RESEARCH ARTICLE Open Access

Formative research for the design of ascalable mobile health program water,sanitation, and hygiene: CHoBI7 mobilehealth programChristine Marie George1* , Fatema Zohura2, Alana Teman1, Elizabeth Thomas1, Tasdik Hasan2, Sohel Rana1,Tahmina Parvin2, David A. Sack1, Sazzadul Islam Bhuyian2, Alain Labrique1, Jahed Masud2, Peter Winch1,Elli Leontsini1, Kelsey Zeller1, Farzana Begum2, Abul Hasem Khan3, Sanya Tahmina3, Farazana Munum3,Shirajum Monira2 and Munirul Alam2

Abstract

Background: The Cholera-Hospital-Based-Intervention-for-7-Days (CHoBI7) is a handwashing with soap and watertreatment intervention program delivered by a health promoter bedside in a health facility and through home visitsto diarrhea patients and their household members during the 7 days after admission to a health facility. In arandomized controlled trial among cholera patient households in Bangladesh, the 7-day CHoBI7 program resultedin a significant reduction in cholera among household members of cholera patients and sustained improvements indrinking water quality and handwashing with soap practices 12 months post-intervention. In an effort to take thisintervention to scale across Bangladesh in partnership with the Bangladesh Ministry of Health and Family Welfare,this study evaluates the feasibility and acceptability of mobile health (mHealth) programs as a low-cost, scalableapproach for CHoBI7 program delivery.

Methods: Formative research for the development of the CHoBI7 mHealth intervention included 40 semi-structuredinterviews, 4 mHealth workshops, 2 group discussions, and a pilot study of 52 households to assess the feasibility andacceptability of the developed mHealth program. Thematic analysis of the interviews and group discussions wasconducted by two individuals separately based on emergent themes, and then themes were compared and discussed.

Results: A theory- and evidence-based approach using qualitative research methods was implemented to design theCHoBI7 mHealth program. Semi-structured interviews with government stakeholders identified perceptions andpreferences for scaling the CHoBI7 mHealth program. Group discussions and semi-structured interviews with diarrheapatients and their family members identified beneficiary perceptions of mHealth and preferences for CHoBI7 mHealthprogram delivery. mHealth workshops were conducted as an interactive approach to draft and refine mobile messagecontent based on stakeholder preferences. The pilot findings indicate that the CHoBI7 mHealth program has high useracceptability and is feasible to deliver to diarrhea patients that present at health facilities for treatment in Bangladesh.Both text and voice messages were recommended for program delivery. Dr. Chobi, the sender of mHealth messages,was viewed as a credible source of information that could be shared with others.

(Continued on next page)

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

* Correspondence: cmgeorge1@jhu.edu1Associate Professor, Department of International Health, Program in GlobalDisease Epidemiology and Control, Johns Hopkins Bloomberg School of PublicHealth, 615 N. Wolfe Street, Room E5535, Baltimore, MD 21205-2103, USAFull list of author information is available at the end of the article

George et al. BMC Public Health (2019) 19:1028 https://doi.org/10.1186/s12889-019-7144-z

(Continued from previous page)

Conclusion: This study presents a theory- and evidence-based approach that can be implemented for the development offuture water, sanitation, and hygiene mHealth programs in low-resource settings.

Keywords: Mobile health, Water, sanitation, and hygiene, Diarrhea, Cholera, Bangladesh, Qualitative researchmethods, Formative research

BackgroundDiarrhea is a leading cause of death in young children glo-bally, causing an estimated 500,000 deaths annually [1].Lack of handwashing with soap and water treatment areimportant risk factors for pediatric diarrheal disease [2–4].Globally, only 26% of the world’s population is estimatedto wash their hands with soap after coming into contactwith human excreta [5]. Previous work in Bangladeshfound that handwashing with soap before eating and feed-ing a child measured by structured observation was lessthan 1% [6]. Water, sanitation, and hygiene (WASH) inter-ventions promoting household chlorination of drinkingwater, safe water storage, and handwashing with soap havethe potential to reduce diarrheal disease incidence in chil-dren less than 5 years of age an estimated 30 to 75% [7].However, there has been limited success in encouraginghouseholds to sustain these behaviors over time [8, 9].Community-based WASH interventions are expensiveand often difficult to implement effectively in an urbancontext in low-resource settings [10]. Effective scalableWASH interventions are urgently needed to reduce diar-rheal diseases globally.Phone-based reminders are an emerging low-cost

intervention approach that has been shown to lead tosignificantly improved case management and diseaseprevention practices [11–15]. Furthermore, mobilehealth (mHealth) presents a scalable approach for whichmessages can be sent to a large number of households atminimal cost and can serve as valuable cues to action tofacilitate behavior change. Phone access and ownershiphave grown tremendously with mobile phone subscrip-tions worldwide more than doubling over the past 10years [16]. In 2017, there were estimated to be 85 mil-lion unique mobile subscribers in Bangladesh--half thecountry’s population [17]. This presents an ideal oppor-tunity for mass mobile phone messaging of public healthinformation. However, there are no published studies, toour knowledge, evaluating delivery of WASH interven-tion programs using mobile health.Most WASH programs focus solely on educational

modules related to diarrhea prevention, and do not tailortheir interventions to their target population [18]. Thisapproach does not take into account the psychosocial,technological, and contextual factors in a given settingthat could influence behavior. There is a growing evi-dence base demonstrating interventions that use health

theory are likely to yield greater behavior change thanthose based on health education alone [19–24]. Theory-based interventions are guided by behavior changetheories and models that provide a framework for whichinterventions can be delivered; examples include theTheory of Planned Behavior, Protection Motivation The-ory, the Integrated Behavioural Model for Water, Sanita-tion, and Hygiene, the Health Belief Model, and theRisks, Attitudes, Norms, Abilities, and Self-regulation(RANAS) Model [25–29]. In Contzen et al. which uti-lized the RANAS model, significant psychosocial factorsrelated to handwashing with soap practices were identi-fied at baseline and then targeted in a subsequent behav-ior change intervention conducted in Ethiopia [30]. Thistheory-based intervention resulted in significantly morehandwashing with soap at stool related events thanhealth education focused messages only [21]. However,despite this growing evidence, the use of theory-basedbehavior change techniques for large scale WASH pro-grams is relatively rare [31].

CHoBI7 programHousehold members of diarrhea patients are at a muchhigher risk of developing diarrheal diseases (> 100 timesfor cholera) than the general population during the 1-week period after the diarrhea patient presents at a healthfacility [32–34]. This is likely because of a shared contami-nated water source and poor hygiene practices [33, 35].However, despite this high risk there is no standard of carefor the household members of diarrhea patients. Further-more, the time patients and their household membersspend at a health facility for the treatment of diarrhea pre-sents the opportunity to deliver WASH interventionswhen perceived severity of diarrhea and perceived benefitsof water treatment and handwashing with soap is likelythe highest [36]. Therefore, in an effort to develop a low-cost standard of care for this population during this 1-week high-risk period, we recently developed a 7 dayhealth facility-based handwashing with soap and watertreatment program entitled CHoBI7 (Cholera-Hospital-Based-Intervention-for-7-Days). “Chobi” means picture inBangla. This name was selected for the pictorial WASHmodules delivered as part of this intervention program.The CHoBI7 program includes: (1) a pictorial module de-livered using a flipbook on how diarrheal diseases spread,and recommendations on proper handwashing with soap

George et al. BMC Public Health (2019) 19:1028 Page 2 of 18

and water treatment; and (2) a diarrhea prevention pack-age containing: chlorine tablets for water treatment, soapywater bottles (a low-cost alternative to bar soap madeusing detergent and water [37]), a handwashing station, awater vessel with tap and lid, and a sticker and cue cardson handwashing with soap and water treatment. A trainedhealth promoter delivers this pictorial module and diar-rhea prevention package to diarrhea patients and their ac-companying family members during a consultationsession bedside in a health facility [38]. This module isthen reinforced through 30min home visits during the 1week period of highest risk for cholera infection amonghousehold members of cholera patients.

