text message or email, inaddition to telephone. The health coach at Omada reportedthat these participants were more interested in texting thanother Omada users that were not part of this study. Partici-pants stated clearly that they were the most satisfied with thedirect support provided by the health coach throughout theprogram.

She’s very nice. . . If I have a question, she tells methe answer. I like talking to her.

In general, about half of participants engaged in theonline social network, though a few participants reportedlimited engagement with participating in online discussionswith other peer cohort members, attributed to a lack of con-nection to cohortmembers or concerns about written literacyskills. Several participants expressed lower confidence withEnglish proficiency and literacy when posting on their own.

I’m not feeling any connection with the peoplethat are in there. There’s no camaraderie in seeingnames in a chatroom.

I don’t feel comfortable doing it on the computer.I’m not the writer. I don’t spell it right, mysentences. . . I have to get my dictionary out.


Table 2: Program modifications to improve usability based on end-user feedback.

Observed challenges Actions/strategies to address challenges

Lower computer literacy skills overall Creating technical assistance tools for various stages of the program. Greaterpromotion of mobile phone interface and use of app for English speakers

Need for additional guidance to complete the sign-upprocess independently

Greater promotion of using Omada support staff for assistance. Creatingtechnical assistance tools, that is, video tutorial, handouts with screenshots,

and instructions for accessing the sign-up website and completing the accountsetup process

Need for additional participant communication abouteach step in the sign-up process

Modifying notification schedule to remind participants more often aboutwhere they were in the sign-up process and the expected launch date of the

group cohortDifficulty logging in to the program after signing up More outreach from the program staff to help participants at the early stage

Desire for more connection with peers in the cohort Offering a conference call session at the beginning of the program to getgroup members oriented and connected to one another offline

3.4. Phase 4: Recommendations for the Next Iteration. Finally,we made several modifications to the procedures, as outlinedin Table 2. These changes provided additional guidanceand assistance to address technical difficulties with signingup for and logging in to the program online, increasedirect communication with the health coach, and facilitatestronger relationships between cohort members and otherworkflow process improvements. For example, we made afew modifications to the existing sign-up and orientationprocedures for enrollment in the program, such as (1) creatingstep-by-step screenshot instructions for registration that allparticipants could take home with them, (2) creating video-based tutorials for interacting with the various features ofthe digital program, (3) creating and sending an additionalemail to participants letting them know more details abouttheir upcoming group start dates (to reduce confusion andincrease engagement before the program even began), and(4) providing the support staff phone number in more placesduring the sign-up process to ensure that participants couldmore easily access phone support at any step of the process.

4. Discussion

In this study, product developers from Omada Health part-nered with health services researchers to utilize a user-centered approach to first adapt the online content of anexisting digital health diabetes prevention program and thentest the usability and feasibility of the modified productin a real-world safety net clinic that cares for underservedpopulations (e.g., low-income people, African Americans,and Latinos) at high risk of diabetes. While preserving thecore elements of the original program, the adapted contentwas more aligned with the needs of the safety net patientpopulation, written at a lower reading level, and incorporatedtips that were more resonant with the preferences and socio-cultural realities of the target population. Patients enrolledin the program remained engaged with high rates of regularuse and reported overall satisfaction despite some substantialtechnical difficulties in signing up and logging into theprogram. This user-centered approach for feasibility testingis significantly underreported in the scientific literature, and

we have provided detailed findings about how to design thisapproach to collect rich data within a short timeframe.

The diabetes prevention program (DPP) has been shownto prevent type 2 diabetes through lifestyle modification.Numerous studies have translatedDPP to real-world commu-nity settings (e.g., YMCA, churches, or workplace) and clini-cal settings that required cultural adaptation of the program[24, 25]. To our knowledge, this is the first digital translationof DPP for an underserved patient population. A recentreview found that among the 15 DPP translation studies,only one was conducted in a primary care clinical setting[24]. Our study makes important additions to the body ofliterature on translation ofDPP in the real-world by providingspecific recommendations about how the recruitment andinitial enrollment process might be better completed withinsafety net clinical sites.

Challenges related to computer and online literacy andaccessibility in signing up and logging into the programposed the biggest barriers for usability in this low-incomepopulation. Despite limiting our study to patients whoreported weekly use of the Internet, a substantial portionof patients needed technical assistance to sign up for theprogram. In fact, even among those who we deemed moretechnically proficient and then attempted to sign up inde-pendently, more than one-third dropped out of the studybecause they could not complete the sign-up process dueto technical difficulties related to poor computer or onlineliteracy and inconsistent computer or Internet access. Thefinding of technical difficulties as a barrier to feasibility is notsurprising. A recent systematic review of the impact of healthIT on patient behavior change reported technical issues as acommon barrier to usability of IT platforms [8].

Our study has uncovered a real opportunity for compa-nies to collaborate with health services researchers embeddedin safety net clinics to adapt and improve usability of digitalhealth products in low-income and underserved populations.This partnership conferred several advantages that are rarelyavailable in the field of digital health. First, the collaborationwith a research center directly integrated with a safety netclinical practice allowed us to recruit a diverse sample ofpatients to test the product in a real-world process thatwas not separate from ongoing care for complex patients.


Second, having the digital health company actively engagedas a partner on this work enabled our team to exploretechnical assistance solutions to improve usability of the tech-nology in real-time (e.g., implementing step-by-step sign-upinstructions within one week of identifying challenges in thisprocess).

There are some limitations to consider in interpreting ourfindings. We designed this study as a feasibility study in asmall sample of patients. Therefore, these findings are notnecessarily generalizable to all high risk patients who receivecare in safety net clinics. It is also possible that this safetynet clinic provides care to a patient population associatedwith lower computer literacy than other safety net clinics.However, our user-centered process to iteratively adapt andtest the Omada Health Program in a uniquely challengingpatient population offers a potentially distinct advantageof refining an intervention toward maximal usability andsustainability. Because the clinic is a university-affiliatedclinic located in San Francisco, its patients may be morereceptive to clinical trials as well as technology solutions.Thiscould have contributed to the level of interest in participatingin the program and the sustained engagement among thosewho enrolled.

5. Conclusion

We documented patient interest, engagement, and satisfac-tionwith a digital health diabetes prevention program amongboth English and Spanish speaking patients at a large safetynet clinic. However, low computer and online literacy forsome of this population presented implementation challengesthat should be considered in digital health adaptations forlow-incomepopulations.Themodel of collaboration betweenresearchers and a digital health company allowed for sub-stantial iterations in the final program that can be scalable,improve usability, and contribute to increasing the overallimpact of the product.

Disclosure

All authors have fulfilled the criteria for authorship estab-lished by the International Committee of Medical JournalEditors and approved submission of the manuscript. Formeraddress of Kelly McDermott was Omada Health, 500 San-some St., Suite 200, San Francisco, CA 94111, USA.

Competing Interests

This work was funded by Omada Health, Inc. a company thatmakes and owns online behavior change programs.

Acknowledgments

This work was also supported by the Agency for HealthcareResearch and Quality (R00HS022408) and the UCSF Centerfor Vulnerable Populations at the Zuckerberg San FranciscoGeneral Hospital.

References

[1] K. Radhakrishnan, “The efficacy of tailored interventions forself-management outcomes of type 2 diabetes, hypertension orheart disease: a systematic review,” Journal of Advanced Nursing,vol. 68, no. 3, pp. 496–510, 2012.

[2] C. A. Chrvala, D. Sherr, and R. D. Lipman, “Diabetes self-management education for adults with type 2 diabetes mellitus:a systematic review of the effect on glycemic control,” PatientEducation and Counseling, vol. 99, no. 6, pp. 926–943, 2016.

[3] T.Deakin, C. E.McShane, J. E. Cade, andR.D.Williams, “Groupbased training for self-management strategies in people withtype 2 diabetes mellitus,” The Cochrane Database of SystematicReviews, no. 2, Article ID CD003417, 2005.

[4] H. Jones, L. Edwards, T. M. Vallis et al., “Changes in diabetesself-care behaviors make a difference in glycemic control: theDiabetes Stages of Change (DiSC) study,”Diabetes Care, vol. 26,no. 3, pp. 732–737, 2003.

[5] M. K. Ali, J. Echouffo-Tcheugui, and D. F. Williamson, “Howeffective were lifestyle interventions in real-world settings thatwere modeled on the diabetes prevention program?” HealthAffairs, vol. 31, no. 1, pp. 67–75, 2012.

[6] R. Whittemore, “A systematic review of the translationalresearch on the Diabetes Prevention Program,” TranslationalBehavioral Medicine, vol. 1, no. 3, pp. 480–491, 2011.

[7] A. L. Albright and E. W. Gregg, “Preventing type 2 diabetes incommunities across the U.S.: the national diabetes preventionprogram,” American Journal of Preventive Medicine, vol. 44, no.4, supplement 4, pp. S346–S351, 2013.

[8] S. Sawesi, M. Rashrash, K. Phalakornkule, J. S. Carpenter, andJ. F. Jones, “The impact of information technology on patientengagement and health behavior change: a systematic review ofthe literature,” JMIR Medical Informatics, vol. 4, no. 1, article e1,2016.

[9] P. Sutcliffe, S. Martin, J. Sturt et al., “Systematic review of com-munication technologies to promote access and engagement ofyoung people with diabetes into healthcare,” BMC EndocrineDisorders, vol. 11, article 1, 2011.

[10] A. P. Cotter, N. Durant, A. A. Agne, and A. L. Cherrington,“Internet interventions to support lifestyle modification fordiabetes management: a systematic review of the evidence,”Journal of Diabetes and Its Complications, vol. 28, no. 2, pp. 243–251, 2014.

[11] S. Harrison, M. Stadler, K. Ismail, S. Amiel, and A. Herrmann-Werner, “Are patients with diabetes mellitus satisfied with tech-nologies used to assist with diabetes management and coping?:a structured review,” Diabetes Technology & Therapeutics, vol.16, no. 11, pp. 771–783, 2014.

[12] C. R. Lyles, U. Sarkar, and C. Y. Osborn, “Getting a technology-based diabetes intervention ready for prime time: a review ofusability testing studies,” Current Diabetes Reports, vol. 14, no.10, article 534, 2014.

[13] D. M. Rabi, A. L. Edwards, D. A. Southern et al., “Association ofsocio-economic status with diabetes prevalence and utilizationof diabetes care services,” BMC Health Services Research, vol. 6,article 124, 2006.

[14] U. Sarkar, A. J. Karter, J. Y. Liu et al., “Social disparities ininternet patient portal use in diabetes: evidence that the digitaldivide extends beyond access,” Journal of the American MedicalInformatics Association, vol. 18, no. 3, pp. 318–321, 2011.

[15] G. Welch and R. Shayne, “Interactive behavioral technologiesand diabetes self-management support: recent research findings


from clinical trials,” Current Diabetes Reports, vol. 6, no. 2, pp.130–136, 2006.

[16] S. C. Sepah, L. Jiang, and A. L. Peters, “Long-term outcomesof a web-based diabetes prevention program: 2-year results ofa single-arm longitudinal study,” Journal of Medical InternetResearch, vol. 17, no. 4, article e92, 2015.

[17] A. E. Shields, P. Shin, M. G. Leu et al., “Adoption of healthinformation technology in community health centers: results ofa national survey,” Health Affairs, vol. 26, no. 5, pp. 1373–1383,2007.

[18] V. A. Lewis, C. H. Colla, K. E. Schoenherr, S. M. Shortell, and E.S. Fisher, “Innovation in the safety net: integrating communityhealth centers through accountable care,” Journal of GeneralInternal Medicine, vol. 29, no. 11, pp. 1484–1490, 2014.

[19] C. LeRouge andN.Wickramasinghe, “A review of user-centereddesign for diabetes-related consumer health informatics tech-nologies,” Journal of Diabetes Science and Technology, vol. 7, no.4, pp. 1039–1056, 2013.

[20] A. L. Knoblock-Hahn, R. Wray, and C. M. LeRouge, “Per-ceptions of adolescents with overweight and obesity for thedevelopment of user-centered design self-management toolswithin the context of the chronic care model: a qualitativestudy,” Journal of the Academy of Nutrition and Dietetics, vol.116, no. 6, pp. 957–967, 2016.

[21] S. Van Der Weegen, R. Verwey, M. Spreeuwenberg, H. Tange,T. Van Der Weijden, and L. De Witte, “The development ofa mobile monitoring and feedback tool to stimulate physicalactivity of people with a chronic disease in primary care: a user-centered design,” Journal of Medical Internet Research, vol. 15,no. 7, article e8, 2013.

[22] H.G.McLaughlin, “SMOGGrading-a new readability formula,”Journal of Reading, vol. 12, no. 8, pp. 639–646, 1969.

[23] How to Write Easy-to-Read Health Materials, 2013, https://medlineplus.gov/etr.html.

[24] R. G. Tabak, K. A. Sinclair, A. A. Baumann et al., “A review ofdiabetes prevention program translations: use of cultural adap-tation and implementation research,” Translational BehavioralMedicine, vol. 5, no. 4, pp. 401–414, 2015.