Previous randomized controlled trial of the CHoBI7interventionTo evaluate the efficacy of the targeted 7-day CHoBI7WASH intervention in reducing cholera among house-hold members of cholera patients during their 1-weekhigh-risk period, we conducted a RCT of CHoBI7. Wecompared the CHoBI7 WASH intervention to the stand-ard recommendation given in Bangladesh to diarrhea pa-tients at discharge on oral rehydration solution (ORS)use. Delivery of the 7-day CHoBI7 intervention resultedin a 47% reduction in overall cholera infections, and asignificant reduction in symptomatic cholera during the1-week high-risk period [38]. Consistent with these find-ings, the odds of handwashing with soap at food andstool related events during structured observation was 14times higher in the CHoBI7 arm compared to the stand-ard recommendation arm, and 94% of CHoBI7 house-holds had free chlorine concentrations in stored drinkingwater greater than the CDC recommended cut-off of 0.2mg/L [39]. There were also no stored drinking water sam-ples in the CHoBI7 intervention arm with V. cholerae.Furthermore, the 7-day CHoBI7 intervention led to sig-nificant sustained improvements in household storeddrinking water quality and observed handwashing withsoap practices up to 12-months post intervention [40].These findings demonstrate that the CHoBI7 interventionpresents a promising standard of care for cholera patienthouseholds during their 1 week high risk period and canresult in sustained WASH practices over time.

Theory based approach for development of the CHoBI7intervention programUsing a theory-based approach, the CHoBI7 interven-tion program was informed by components of ProtectionMotivation Theory, IBM-WASH, and the RANAS Model[25, 27, 29]. Using these theories and models, we devel-oped behavior change techniques for the CHoBI7 inter-vention program targeting remembering, perceivedsusceptibility and severity, cholera awareness, disgust, re-sponse efficacy, convenience, and self-efficacy (based on

previous studies [18, 21, 30, 41–43]), and used Likert scalestatements to measure these psychosocial factors(methods are published in George et al. [44]). By measur-ing these psychosocial factors during our recent RCT ofCHoBI7, we were able to investigate the underlying mech-anism of change that led to the high handwashing withsoap behavior observed among intervention participants.Response efficacy (judgments about the efficacy of a pre-ventive response that will avert the perceived threat [29])was found to mediate the intervention’s effect on hand-washing with soap habit formation at the 1-week follow-up, whereas disgust, convenience, and cholera awarenesswere mediators of habit maintenance at the 6 to 12-monthfollow-up [44]. Our study was the first RCT of a WASHintervention that conducted a mediation analysis to inves-tigate the underlying mechanism of change in a low-in-come country.

Rationale for the development of the CHoBI7 mobilehealth programBased on this previous work, the Bangladesh Ministry ofHealth and Family Welfare has expressed an interest in in-cluding the CHoBI7 program in the upcoming NationalOperational Plan for Communicable Disease Control, andhas requested evidence be provided on low-cost, scalableapproaches that could be implemented to deliver theCHoBI7 program across Bangladesh. Therefore, thiscurrent study building on our previous work evaluates thefeasibility and acceptability of implementing a mHealthprogram as a low-cost, scalable approach for CHoBI7 pro-gram delivery that does not involve frequent in-personvisits. This paper reports the formative research findingsfrom this study, which have informed the design of theCHoBI7 mHealth program. The primary objective of thisformative research was to develop a WASH mHealth pro-gram that could be taken to scale by the Bangladesh Min-istry of Health and Family Welfare to serve as cues toaction to facilitate sustained WASH behavior change.In this study we take a theory- and evidence-based ap-

proach using qualitative research methods to design andtailor the CHoBI7 mHealth intervention program forour target population in Bangladesh. The CHoBI7mHealth program developed through these formative re-search activities is currently being evaluated in a RCT.This is the first RCT, to our knowledge, evaluating theimpact of a WASH mHealth program.

MethodsFormative research activitiesThe formative research activities had three components:(1) intervention development through mHealth work-shops; (2) semi-structured interviews and group discus-sions; and (3) a pilot study of the CHoBI7 mHealthprogram. An overview of formative research activities is

George et al. BMC Public Health (2019) 19:1028 Page 3 of 18

shown in Fig. 1. Research activities were a partnership be-tween the Johns Hopkins Bloomberg School of PublicHealth, the Bangladesh Ministry of Health and FamilyWelfare, and the International Centre for DiarrhoealDisease Research, Bangladesh (icddr,b). The formative re-search aims were the following: (1) identify governmentstakeholders perceptions and preferences for scaling theCHoBI7 mHealth program in Bangladesh; (2) identifybeneficiary perceptions and preferences for delivering thisprogram (e.g. mobile message type, message content, andmessage timing and frequency); and (3) determine thefeasibility of implementing this program (e.g. mobilephone access and message sharing).

Intervention developmentThe CHoBI7 mHealth intervention program develop-ment targeted five key behaviors: (1) preparing soapywater using water and detergent powder; (2) handwash-ing with soap at food- and stool-related events; (3)

treating household drinking water using chlorine tabletsduring the one-week high-risk period after the diarrheapatient in the household was admitted to the health fa-cility; (4) safe drinking water storage in a water vesselwith cover; and (5) heating of household drinking wateruntil it reaches a rolling boil after the one-week high-risk period. The CHoBI7 mHealth program was devel-oped through a theory-based approach, which was in-formed by IBM-WASH and Protection MotivationTheory [25, 29]. Intervention development was guidedby IBM-WASH to target the habitual, individual, inter-personal/household, community, and structural/societal-level factors that are drivers of our target WASH behav-iors [25]. For example, higher-level contextual andtechnological factors, such as national policies onWASH and delivery of mHealth messages to the publicwere considered through the engagement of governmentstakeholders in intervention development; and habituallevel factors such as frequency of exposure to mobile

Government Stakeholder Interviews

1st Mobile Health Intervention Development Workshop

1st Group Discussion

Pilot-Testing and Revisions to the

CHoBI7 Mobile Health Intervention Package

2nd Group Discussion

Formative Research Stage vretnIytivitcA ention Revisions

Development of the CHoBI7 Mobile Health Intervention Package

• Added Bangla script• Both voice and text messages included• Added photo of female Dr. Chobi in

her 30s• Promoters helped diarrhea patients to

save Dr. Chobi’s number in their mobile phone in the health facility

• All text messages show Dr. Chobi as the sender

• Voice messages shortened to 2 minutes• Added summary text messages after

voice messages were sent• Text and voice messages were sent out

at 5 PM in the afternoon

• All IVR messages stated "There is no charge for your reply."

• Tutorial on how to respond to quiz IVR messages

• Module in health facility on target person for message delivery and message sharing

• All text and voice messages state "Please share this message.“

• Voice and text messages are sent at least once every two weeks for 1 year

• Added mobile messages targeting gender norms, nurture, and remembering

Government Stakeholder and Patient Household Interviews

2nd Mobile Health Intervention Development Workshop

3rd Mobile Health Intervention Development Workshop

4th Mobile Health Intervention Development Workshop

Fig. 1 Formative Research Activities for CHoBI7 Mobile Health Program Development

George et al. BMC Public Health (2019) 19:1028 Page 4 of 18

messages were explored by engaging pilot participants ininterviews and discussion groups. This multi-level ap-proach allowed us to develop behavior change tech-niques to regulate factors identified during formativeresearch that facilitated or impeded the targetedbehaviors.mHealth messages were developed to target the follow-

ing psychosocial factors: response efficacy of the behav-ioral recommendations, remembering of handwashingwith soap, water treatment, and safe water storage, con-venience (e.g. promoting the use of enabling hardware),awareness of diarrhea transmission and prevention,nurture towards children in the household, self-efficacy ofbehavioral recommendations, feelings of disgust towardsfeces, perceived severity and susceptibility, and gendernorms and roles (e.g. acceptability of the female householdmembers receiving mHealth messages) [25, 29]. Table 1includes examples of the mHealth Behavior Change Tech-niques included in the CHoBI7 mHealth program to tar-get these psychosocial factors and IBM-WASHdimensions. These psychosocial factors were identifiedthrough: (1) the formative research findings presented inthis publication (gender norms/roles, nurture, and re-membering); (2) psychosocial factors associated with habitformation and maintenance in our previous RCT of theCHoBI7 intervention (response efficacy, disgust, conveni-ence, and diarrhea disease awareness) [44]; and (3) factorsfound in previous studies to be associated with WASH be-havior change (self-efficacy and descriptive norms [25]and perceived severity and susceptibility [29, 30]).The CHoBI7 mHealth messages were developed during 4

mobile health workshops. The initial workshop in June2016 was led by a mHealth expert that trained the studyteam on techniques for delivering voice, text, and inter-active voice response (IVR) messages. The second mHealthworkshop was later in June 2016, the third was in July2016, and the fourth in September 2016. There were 6–10participants in each mHealth workshop, and workshopswere 2 to 4 h in length. Workshop participants wereCHoBI7 research team members including the study inter-vention and project coordinator, health promoters, andstudy investigators. During the workshops, contextual, psy-chosocial, and technological factors that emerged duringthe qualitative research were discussed using the IBM-WASH framework (Table 2). Mobile messages targetingthese factors were than written on a flip chart by teammembers and discussed and refined as a group.The VIAMO platform was used to deliver text,

voice, and IVR messages to pilot households (https://viamo.io/). IVR messages were sent as “quiz” ques-tions for recipients to assess their knowledge of theWASH key behaviors recommended by the CHoBI7intervention. There was no call- or text-back optionavailable, or hotline.