[25] D. L.Hall, E. G. Lattie, J. R.McCalla, and P.G. Saab, “Translationof the diabetes prevention program to ethnic communities inthe United States,” Journal of Immigrant and Minority Health,vol. 18, no. 2, pp. 479–489, 2016.

https://medlineplus.gov/etr.html

https://medlineplus.gov/etr.html

Research ArticleLinking High Risk Postpartum Women with a TechnologyEnabled Health Coaching Program to Reduce Diabetes Risk andImprove Wellbeing: Program Description, Case Studies, andRecommendations for Community Health Coaching Programs

Priyanka Athavale,1,2 Melanie Thomas,3 Adriana T. Delgadillo-Duenas,1

Karen Leong,4 Adriana Najmabadi,4 Elizabeth Harleman,5,6 Christina Rios,2,4

Judy Quan,2,4 Catalina Soria,2,4 and Margaret A. Handley1,2,4

1Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, CA, USA2UCSF Center for Vulnerable Populations, Zuckerberg San Francisco General Hospital, San Francisco, CA 94110, USA3Department of Psychiatry, UCSF/Zuckerberg San Francisco General Hospital, San Francisco, CA, USA4Division of General Internal Medicine, UCSF/Zuckerberg San Francisco General Hospital,University of California San Francisco, San Francisco, CA, USA5Department of Obstetrics and Gynecology, University of California San Francisco, San Francisco, CA, USA6San Francisco General Hospital, San Francisco, CA, USA

Correspondence should be addressed to Priyanka Athavale; [email protected]

Received 6 May 2016; Revised 1 August 2016; Accepted 14 August 2016

Academic Editor: Shari Bolen

Copyright © 2016 Priyanka Athavale et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Background. Low-income minority women with prior gestational diabetes mellitus (pGDM) or high BMIs have increased risk forchronic illnesses postpartum. Although the Diabetes Prevention Program (DPP) provides an evidence-based model for reducingdiabetes risk, few community-based interventions have adapted this program for pGDM women. Methods. STAR MAMA is anongoing randomized control trial (RCT) evaluating a hybrid HIT/Health Coaching DPP-based 20-week postpartum program fordiabetes prevention compared with education from written materials at baseline. Eligibility includes women 18–39 years old, ≥32weeks pregnant, and GDM or BMI > 25. Clinic- and community-based recruitment in San Francisco and Sonoma Counties targets180 women. Sociodemographic and health coaching data from a preliminary sample are presented. Results. Most of the 86 womenincluded to date (88%) have GDM, 80% were identified as Hispanic/Latina, 78% have migrant status, and most are Spanish-speaking. Women receiving the intervention indicate high engagement, with 86% answering 1+ calls. Health coaching callbackslast an average of 9 minutes with range of topics discussed. Case studies presented convey a range of emotional, instrumental, andhealth literacy-related supports offered by health coaches. Discussion. The DPP-adapted HIT/health coaching model highlightsthe possibility and challenge of delivering DPP content to postpartum women in community settings. This trial is registered withClinicalTrials.gov NCT02240420.

1. Introduction

Following pregnancy, low-income, minority women with ahistory of GDM or high BMIs are at high risk for chronicillnesses, particularly obesity, type 2 diabetes mellitus (DM),and postpartum depression [1, 2]. Racial/ethnic disparities

exist across a variety of postpartum health outcomes, andthese disparities may be widened by less postpartum clinicalfollow-up by Hispanic/Latina and African American patients[3]. These disparities include a higher prevalence of gesta-tional diabetes (GDM) during pregnancy, an increased riskfor DM postpartum [4, 5], and a lack of postpartum diabetes


http://dx.doi.org/10.1155/2016/4353956


screening for those with GDM. Only one in five Latinawomen with prior GDM returns for postpartum diabetescheckups, the lowest follow-up frequency of any group [6].These women are lost to follow-up despite recommendationsfor six-week postpartum screening as well as annual diabeteschecks [7].

In California, the prevalence of GDM is 5.7% amongLatina women, compared to 4% in non-Hispanic whites[8, 9]. Women born outside the US make up a majority ofpregnancies in California and may be at higher risk for GDMand subsequent DM for reasons including disruption totheir normal eating habits, unhealthy dietary acculturation,barriers to physical activity, and rapid weight gain aftermigration to high-income countries [10, 11]. Results from theDiabetes Prevention Program (DPP) suggest that the long-term risk for diabetes postpartum can be reduced throughbehavior change [12–14]. Current strategies to expand post-partum care management and to reduce chronic disease riskfor ethnic minority women include tailored interventionsthat address health literacy and provide accessible resourcesor health coaching on lifestyle changes, but there are fewprograms that focus on delivering the DPP content for earlypostpartum women in the period immediately followingbirth.

Health information technology (HIT) can be an impor-tant tool to tailor health communication efforts, and inparticular if developed with end-users, it can be effectivewith low-literacy and low-income populations [15–19]. Areview of literacy-focused interventions shows that inter-ventions using health information technologies have highpotential to reach low-literacy, high risk populations sincethey have flexibility of access in the home and duringconvenient times [20, 21]. Interventions providing longitu-dinal care and support through health coaching or coun-seling have also been effective in reducing chronic illness[22].

Approximately 85% of adults in the US are cell phoneusers, and cell phone use does not vary by race or ethnicity.Majority of low-income users have basic cell phones used forvoice messaging and texting, and while this type of phoneusage does not have the functionality of a smart phone(Web browsing and mobile applications), it provides accessto the broadest range of users across SES status [16]. Toimprove women’s health postpartum, several HITmodels areeffective in enabling communication and reaching womenwhen the clinic-based model is less convenient. Postpartumlow-income women may lose Medi-Cal eligibility, encounterbarriers in accessing health care, and face new demandsat home. Postpartum calls can reinforce messages receivedin health care encounters as well as facilitate uptake ofpreventive services for both mother and child/children inthe 6–9 months after delivery (e.g., 6-week glucose testingafter delivery and vaccinations in infants). In the case ofdiabetes, there is growing literature indicating that diabetesself-management support can be improved with improve-ments in patient satisfaction aswell as diabetes-specific healthoutcomes through health technology (Interactive Voice Mes-saging; Internet-Based Systems).Our teamhas used a tailoredcombination of short (4-5 minutes) automated calls with

queries and narratives in diabetes self-management andhave found such an approach, Automated Telephone Self-Management Support (ATSM), effective for reaching andengaging patients with low health literacy and limited Englishproficiency and it can be cost effective and can improve healthoutcomes [23–25].

Another proven strategy for diabetes prevention isthrough health coaching. Health coaching involves coun-seling patients with chronic conditions to improve theirown health by increasing their knowledge, skills, and con-fidence in managing their own health behaviors [26]. Theeffectiveness of health coaching in motivating, empowering,and enabling patients to improve health behaviors is nowwell established [27–31]. In particular, the peer health coachmodel has been successful in helping patients self-managediabetes [32, 33]. Goldman et al. describe successfully usingpeer coaches, who were diabetic patients themselves, toprovide support to other patients through three key roles:advisor, supporter, and role model [34]. Another studyshowed that the peer coach model is particularly effectiveamong patients with worse medication adherence and higherHbA1c levels [35]. Some evidence suggests no difference inpatient outcomes when comparing peer coaches and healthprofessionals in the counsel of diabetes patients [36]. Heisleret al. randomized patients with diabetes to a diabetes caregroup managed by a nurse practitioner either with a peercoach or with a nurse practitioner alone and found thatparticipants with the peer coach had greater improvementsin HbA1c levels after intervention [30].

Despite the success of peer health coaching programs inprimary care and clinical settings, there are few examples ofsimilar programs implemented in community-based settings,such as federally funded programs or nonprofit organiza-tions. There is also limited data on how such a model mightaddress language and literacy challenges faced by low-incomewomen with recent histories of migration to the US. Onewell-known and relevant community program that offershealth coaching for this population is the Women, Infants,and Children (WIC) Program, which provides supplementalfoods, health care referrals, and nutrition education for low-income pregnant and postpartum women [37]. In someWIClocations, peer-supported health coaching is available forbreastfeeding support [38].

In this paper, we present a unique health coachingmodel,the STAR MAMA program (support via Telephone Adviceand Resources/Sistema Telefonico de Apoyo y Recursos-MAMA), which combines HIT-based queries and narratives,with follow-up by trained health coaches, to deliver adaptedDPP content. The goals of this paper are to (1) illustrate aunique model of health coaching for high risk, low-incomepostpartum women which has relevance in community andclinic-based systems and (2) present case studies of exem-plary STAR MAMA health coached participants stories.

2. Methods

STAR MAMA is an ongoing randomized clinical trial com-paring a HIT-based health coaching program with a usualcare arm providing an educational resource guide covering


approaches to improving postpartum diabetes risk behaviorsamong GDM women based on DPP content. The studypartnered with four key community sites: Zuckerberg SanFrancisco General (ZSFG), Santa Rosa Community HealthCenters (Vista and Lombardi clinics) in Sonoma County, andSan Francisco and SonomaWIC offices. We chose to partnerwith WIC, as it is one of the few programs that provideservices to low-income women spanning from pregnancyinto the postpartum period and because migrant womencan access most services on behalf of themselves or theirinfant. Additionally, WIC’s program content, which providesinformation and resources on healthy eating and referralsto health care, aligns with the main objectives of the STARMAMA program [37].

2.1. STAR MAMA Study Design and Evaluation. The STARMAMA program is evaluated through a randomized controltrial design in which women are assigned to one of twoarms: (1) HIT arm: participants receive weekly phone callsfrom the automated telemedicine self-support system onvarious diabetes preventive topics and arematched to a healthcoach for longitudinal follow-up (2); usual care/educationarm: participants receive an education resource guide (infor-mation and links to nearby resources) about postpartumcare for themselves and their baby along with communityresources for diet, physical activity, and so forth. Recruitmentat clinic sites was monitored through a thorough list ofGDM women who were approached based on eligibilitycriteria to enroll in the study. For the WIC sites, womenwere targeted based on delivery dates and either havingGDM or a high BMI status (>25). WIC staff recruitedwomen through phone calls prior to enrollment in thestudy. Women were randomized prior to conducting thebaseline survey at their enrollment visit to one of thetwo study arms (using an envelope sealed randomizationassignment), following administration of the baseline survey.Primary outcomes include self-reported weight, body massindex (BMI) based on chart review, receipt of recommendedpostpartum glucose testing, changes in dietary patterns, suchas consumption of fruits and vegetables and foods high infats or sugars, and physical activity (minutes per week). Priorto enrollment, all women completed a written or verbal con-sent.

2.2. STAR MAMA: Intervention Description

2.2.1. STAR MAMA HIT Component. STAR MAMA partici-pants randomized to the intervention armof the study receiveweekly phone calls through our automated telemedicineself-support system and are paired with a health coachfor longitudinal follow-up. The phone calls start 6 weekspostpartum and continue for 20 weeks. AHIT enabledmodelwas selected at the outset because it allows participantsto receive weekly content and health coaching support intheir primary language while remaining in their homes, astraveling to appointments and group sessions is known to bea key barrier to receiving preventive services in communitysettings [39]. The calls focus on DPP topics including diet,

physical activity, encouraging partner support, balancingself-care postpartum, healthy eating tips, importance ofreceiving a blood sugar checkup, and baby care. Informationwas delivered through recorded narratives and text tipsin the automated telephone system. Women also receivedinformation from “live” health coach call backs [40]. Forexample, if a participant pressed “1” (yes) to a query asking ifshe was feeling stressed about her baby crying, shemight heara story about a new mom like her, facing similar challenges,reassuring her that it is ok to ask for help. The informationfrom this question would then be delivered to her STARMAMA health coach, who would also call her back andprovide her with support. Figure 1 describes the content andmethod of delivery of the range of STAR MAMA topicscovered from weeks 1 through 20.

An enrolled participant can trigger a response severaltimes during each phone call. Triggers are classified bypredetermined values which determine whether a healthcoach callback is required. Daily and weekly reports fromthe HIT calls provide context for the health coaching callandmotivational interviewing andhelp the coach understandwhat issues to focus on when developing an action plan orgoals with the participant.

2.2.2. STAR MAMA Health Coaching Component. A partic-ipant receiving the weekly phone calls is also matched toa health coach who monitors her response to the calls andregularly follows up with the participant on relevant issues.Follow-up topics vary from specific concerns, longitudinalsupport, empowerment, or resources for the mother. Forexample, in the sixth week of calls, the participant is queried:“If you have questions about feeding your baby or how to dealwith the pressure you are feeling from your family and friends,press 1 and a health coach will call you back. If not, press 2.”Health coaches at ZSFGH were from bicultural backgroundsand had previous experience in the clinical setting as eitherhealth coaches or para-health professionals. Coaches at theWIC sites were trained nutritionists or registered dieticiansalready working in the WIC system. At minimum, eachhealth coach received a two-day training at the Universityof California San Francisco (UCSF) Center for Excellence inPrimary Care and was given a one-day STARMAMA specifictraining that focused on delivery of the DPP, for a totalof three days of health coaching training. Ongoing healthcoaching training also included review of the health coachingmanual adapted from the DPP and biweekly review of casesat staffmeetings. Additionally, coaches had ongoingmeetingsto discuss and reflect on common topics and support eachother in the coaching process. Health coaches kept detailednotes based on call summary and trigger reports, which allcoaches could access for internal resources.The health coach-ing curriculum was codeveloped with relevant stakeholders,including participants and care providers familiar with theethnic diversity of the local populations. The following listtitled “Health Coaching Training Topics, Example Query,Narrative, and Health Coach Script” describes a summary ofthe health coaching training, an example of a STAR MAMAautomated telephone call, and a guideline for the subsequenthealth coaching topics.


Week 1 Week 2 Week 3 Week 4

Mom

Baby

Q: Had a doctor’scheckup?

Q: Had a blood sugarcheckup?

Edu: Healthiersweets & dessertsNarrative; Tip + Text

for recipe

Q: Feeling able totake care of yourself

and your baby?

Q: Breastfeedingsupport?

Q: Bottle-feeding andformula?

Q: Someone to turnto for practical help

with your baby?

Narrative: Engaginghusband in baby care

and self-care

Q: Usefulness of WICservices

Q: Need helpidentifying baby cues?

Q: Fussy baby? + Narr.Q: Breastfeeding

support?Q: Bottle-feeding and

formula?

Q: Had a health checkup?

Q: Had a doctor’scheckup?

Narrative: Discussinghealthy food optionswith your husband

Tip + Text for familyon healthy food

recipes

Narrative: Safe sleepingenvironment for baby

Q: Breastfeedingsupport?

Q: Bottle-feeding andiron formula info.

Q: In last week,number of days with

sweet drinks?

Tip and Narrative onlow sugar drinks

Edu: Facts aboutcarbohydrates

Q: Usefulness of WICservices

Q: 2-monthimmunizations?