CHoBI7 mHealth pilotThe objective of the pilot study was to identify thefeasibility of implementing the CHoBI7 mHealth pro-gram, and potential challenges for successful programimplementation. Fifty-two households of diarrhea pa-tients (7 Control Arm, 23 mHealth with no homevisits arm, and 22 Intervention Arm 2) were enrolledin the pilot study of the CHoBI7 mHealth program.Diarrhea patients presenting at icddr,b Dhaka Hospitalor Mugda Hospital in Dhaka, Bangladesh with threeor more loose stools over a 24 h period were re-cruited for the pilot study. Households of diarrhea pa-tients were eligible for the study if they had a childunder 5 years of age and at least one household mem-ber owned an active mobile phone. Household mem-bers were defined as those individuals sharing thesame cooking pot with the diarrhea patient for thepast three days. Households were followed up to 3months. The Control Arm (Standard Message Arm)received the standard message in Bangladesh given todiarrhea patients on the use of ORS (Additional file 1:Table S1). The mHealth with no home visitsarm households received: (1) the standard message;(2) one visit by a promoter to deliver the CHoBI7pictorial module during the patient’s visit for diarrheatreatment at the health facility; and (3) CHoBI7mHealth voice and text messages at least once everytwo weeks. The mHealth with home visits arm house-holds received the same activities as the mHealthwith no home visits arm plus two – 30 min homevisits by a health promoter during the first week ofintervention delivery. A diarrhea prevention packagewas given to diarrhea patients in the interventionarms by promoters during treatment at the health fa-cility. This package contained chlorine tablets forwater treatment, a soapy water bottle, a handwashingstation, a water vessel with lid, and a sticker and cuecards on handwashing with soap and water treatment.Participants were assigned to study arms based onthe day of the week they were admitted to eachhealth facility. The pilot study was conducted in aniterative manner where changes to the CHoBI7mHealth message content and delivery were madebased on participant feedback from semi-structuredinterviews and group discussions between May to No-vember 2016.

Semi-structured interview and group discussionsForty semi-structured interviews were conducted fromMay to November 2016. Respondents were selectedthrough convenience-based sampling [45]. There were12 government stakeholder interviews to identify gov-ernment stakeholder perceptions and preferences forscaling the CHoBI7 mHealth program in Bangladesh;

George et al. BMC Public Health (2019) 19:1028 Page 5 of 18

Table

1Exam

pleof

Theo

ry-based

App

roachforDevelop

men

tof

CHoB

I7mHealth

Message

s:Psycho

socialFactorsandIBM-W

ASH

Dim

ension

s

Psycho

social

Factor

Factor

Definition

Behavior

Chang

eTechniqu

eIBM-W

ASH

Dim

ension

Exam

pleMessage

sHypothe

sized

Chang

ewith

Interven

tion

Self-Efficacy

Thebe

liefin

one’scapabilitiesto

organize

andexecutethecoursesof

actio

nrequ

iredto

manage

prospe

ctivesituations.(Band

uraet

al.

1997)

mHealth

message

sprovidinginform

ationon

actio

nabletasks

Structural

(Gen

der

Roles)/

Individu

al

VoiceMessage

Dr.Ch

obiA

pa:H

owareyoubrother?How

isAklim

aan

dyour

children?

Higheracceptability

ofWASHbeha

vioral

recommendations

amon

gmen

Husband

:Yes,good!

Mychildrengotfrequentdiarrhea

andsometimes

weneeded

toho

spitalizethem

before.

Now

,theyremainhealthy.

Higherself-efficacy

amon

gmale

householdmem

bers

Dr.Ch

obiA

pa:W

hydo

youthinkthey

remainhealthy

now?

Husband

:Because

myfamily

andIn

owboilan

dsafely

storeourdrinking

wateran

dwashourha

ndswith

soapywaterafterdefecation,aftercleaning

achild’s

fecesan

dan

us,beforepreparingfood,and

before

eatingor

feedingchildren.See,youph

oned

andtexted

mewhich

Ihaveshared

with

myfamily

andfollowed

allthe

instructions.

mHealth

message

sen

couragingthe

useof

enablingtechno

logy

Habitu

alText

Message

Tokeep

your

watersafeto

drink,alwaysputyour

boiledwaterin

theblue

bucket

with

thelid

on.N

ever

putyour

hand

sinto

thiswater,alwaysusethetap.

Keep

your

family

happyan

dhealthy!

Higherself-efficacy

andconvenienceof

WASHbeha

vioral

recommendations

Descriptive

Norms

Percep

tions

abou

twhich

behaviors

aretypically

perfo

rmed

byothe

rs(Cialdinietal.2006)

mHealth

message

sde

scrib

ingtheprop

ortio

nof

othe

rsin

thecommun

itype

rform

ingthe

samebe

havior

Com

mun

ityVo

iceMessage

Iam

Dr.Ch

obiA

pafro

mMoh

akha

liCh

oleraHospital.

How

areyou?

Weha

vefoundthat

80%

ofindividuals

inyour

neighb

ourhoodreported

washing

theirha

nds

with

soap

beforefeedingtheirchild

thisweek.Like

othersin

your

neighb

orho

od,alwayswashyour

hand

swith

soap

beforepreparingfood

andfeedingyour

child.M

akesurethereisalwayssoapywaterpresent

within

10stepsof

your

cookingarea.Keepyour

family

healthyan

dha

ppy!

Higherdescriptive

norm

sof

WASH

beha

vioral

recommendations

mHealth

message

sfro

mAklim

a,awom

anwho

brou

ghthe

rchild

toahe

alth

facilityfordiarrhea

treatm

ent,andwho

learne

dprop

erhand

washing

andwater

treatm

entbe

haviorsfro

mDr.Cho

bi.

InterpersonalVo

iceMessage

ThisisDr.ChobiApa.Aklimaisherewith

meagaintoday

andwantstoshareherstorywith

you.Aklim

a:Mysonwas

verysick;he

almostdied.Wetook

him

tothehospitaland

thedoctorsgave

him

treatment.BeforeIleftthe

hospital,

they

gave

meanice

redhandwashing

bucketandblue

waterbucket.D

r.ChobiApa

toldmeaboutw

ashing

my

handswith

soap

atfourkeytim

es:beforeIpreparefood,

beforeIeat,and

afterIusethetoiletorclean

mychild’s

fecesoranus.Ihave

followed

alltheseinstructions

allthe

time,andmyfamilyisnowhealthyandhappy!

Dr.Ch

obi:Than

kyouAklim

aforsharingyour

story!Did

youhear

howwellA

klimaisdoingno

w?Areyoualso

practicingthesesafepractices

andstayinghealthy?

Higherdescriptive

norm

sof

WASH

beha

vioral

recommendations

George et al. BMC Public Health (2019) 19:1028 Page 6 of 18

Table

1Exam

pleof

Theo

ry-based

App

roachforDevelop

men

tof

CHoB

I7mHealth

Message

s:Psycho

socialFactorsandIBM-W

ASH

Dim

ension

s(Con

tinued)

Psycho

social

Factor

Factor

Definition

Behavior

Chang

eTechniqu

eIBM-W

ASH

Dim

ension

Exam

pleMessage

sHypothe

sized

Chang

ewith

Interven

tion

Emotionof

Nurture

Thede

sire

tocare

forsomeo

ne,and

seethem

developor

grow

.mHealth

message

sen

couragingnu

rtureof

youn

gchildrenas

amotivator

topracticethekey

behaviorsprom

oted

InterpersonalVo

iceMessage

Diarrheaespeciallyhu

rtssm

allchildren.

Toprotectyour

child

from

thesegerm

s,makesureyour

who

lefamily

washestheirha

ndsat

4keytim

es:after

defecation,aftercleaning

achild’sfecesan

dan

us,

beforepreparingfood,and

beforeeatingor

feeding

children.Use

soapywatereach

time!