Q: Breastfeeding andeducation

Q: Bottle-feedingEdu: Baby hunger

cues


Mom

Baby

Q: Number of days eatout? Narrative on

salads and veggies atmeal

Edu: Folic acid andtaking multivitamin

Narrative: Substitutehealthy foods and

reduce portion size

Q: Breastfeeding withno water Education

Q: Bottle-feedingwith no water

Education

Q: Last week, #of days with high fat

foods?Edu + Tips to reduce

cooking high fat foods

Q: Feel pressure tofeed baby solids?

Q: Feeling able to takecare of yourself and

baby?Narrative + Tips on

support/mood

Q: Breastfeeding +support? Narrative

and comfort suckingEducation

Q: Bottle-feeding?

Narrative: Normal tobe sad + ask for

help; need support +want to talk with

coach?

Edu: Facts aboutcarbs. Difference

between simple andcomplex carbs

Q: Need baby checkupinfo? Growth charts?

Q: Breast/bottle-feeding? No early

solids

Q: Do you want totalk about healthyprograms/plans?

Q: # of days eaten fruit/vegetables

in the past week?

Q: Last week-#of days exercise? Text

link to Latin music

Q: Want link to websiteabout baby’s health?

Q: Do you feel ontrack for healthy

weight? + Narrativeabout exercise support

Text link to Latinmusic

Q: Breastfeedingsupport + hunger

cuesBottle-feeding

support + hungercues


Mom

Baby

Q: How many days inthe past week of

exercise? Text linkwith audio

Edu: Sports andenergy drinks

Text: Infused water

Edu: Understandingfood labels and

serving sizes

Q: Doctor visits beforepregnancy?

Text link: Music forexercise

Q: Breastfeeding?frequency?

Q: Bottle-feeding?

Q: How many daysin the past week of

exercise?

Edu: PregnancyWeight

Narrative: Myths

Q: Do you needsupport for feeling

worried or sad?

Q and Edu:Emotional Eating

Q: Do you haveproblems affording

fresh produce?Narrative

Edu: Exercisecommunity resources

Q: Baby’s 4-monthimmunizations?Q: Breastfeeding

support + hungercues

Bottle-feedingsupport + hunger

cues

Q: Breast/bottle-feeding?No early solids

Q: Breastfeeding?Frequency?

Q: Bottle-feeding?

(a)

Figure 1: Continued.



Mom

Baby

Narrative:Carbohydrate cravings

Q: Breastfeeding withno water Education

Q: Bottle-feedingwith no water

Education

Narrative: Caloriecontrol and healthy

substitutes

Edu: Empty calories,added sugar, eatingsweets Narrative +

Text

Education: Childspacing. Planning ahealthy pregnancy

Q: How many days inthe past week ofexercise? Seeking

support with othermothers


Q: Bottle-feedingwith formula?


exercise?Narrative: Finding

time to exerciseNarrative: Organic

produce

Q: How many days inpast week with high fat

foods? Narrative

Narrative: Partnerengagement with baby

care

Q: Breastfeeding? Frequency?

Q: Bottle-feedingwith formula?

Q: How many days inthe past week did you

exercise?Text link for music


Q: Bottle-feeding withformula?

Q: Prenatal vitamins?Education


Mom

Baby

Q: Making smallchanges to manage exercise? Narrative

Narrative: Healthysnacks

Narrative: Emotionaleating; cravings and

feeling down

Narrative: Irondeficiency and

anemia

Q: Breastfeeding?Bottle-feeding?

Edu: Minimizingsugary foods for the

baby

Q: Affordability offresh produce?

Education: Tips foreating out at restaurants

Q: What to do whenyour baby is sick?

Education with Link

Q: How many daysdid you cook foods

with high fats? Narrative

Q: Family support?

Education: Familyplanning and spacing

children


exercise? Edu: Healthyweight, intensity, body

image

Education: Understandingnutrition labels

Narrative: Managingpressure and stress

Q: Breastfeeding? Frequency?

Q: Bottle-feeding with formula? Q: Breastfeeding?

Frequency? Q: Bottle-feeding with

formula?

BabyMomEngagement withhealth care

Exercise and weight loss Breastfeeding Behavior

ImmunizationsFeedingStress/depression/social support

Nutrition and foodinsecurities

(b)

Figure 1: STARMAMAHIT automated telephone messages content and mode of delivery: Maternal and Child Information (Edu), Queries(Q), Narrative, or Tip/Text.

Health Coaching Training Topics, Example Query, Narrative,and Health Coach Script

Health Coach Training Topics(i) Communication

(a) Ask-tell-ask, how to receive information fromthe participant

(b) Closing the loop(c) Setting the agenda(d) Understanding your current health status, num-

bers

(ii) Motivational interviewing

(a) Exploring patients motivations and barriers


(iii) Postpartum risk specific knowledge

(a) Knowledge regarding risk of DM postpartum

(iv) Community and clinic specific resources

(a) Community related risks: safety, access toaffordable produce, and primary care

(b) Community-based resources: food banks, WICagencies, free health clinics, and physical activ-ity groups

Health Coaching Example

Query to Participant. “In the last 7 days, how many days didyou drink sweetened drinks like sodas, aguas frescas, fruitjuices, coffee with sugar or condensed milk, sports drinks orenergy drinks? Press the number of days.”

Callback Trigger. Callback trigger ≥3 days (this trigger isdetermined by the participant pressing 3–9 on her phone, as aresponse to the question about the number of days she dranksugar sweetened drinks).

Narrative, Heard by Participant. “Sweetened drinks taste goodbut don’t have healthy calories. A small can of soda can haveas many as 10 sugar packets in it! Even aguas frescas can haveextra calories and loads of sugar without giving you muchnutrition. You don’t have to stop drinking them, but you can tryhaving them less often and making your own with less sugar.First try preparing them with half the amount of sugar. Forexample, if you like preparing agua frescawith strawberries andusually add two large spoonfuls of sugar, try just adding one.At first it might taste less sweet but it is something that you andyour family can get used to, and is part of showing them yourcommitment to health- theirs and yours!”

Health Coach Script and Topic Guide

Verify Report

(i) “On your call you answered that in the past 7 days youdrank sweetened drinks days. Is that correct?”

Open-Ended Question

(i) “How many sugary drinks do you typically have inone day?”

(ii) “What do you like about drinking sweetened drinks?”(iii) “Is there something you don’t like about drinking

them?”(iv) “How do you think drinking sweetened drinks affects

your health?”(v) “How do you think it could affect your risk of

developing T2DM?”

Provide Education

(i) Sugary drinks contribute to high calorie intake andcan lead to obesity.

(ii) Drinks high in sugar make your blood sugar spikewithin a few minutes of drinking it.

(iii) People who consume sugary drinks regularly—1 to 2cans a day or more—have a greater risk of developingtype 2 diabetes than people who rarely have suchdrinks.

Help Participant Make Action Plan

(i) “What step would you like to take to start reducingyour intake of sugary drinks?”

(ii) “When are you going to do it?”(iii) “How much are you going to decrease them?”(iv) “How often are you going to do it?”(v) “On a scale of 1–10, 1 being not sure at all and 10 being

completely sure, how sure are you that you canby ?”

If less than 7, encourage participant to modify action plan tomake it achievable.

While the automated phone calls provide participantswith passive information and support through narratives,the health coaches directly reach participants, explore theirneeds, build on their strengths, and set goals to help themreach their health targets. As such, the health coach serves asa bridge between the participant and the primary care clinicand as a source of support, resources, and accountability.Figure 2 illustrates the relationship between the health coach,the ATSM service, primary care providers, the community,and the participant within the STAR MAMA study. Inthis patient-centered model, the health coach is an integralsource of tangible preventive information and longitudinalcommunication with the participant and health care setting,community, or clinic.

2.3. STARMAMA Participants. This paper includes a sampleof women who have either completed STAR MAMA or arecurrently enrolled representing half of the targeted recruit-ment sample. Eligibility criteria for STARMAMA include 18–39 years of age, at least 32 weeks pregnant, and either a GDMdiagnosis or BMI > 25. Participants were recruited from ourfour community sites through either physician referral, WICreferral, or direct communication during scheduled prenatalappointments.

2.4. Assessment of Engagement with STAR MAMA Pro-gram. Participant engagement was assessed using our onlinedatabase tracking system, which monitors participant weeklycall responses. Different levels of engagement were deter-mined based on measures in previous studies: (1) participanthad completed at least one of the weekly phone calls; (2)average number of calls completed out of the 20 weeks ofcalls, and (3) among those completing one or more call, thepercentage of calls completed over the intervention period[26].

All women receive a baseline visit, 3-month short phonesurvey and a 9-month postpartum follow-up survey and willhave their medical charts reviewed over the study period.The


(1) Postpartum GDM women

(4a) Community(resources, support groups, organizations,etc.)

(4b) Community health provider at community clinic

(3) Health coach: trained individualto provide support, follow-up, and longitudinal care and goal setting withpatient

(2) Health-IT (weekly automated phone calls)

Figure 2: STAR MAMA health-IT intervention linkage model: using the health coach as a bridge between the community and hospitalinfrastructure for postpartum GDM women. (1) A woman is enrolled into the STAR MAMA study based on her eligibility. See Table 1 forbaseline demographics. Eligible WIC participants were referred to the STAR MAMA study by their respective coordinators. (2) Enrolledparticipants select call times to receive proactive calls or call in toll-free from the automated telemedicine system. Each week participantsreceive a rotating set of prevention-focused queries, narratives, and texts (e.g., on diet, exercise, breastfeeding, and baby care). If a participantenters a value predefined as “out of range,” participants also hear recorded first person supportive narratives related to their “out-of-range”reply encouraging behavior change as well as narratives offering shorter tips. (3) Each participant is matched with a health coach, a trainednonprofessional individual recruited for this study. The health coach is trained on health coaching protocol and diabetes prevention (Centerfor Excellence in Primary Care). The coach receives automatically downloaded daily reports from the ATSM calls and participant responses.Depending on the participant’s needs, the health coach calls back to provide participant with emotional support and engage participant ingoal setting/action and provides information about community resources. (4) ((4a) and (4b)) The health coach can connect the patient withcommunity programs, food banks, farmers markets, WIC counselors, mental health support groups, and so forth. Additionally, the coachmay send a notification to a patient’s clinic and/or clinician if deemed urgent, based on predetermined protocols.

follow-up surveys are used to assess feasibility, acceptability,and health related outcomes (e.g., weight loss, physical activ-ity, consumption of healthy foods, breastfeeding, replacementof water for sugar sweetened drinks, and glucose screening).For the women enrolled in the health-IT arm, engagementis tracked through the HIT system, in which we monitor thecalls women are responding to and the queries they triggerfor. A health coach also monitors the participant’s progressthrough extended phone calls for resources and support.

To achieve empirical results regarding the success of theprogram, key stakeholders in our partner sites were consultedto iteratively assess the implementation of the program. Agroup of regional and national advisors was assembled to helpassess the challenges in integrating the STAR MAMA modeland to critically evaluate the feasibility and acceptability ofthis hybridHIT andhealth coachingmodel in the communitysetting. The primary advisors were from WIC, and theyincluded research staff,management, and nutritionists. Basedon discussions, reflections, and informant interviews withour advisors and partners, we were able to articulate keybarriers to our model and assess the scope of scalability.

2.5. Selection of Case Studies. Case studies were selected fromWIC sites to represent health coaching calls conducted in

community settings. Two participants from each the SanFrancisco (SF) and SonomaCountyWIC sites were chosen toreflect the diversity of women’s experiences and the diversityof coaching content.

3. Results

3.1. Participant Sociodemographic Characteristics. Table 1describes key demographic characteristics of the study pop-ulation (𝑁 = 86). Women enrolled were on average 30years old; 78% were identified as Latina or Hispanic and werenot born in the US. Migrant women lived an average of 10years in the US and 63% listed Spanish as their preferredlanguage. Of the subsample, 23% were members of the WICprogram. The women had, on average, two children belowthe age of 18 currently living in their household. Eighty-sevenpercent of women were diagnosed with gestational diabetespreviously, of which 97% were diagnosed during their mostrecent pregnancy. Thirty-six percent were obese, overweight,or experience unhealthy weight gain during their pregnancy.

3.2. STAR MAMA Engagement. Since the STAR MAMAprogram enrollment is ongoing, engagement data is represen-tative of the first wave of enrollment, accounting for almost


Table 1: Sociodemographic characteristics of currently enrolled or completed STAR MAMA participants (𝑁 = 86).

Both study arms (educationresources and HIT arm)

(𝑁 = 86)𝑁%Age (in years), mean (SD) 30.05 (5.16)Race/ethnicity, 𝑛 (%)

Asian or Pacific Islander 7 (8.2%)Black or African American 6 (7.1%)White or Caucasian 4 (4.7%)Latino (a) or Hispanic 67 (78.8%)Others 1 (1.2%)

Children currently in household under 18 years of age, mean (SD) 1.68 (1.31)Born outside US, 𝑛 (%) 66 (77.6%)

If not born in US, total years living in US, mean (SD) 10.27 (6.57)Previously diagnosed with gestational diabetes, 𝑛 (%) 72 (86.7%)

Of those with GDM, diagnosed during this pregnancy, 𝑛 (%) 67 (97.1%)Previously diagnosed with overweight, obese, or unhealthy weight gain, 𝑛 (%) 30 (36.1%)

Of those overweight, obese, or unhealthy weight gain, diagnosed during this pregnancy, 𝑛 (%) 10 (71.4%)Preferred language, 𝑛 (%)

English 32 (37.2%)Spanish 54 (62.8%)

WIC status, 𝑛 (%)Non-WIC 66 (76.7%)WIC 20 (23.3%)

half of the recruitment target. Among the study subsample(𝑁 = 86), twenty-eight women have completed all 20 weeksof the HIT and health coaching program. Of those women,89% answered at least 1 phone call, with an average of 12 totalphone calls answered out of 20 weekly calls. On average, thewomen answered 61% of calls during the intervention period.Among all women randomized to the STARMAMAHIT andhealth coaching arm (𝑛 = 43), excluding those who withdrewfrom the study or are lost to follow-up, 86% have answered atleast one phone call till date (Figure 4).