Highernu

rturefor

youn

gchildren

Emotionof

Disgu

stRevulsionthat

isoccasion

edby

the

sigh

tof

excreta,rotten

food

,slim

e,andbu

gs.(Curtis

etal.2001)

mHealth

message

sen

couragingtheem

otionof

disgustas

amotivator

topracticethekeybe

haviors

prom

oted

Individu

alText

Message

Ifyoudon’twashyour

hand

safterusingthetoilet,

you’reeatingfeceswhenyou’reeatingyour

food.There

arelotsof

germ

sin

feces,an

dyoumay

have

diarrhea

again.

Higherdisgust

Remem

berin

gTo

perfo

rmabe

havior,ithasto

beremem

beredat

therig

httim

e/situation.(Tob

iaset

al.2009)

mHealth

message

sreminding

househ

olds

tope

rform

thekeybe

haviorsprom

oted

Habitu

alText

Message

Haveyoumadeaha

bitof

drinking

safewateran

dwashing

your

hand

swith

soap

tocreate

ashieldfor

good

health?Ifno

t,youcanstartno

w!

Higherdaily

remem

beringof

beha

vioral

recommendations

George et al. BMC Public Health (2019) 19:1028 Page 7 of 18

and 28 intervention arm pilot participant interviews toidentify beneficiary perceptions and preferences for de-livering this program and to determine the feasibility ofimplementing this program. Government stakeholderswere selected if they were involved in mHealth orWASH activities at the Bangladesh Ministry of Healthand Family Welfare. All government stakeholders wereresponsible for the implementation of government pro-grams related to health in Bangladesh. Participants fromdiarrhea patient households were selected based on will-ingness to participate and availability of time. One pilothousehold participant was interviewed twice. All pilotparticipant interviews were in Bangla. Eleven govern-ment stakeholder interviews were in English and onewas in Bangla. One government stakeholder responsiblefor health policy on diarrhea prevention in Bangladeshwas interviewed three times from May to September2016. During government stakeholder interviews theflipbook and diarrhea prevention package were shown,and feedback was requested. Thirty-nine of the inter-views were audio-recorded, one respondent asked to not

be recorded. Twenty three of the semi-structured inter-views were transcribed verbatim, and summary fieldnotes were written for 39 interviews. One interviewcould not be transcribed because of the poor quality ofthe recording. The transcripts and recordings for thepilot study interviews were used to develop a summaryframework focused on the key research questions for theanalysis. The summary framework was then completedfor all pilot participant interviews.Two group discussions were conducted. The first

group discussion had 7 participants and the secondgroup discussion had 12 participants. These individ-uals were selected using convenience sampling. Bothgroup discussions included diarrhea patients admittedto icddr,b Dhaka Hospital and Mugda Hospital in thepast 3 months and household members of these diar-rhea patients. Participants in the first group discus-sion, did not have previous exposure to our CHoBI7mHealth messages. The objective of the first groupdiscussion was to obtain stakeholder feedback on themobile messages that had been developed during the

Table 2 The IBM-WASH Framework Applied to the Development of the CHoBI7 Mobile Health Intervention Based on QualitativeFindings

BM-WASH Dimension Contextual Factors Psychosocial Factors Technological Factors

Structural/ Societal • Existing government mobile healthprograms send out health-relatedmessages on government healthdays, these include voice and textmobile messages

• Potential inclusion of CHoBI7intervention in the NationalOperational Plan in Bangladesh

• Potential integration of CHoBI7in existing mobile health programsin Bangladesh

• Government commitment tomobile health as a method todeliver public health information

• Existing government mobile platform hasthe potential to be used for CHoBI7intervention delivery at reduced cost

• Health Education Bureau in the Ministry ofHealth and Family Welfare currentlydevelops mobile health messages, and canbe potentially engaged for CHoBI7intervention development

Community • High household mobile phoneaccess and ownership in Bangladesh

• Sharing of mHealth messages withneighbors

• High mobile network coverage in Dhaka,Bangladesh

• Most feature phones available in Bangladeshallow for viewing of Bangla script

Interpersonal • Females in the households are oftenthe ones responsible for caring foryoung children

• Male household members may notwant female household members toreceive text and voice messages froman unknown sender

• Female caregivers requestingaccess to CHoBI7 mHealth messagesto allow them to better care fortheir children

• Text messages allow for sharing with othersat a later time

• Male household members do not alwaysshare mobile messages or their mobilephones with other household members

• Timing for mobile message delivery whenall household members are present

• Adult male household members typicallyhave primary mobile phone ownership inhousehold

• Lower female access to mobile phones

Individual • Literacy rate of household members• Limited mobile message sharing bythose working outside of the home

• Self-efficacy to open textmessages, and respond tointeractive voice responsemessages

• Concerns about being charged a fee forviewing or listening to mobile messages

Habitual • Frequency of exposure to mobilemessages

• Outcome expectancy thatfollowing recommendationscontained in mobile messageswill reduce disease

• Voice and text messages as reminders toperform the promoted water, sanitation,and hygiene behaviors

George et al. BMC Public Health (2019) 19:1028 Page 8 of 18

initial two mhealth workshops. These messages areshown in Additional file 1: Table S2. This feedbackwas used to refine mobile messages before delivery inthe pilot study. All participants in the second groupdiscussion resided in households that had participatedin the pilot of the CHoBI7 mHealth program and hadreceived these voice and text messages for up to 6weeks prior to the group discussion. The objective ofthe second group discussion was to assess the usabil-ity and acceptability of the developed mHealth mes-sages delivered in the pilot. A notetaker summarizedthe feedback from both group discussions.The semi structured interviews and group discussions

were conducted with a guide covering the followingtopics: current sources of health related information, ex-periences with the current and previous mobile healthprograms, and recommendations for delivery of CHoBI7WASH mHealth program. Thematic analysis for the in-terviews and group discussions was conducted by twoindividuals separately based on emergent themes. Codesbased on emergent themes and the IBM WASH frame-work were then compared and discussed. The themesthat emerged where put in the following categories:current mobile health activities in Bangladesh, use ofmobile phone messages for CHoBI7 program delivery,mobile message type preference for CHoBI7mHealth program, sender of CHoBI7 mHealth pro-gram messages, message delivery and content, access toCHoBI7 mHealth messages, and timing and frequencyfor CHoBI7 mHealth message delivery.

ResultsBangladesh Government’s use of mobile healthGovernment stakeholdersAll government stakeholders interviewed describedprevious mHealth activities implemented by theBangladesh government for health awareness andhealthcare capacity building. Most of these mHealthmessages were developed in partnership with theHealth Education Bureau in the Ministry of Healthand Family Welfare. Respondents described a range ofmHealth activities including mobile messages sent tothe public during epidemic disease, and messages sentusing the voice of the Prime Minster for NationalImmunization Day, Community Clinic Day, andWorld Health Day. Health call centers were men-tioned where a “16263” phone number is used for thepublic to talk directly to doctors, and to request am-bulance services for a fee. A telemedicine programwas described for physicians in Dhaka to connectwith patients across the country. In addition, govern-ment stakeholder stated that all government healthfacilities receive a mobile phone for patients to texttheir complaints and suggestions 24 h a day.

Acceptability of mobile phone messages for CHoBI7program deliveryGovernment stakeholdersAll government stakeholders were receptive to theuse of mobile messages for CHoBI7 program deliv-ery. The high phone coverage in Bangladesh wassaid to be an advantage, allowing health awarenessmessages to reach many individuals across thecountry:

In my country right now 90% of our people areconnected with mobile phones. Using mobile phones isa very smart behavior for sharing information andknowledge. My people starting from the very poor tothe high [wealthy] people you just give them themessage, it is not a big challenge. They will take it.

Text messages were mentioned to be low in cost, andcould potentially be funded by the government:

For generating text messages, funding will be required.Since the funding is not big, if external funding isavailable, then better. But it is possible to do [this] bythe government funding.