3.3. Case Studies from San Francisco WIC and SonomaCounty WIC. Table 2 presents four case studies from SF andSonoma country WIC sites. The women receiving the STARMAMA HIT program through weekly calls were at high riskand triggered for poor physical activity, high carbohydrate,fat or sugar consumption, signs of depression of feeling“overwhelmed,” and more. The length of the health coachingcallbacks ranged from short follow-ups of 3 minutes to upto 45 minutes in some cases, with an average of 9 minutesper call. Topics covered ranged from the health-IT phone callnarrative topics (diet control, physical activity, depression,cutting back on high fat and sugary foods, breastfeeding,and bottle-feeding) to miscellaneous health needs of theparticipants. In Table 2, we describe synopses of coachingcalls with relevant actionable items, such as follow-up topics,and community or clinic implications.

These stories illustrate accounts of coaching to womenwho are representative of the enrolled participants in theSTAR MAMA program. It is evident that the health coachserves as a connection between the woman postpartum andcritical resources, including information, basic knowledgeand tips about postpartum care, and links to their preferredprimary care system. Moreover, the health coach is a keypoint of support not only for the health-IT call-based topics,but also for miscellaneous questions the new mother hasdoubts about. Figure 3 demonstrates how theHIT componentintegrates with health coaching and how the coaches usethe participant-driven triggers to direct coaching calls anddiscuss specific and relevant topics during a session.

These vignettes illustrate the depth and breadth of issuescovered by health coaches during their interactions withparticipants. They highlight major themes and barriers toself-management postpartum including (but not limited to)need for improved resources for child care after delivery,reinforcement for reduction of sugar and fat consumption,goal setting and action planning to improve physical activity,reminders about the importance of follow-up blood sugartesting postpartum and postpartum depression screening.

4. Discussion

Because high risk and low-income postpartum women oftendo not receive the longitudinal care and support they need to


4321 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Weeks (1–20)

Participant trigger

Week Reason Health coaching key messages

5 (i) Participant was not able to exercise at least 4 times during the week

(ii) Participant had fast foods more than 3 times in the past week

(i) Review importance of physical activity to improve cardiovascular health and create exercise action plan

(ii) Provide quick and easy recipes as a tool to avoid fast food

6 (i) Participant had a hard time cutting back on fatty foods

(ii) Breastfeeding follow-up

(i) Review substitutes for fat foods(ii) Assess breastfeeding progress, frequency, baby’s feeding

pattern

8 (i) Breastfeeding follow-up call

9 (i) Participant was not able to exercise at least 4 times during the week (i) Review barriers in exercise and identify possible solutions

11 (i) Child immunizations(ii) More information about exercise tips

(i) Remind participant which immunizations baby needs

14 (i) Breastfeeding follow-up call

Figure 3: HIT enabled phone call system, participant triggers, and context-based health coaching messages: summary for Ms. C. at SanFrancisco WIC. Timeline of calls and weekly triggers indicated by Ms. C. The timeline displays the weekly phone calls to Ms. C., from weeks1 through 20 by the ATSM system. The diamonds indicate triggers and actionable events, and the table summarizes the reason for triggerseach week. A health coach monitors the daily and weekly reports from the HIT system to follow-up with the participants through a triggerbased approach.

reduce their risk of DM and other chronic illnesses, this is acritical windowof opportunity to intervene and providemax-imal resources, support, and tools for prevention of chronicillnesses such as diabetes and also help self-manage existingchronic conditions. In this paper, the example coaching callssuggest that health coaches in STAR MAMA act as a bridgebetween a participant and the primary care system to emulatea continuum of care even after delivery.

Scalable implementation of health coaching, a HITenabled model, or a hybrid HIT and coaching model canhave community and clinic specific benefits but there areseveral identified barriers to such an attempt at integration.We investigated these potential barriers with communityprograms through discussions with our regional advisors atSan Francisco and Sonoma WIC. In the following, we artic-ulate three core limitations to health coaching specificallyidentified within this specific community setting.

First, community WIC programs have boundariesregarding the services they can offer and may face restrictivefunding for programs like health coaching and low capacityto train and hire coaches. In the context of WIC, while thereis a currently funded and high functioning Peer Coachingprogram (Loving Support Peer Counseling) it focusesprimarily on breastfeeding and lactation support. As such,peer coaches within the WIC infrastructure are not trained

to coach women on critical postpartum topics, such as diet,physical activity, postpartum depression, healthy eatingtips, and family support. Moreover, not all counties withinstates have the funding allocated for the Loving SupportPeer Counseling Program, and absorbing a HIT or healthcoaching component can strain their budget.

A secondmajor barrier for health coaching in communitysettings is the limited availability of coaches who haveexpertise required within the program structures. For exam-ple, within WIC, those who provide nutrition counselingare often expert health professionals: dieticians, nurses, ortrained diabetes educators. In most sites, though there arefew of these expert health professionals available to receiveadditional training as health coaches for a broader set ofconcerns outside of nutrition. On the other hand there areoften peer coaches, who are more numerous but less welltrained, and they are not able to coach beyond a morelimited scope, as with an emphasis on breastfeeding andlactation.

Lastly, a major limitation in prevention in such high risk,low-income populations is that they are very hard to reachand follow-up with [16, 41]. Women who are at most risk forchronic illnesses like type 2 diabetes, obesity, or depressionpostpartum are often from the most vulnerable populations,who historically have transient housing situations and may


(2) HIT enabled phonecalls (3) Health coaching

(1) HIT & health coaching blend(STAR MAMA)

Clinic Community

Calls (with narratives, texts, or adapted content) can be independently pushed out through an HIT system

Interactive calls can allow patients to respond to queries, send feedback, or trigger for support

Unidirectional counseling on specific topics, pertinent to the patient’s needs

HIT phone calldependentcounseling, in which coach follows up with patient on triggered responses during ATSM phone calls

Patient

Figure 4: Multimodal adaptation of the STAR MAMA HIT/health coaching hybrid model to meet community needs. This model breaksdown the differentmodes of implementation of the STARMAMAmodel to illustrate the flexibility of supporting self-managed care within theclinic to community spectrum. Both components of the model, HIT and health coaching, have the capacity to interact uni- or bidirectionallywith the patient in the clinic or community setting. (1) STAR MAMA: the STAR MAMA intervention is a blend of weekly HIT phone callsto eligible patients and health coaching calls for support and follow-up. (2) ATSM: this is one component of the STAR MAMA model, inwhich patients receive weekly phone calls for 20 weeks on various topics regarding postpartum health. The calls can be implemented in thecommunity setting unidirectionally, in which the patient listens to educational narratives, or the phone calls can be programmed to offer aninteractive component. (3) Health coaching: trained health coaches can provide topic-based counseling to patients regarding specific topicstailored to the patient’s needs. Or the health coach can receive triggers from an HIT system (if both are used in conjunction) to follow-upwith patient on high risk issues.

have difficulty engaging in such a program. Even with a HITblended approach such as STAR MAMA, where a healthcoach counsels patients over the phone, reaching patients isthe biggest barrier in coaching. In particular, women whostart working after delivery are the most difficult to contact,with their irregular and often hectic schedules. However, theperiod after delivery is a very critical time, when womenrequire the most support to adjust to their physical andmental changes after having a baby.

5. Conclusion

It is critical to consider the positive impacts of healthcoaching and health-IT interventions in the clinical settingsand develop techniques to execute these strategies in thecommunity. Our preliminary conversations with key stake-holders (County WIC staff, STAR MAMA health coaches,and National WIC advisors) have outlined the scope ofintegration of these interventions in the community set-tings and have addressed the need to expand care through

methods like telemedicine. A model like STAR MAMA maybe daunting to implement within a community structure;particularly when funding is limited, scopes of practiceare restricted, or participants are hard to reach. However,frameworks of self-care management and behavior changeusing HIT, health coaching, or both will be extremely ben-eficial in the future to improve preventive health practicesin communities and potentially mitigate the disease burdenthat a safety net hospital or community clinic may face. Weneed more innovative programs to bridge counseling andresources between the patient and provider, facilitated byhealth information technology. While challenges persist, theflexibility of these interventions and evidence-based successin the clinical setting urges expanding care to communityprograms.

Disclosure

Thecontent of this publication does not necessarily reflect theviews or policies of the US Department of Agriculture nor


Table2:Health

coaching

case

studies:San

FranciscoandSono

maW

ICparticipants.

Site

Case

studies:sum

marieso

fcoachingcalls

with

wom

enenrolledin

theH

ITandcoaching

arm

Health

coaching

actio

nsEx

amples

ofrangeo

feffo

rtsu

ndertakento

addressc

omplex

emotional,health

literacy,and

instr

umentaln

eeds

SanFranciscoWIC

(1)M

s.C.,a

33-year-old,Latin

awom

anwith

arecent

histo

ryof

diet-con

trolledGDM

received

herfi

rsth

ealth

coachcalldu

ringweek1o

fthe

STARMAMAstu

dy,w

henshew

as8weeks

postp

artum.She

isah

omem

aker

andno

tmarrie

d,bu

tlivingwith

apartner

ina

marria

ge-like

relationship.Ms.C.

delivered

her

baby

at39

weeks.W

hilesheu

ndersto

odkeyste

psandrequ

irementsforb

abycare,she

repo

rted

feeling

overwhelm

eddu

etothes

tressof

carin

gforh

erthreeo

ther

child

renas

well.In

additio

nto

repo

rted

practic

alsupp

ortand

feeling

likes

hehad

someone

tolistento

her;theh

ealth

coach

provided

supp

ortand

identifi

edthatshew

asno

tsufferin

gfro

mdepressio

n.In

thethird

week,Ms.

C.repo

rted

mixed-fe

edingforh

erbaby

with

breastmilk

andpu

mpedmilk

,ashift

from

her

exclu

siveb

reastfe

edingin

thep

asttwoweeks.H

erhealth

coachencouraged

hertoexclu

sively

breastfeed

wheneverp

ossib

leandreview

edthe

impo

rtance

ofbreastmilk

fora

grow

inginfant.

With

thisandothersup

ports,Ms.C.,w

horepo

rted

high

intentionto

breastfeed

priorto

pregnancy,was

ableto

eventuallycontinue

with

outformulafor

thefi

rst6

mon

ths.Duringthe

fifthweek,Ms.C.

repo

rted

bing

ingon

unhealthy

snacks:sod

a,sw

eets,

andfood

sfrom

thelocal

taqu

eria.A

fterq

ueryingabou

ther

symptom

s,her

health

coachwas

concernedthatshed

isplay

edsig

nsof

elevatedbloo

dsugar.Shed

iscussedthe

dangerso

fahigh

fatand

high

sugard

ietand

encouraged

hertoreplaces

odaw

ithwater.

Together,theysetgoalsandherh

ealth

coach

follo

wed

upweeklyto

assessherimplem

entatio

nof

hera

ctionplan.A

ddition

ally,

herc

oach

helped

Ms.C.

makea

nappo

intm

entw

ithprim

arycare

provider

togeth

erbloo

dsugarrechecked

(i)Supp

ortiv

ecou

nselingandprovision

ofpo

stpartum

stressm

anagem

entstrategies

(ii)E

ncou

ragemento

fsustained

breastfeedingfor

first6mon

thsa

ndinstructionforsafe

bottle-feeding

(iii)As

sessmento

fpartic

ipant-s

pecific

barriersto

redu

cing

ahighsugara

ndhigh

fatd

iet;developa

seto

fresou

rces

andtoolstosharew

ithparticipant

form

anagingpo

stpartum

eatin

ghabitsforself

andfamily

(iv)A

ssistance

with

reconn

ectin

gwith

the

prim

arycare

setting

(com

mun

ityclinico

rho

spita

l)to

follo

w-upon

herh

ealth

status

(i)Participantw

asun

certainabou

tthe

meaning

ofherscreening

testresultandwhattodo

next

(ii)P

artic

ipantb

enefitedfro

mtailo

redbreast

feedingsupp

ortw

ithin

thec

ontext

ofthes

tresses

encoun

teredwith

multip

lechild

ren’s

needsto

addressb

esides

baby


Table2:Con

tinued.

Site

Case

studies:sum

marieso

fcoachingcalls

with

wom

enenrolledin

theH

ITandcoaching

arm

Health

coaching

actio

nsEx

amples

ofrangeo

feffo

rtsu

ndertakento

addressc

omplex

emotional,health

literacy,and

instr

umentaln

eeds

SanFranciscoWIC

(2)M

s.F.,

a21-y

ear-oldLatin

awom

anwith

arecent

histo

ryof

diet-con

trolledGDM

durin

gher

pregnancy.Duringpregnancyshew

orked

part-time(<20

hours)in

food

deliveryandwas

notm

arrie

dbu

tlivingwith

onlyherp

artner

and

pregnant

with

herfi

rstchild.She

received

herfi

rst

health

coaching

calldu

ringherw

eek1o

fenrollm

entinST

ARMAMA,8

weeks

postp

artum,w

hensher

eportedfeelinglik

eshe

couldno

tdoallthe

things

shen

eededforh

erbaby.She

was

occupied

with

herb

aby’s

belly

butto

n,which

shetho

ught

looked

abitabno

rmal,

andwas

denied

afollow-upappo

intm

entsince

the

baby’sMediCalwas

inactiv

e.Her

health

coach

inform

edMs.F.abou

tthe

different

MediCal

managed

care

plansa

ndadvisedhero

nho

wto

commun

icatew

ithMediCalandsw

itchherb

aby

toag

oodplan.D

uringpregnancy,Ms.F.repo

rted

high

intentionto

breastfeedandmainlybreastfed

herb

aby,supp

lementedsometim

eswith

form

ula.