Government stakeholders recommended CHoBI7mobile messages be sent through the ManagementInformation System Department in the Ministry ofHealth and Family Welfare and through the BangladeshTelecommunication Regulatory Commission. Severalgovernment stakeholders also mentioned that sendingCHoBI7 mobile messages through the governmentsystem would reduce the cost of message delivery.Government stakeholders recommended the CHoBI7

program be included in the upcoming National Oper-ational Plan for the Communicable Diseases Control(CDC) Department at the Ministry of Health andFamily Welfare. Integration of the CHoBI7 mHealthprogram into existing mHealth programs such as theHealth Call Centers was recommended. Mobile appli-cations were also recommended to deliver CHoBI7messages since this would align with componentsalready planned for the upcoming National Oper-ational Plan. Several government stakeholders recom-mended that CHoBI7 mobile messages be developedin partnership with the Health Education Bureau:

Request the Health Education Bureau, they will formthe message, taking the participation [of] you andother experts … who [will] prepare this message. And[they] will disseminate this message through [out] thewhole country and [the] funding required for thismessage will be, given from this [National]Operational Plan.

George et al. BMC Public Health (2019) 19:1028 Page 9 of 18

Patient householdsCHoBI7 mHealth messages were well-received by groupdiscussion and pilot interview participants. CHoBI7voice and text messages served as important remindersfor patients and their household members to performthe recommended behaviors after returning from thehealth facility. Participants liked that the messages werecoming from the hospital, and mentioned this broughtcredibility to the messages being sent:

Both IVR [voice] and [text] messages are good. Manypeople can learn from this and by saying thisinformation is from the hospital, we can let otherpeople know about this information as well.

It was really good to get a call after coming fromhospital.

A few participants said CHoBI7 messages would bebetter understood and well-received through in-personvisits or a combination of in-person visits and phonecalls, with one participant mentioning that not all prob-lems can be solved over the phone. A few participantsrecommended pictures or videos be added to mobilephone messages.

Voice of the CHoBI7 program mHealth messages: Dr.Chobi and AklimaGovernment stakeholdersGovernment stakeholders recommended a wide range ofindividuals to be the sender for CHoBI7 mobile mes-sages including the Prime Minster, the Health Minister,a parliament member, a health care provider, a high levelperson from icddr,b, a socialite, or celebrities such as acricket player or singer. A female voice was said to be at-tractive for message delivery. One respondent mentionedthat the visibility of the person sending the messageswas very important, and that the government is planningfor people to see the person who is sending voice andtext messages in future programs.

Character developmentTwo characters were developed to deliver themHealth messages: Dr. Chobi and Aklima. Dr. Chobiis a doctor at icddr,b hospital who called and textedparticipants to share information and reminders onhandwashing with soap and water treatment behav-iors. She was sometimes also called “Dr. Chobi Apa,”meaning “Sister Dr. Chobi.” Dr. Chobi introducedAklima, a woman who brought her child to a healthfacility for diarrhea treatment, and who learnedproper handwashing with soap and water treatmentbehaviors from Dr. Chobi. Aklima shared how sheused Dr. Chobi’s advice to successfully keep her

family safe from diarrheal disease. While Dr. Chobiwas introduced as a guide, Aklima served as a peerrole model for the participants designed to target de-scriptive norms—they could learn anything she couldlearn and also keep their families safe and healthy.

Patient householdsThere was a positive response for Dr. Chobi:

I could feel her [Dr. Chobi’s] words from my heart.

Participants in the first group discussion helped deter-mine Dr. Chobi’s age and sex. In the first group discus-sion, a recording was provided of a male and femalevoice for Dr. Chobi. Participants liked both voices. Par-ticipants in the first group discussion were also shown aphoto of a female Dr. Chobi in her 20s and a secondimage in her 30s. The majority of participants preferredthe image of the woman in her 30s, stating she had moregravitas. Based on the feedback from the first group dis-cussion, the Dr. Chobi in her 30s was selected as thecharacter for the CHoBI7 program, and images of herwere showed to participants in the pilot.Participants said Dr. Chobi’s name was easy to remem-

ber, and thought that all of the CHoBI7 program mes-sages should come from her. Participants alsorecommended that her name should show up on thecaller ID when phone messages were sent. Voice mes-sages were viewed as a more effective way to popularizeDr. Chobi because individuals could interact directlywith her and hear her voice. Participants in the pilot re-quested weekly sessions with Dr. Chobi to discuss theirhealth problems.Aklima was also well-received, with one participant

stating “I am Aklima—it’s me, not Aklima”, and anotherstating “It’s my life story.” Participants recommendedAklima share her knowledge with others:

Aklima was a patient like me and it is good that she istaking help from Chobi apa [sister] like us … Sheshould promote all this knowledge to others.

Text and voice messages for delivery of CHoBI7 programGovernment stakeholdersVoice and text messages were recommended by all gov-ernment stakeholders for CHoBI7 program delivery, ex-cept for one who recommended mobile messageshowever did not specify a type. Voice messages were rec-ommended because they could be more easily understoodthan text messages by individuals that could not read, withnon-literate individuals described as the ones needing theCHoBI7 program messages the most. Respondents recom-mended voice messages come as calls without requiring

George et al. BMC Public Health (2019) 19:1028 Page 10 of 18

files to be opened. Text messages had been used in previ-ous government programs, and were said to have a posi-tive impact and be important. Phonetic Bangla wasrecommended for text messages because many individ-uals were stated to not have smart phones.

Patient householdsThe response was positive for both voice and text mes-sages, with some participants recommending both voiceand text messages be sent:

Both voice and text [messages] are good. From calls wecan learn and if sometimes we forget, we can see the[text] message or tell others to read them for us. Thatis why both together are good.

Voice messages were viewed as better than text mes-sages for understanding message content. While theability to save text messages was mentioned as an advan-tage over voice messages that could not be saved. Textmessages were also described as better for sharing mes-sage content with others:

I can show this [text] message to 10 more people.

Some participants said text messages were faster toread and access than voice messages. However, text mes-sages were problematic for those that were not literate.Some participants said no one in their home could readmessages aloud to them. While other participants men-tioned that household members could read them CHoBI7program text messages.In the first group discussion, phonetic Bangla (Bangla

written in English letters) was used to send CHoBI7mHealth messages to participants based on the recom-mendations from government stakeholder interviews.Some Bangla words spelled using English characterssuch as the word for diarrhea, “patla paikhana”, werechallenging for participants to understand. Participantsrecommended all text messages be in Bangla script somessages were more easily understood.Participants recommended having text messages sum-

marizing the content of voice messages (summary textmessages), so if a voice call was dropped individualscould still receive CHoBI7 messages:

If the phone [voice message] comes first and themessage [text message] comes later this would be good.Then if you cannot remember the content of the call,you can read the message [text message].

The CHoBI7 IVR quiz questions had a positiveresponse:

I feel so happy when Chobi apa [sister] says-yes! Youare right [IVR quiz message response].

Participants enjoyed taking these quizzes, andmentioned discussing the content of these IVRquiz messages with neighbors. Some participants de-scribed previous experiences where money was “cut”from their phones when they replied to text mes-sages from health information lines and herbal com-panies. Participants mentioned being comfortablesubmitting quiz IVR responses because Dr. Chobistated it was free to reply during these messages.A few participants mentioned a preference for direct

phone calls over recorded voice messages. One partici-pant expressed concern that some people may not re-ceive CHoBI7 voice calls because they will think they areadvertisements from different companies. Some partici-pants had full inboxes and could not receive CHoBI7text messages, or needed to delete CHoBI7 text mes-sages to free space in their inboxes. Additional chal-lenges included lost phones, broken phones that didn’tallow participants to view messages, not knowing how toopen text messages, and not checking text messagesregularly. Some of the participants in the first group dis-cussion mentioned that voice messages were too longand should be shorten.

CHoBI7 mHealth message delivery and contentGovernment stakeholdersFor the CHoBI7 mHealth program, government stake-holders recommended giving only a few key messages,and recommended that these messages be very specificand based on scientific evidence. This was consideredimportant given the limited time people would likelygive for the CHoBI7 program once they returned fromthe hospital:

During their hospital stay they will give 80% of theirtime to health related issues but in their house theymay give 1% of their time …[give] a few messages …what you want to emphasize.

Government stakeholders recommended that CHoBI7mHealth messages be integrated with other health caremessages, such as those on antenatal care. Governmentstakeholders also recommended tailoring CHoBI7mHealth messages to high-risk populations to ensurethe messages were relevant to those receiving them. De-livering mobile messages during diarrhea outbreaks wasgiven as an example:

Sending messages out to everyone would be a wastageof money … [this] should be for times like during

George et al. BMC Public Health (2019) 19:1028 Page 11 of 18

diarrhea outbreaks and should be areas-based [and]people-based not general.

However, it was mentioned that individuals should bepracticing handwashing with soap always not only dur-ing outbreaks.

Patient householdsMessage content was found to be clear and participantssaid messages from Dr. Chobi were stated to be cultur-ally appropriate. Mobile messages from the CHoBI7program were considered to be important for spreadingawareness and improving health:

Everybody will see these messages and learn fromthem. They would then be free from cholera and otherdiseases.

During the first group discussion participantsreacted strongly to Mobile Message 4 in Additionalfile 1: Table S2 on “eating poop.” When asked if thismessage content should be changed participants saidthese words were true and should remain:

These are not bad words, and people will be morecareful after reading the text message. (MobileMessage 4 in Additional file 1: Table S2).

Participants understood and liked the messages pro-moting handwashing with soapy water, mentioning thesemessage were important for reducing diseases:

We always have to be clean to stay away fromdiseases. Like Dr. Chobi says, we should always washour hands. We should wash our hands right aftercleaning up children after toileting [defecating]. If wedo this we would be free from all diseases.

Access to CHoBI7 mHealth messagesPatient householdsParticipants shared messages from Dr. Chobi withspouses, family members, and neighbors. Participantsrecommended that those receiving CHoBI7 mobile mes-sages should share them with others including neighborsand family members to spread these messages:

Chobi apa [sister] taught us. We are Aklima, we willtell other people … They will gain a benefit from this.

Participants also recommended that CHoBI7 mobilemessages be sent to both husbands and wives. Femaleparticipants emphasized the importance of messagesfrom Dr. Chobi coming to their phone since they were

often the ones responsible for taking care of childrenand would like to maintain the promoted hygienepractices:

The person who is in charge of maintaining thehome--you should send [mobile messages] to her.

One participant said that although she has a mobilephone, these messages should be sent to her hus-band’s phone, because sometimes husbands may besuspicious of wives if an unknown person calls orsends them a text message. Another participantexpressed concern about the CHoBI7 program poten-tially sending messages to his wife’s phone.Several female pilot participants did not have their

own mobile phone. Some of these participants saidthat their husbands or other household membersshared CHoBI7 messages with them. Others reportedlooking at their husband’s phone to see these mes-sages. This was often at lunch or in the evening whenmale household members were home. However, sev-eral female pilot participants without mobile phonessaid that male household members were often notsharing the content of CHoBI7 messages with them:

If you send this information by mobile phone, itwould be good, but not everyone will come to knowthis [information]. Like, I don’t have a mobilephone. Then how would I know that you sent amessage? Suppose my husband has a mobile phoneand receives and reads those [text] messages butdoesn’t tell me anything about them? Don’t we haveto know about these [messages]? Of course, we haveto know about these messages.

Some female participants mentioned not being able tolisten to CHoBI7 voice messages because their husbandhad the only mobile phone in the household and wasaway at work when the messages were sent.

Timing and frequency for CHoBI7 mHealth messagedeliveryGovernment stakeholdersOne government stakeholder recommended CHoBI7mHealth messages be sent every two weeks for aduration of three months. No recommendations weregiven on the timing for message delivery.

Patient householdsEvening time was preferred by the most participants forreceiving mobile messages from the CHoBI7 program.The most popular time window was 4:00–8:00 PM. Thiswas the time when participants and their family

George et al. BMC Public Health (2019) 19:1028 Page 12 of 18

members, including husbands, were often at home andeating and bathing:

At night the phone remains with us. We do not goanywhere then, do not remain busy, and the phoneremains nearby. If a call comes we can receive itimmediately. And if a [text] message comes we can seeit. So if [the message] is sent at night it would be good.

Evening was also mentioned as an important time tostay clean. Participants described sharing text messageswith each other in the evening:

We received [text] messages at 5 pm ... We all readthose together in the night.

Most pilot participants preferred to receive CHoBI7phone messages at least once per week, and all pilot par-ticipants preferred to receive at least one mobile phonemessage monthly. Frequent phone messages were con-sidered important to make sure promoted behaviorswere maintained and not forgotten, and to make individ-uals more aware:

If you don’t send us text messages frequently, we willforget everything within 6 months. So, we will be ableto keep them in our mind if you send them frequently.

However it was also emphasized that messages shouldnot be sent too frequently, and should not cause inter-ruptions at work. One participant said that more thantwice monthly could be irritating. Participants’ recom-mended CHoBI7 mHealth messages be sent for 6months to 1 year.

Intervention developmentThe IBM-WASH framework was applied to the develop-ment of the CHoBI7 mHealth intervention based on quali-tative findings. This framework is shown in Table 2.Identifying the contextual, psychosocial, and technologicalfactors at each IBM-WASH level has allowed us to design amHealth intervention that targets the multi-level, multidi-mensional factors that are facilitating or impeding our rec-ommended WASH behaviors. An overview of revisionsmade to the CHoBI7 mHealth program based on formativeresearch findings is shown in Fig. 1. Additional file 1: TableS3 provides a detailed description of each intervention com-ponent added.

CHoBI7 mobile message typeBoth voice and text messages are included in the CHoBI7mHealth program based on the recommendations frominterview and group discussion participants. This includessummary text messages sent out after each voice message to

serve as reminders to perform the key behaviors (targetingremembering). Mobile messages are also sent out usingBangla script based on the recommendations from the firstgroup discussion. It was found that Bangla script can beviewed on feature phones without smart phone capabilities.

CHoBI7 mobile message senderBased on the recommendations of government stake-holders and participants in the first group discussion, afemale health care provider, Dr. Chobi, was selected tobe the sender of voice and text messages for the CHoBI7mHealth program. Given the iterative nature of messagedevelopment, using the voice of a government officialwas explored, however, was found not to be feasible be-cause of the repeated recordings of voice messages thatwas needed. A celebrity was also found to not be a feas-ible option for the same reason and because of the highcost for their time.Promoters at health facilities ask diarrhea patients and

their family members to save the number for Dr. Chobiin their phones upon enrollment in the CHoBI7mHealth program. In addition, all text messages showDr. Chobi as the sender. This was included to ensure in-dividuals receiving calls and text messages from the pro-gram know the sender. A tutorial is also given by healthpromoters in the health facility on how to open textmessages and respond to an IVR quiz message to re-spond to concerns about difficulties opening text mes-sages and responding to quizzes.

CHoBI7 mobile message deliveryIn an effort to increase access to mHealth messagesamong female caregivers, we added a module deliveredby a promoter in the health facility during the patient’sstay in the hospital for illness on mHealth message deliv-ery and access. During this module the promoter startsby asking diarrhea patients and their family memberswhich phones in the household should receive CHoBI7mHealth messages. The health promoter then recom-mends mobile messages be sent to the phone of the pri-mary caregiver of the child under 5 years in the home,when possible, explaining the importance of caregiversreceiving these messages for the health of their children(targeting nurture). Promoters also explain that a femalehealth provider will be contacting them by phone, nevera male. This module was included to try to alleviate con-cerns from husbands about voice and text messages be-ing sent to their wife’s phone, and based onrecommendations to send phone messages to the femalehousehold members.All CHoBI7 text and voice messages state “Please

share this message” to promote message sharing amongthose in the household without personal mobile phonesor not receiving program messages. This was added to

George et al. BMC Public Health (2019) 19:1028 Page 13 of 18

address challenges with message sharing identified in thepilot. Voice and text messages are sent at 5 PM, this wasrecommended as a time when most individuals werehome. CHoBI7 voice and text messages are sent tohouseholds at least once every 2 weeks for 1 year, basedon the recommendations from interviews. Voice mes-sages were shortened to approximately 2 minutes basedon feedback from the first group discussion. This mes-sage length was reported to be fine in the second groupdiscussion. All IVR quiz messages state “There is nocharge for your reply” based on pilot participant con-cerns that a fee would be charged for responding to IVRmessages.