Her

health

coachreinforced

theimpo

rtance

ofexclu

siveb

reastfe

edingandoff

ered

abreastp

ump

from

WIC

forM

s.F.to

borrow

.Inthefollowing

weeks,M

s.F.contactedMediCalandwas

ableto

geth

ercase

review

ed.Th

ough

shew

antedto

start

herb

abyon

solid

food

s,herh

ealth

coach

suggestedwaitin

gun

tiltheb

abywas

approaching

5-6mon

thsa

ndsher

eviewed

ther

iskso

fstarting

solid

food

spreem

ptively

.Ms.F.was

motivated

tofollo

wtheser

ecom

mendatio

nsandtake

full

precautio

nwhenfeedingherb

aby.Shea

lsocle

ared

herd

oubtsa

bout

babies

burpingandfat

consum

ptionwith

herh

ealth

coach

(i)Provides

uppo

rtandkn

owledgeo

npo

stpartum

baby

care

andtim

emanagem

ent

(ii)G

uide

participantw

ithactiv

ationand

follo

w-upof

baby’sMediCalplan

(iii)Re

view

andinstr

uctp

artic

ipanto

nprop

ercomplem

entary

feedingpractic

es(timing,whatto

start,fre

quency,assessin

ghu

nger

cues)

(iv)A

nswer

misc

ellaneou

squestions

andprovide

long

itudinalsup

porton

vario

usdo

ubts(belly

butto

n,baby

burping,etc.)

thatparticipantm

ayhave

asan

ewmother

(i)Participantn

eededsupp

ortand

guidance

onho

wto

renewherb

aby’s

MediCalplan

tofacilitate

continuedcare

(ii)P

articipantreceivedextras

uppo

rtand

resourcesfrom

theh

ealth

coach(i.e.,

breastpu

mp

from

WIC)toenableexclu

siveb

reastfe

edingand

instructions

onho

wandwhento

start

complem

entary

feeding


Table2:Con

tinued.

Site

Case

studies:sum

marieso

fcoachingcalls

with

wom

enenrolledin

theH

ITandcoaching

arm

Health

coaching

actio

nsEx

amples

ofrangeo

feffo

rtsu

ndertakento

addressc

omplex

emotional,health

literacy,and

instr

umentaln

eeds

Sono

maC

ountyWIC

(3)M

s.H.,a3

3-year-old

Latin

awom

anwas

enrolledin

thes

tudy

at7weeks

postp

artum

and

received

health

coaching

calls

durin

gherfi

fthweekin

thes

tudy.She

isah

omem

aker

and

marrie

danddelivered

herfou

rthchild

atfull-term

.Her

health

coachgave

herideas

abou

tencouragingchild

rento

eath

ealth

ierfoo

ds,that

is,fruitsandvegetables,since

sheh

adyoun

ger

child

renin

herfam

ily(tw

ochild

ren<5yearso

ld,

inadditio

nto

then

ewbaby)w

howerefussy

eaters.She

was

nottoo

keen

onexercisin

g,bu

ther

health

coachapplaudedherfor

trying

atleast

once

aweekandencouraged

hertoexercise

more

frequ

ently

which

was

agreataccomplish

ment;

thou

ghatho

meM

s.H.felt

thatsheh

adsomeone

tolistento

andcomforther,shed

idno

tfeelthat

sheh

adsupp

ortw

ithpracticalhelp.D

uringthe

15th

week,Ms.H.strug

gled

with

cutting

back

onfatty

food

sand

sugary

drinks,likes

odas.H

erhealth

coachworkedwith

hertomakea

plan

and

incorporateq

uick

tipstoaddressthese

issues,

such

asdraining

fatd

uringcook

ingandmaking

homem

adea

guafrescas.H

erhealth

coachalso

provided

herw

ithmanylocal,commun

ity-based

food

resources.Ms.H.w

asleftfeelingsupp

orted,

motivated,and

confi

dent

inhera

bilityto

make

changes(seeF

igure3

)

(i)Providetipso

nmanagingotherc

hildren,

especiallyencouraginghealthyhabitsforfussy

eaters

(ii)Improvek

nowledgea

bout

exercise

benefits

andsuggestrecom

mendatio

nsandstrategies

tointegratep

hysic

alactiv

ityinto

participant’s

daily

schedu

le(iii)Disc

usstechn

iquestosubstituteh

igh

fat/s

ugar

food

swith

healthiera

lternatives

(i)Participantn

eededinform

ationandsupp

ort

onho

wto

controlthe

dietso

ffussy

eaters

(ii)P

artic

ipantn

eededreinforcem

ento

nexercise

andhealthiersub

stitutes

(iii)Participantreceivedrelevant

inform

ationon

localfoo

dbank

sand

commun

ityresourcesto

improves

elf-m

anagem

entand

care


Table2:Con

tinued.

Site

Case

studies:sum

marieso

fcoachingcalls

with

wom

enenrolledin

theH

ITandcoaching

arm

Health

coaching

actio

nsEx

amples

ofrangeo

feffo

rtsu

ndertakento

addressc

omplex

emotional,health

literacy,and

instr

umentaln

eeds

Sono

maC

ountyWIC

(4)M

s.G.isa

24-year-oldLatin

awom

an.She

isa

homem

aker,livingwith

apartner

inalarge

householdof

12individu

als.Th

ough

sher

eports

feeling

likeo

nces

hehadtheb

abyshew

asableto

geth

elpwith

cook

ingandothertasks,she

didno

tfeelsupp

ortedem

otionally.H

erbaby

was

born

full-term

inav

aginalbirth.Sh

ewas

enrolledinto

thes

tudy

andsta

rted

receivinghealth

coaching

calls

at7weeks

postp

artum.D

ueto

poor

latching

,Ms.G.breastfe

dforjust3

weeks

andstartedwith

form

ulam

ilkforh

erbaby.H

erhealth

coach

instructed

hero

nsafebo

ttle-feedingpractic

esand

review

edhu

nger

cues

toprevento

verfe

edingand

recogn

izes

igns

offulln

ess.Aroun

dthes

ixth

week,Ms.G.reportedfeeling

overwhelm

ed,tire

d,anddepressed.Her

health

coachdiscussedthe

common

nessof

baby

bluesa

fterd

elivery

and

relevant

symptom

sand

encouraged

Ms.G.to

speakwith

herp

rimarycareprovider

abou

tthisin

heru

pcom

ingappo

intm

ent.In

follo

w-upcalls,

Ms.G.w

asableto

approp

riatelyrecogn

izeh

unger

cues

andherm

oodim

proved

after

taking

her

multiv

itaminsa

ndeatin

gprop

erly

(i)Re

view

safebo

ttle-feedingpractic

esand

supp

orto

nadaptin

gto

baby

hunger

cues

(ii)P

rovide

emotionalsup

portandreassurance

abou

tbabyblues

(iii)Facilitatec

onnectionbetweenparticipantand

prim

arycare

provider

postp

artum

toackn

owledgeissueso

fpostpartum

depressio

n

(i)Participantreceivedpreliminarypo

stpartum

depressio

nscreening,so

shec

ould

beapprop

riatelydirected

toap

rimarycare

provider

(ii)P

articipantw

asun

certainabou

tbabyhu

nger

cues

andprop

erbo

ttle-feedingpractic

es


mention trade names, commercial products, or organizationswhich implies endorsement by the US Government. Thecontent is solely the responsibility of the authors and does notnecessarily represent the official views of the NIH.

Competing Interests

All of the authors declare that they have no competinginterests.

Acknowledgments

The authors acknowledge the National Institutes of HealthsNational Center on Minority Health and Health DisparitiesP60MD006902 and the University of California Institute forMexico and the United States (UC MEXUS) and ResearchProgram on Migration and Health (PIMSA) for fundingthis paper. This project has been funded at least in partby federal funds from the US Department of Agriculture;the National Institutes of Healths National Institute of Dia-betes and Digestive and Kidney Diseases P30DK092924;the National Center for Advancing Translational Sciences,National Institutes of Health, through UCSF-CTSI Grantno. UL1 TR000004. The authors want to acknowledge thesupport of the following individuals for facilitating theimplementation of the study: Jennifer Cherry, Chris Bekins,Patricia Ibarra, Emily Melaugh, Rubi Luna, Rebecca Jones-Munger,Magdalene Louie,Monika Rubin, and all supportingstaff andmanagers at the SonomaCounty and SFWICoffices.

References

[1] J. D. Parker and B. Abrams, “Differences in postpartum weightretention between black and white mothers,” Obstetrics andGynecology, vol. 81, no. 5, pp. 768–774, 1993.

[2] L. O.Walker, J. H. Freeland-Graves, T.Milani et al., “Weight andbehavioral and psychosocial factors among ethnically diverse,low-income women after childbirth: I. Methods and context,”Women & Health, vol. 40, no. 2, pp. 1–17, 2004.

[3] I. P. Mathieu, Y. Song, and S. M. Jagasia, “Disparities in post-partum follow-up in women with gestational diabetes mellitus,”Clinical Diabetes, vol. 32, no. 4, pp. 178–182, 2014.

[4] E. Reece, G. Leguizamon, and A. Wiznitzer, “Gestationaldiabetes: the need for a common ground,” Obstetric AnesthesiaDigest, vol. 30, no. 2, pp. 84–85, 2010.

[5] C. Kim, “Gestational diabetes: risks, management, and treat-ment options,” International Journal of Women’s Health, vol. 2,pp. 339–351, 2010.

[6] M. Stasenko, Y. W. Cheng, T. McLean, A. C. Jelin, L. Rand,and A. B. Caughey, “Postpartum follow-up for women withgestational diabetes mellitus,”American Journal of Perinatology,vol. 27, no. 9, pp. 737–742, 2010.

[7] A. Tovar, L. Chasan-Taber, E. Eggleston, and E. Oken, “Post-partum screening for diabetes among women with a history ofgestational diabetesmellitus,” Preventing Chronic Disease, vol. 8,no. 6, article A124, 2011.

[8] California Department of Public Health (CDPH), “Maternal,child and adolescent health program, California diabetes andpregnancy program: sweet success,” in Gestational DiabetesMellitus, Fact Sheet, CDPH, Sacramento, Calif, USA, 2008.

[9] A. B. Caughey, Y. W. Cheng, N. E. Stotland, A. E. Washington,and G. J. Escobar, “Maternal and paternal race/ethnicity areboth associated with gestational diabetes,” American Journal ofObstetrics and Gynecology, vol. 202, no. 6, pp. 616.e1–616.e5,2010.

[10] M. M. Hedderson, J. A. Darbinian, and A. Ferrara, “Disparitiesin the risk of gestational diabetes by race-ethnicity and countryof birth,” Paediatric and Perinatal Epidemiology, vol. 24, no. 5,pp. 441–448, 2010.

[11] A. M. N. Renzaho, H. Skouteris, and J. Oldroyd, “Prevent-ing gestational diabetes mellitus among migrant women andreducing obesity and type 2 diabetes in their offspring: a callfor culturally competent lifestyle interventions in pregnancy,”Journal of the American Dietetic Association, vol. 110, no. 12, pp.1814–1817, 2010.

[12] L. Bellamy, J. Casas, A. Hingorani, and D. Williams, “Type 2diabetes mellitus after gestational diabetes: a systematic reviewand meta-analysis,”Obstetric Anesthesia Digest, vol. 30, no. 2, p.85, 2010.

[13] A. Ferrara and S. F. Ehrlich, “Strategies for diabetes preventionbefore and after pregnancy in women with GDM,” CurrentDiabetes Reviews, vol. 7, no. 2, pp. 75–83, 2011.

[14] A. Ferrara, M. M. Hedderson, and C. L. Albright, “A pregnancyand postpartum lifestyle intervention in women with gesta-tional,” Diabetes Care, vol. 34, no. 7, pp. 1519–1525, 2011.

[15] S. A. Boren, “A review of health literacy and diabetes: opportu-nities for technology,” Journal of Diabetes Science and Technol-ogy, vol. 3, no. 1, pp. 202–209, 2009.

[16] R. M. Davis, A. D. Hitch, M. M. Salaam, W. H. Herman, I. E.Zimmer-Galler, and E. J. Mayer-Davis, “TeleHealth improvesdiabetes self-management in an underserved community: dia-betes telecare,” Diabetes Care, vol. 33, no. 8, pp. 1712–1717, 2010.

[17] B. B. Brodey, C. S. Rosen, I. S. Brodey, B. Sheetz, and J. Unutzer,“Reliability and acceptability of automated telephone surveysamong Spanish- and English-speaking mental health servicesrecipients,”Mental Health Services Research, vol. 7, no. 3, pp. 181–184, 2005.

[18] D. S. Mastrogiannis, E. Igwe, and C. J. Homko, “The role oftelemedicine in the management of the pregnancy complicatedby diabetes,”CurrentDiabetes Reports, vol. 13, no. 1, pp. 1–5, 2013.

[19] J. M. Nicklas, C. A. Zera, E. W. Seely, Z. S. Abdul-Rahim, N. D.Rudloff, and S. E. Levkoff, “Identifying postpartum interventionapproaches to prevent type 2 diabetes in women with a historyof gestational diabetes,” BMC Pregnancy and Childbirth, vol. 11,no. 1, article 23, 2011.

[20] L. R. Buis, L. Hirzel, S. A. Turske, T. R. Des Jardins, H. Yarandi,and P. Bondurant, “Use of a text message program to raise type2 diabetes risk awareness and promote health behavior change(Part I): assessment of participant reach and adoption,” Journalof Medical Internet Research, vol. 15, no. 12, article e281, 2013.

[21] C. Y. Osborn and S. A. Mulvaney, “Development and feasibilityof a text messaging and interactive voice response interventionfor low-income, diverse adults with type 2 diabetes mellitus,”Journal of Diabetes Science and Technology, vol. 7, no. 3, pp. 612–622, 2013.

[22] Diabetes Prevention ProgramResearch Group, “10-Year follow-up of diabetes incidence and weight loss in the DiabetesPrevention Program outcomes study,” The Lancet, vol. 374, no.9702, pp. 1677–1686, 2009.

[23] O. Verier-Mine, “Outcomes in women with a history of gesta-tional diabetes. Screening and prevention of type 2 diabetes.


Literature review,” Diabetes & Metabolism, vol. 36, no. 6, pp.595–616, 2010.

[24] D.Thom, H. Hammer, F.Wang et al., “Seeing in 3-D: examiningthe reach of diabetes self-management support strategies in apublic health care system,” Health Education & Behavior, vol.35, no. 5, pp. 664–682, 2006.