DiscussionA theory- and evidence-based approach using qualitativeresearch methods was conducted to design the CHoBI7mHealth program. The formative research findings haveshown that the CHoBI7 mHealth program has high useracceptability and is feasible to deliver to diarrheapatients that present at health facilities for treatment inBangladesh. Semi-structured interviews with govern-ment stakeholders identified their perceptions and pref-erences for scaling the CHoBI7 mHealth program.mHealth workshops were an interactive approach todraft and refine mobile message content based on stake-holder preferences. Group discussions and semi-struc-tured interviews with diarrhea patients and their familymembers identified beneficiary perceptions of mHealthand preferences for delivery of the CHoBI7 mHealthprogram. The pilot of the CHoBI7 mHealth programallowed for an assessment of program feasibility and toidentify potential barriers to successful program imple-mentation. Applying the IBM-WASH framework tothe development of the CHoBI7 mHealth programallowed us to gain an understanding of the multi-level,multidimensional contextual, psychosocial, and techno-logical factors that were influencing program deliveryand the targeted WASH behaviors. This is the first pub-lished study, to our knowledge, to conduct formative re-search to design a WASH mHealth interventiondelivered directly to households to increase handwashingwith soap and water treatment behavior.Through government stakeholders interviews we

identified contextual and technological factors that willbe important to consider for scaling the CHoBI7mHealth program across Bangladesh. Integration of theCHoBI7 program into existing mHealth programs beingdelivered by the government, such as mobile messagingpromoting child and maternal health and health callcenters was recommended by government stakeholders.This could help to reduce the cost of program imple-mentation. Government stakeholders also stated thepotential for implementation of the CHoBI7 mHealth

program without external donor support, and recom-mended the CHoBI7 program be included in the up-coming National Operational Plan for the CDCDepartment. Inclusion of the CHoBI7 program in theNational Operational Plan and government fundingwithout external donor support will be crucial to thesustainability of this program if taken to scale acrossBangladesh.Government stakeholders also recommended deliv-

ering CHoBI7 mobile messages to high risk popula-tions during the times when they were mostsusceptible, such as diarrhea outbreaks. This is con-sistent with the CHoBI7 program approach to initiallydeliver mobile messages in the health facility to diar-rhea patients and their family members when they areat highest risk for diarrhea. This high-risk time iswhen WASH messages are likely to resonate the mostwith individuals and facilitate sustained behaviorchange [36, 40].There was a positive response for Dr. Chobi as the

sender of the CHoBI7 voice and text messages. Shewas viewed as a credible source of information, withparticipants reporting they could share the messagesthat they received from her with others since she wasa doctor at the well-known “Cholera Hospital”.Aklima was also well liked, with participants identify-ing with her as also being a mother who came withher child to the health facility for diarrhea treatment.Through the formative research it emerged that itwas important to have the sender of text and voicemessages be a known individual and female. This isconsistent with a previous study in Ghana that founda positive response with a female voice being used forIVR message delivery [46]. In Bangladesh, Pakistan,and Nepal, the use of the character Meena, a youngSouth Asian girl, for a UNICEF health awarenesscampaign, delivered by both television and printedcommunication materials, was immensely successfulin increasing handwashing with soap and water treat-ment [47, 48]. The success of Meena serves as an ex-ample of how female characters could be used tofacilitate WASH behavior change. Future studiesshould investigate how the identity of the sender ofmobile messages influences behavior for WASHprograms.Both text and voice messages were considered to be

important by beneficiaries and government stake-holders for delivery of the CHoBI7 mHealth program.Voice messages were seen as beneficial because theycould be understood by those that could not read.While text messages were viewed as being valuablebecause they could be saved and viewed later andshared messages with others. Text messages were alsostated to serve as important reminders to perform the

George et al. BMC Public Health (2019) 19:1028 Page 14 of 18

promoted WASH behaviors. This finding lead to thedelivery of voice messages followed by a summary textmessage in the CHoBI7 mHealth program. MostmHealth programs rely on either text or voice message,and rarely employ both [13]. Future studies should in-vestigate preferences around voice and text messagesfor delivery of WASH programs in other settings.Gender norms around mobile phone access and

message sharing emerged as important contextual andtechnological factors for delivery of the CHoBI7mHealth program. Some female household membersdid not have consistent access to mobile phones, andreported that male household members were not al-ways sharing CHoBI7 mHealth messages. Female par-ticipants emphasized the importance of receivingprogram messages since they were often the ones re-sponsible for taking care of young children, andwanted to keep their child healthy. This is an import-ant challenge to address, particularly if no home visitsare conducted to reinforce mHealth messages, andgiven our setting where female household membershave lower rates of phone ownership than malehousehold members. In Bangladesh, it is estimatedthat 82% of male adults are mobile owners comparedto only 55% of adult females [49]. This gender in-equity in phone ownership is likely driven by thepatriarchal family structure which is common inBangladesh [50]. The challenge of phone access isconsistent with findings from the Aponjon mHealthprogram [51]. This program in Bangladesh deliverstext messages to women during pregnancy on ante-natal care. Aponjon subscribers reported that callswere sometimes missed because someone else in thehousehold had the phone. Through the CHoBI7 for-mative research we were able to identify potential so-lutions to address the challenge of message sharingand access by delivering messages at a time whenmost household members were present, sending mes-sages to the phones of both female and male house-hold members (when possible), and throughdiscussing message sharing in the health facility andin mobile message content. The effectiveness of theseapproaches will be evaluated in our RCT of theCHoBI7 mHealth progarm. These findings will be im-portant given that previous studies have emphasizedthe importance of usability and accessibility for theeffectiveness and sustainability of mHealth interven-tions [52]. Our results highlight the importance ofconsidering gender dynamics during message develop-ment, and the need for intervention approaches thatensure gender equity in access to message content.An important contextual factor for delivery of text

messages for the CHoBI7 program is the literacy rate inthe target population. In Bangladesh, the literacy rate for

females over 15 years of age is 70%, and 76% for males[53]. Furthermore, in our current work we found that92% of households in slum areas of Dhaka have at leastone person in the home that can read and write (Per-sonal Communication: Christine Marie George). Giventhis high household literacy rate, text messages presentsa feasible approach for intervention delivery in our studysetting.This is the first study, to our knowledge, to use

IVR for sending quiz questions to assess knowledgeof WASH practices in a low and middle-incomecountry setting. IVR quizzes were well-received byboth group discussions and pilot interview partici-pants, with some participants even reporting sharingthe content of quiz questions with neighbors. Previ-ous studies have found that two-way text messagingwas associated with improved medication adherencepractices compared to one-way texting [46, 54]. Thecurrent RCT of the CHoBI7 trial will investigate ifresponding to IVR quiz questions and correct quizresponses are associated with increases in the pro-moted WASH behaviors and sustained behaviorchange.The use of mHealth for promoting WASH behaviors

presents a low-cost approach for program delivery thatdoes not involve the cost of frequent in-person visits byhealth promoters. The VIAMO platform was 0.028 USDper minute for recorded voice messages and 0.041 USDper text message in Bangladesh at the time the pilotstudy was conducted. Therefore, if text and voice mes-sages are delivered every 2 weeks over a 1 year period,the cost of message delivery would be less than 2 USDper phone. Previous formative research from the Apon-jon mHealth program found that the majority of preg-nant women and new mothers were willing to pay 0.025USD per mobile message [51]. Similar research is neededto evaluate willingness to pay for CHoBI7 mHealth mes-sages. Government stakeholders mentioned that thecost could be less for message delivery if mobile mes-sages are sent through the government system. Thisoption should be further explored.mHealth messages have the potential to serve as

important reminders to facilitate WASH behaviorchange that do not require home visits. However,there are no published studies to date, to our know-ledge, that have evaluated the impact of a WASHmHealth programs on WASH behavior change. Previ-ously, mHealth RCTs in low- and middle-incomecountries have focused mostly on reminders for medi-cation adherence and immunization visits [11–15].Our current RCT of the CHoBI7 mHealth programwill evaluate the impact of this intervention on sus-tained handwashing with soap and stored drinkingwater quality.

George et al. BMC Public Health (2019) 19:1028 Page 15 of 18

This study has several strengths. First, the engage-ment of both government stakeholders and beneficiar-ies during the formative research that informedintervention development. Through engaging both ofthese key stakeholders, both scalability and acceptabil-ity of the CHoBI7 mHealth program were assessed.Second, the 7-month length of our formative researchphase, which allowed for intervention developmentthrough an iterative process where changes weremade to CHoBI7 mHealth messages based on findingsfrom the pilot study. Third, the use of both groupdiscussions and semi-structured interviews, whichprovided complementary information. Fourth, the useof mHealth workshops, which were valuable to dis-cuss findings from semi-structured interviews andgroup discussions, and to draft and refine mobilehealth message content. Fifth, the use of health the-ory for intervention development.One limitation is that this study focused only on

households that reported mobile phone ownership,and therefore is not generalizable to households shar-ing phones with other individuals outside the home.In our current RCT, over 90% of diarrhea patienthouseholds screened for eligibility reported householdphone ownership (Personal Communication: ChristineMarie George). Future studies should evaluate the im-pact of the CHoBI7 mHealth program on householdsthat report sharing a mobile phone. A second limita-tion is that the pilot study was conducted only inDhaka, Bangladesh. Future studies should pilot theCHoBI7 mHealth program across all divisions ofBangladesh. A third limitation is that messages werenot sent out using the government’s mobile platform.The VIAMO platform was used given the iterative na-ture of message development. Future work should usethe government’s mobile platform.