[25] M. A. Handley, M. Shumway, and D. Schillinger, “Cost-effectiveness of automated telephone self-management supportwith nurse care management among patients with diabetes,”Annals of Family Medicine, vol. 6, no. 6, pp. 512–518, 2008.

[26] A. Ghorob, “Health coaching: eaching patients to fish,” FamilyPractice Management, vol. 20, no. 3, pp. 40–42, 2013.

[27] J. A. Long, E. C. Jahnle, D. M. Richardson, G. Loewenstein,and K. G. Volpp, “Peer mentoring and financial incentives toimprove glucose control in African American veterans,” Annalsof Internal Medicine, vol. 156, no. 6, pp. 416–424, 2012.

[28] D. H. Joseph, M. Griffin, R. F. Hall, and E. D. Sullivan,“Peer coaching: an intervention for individuals struggling withdiabetes,”TheDiabetes Educator, vol. 27, no. 5, pp. 703–710, 2001.

[29] T. C. Keyserling, C. D. Samuel-Hodge, A. S. Ammerman etal., “A randomized trial of an intervention to improve self-carebehaviors of African-American women with type 2 diabetes:impact on physical activity,” Diabetes Care, vol. 25, no. 9, pp.1576–1583, 2002.

[30] M. Heisler, S. Vijan, F. Makki, and J. D. Piette, “Diabetes controlwith reciprocal peer support versus nurse care management: arandomized trial,” Annals of Internal Medicine, vol. 153, no. 8,pp. 507–515, 2010.

[31] D. H.Thom, A. Ghorob, D. Hessler, D. de Vore, E. Chen, and T.A. Bodenheimer, “Impact of peer health coaching on glycemiccontrol in low-income patients with diabetes: a randomizedcontrolled trial,” The Annals of Family Medicine, vol. 11, no. 2,pp. 137–144, 2013.

[32] T. S. Tang, A. K. Garg, P. S. Sohal, and E. Ur, “Training peersto provide ongoing diabetes self-management support (DSMS)in the english and punjabi-speaking south asian community,”Canadian Journal of Diabetes, vol. 36, no. 5, supplement, p. S7,2012.

[33] E. B. Fisher, J. A. Earp, S. Maman, and A. Zolotor, “Cross-cultural and international adaptation of peer support for dia-betes management,” Family Practice, vol. 27, supplement 1, pp.I6–I16, 2010.

[34] M. L. Goldman, A. Ghorob, S. L. Eyre, and T. Bodenheimer,“How do peer coaches improve diabetes care for low-incomepatients?: a qualitative analysis,” Diabetes Educator, vol. 39, no.6, pp. 800–810, 2013.

[35] A. Ghorob, M. M. Vivas, D. D. Vore et al., “The effectivenessof peer health coaching in improving glycemic control amonglow-income patients with diabetes: protocol for a randomizedcontrolled trial,” BMC Public Health, vol. 11, no. 1, article 208,2011.

[36] D. Margolius, T. Bodenheimer, H. Bennett et al., “Healthcoaching to improve hypertension treatment in a low-income,minority population,” Annals of Family Medicine, vol. 10, no. 3,pp. 199–205, 2012.

[37] “Women, Infants, and Children (WIC),” Women, Infants, andChildren (WIC), 2016.

[38] F. Castellanos-Brown, K. Service, N. Epstein, C. Collins, and A.Associates, “WIC Breastfeeding Peer Counseling Study Phase II: Follow-up Implementation Report,” 2015.

[39] K. K. Nielsen, A. Kapur, P. Damm, M. de Courten, and I.C. Bygbjerg, “From screening to postpartum follow-up—thedeterminants and barriers for gestational diabetes mellitus(GDM) services, a systematic review,” BMC Pregnancy andChildbirth, vol. 14, no. 1, article 41, 2014.

[40] M.A.Handley, E.Harleman, E.Gonzalez-Mendez et al., “Apply-ing the COM-B model to creation of an IT-enabled healthcoaching and resource linkage program for low-income Latinamoms with recent gestational diabetes: the STAR MAMAprogram,” Implementation Science, vol. 11, article 73, 2015.

[41] M. J. Baker-Ericzen, C. D. Connelly, A. L. Hazen, C. Duenas,J. A. Landsverk, and S. M. Horwitz, “A collaborative caretelemedicine intervention to overcome treatment barriers forLatina women with depression during the perinatal period,”Families, Systems & Health, vol. 30, no. 3, pp. 224–240, 2012.

Research ArticlePilot Study of a Web-Delivered Multicomponent Interventionfor Rural Teens with Poorly Controlled Type 1 Diabetes

Amy Hughes Lansing,1 Catherine Stanger,1 Alan Budney,1

Ann S. Christiano,2 and Samuel J. Casella2

1Department of Psychiatry, Geisel School of Medicine at Dartmouth, Lebanon, NH 03766, USA2Children’s Hospital at Dartmouth, Dartmouth-Hitchcock Medical Center, Lebanon, NH 03766, USA

Correspondence should be addressed to Amy Hughes Lansing; [email protected]

Received 3 May 2016; Accepted 25 July 2016

Academic Editor: Hiroshi Okamoto

Copyright © 2016 Amy Hughes Lansing et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Objective. The purpose of this study was to examine the feasibility and effectiveness of a web-delivered multicomponent behavioraland family-based intervention targeting self-regulation and self-monitoring of blood glucose levels (SMBG) and glycemic control(HbA1c) in teens with type 1 diabetes (T1DM) living in rural US.Methods. 15 teens with poorly controlled T1DM participated in a25-week web-delivered intervention with two phases, active treatment (weekly treatment sessions and working memory trainingprogram) and maintenance treatment (fading of treatment sessions). Results. Almost all (13 of 15) participants completed at least 14of 15 treatment sessions and at least 20 of 25 working memory training sessions. SMBG was increased significantly at end of activeand maintenance treatment, and HbA1c was decreased at end of active treatment (𝑝’s ≤ 0.05). Executive functioning improved atend of maintenance treatment: performance on working memory and inhibitory control tasks significantly improved (𝑝’s ≤ 0.02)and parents reported fewer problems with executive functioning (𝑝 = 0.05). Improvement in inhibitory control was correlatedwith increases in SMBG and decreases in HbA1c. Conclusions. An innovative web-delivered andmulticomponent intervention wasfeasible for teens with poorly controlled T1DM and their families living in rural US and associated with significant improvementsin SMBG and HbA1c.

1. Introduction

Management of type 1 diabetes involves the completion ofmultiple daily adherence behaviors that may be complex andoften disruptive to daily life (e.g., blood glucose checking atleast four times per day, correctly calculating, and admin-istering insulin doses). Adolescents with type 1 diabetesoften struggle to maintain adherence to the recommendedfrequency of self-monitoring of blood glucose (SMBG) andachieve optimal metabolic control (HbA1c). Interventionstargeting adherence in youth with type 1 diabetes, includingthose with coping skills, motivational, cognitive behavioral,and family systems components, have typically shown onlysmall to moderate effect size improvements in adherencebehaviors and HbA1c [1, 2]. Research has suggested thatdeficits in self-regulation and underlying executive functionare associated with poorer adherence and higher HbA1c in

teens with type 1 diabetes [3, 4]. The development of inter-ventions that specifically target self-regulation and executivefunction may be particularly beneficial for youth with type 1diabetes [5]. Further, among studied interventions targetingadherence, none have specifically targeted a rural populationand only one has been translated into a web-delivered format,behavioral family systems therapy for diabetes [6, 7]. Thus,there is a need to develop more effective behavioral interven-tions for nonadherence and, in particular, interventions thatcan be delivered to families in rural regions with limitedaccess to specialized behavioral health and pediatric endo-crinology services [8].

This paper describes a pilot study of a multicomponentintervention that integrated a behavioral and family-basedtreatment [9] with a working memory training program topromote youths’ developing self-regulation, encourage opti-mal diabetes management, and improve glycemic control.


http://dx.doi.org/10.1155/2016/7485613


Specifically, this intervention provided (1) behavior economicincentives to teens for daily self-monitoring of blood glucose(SMBG) and to parents for parental monitoring of SMBGand (2) cognitive behavioral andmotivational therapy [9, 10],along with (3) a working memory training program forteens [11].The intervention included both an active treatment(ActiveTX) phase where incentives and therapy were deliv-ered weekly along with the working memory training pro-gram, as well as a maintenance treatment phase (MaintTX)where incentives and therapy were faded, with an increasinglength of time between sessions.

This intervention was based on a previously pilotedbehavioral and family-based intervention for teens withpoorly controlled type 1 diabetes [9]; however, this prior studydid not include a maintenance phase or working memorytraining and was delivered through face-to-face sessions ina metropolitan region in southern US. In this previous pilotstudy, an inclusion criterion for participation was livingwithin 30miles of the treatment center to ensure that familiescould reasonably and regularly attend the weekly therapysessions. In translating this interventionmodel for delivery ina more rural region of the US, it was necessary to implementthe intervention without using a clinic-based model. Web-delivery of the entire intervention, including both the treat-ment sessions and working memory training program, wasselected as broadband Internet penetration in the region wasaround 75% of households and expected to increase duringthe course of the trial. The distance many families lived fromthe pediatric endocrinology clinic and the treatment intensity(3 months of weekly sessions and then 3 months of fadingtreatment sessions) made a home-based visit model incom-patible with providing a time and cost-effective treatment.

The primary components of this multicomponent inter-vention were selected for efficacy in targeting self-regulationof health behaviors in teens. Behavior economic interventionapproaches are promising new methods for improving self-regulation of health behaviors. Behavioral economic incen-tives (BEI) are grounded in neuroeconomic theory [12],which purports that self-regulation failure involves overlyvaluing immediate rewards and devaluing future rewards.There is an expanding literature on incentive use to increase awide range of health behaviors [13] such as increasing SMBGamong teenswith type 1 diabetes [9, 14] and losingweight [15].Interventions like BEI that reward the initiation of healthybehaviors that lack immediate inherent reward, for example,SMBG, may improve self-regulation and facilitate the devel-opment of more habitual completion of healthy behaviors.Further, adolescence may be an ideal time to use incentivesto improve adherence. One study has shown that the useof rewards facilitates self-regulation among adolescents to agreater extent than for adults [16], suggesting that adolescents’neural functions might be more influenced by immediaterewards than adults. Thus, incentivizing SMBG, an oftenunpleasant and inherently unrewarding health behavior maybe an important tool for improving self-regulation as well asincreasing SMBGand in turn improvingHbA1c in youthwithpoorly controlled type 1 diabetes.

This intervention model also provided behavior eco-nomic incentives to parents towards increasing parental

monitoring of adolescents’ SMBG behaviors. Parents arecritical to the development of adolescent self-regulation, andin particular, parental monitoring has been identified as a keypredictor of adherence and glycemic control in youth withtype 1 diabetes [17–19]. Incentivizing parentalmonitoring andparental implementation of behavioral contracts for youthSMBG may help to establish family environments that aremore supportive of the development of youth self-regulatoryskills for diabetes management. Such family environmentswould be more conducive to effective parent-child collab-oration in diabetes management that is seen in families ofadolescents with more optimal glycemic control [17, 20].

Cognitive training interventions, in particular workingmemory training, are associated with improvements in exec-utive functioning in youth and may also benefit teens withtype 1 diabetes. Working memory is a cognitive system thatactively holds information in the mind permitting verbaland nonverbal activities such as reasoning and compre-hension processing [21]. Working memory is an executivefunction that involves goal-oriented active monitoring ormanipulation of information. Further, performance deficitson inhibitory control and risky decision-making tasks are alsorelated to working memory capacity [22]. Studies show thatexecutive function deficits, including those related toworkingmemory, are associated with adherence and metabolic con-trol in youth with type 1 diabetes [3, 23].

Working memory training involves practice of increas-ingly difficult verbal and visuospatial tasks requiring the tem-porary storage and manipulation of information. Workingmemory training aims to improve executive function anddecision-making by strengthening working memory neuro-cognitive processes through practice.There is also increasingevidence demonstrating that commercially available com-puterized working memory training programs can reliablyenhance executive function in youth with ADHD, teens withextremely low birth weight history, and teens with moderatecognitive deficits [24–26]. Working memory training notonly improves working memory performance, but has alsobeen found to enhance performance on other cognitive tasksthat have not been trained including decision-making [27].Thus, working memory training may be a useful tool forimproving executive function and adherence in youth withpoorly controlled type 1 diabetes and was integrated into thispilot intervention model.

A primary aim of this pilot study was to examine the fea-sibility of providing a web-delivered multicomponent inter-vention in a rural region of the US. Frequency of exclusiondue to lack of broadband Internet as well as treatment com-pletion rates is explored. With regard to our secondary aimto examine treatment efficacy, we hypothesized that (1) teenand parent participation in the intervention would increaseteen SMBG and decrease HbA1c and that changes in SMBGand HbA1c at the end of ActiveTX would be maintained atthe end of MaintTX, (2) participation in the interventionwould improve teen executive functioning on both objectivebehavioral and parent report measures, and (3) changes inexecutive functioning would be associated with changes inSMBG and HbA1c.


2. Methods

Dartmouth College’s Institutional Review Board approvedthe study. Fifteen teens (47% female, 60% using an insulinpump, average age = 15.8 years, range = 13.6–17.5 years,average length of diagnosis = 6.4 years) with poorly con-trolled type 1 diabetes were recruited from a pediatric type 1diabetes clinic. 12 of 15 adolescents lived in Health Resourcesand Services Administration (HRSA) defined rural regions.Although the 3 remaining participants lived just outside ofa HRSA defined rural region, they each travelled at least 90miles to reach their pediatric endocrinology clinic. Inclusioncriteria included age 13–17, average HbA1c ≥ 64mmol/mol(8%) for past 6 months, most recent HbA1c ≥ 64mmol/mol(8%), type 1 diabetes duration >18 months, at least oneparent/guardian participant, and a computer with broadbandInternet at home. Exclusion criteria included pregnancy andsevere medical or psychiatric illness. All teens were providedwith a meter and testing strips during the course of theirparticipation in the study. There were 9 teens screened forthe program that were eligible but declined to participatedue to lack of interest or being too busy at the time. Therewas not a significant difference in HbA1c at the date ofscreening between those teens that participated and thosethat declined (𝑡(22) = −0.73, p = 0.47). Additionally, therewere 3 teens that were otherwise eligible for the study butwere excluded due to not having a computer with broadbandInternet at home. Families were loaned a web camera if theydid not have working cameras on their PCs or laptops. Intakeand follow-up assessments were conducted in person at theendocrinology clinic.

2.1. Intervention

2.1.1. Active Treatment (ActiveTX). The 11-week ActiveTXincluded weekly behavior economic incentives (BEI), briefmotivational enhancement and cognitive behavioral therapy(MET/CBT) sessions, and workingmemory training (WMT)all delivered over the Internet.

For teens, BEI involved a 2-week baseline phase whenteens received $10 per week for uploading their blood glucosemeters to a personal blood glucose datamanagement website,Carelink. From week 3 to week 11, weekly incentives wereearned for meeting the SMBG goal, testing ≥5 times daily(>2 hours apart), on an increasing number of days per week.Incentives escalated from $10 to a maximum of $30 and aweekly $5 bonus for exceeding the number of days meetingthe goal. In week 3, the initial target for days meeting theSMBG goal to earn incentive was individualized, set at oneday more than achieved during week 2. During weeks 4–7,the target number of days increased by 1 if the prior targetwas met, up to 5 days per week, and then in weeks 8–11 thetarget was set at 5 days per week for all participants.

Parents also participated in BEI. Weekly web-deliveredsessions were used to develop and implement a home SMBGcontract specifying small daily rewards and consequencesfor teens meeting SMBG goals. To encourage parent mon-itoring and implementation of the contract, parents earnedincentives for providing daily reports to the clinic. Parent

reporting goals were always set at 5 days per week, with thesame escalating earning system, bonuses, and dollar amountsas teens.

In weeks 1–11, teens also received weekly 20-minute web-deliveredMET/CBT sessions, which coincidedwith awardingof incentives, focused on improving SMBG and other self-care behaviors usingmotivational interviewing and cognitivebehavioral principles.

Beginning in week 3 of ActiveTx, teens completedWMT,via Cogmed-RM v.2 [11]. Teens were to complete 25 WMTsessions during active treatment, 5 per week optimally. Eachsession lasted about 1 hour and included 8 different trainingtasks and then a game could be played at the end of thesession. Youth could earn up to $10 for each WMT sessioncompleted. This included $5 for completing the session ina single day and a $5 bonus for good performance in thesession, indexed as improving or maintaining performanceon 3 out of 8 training tasks. Weekly coaching calls fromresearch staff provided feedback and motivational support toteens and parents to facilitate continued improvements andcompletion of sessions.

2.1.2. Maintenance Treatment (MaintTX). The 14-weekMaintTX included fading of BEI and MET/CBT sessions.Incentives were awarded and MET/CBT sessions occurredonly on weeks 13, 16, 20, and 25. The weekly BEI rewardmagnitude remained the same ($30 per week, $5 bonus). Toencourage weekly family review of SMBG, teens and parentsearned $5 per weekly upload.

Across ActiveTX and MaintTX, the maximum BEI earn-ings for teens and parents were $845 each. The teen couldearn an additional $245 fromWMT. Incentives were remotelyloaded onto a study-provided debit card. Also, families wereencouraged, but not required, to contact their diabetes edu-cator. Educators also communicated back to the counselorif there were specific concerns and/or goals for individualpatients. Table 1 provides an overview of the treatmentmodel.

2.2. Measures

2.2.1. SMBG. To assess SMBG frequency, the total numberof SMBGs a day during the 14 days prior to each assessmentpoint was recorded to calculate an average daily frequency.Blood glucose data were gathered from blood glucosemeters.SMBGwas assessed PreTX, at the end of ActiveTX, and at theend of MaintTX.

2.2.2. HbA1c. HbA1c was assessed during endocrinologyclinic and study assessment visits.HbA1cwas assessedPreTX,at the end of ActiveTX, and at the end of MaintTX.

2.2.3. Executive Functioning. Executive functioning wasassessed pretreatment (PreTX) and at the end of MaintTX,but not at the end of ActiveTX. Measures to assess executivefunctioning were selected to capture changes in workingmemory and related changes in inhibitory control as wellas more global parent reports of executive functioning. Toassess workingmemory, the digit span subtest of theWechsler


Table 1: Treatment schedule.

Week Adolescent Parent

Active treatment(ActiveTX)

1 Incentives, CBT/MET Session Incentives, Parent Session2 Incentives, CBT/MET Session, WMT Incentives, Parent Session3 Incentives, CBT/MET Session, WMT Incentives, Parent Session4 Incentives, CBT/MET Session, WMT Incentives, Parent Session5 Incentives, CBT/MET Session, WMT Incentives, Parent Session6 Incentives, CBT/MET Session, WMT Incentives, Parent Session7 Incentives, CBT/MET Session Incentives, Parent Session8 Incentives, CBT/MET Session Incentives, Parent Session9 Incentives, CBT/MET Session Incentives, Parent Session10 Incentives, CBT/MET Session Incentives, Parent Session11 Incentives, CBT/MET Session Incentives, Parent Session

Maintenance treatment(MaintTX)

1213 Incentives, CBT/MET Session Incentives, Parent Session141516 Incentives, CBT/MET Session Incentives, Parent Session17181920 Incentives, CBT/MET Session Incentives, Parent Session2122232425 Incentives, CBT/MET Session Incentives, Parent Session

Adult Intelligence Scale (WAIS-IV) [28], for youth age 16 orolder, or theWechsler Intelligence Scale for Children (WISC-IV) [29], for youth younger than 16, was administered. Thescaled score from the digit span subtest was utilized, wherehigher scores indicated better working memory capacity. Toassess inhibitory control, theDelis-Kaplan Executive FunctionSystem [30], color word interference subtest was adminis-tered. Specifically, the Condition 3 Inhibition CompletionTime Scaled Score and Errors Scaled Score were utilized,where higher scores indicated better inhibitory control. Par-ents completed the Behavior Rating Inventory of ExecutiveFunction [31] to provide ratings of teen’s executive functioningin everyday life. The Global Executive Composite (GEC)𝑇-Score was utilized, with higher scores indicating greaterproblems in executive functioning.

2.3. Statistical Analyses. Change in SMBG, HbA1c, andexecutive functioning were assessed through paired t tests.Cohen’s d effect sizes were also calculated. For SMBG andHbA1c changes fromPreTX toActiveTX, PreTX toMaintTX,andActiveTX toMaintTXwere assessed. Executive function-ing only included measures at PreTX and MaintTX. Anal-yses to examine associations between changes in executivefunctioning, SMBG, and HbA1c utilized PreTX to MaintTXchange scores that were calculated for each measure. Pearson

correlations were used to examine associations betweenchange scores. Participants were included in all analysesregardless of the number of weeks of treatment completed. Atthe end of ActiveTX all 15 participants had HbA1c measured,while 1 participant did not provide SMBG data. At the end ofMaintTX all 15 participants had HbAlc measured and all 15provided SMBG data. One participant did not complete theexecutive functioning assessment at the end of MaintTX.

3. Results

3.1. Treatment Adherence and Incentive Earnings. Almost all(13 of 15) participants completed at least 14 out of 15 ofthe BEI + MET/CBT treatment sessions. For WMT, 14 of15 participants completed at least 20 of 25 sessions. Youthtrained an average of 3.71 times per week (SD = 1.23)and improved an average of 60% of tasks relative to priorperformance on the same task. On average, teens earned $419from BEI and $181 fromWMT, while parents earned $445.

3.2. Self-Monitoring of Blood Glucose. Figure 1(a) showsSMBG data for each participant. Compared to PreTX (3.73 ±1.70), SMBG was significantly increased at end of ActiveTX(6.92 ± 1.26; 𝑡(13) = −7.78, p < 0.001, d = 1.56), and theeffect size was large. SMBG was also significantly increased


0

2

4

6

8

10

Pretreatment Postactive Postmaintenance

Average frequency of daily SMBG

(a)

405060708090

100110120

HbA1c (mmol/mol)

Pretreatment Postactive Postmaintenance(b)

6

8

10

12

14

Digit span Inhibition completionTime

Inhibition errors

Cognitive assessments

PretreatmentPostmaintenance

(c)

555759616365

Parent reportGlobal executive composite

Pretreatment Postmaintenance

(d)

Figure 1: Individual participants’ frequency of self-monitoring of blood glucose (SMBG (a)), glycosylated hemoglobin levels (HbA1c (b))and average scores for all participants on cognitive functioning assessment and scaled scores (c) and 𝑇-scores (d).

compared to PreTX at the end of MaintTX (4.87 ± 2.45;𝑡(14) = −2.19, p = 0.05, d = 0.54), with a medium effectsize. Although significantly higher than at PreTX, SMBG atend of MaintTX was significantly decreased from the end ofActiveTX (𝑡(13) = 4.00, p = 0.002).

3.3. Glycemic Control. Figure 1(b) shows HbA1c data for eachparticipant. Compared to PreTX (80 ± 15mmol/mol (9.5 ±1.4%)), HbA1c was significantly decreased at end of ActiveTX(71 ± 15mmol/mol (8.7 ± 1.4%); 𝑡(14) = 2.74, p = 0.02. d =0.62), with a medium effect size. Six participants achieved>11mmol/mol (1%) decrease in HbA1c at end of ActiveTX.HbA1c was not significantly decreased at end ofMaintTX (76± 15mmol/mol (9.1 ± 1.4%); 𝑡(14) = 0.89, p = 0.39, d = 0.29).Of note, 11 of 15 teens did maintain a lower HbA1c at end ofMaintTX compared to PreTx, and 5 participants maintained>11mmol/mol (1%) decrease inHbA1c at the end ofMaintTX.HbA1c at the end ofMaintTX did not significantly differ fromHbA1c the end of ActiveTX (𝑡(14) = −1.43, p = 0.17).

3.4. Executive Functioning. Participants improved on per-formance tasks measuring working memory and inhibitorycontrol and parents reported fewer problems with executivefunctioning at the end ofMaintTX. Figure 1(c) shows averagescores on executive functioning task assessments. Teensimproved on digit span scaled scores (PreTX = 8.60 ± 2.41;MaintTX = 11.36 ± 3.10; 𝑡(13) = −3.24, p = 0.006, d = 0.95),and these changes reflected a large effect size for workingmemory training. Participants also improved on measuresof inhibitory control. Inhibition Completion Time Scaled

Scores (PreTX = 11.2 ± 1.97; MaintTX = 12.07 ± 1.38; 𝑡(13) =−2.75, p = 0.02, d = 0.51) and inhibition errors scaled scores(PreTX = 9.21 ± 2.67; MaintTX = 11.29 ± 2.16; 𝑡(13) =−2.61, p = 0.02, d = 0.86) improved reflecting medium tolarge effect sizes for inhibitory control. In addition, parentsreported fewer problems with executive functioning at theend of MaintTX (Figure 1(d)) and GEC 𝑇-Score (PreTX =61.79 ± 11.36; MaintTX = 56.71 ± 3.63; 𝑡(13) = 2.14, p = 0.05,d = 0.40).

3.5. Association of Changes in Executive Functioning, SMBG,and HbA1c. Correlations between PreTX-Post MaintTXchange scores for measures of executive functioning, SMBG,and HbA1c are provided in Table 2. Improvements ininhibitory control, indexed via participants making fewererrors on the color word interference task (inhibition errors),were associated with increased frequency of SMBG (r = 0.78,p = 0.001) and decreased HbA1c (r = 0.59, p = 0.03). Bothassociations are consistent with large effect sizes. Changesin the other task and parent report measures of executivefunctioning were not significantly associated with SMBG orHbA1c.

4. Discussion

Findings suggest that this web-delivered multicomponentintervention is feasible to deliver in a rural region and mightbe an effective intervention to increase SMBG, decreaseHbA1c, and improve executive functioning in teens withpoorly controlled type 1 diabetes. Good compliance with


Table 2: Correlations between change scores for self-monitoring of blood glucose (SMBG), glycemic control (HbA1c), and executivefunctioning.

ΔHbA1c ΔDS ΔComp ΔErrors ΔGEC MΔ (SD)ΔSMBG −0.76∗∗ −0.21 −0.19 0.78∗∗ 0.04 1.14 (2.01)ΔHbA1c 0.08 0.15 −0.59∗ −0.20 0.36 (1.57)ΔDigit span (DS) 0.19 −0.02 −0.20 2.64 (3.05)ΔInhibition completion time (comp) −0.33 −0.05 4.86 (5.78)ΔInhibition errors (errors) 0.08 1.64 (2.06)ΔParent report GEC 5.07 (8.88)∗

𝑝 < 0.05; ∗∗𝑝 < 0.01.

the intervention protocol, evidenced through high rates ofcompletion of therapy sessions and WMT, suggests that thismulticomponent intervention was acceptable for the familiesand provides initial support for the acceptability of utilizingWMT in conjunction with therapy in youth with poorlycontrolled type 1 diabetes.

More specifically, this intervention showed large effectson SMBG (effect size at the end of ActiveTX for increasesin SMBG was 1.56), with teens on average increasing SMBGto over 6 times per day within 3 months’ time. IncreasedSMBG as a result of the intervention may not only helpadolescentsmore effectivelymanage their daily blood glucoselevels, but provide endocrinologists and diabetes educatorswith sufficient data to assist families in making appropriatechanges, if needed, to their insulin dosing. Anecdotally,we found encouraging families to communicate about theirnewly available blood glucose data with their provider to beimportant to the treatment process. When compared withavailable psychosocial interventions on SMBG, which haveevidenced small and nonsignificant effect sizes (ES = −0.44–0.13) [2], the current intervention shows promise. In fact,this pilot shows effect sizes similar to intensive multisystemictherapy for diabetes management (ES = 1.09) [32]. Withregard to improving HbA1c, this intervention also showspromise. There was a moderate effect size for changes inHbA1c at the end of ActiveTX (d = 0.62), which is largerthan or similar to the effect sizes for available psychosocialinterventions (ES = −0.55–0.59) [1].

The findings from this pilot are also consistent withother intervention studies that have utilized BEI to increaseengagement in a health behavior that does not have inherentimmediate rewards. Two other pilot trials have utilized BEIto increase SMBG in youth with type 1 diabetes and ourfindings are consistent with those studies. Stanger et al., 2013,utilized a similar protocol to the BEI intervention pilotedhere but did not include a maintenance phase of treatmentand did not include working memory training [9]. The effectsize of that pilot on SMBG (d = 1.00) was similar to thelarge effect size seen in this study. Petry et al., 2015, alsopiloted a BEI intervention that utilized a different incentiveschedule, $.10 per test with bonuses for ≥4 tests with smalleraverage earnings of $122, and no other counseling or workingmemory training [14]. The effect size of that pilot on SMBG(d = 3.10) is also large; however, only youth testing <4 timesper day were recruited into that study, while the current

study recruited youth regardless of baseline SMBG testingfrequency. Thus, our findings extend prior work on BEIand SMBG, supporting the use of BEI even for youth whoare already blood glucose checking >4 times per day butare still experiencing poor glycemic control. These findingssupport continued research into multicomponent treatmentsthat integrate self-regulatory interventions such as BEI andcognitive training with counseling to facilitate increasedSMBG and improved HbA1c.

This pilot provides support for the use of a web-deliverymodel of a multicomponent family intervention in a ruralregion, which is important as web-delivery might decreaseobstacles to accessing specialized behavioral healthcare (e.g.,distance to pediatric endocrinology clinic, few communitybehavioral health providers with training in teen type 1diabetes nonadherence). There has been one previous trans-lation of a family-based intervention for teens with poorlycontrolled type 1 diabetes into a web-delivered format [7].In those trials, researchers reported similar effect sizes forchanges in adherence and glycemic control and similar parentand teen reported working alliance with the therapist inthe face-to-face versus web-delivered formats [7, 33]. Thatparticular intervention was delivered from a metropolitancity center and likely reached some families living in ruralregions but was not specifically targeting a rural population.The current intervention model delivered in a rural regionshows comparatively greater effects on adherence (d = 1.56 vs.d = 0.45) and slightly greater effects on HbA1c (d = 0.62 vs.d = 0.40). These interventions provide a promising frame-work for the delivery of efficacious interventions via the webfor youth with poorly controlled type 1 diabetes living in ruralregions.

In addition, this is the first trial to our knowledge toutilize cognitive training in youth with type 1 diabetes.Participants evidenced significant improvements in workingmemory, inhibitory control, and parent reported executivefunctioning, with large effect sizes for improvements on thedigit span task and decreases in errors on the color wordinterference task. Improvements in inhibition that are notdirectly trained in the WMT tasks are consistent with a neartransfer of executive skill to domains other than workingmemory [34] and provide further support for the possiblyutility in WMT to address executive functioning and self-regulation in youth with poorly controlled type 1 diabetes.Given neuroimaging findings that type 1 diabetes is associated


with disrupted brain structure and function [5], interventionstargeting cognitive functioning may be particularly impor-tant to pursue.

This pilot study suggests multiple directions for futureresearch. Since improvements in SMBG and HbA1c werelarger at the end of ActiveTX compared to MaintTX, futureiterations of this intervention should focus on strategies forbetter sustaining the positive change. For example, the dura-tion of ActiveTX could be increased to provide more timefor new SMBG habits to develop. Future interventions mightalso use BEI targeting other self-care behaviors such as car-bohydrate counting and insulin dosing that are challengingfor teens with poorly controlled type 1 diabetes, or targetingthemaintenance of an increasing percentage of blood glucosevalues in a healthy range. Assessment of other mediators ofintervention effects beyond executive functioning, such asself-efficacy, coping skills, psychopathology, and family func-tioning would also be important. Further, while this studyutilized a general working memory training program, someemerging research suggests that domain-specific cognitivetraining may have a greater effect on modifying executivefunctioning and self-regulation for specific health behaviors[35]. The development of diabetes-specific cognitive traininginterventions where the training tasks include stimuli that areassociated with diabetes such as meters, insulin injections,pumps, and carbohydrate counting may help to improvethe efficacy of cognitive training interventions targetingadherence.

While promising, these preliminary findings require fur-ther validation in a larger sample with a randomized controlmethodology, not only to better assess the interventioneffects on SMBG and HbA1c, but also to examine changesin objective metrics of executive functioning where practiceeffects may be evident. There is also a need for furtherresearch examining the cost effectiveness of integratingincentives into our current healthcare delivery models foryouth with type 1 diabetes. Given the high cost of hospitaladmissions for hyper- and hypoglycemic events, as well asthe long-term costs associated with vascular disease intoadulthood, incentive interventions may reduce overall costsof care. Accordingly, the intervention described here is beingevaluated in a randomized control trial, examining efficacyand cost effectiveness.

Competing Interests

The authors have no competing interests related to this paperto disclose.

Acknowledgments

This study was supported by a grant from NICHD-DP3HD076602. Drs. Stanger and Budney were also supportedby a grant from NIDA-P30 DA029926 (PI: Lisa A. Marsch).Cogmed and Cogmed Working Memory Training are trade-marks, in the US and/or other countries, of Pearson Educa-tion, Inc., or its affiliate(s).

References

[1] K. K. Hood, J. M. Rohan, C. M. Peterson, and D. Drotar,“Interventions with adherence-promoting components in pedi-atric type 1 diabetes: meta-analysis of their impact on glycemiccontrol,” Diabetes Care, vol. 33, no. 7, pp. 1658–1664, 2010.

[2] S. E. Hampson, T. C. Skinner, J. Hart et al., “Behavioral inter-ventions for adolescents with type 1 diabetes: how effective arethey?” Diabetes Care, vol. 23, no. 9, pp. 1416–1422, 2000.

[3] C. A. Berg, D. J. Wiebe, Y. Suchy et al., “Individual differ-ences and day-to-day fluctuations in perceived self-regulationassociated with daily adherence in late adolescents with type 1diabetes,” Journal of Pediatric Psychology, vol. 39, no. 9, pp. 1038–1048, 2014.

[4] A. H. Lansing, C. A. Berg, J. Butner, and D. J. Wiebe, “Self-control, daily negative affect, and blood glucose control inadolescents with Type 1 diabetes,”Health Psychology, vol. 35, no.7, pp. 643–651, 2016.

[5] E. Litmanovitch, R. Geva, and M. Rachmiel, “Short and longterm neuro-behavioral alterations in type 1 diabetes mellituspediatric population,” World Journal of Diabetes, vol. 6, no. 2,pp. 259–270, 2015.

[6] T. Wysocki, M. A. Harris, L. M. Buckloh et al., “Randomizedtrial of behavioral family systems therapy for diabetes: mainte-nance of effects on diabetes outcomes in adolescents,” DiabetesCare, vol. 30, no. 3, pp. 555–560, 2007.

[7] M. A. Harris, K. A. Freeman, and D. C. Duke, “Seeing isbelieving: using skype to improve diabetes outcomes in youth,”Diabetes Care, vol. 38, no. 8, pp. 1427–1434, 2015.

[8] P. W. Goss, M. A. Paterson, and J. Renalson, “A ‘radical’ newrural model for pediatric diabetes care,” Pediatric Diabetes, vol.11, no. 5, pp. 296–304, 2010.

[9] C. Stanger, S. R. Ryan, L. M. Delhey et al., “A multicomponentmotivational intervention to improve adherence among ado-lescents with poorly controlled type 1 diabetes: a pilot study,”Journal of Pediatric Psychology, vol. 38, no. 6, pp. 629–637, 2013.

[10] C. Stanger, A. J. Budney, J. L. Kamon, and J. Thostensen, “Arandomized trial of contingency management for adolescentmarijuana abuse and dependence,” Drug and Alcohol Depen-dence, vol. 105, no. 3, pp. 240–247, 2009.

[11] Cogmed Coaching Manual, Pearson, London, UK, 2011.[12] C. Stanger, A. J. Budney, and W. K. Bickel, “A developmental

perspective on neuroeconomic mechanisms of contingencymanagement,” Psychology of Addictive Behaviors, vol. 27, no. 2,pp. 403–415, 2013.

[13] K. Sutherland, J. B. Christianson, and S. Leatherman, “Impactof targeted financial incentives on personal health behavior: areview of the literature,”Medical Care Research and Review, vol.65, no. 6, pp. 36S–78S, 2008.

[14] N. M. Petry, E. Cengiz, J. A. Wagner, K. Weyman, E. Tichy, andW.V. Tamborlane, “Testing for rewards: a pilot study to improvetype 1 diabetes management in adolescents,” Diabetes Care, vol.38, no. 10, pp. 1952–1954, 2015.

[15] K. G. Volpp, L. K. John, A. B. Troxel, L. Norton, J. Fassbender,and G. Loewenstein, “Financial incentive-based approaches forweight loss: a randomized trial,” Journal of theAmericanMedicalAssociation, vol. 300, no. 22, pp. 2631–2637, 2008.

[16] S. Jazbec, M. G. Hardin, E. Schroth, E. McClure, D. S. Pine, andM. Ernst, “Age-related influence of contingencies on a saccadetask,” Experimental Brain Research, vol. 174, no. 4, pp. 754–762,2006.


[17] C. A. Berg, J. M. Butler, P. Osborn et al., “Role of parental mon-itoring in understanding the benefits of parental acceptance onadolescent adherence and metabolic control of type 1 diabetes,”Diabetes Care, vol. 31, no. 4, pp. 678–683, 2008.

[18] D. L. Palmer, P. Osborn, P. S. King et al., “The structure ofparental involvement and relations to disease management foryouth with type 1 diabetes,” Journal of Pediatric Psychology, vol.36, no. 5, pp. 596–605, 2011.

[19] D. A. Ellis, C.-L. Podolski, M. Frey, S. Naar-King, B. Wang, andK. Moltz, “The role of parental monitoring in adolescent healthoutcomes: impact on regimen adherence in youth with type 1diabetes,” Journal of Pediatric Psychology, vol. 32, no. 8, pp. 907–917, 2007.

[20] D. C. Duke, G. R. Geffken, A. B. Lewin, L. B. Williams, E. A.Storch, and J. H. Silverstein, “Glycemic control in youth withtype 1 diabetes: family predictors and mediators,” Journal ofPediatric Psychology, vol. 33, no. 7, pp. 719–727, 2008.

[21] J. T. Becker and R. G. Morris, “Working memory(s),” Brain andCognition, vol. 41, no. 1, pp. 1–8, 1999.

[22] A. Bechara and E. M. Martin, “Impaired decision makingrelated to working memory deficits in individuals with sub-stance addictions,” Neuropsychology, vol. 18, no. 1, pp. 152–162,2004.

[23] D. C. Duke and M. A. Harris, “Executive function, adherence,and glycemic control in adolescents with type 1 diabetes: aliterature review,”CurrentDiabetes Reports, vol. 14, no. 10, article532, 2014.

[24] J. Holmes, S. E. Gathercole, M. Place, D. L. Dunning, K. A.Hilton, and J. G. Elliott, “Working memory deficits can beovercome: impacts of training and medication on workingmemory in childrenwithADHD,”AppliedCognitive Psychology,vol. 24, no. 6, pp. 827–836, 2010.

[25] G. C. C. Løhaugen, I. Antonsen, A. Haberg et al., “Com-puterized working memory training improves function inadolescents born at extremely low birth weight,” Journal ofPediatrics, vol. 158, no. 4, pp. 555–561.e4, 2011.

[26] M. J. Van der Molen, J. E. H. Van Luit, M. W. Van der Molen, I.Klugkist, and M. J. Jongmans, “Effectiveness of a computerisedworking memory training in adolescents with mild to bor-derline intellectual disabilities,” Journal of Intellectual DisabilityResearch, vol. 54, no. 5, pp. 433–447, 2010.

[27] T. Klingberg, “Training and plasticity of working memory,”Trends in Cognitive Sciences, vol. 14, no. 7, pp. 317–324, 2010.

[28] D. Wechsler, Wechsler Adult Intelligence Scale, Pearson, SanAntonio, Tex, USA, 4th edition, 2008.

[29] D. Wechsler, Wechsler Intelligence Scale for Children-FourthEdition, Pearson, San Antonio, Tex, USA, 4th edition, 2003.

[30] D. Delis, E. Kaplan, and J. Kramer, Delis-Kaplan ExecutiveFunction System: Examiner’s Manual,The Psychological Corpo-ration, San Antonio, Tex, USA, 2001.

[31] G. A. Gioia, P. K. Isquith, S. C. Guy, and L. Kenworthy, BehaviorRating Inventory of Executive Function (BRIEF): ProfessionalManual, Psychological Assessment Resources, Lutz, Fla, USA,2000.

[32] D. A. Ellis, M. A. Frey, S. Naar-King, T. Templin, P. Cunning-ham, and N. Cakan, “Use of multisystemic therapy to improveregimen adherence among adolescents with Type 1 diabetes inchronic poor metabolic control,” Diabetes Care, vol. 28, no. 7,pp. 1604–1610, 2005.

[33] K. A. Freeman, D. C. Duke, and M. A. Harris, “Behavioralhealth care for adolescents with poorly controlled diabetes via

Skype: doesworking alliance remain intact?” Journal of DiabetesScience and Technology, vol. 7, no. 3, pp. 727–735, 2013.

[34] E. N. Kerr and M. C. Blackwell, “Near-transfer effects fol-lowing working memory intervention (Cogmed) in childrenwith symptomatic epilepsy: an open randomized clinical trial,”Epilepsia, vol. 56, no. 11, pp. 1784–1792, 2015.

[35] R. B. Lopez, K. D. Vohs, D. D. Wagner, and T. F. Heatherton,“Self-regulatory strength: neural mechanisms and implicationsfor training,” in Handbook of Biobehavioral Approaches to Self-Regulation, G. H. E. Gendolla, M. Tops, and S. L. Koole, Eds.,pp. 43–54, Springer, New York, NY, USA, 2015.

mhealth and health information technology tools for...

Documents