ConclusionThis study presents a theory- and evidence-basedapproach that can be used for the development offuture WASH mHealth programs. The CHoBI7mHealth program presents a promising scalableapproach for which mobile messages can be sent to alarge number of households at minimal cost and canserve as valuable cues to action to facilitateWASH behavior change. The ongoing RCT of theCHoBI7 mHealth program is the first to evaluate theimpact of voice and text message reminders sent tohouseholds promoting handwashing with soap andwater treatment behaviors. The evidence generated bythis study will be used by the Bangladesh Ministry ofHealth and Family Welfare to determine the feasibilityof scaling the CHoBI7 mHealth program acrossBangladesh.

Additional files

Additional file 1: Table S1. Overview of Intervention Activities by Arm.Table S2. Mobile Messages Played During the First Group Discussion.Table S3. Overview of CHoBI7 Mobile Health Program Development:Themes, Formative Research Findings, and Intervention Components.(DOCX 38 kb)

AbbreviationsCHoBI7: Cholera-Hospital-Based-Intervention-for-7-Days; IBM-WASH: TheIntegrated Behavioural Model for Water, Sanitation, and Hygiene;icddr,b: International Centre for Diarrhoeal Disease Research, Bangladesh;IVR: Interactive voice response; mHealth: Mobile health; ORS: Oral rehydrationsolution; RANAS model: Risks, Attitudes, Norms, Abilities, and Self-Regulationmodel; RCT: Randomized controlled trial; WASH: Water, sanitation, andhygiene

AcknowledgementsWe thank the study participants and the following individuals for theirsupport with the implementation of this study: Professor Abul KhairMohammad Shamsuzzaman, Professor Be-Nazir Ahmed, Fosiul Alam Nizame,Khobair Hossain, Jamie Perin, Ismat Minhaz Uddin, Rafiqul Islam, Al-Mamun,Maynul Hasan, Kalpona Akhter, Khandokar Fazilatunnessa, Sadia Afrin Ananya,Akhi Sultana, Sohag Sarker, Abul Sikder, Shirin Akter, and Laki Das.

Authors’ contributionsCMG conceived of the study design, directed the study, conducted dataanalysis, and wrote the first draft of the manuscript. MA and SM directeddata collection for the study and reviewed the draft manuscript. FZ, AT, ET,TH, SR, TP, DAS, JH, FB, SM, MA contributed to the study design, datacollection and analysis, and reviewed the draft manuscript. AL and KZsupported the design of the study and reviewed the draft manuscript.MAHK, FM, and SA reviewed the draft manuscript. PW and EL reviewed themanuscript. All authors have read and approved the manuscript.

FundingThis study received financial support from the United States Agency forInternational Development (USAID-OAA-F-15-00038). The funder did nothave a role in the design, implementation, or evaluation of the study.

Availability of data and materialsThe transcripts analyzed during the current study are not publicly availablebecause they contain identifying participant data.

Ethics approval and consent to participateAll study participants provided written informed consent; consent comprisedadult participants (> 18 years of age) signing an informed consent and/orparental consent form, and children 12–17 years of age signing an assentform. If a study participant could not read, the consent form was read tohim or her, and the participant then was asked to document his or herconsent with an 'X' in the presence of a witness. All study procedures wereapproved by the research Ethical Review Committee of icddr,b (Dhaka,Bangladesh), and the Institutional Review Board of the Johns HopkinsBloomberg School of Public Health (Baltimore, MD, USA).

Consent for publicationWritten consent to publish this information was obtained from all studyparticipants.

Competing interestsCMG is a member of the editorial board of this journal. No authors havecompeting interests.

Author details1Associate Professor, Department of International Health, Program in GlobalDisease Epidemiology and Control, Johns Hopkins Bloomberg School of PublicHealth, 615 N. Wolfe Street, Room E5535, Baltimore, MD 21205-2103, USA.2International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b),Dhaka, Bangladesh. 3Ministry of Health and Family Welfare, Dhaka, Bangladesh.

George et al. BMC Public Health (2019) 19:1028 Page 16 of 18

Received: 8 November 2018 Accepted: 10 June 2019

References1. Estimates of global, regional, and national morbidity, mortality, and

aetiologies of diarrhoeal diseases. a systematic analysis for the GlobalBurden of Disease Study 2015. Lancet Infect Dis. 2017;17(9):909–48.

2. Esrey SA. Water, waste, and well-being: a multicountry study. Am JEpidemiol. 1996;143(6):608–23.

3. Saha D. Acute diarrhoea in children in rural Gambia: Knowledge, attitudeand practice, aetiology, risk factors and consequences among children lessthan five years of age. Dunedin: The University of Otago; 2013.

4. George CM, Perin J, Neiswender de Calani KJ, Norman WR, Perry H, Davis TPJr, Lindquist ED. Risk factors for diarrhea in children under five years of ageresiding in peri-urban communities in Cochabamba, Bolivia. Am J Trop MedHyg. 2014;91(6):1190–6.

5. Wolf J, Johnston R, Freeman MC, Ram PK, Slaymaker T, Laurenz E, Pruss-Ustun A. Handwashing with soap after potential faecal contact: Global,regional and country estimates for handwashing with soap after potentialfaecal contact. Int J Epidemiol. 2018. https://doi.org/10.1093/ije/dyy253.

6. Halder A, Tronchet C, Akhter S, Bhuiya A, Johnston R, Luby S. Observedhand cleanliness and other measures of handwashing behavior in ruralBangladesh. BMC public health. 2010;10(1):545.

7. Wolf J, Hunter PR, Freeman MC, Cumming O, Clasen T, Bartram J, Higgins JPT,Johnston R, Medlicott K, Boisson S, et al. Impact of drinking water, sanitationand handwashing with soap on childhood diarrhoeal disease: updated meta-analysis and meta-regression. Trop Med Int Health. 2018;23(5):508–25.

8. Luby SP, Agboatwalla M, Bowen A, Kenah E, Sharker Y, Hoekstra RM.Difficulties in maintaining improved handwashing behavior, Karachi,Pakistan. Am J Trop Med Hyg. 2009;81(1):140–5.

9. Zwane AP, Kremer M. What works in fighting diarrheal diseases in developingcountries? A critical review. World Bank Res Obs. 2007;22(1):1–24.

10. Qadri F, Ali M, Chowdhury F, Khan AI, Saha A, Khan IA, Begum YA,Bhuiyan TR, Chowdhury MI, Uddin MJ, et al. Feasibility and effectivenessof oral cholera vaccine in an urban endemic setting in Bangladesh: acluster randomised open-label trial. Lancet. 2015;386(10001):1362–71.

11. Zurovac D, Sudoi RK, Akhwale WS, Ndiritu M, Hamer DH, Rowe AK, SnowRW. The effect of mobile phone text-message reminders on Kenyan healthworkers' adherence to malaria treatment guidelines: a cluster randomisedtrial. Lancet. 2011;378(9793):795–803.

12. Cole-Lewis H, Kershaw T. Text messaging as a tool for behaviorchange in disease prevention and management. Epidemiol Rev. 2010;32:56–69.

13. Free C, Phillips G, Galli L, Watson L, Felix L, Edwards P, Patel V, Haines A. Theeffectiveness of mobile-health technology-based health behaviour changeor disease management interventions for health care consumers: asystematic review. PLoS Med. 2013;10(1):e1001362.

14. Higgs ES, Goldberg AB, Labrique AB, Cook SH, Schmid C, Cole CF, ObregonRA. Understanding the role of mHealth and other media interventions forbehavior change to enhance child survival and development in low- andmiddle-income countries: an evidence review. J Health Commun. 2014;19(Suppl 1):164–89.

15. Gibson DG, Ochieng B, Kagucia EW, Were J, Hayford K, Moulton LH, LevineOS, Odhiambo F, O'Brien KL, Feikin DR. Mobile phone-delivered remindersand incentives to improve childhood immunisation coverage andtimeliness in Kenya (M-SIMU): a cluster randomised controlled trial. LancetGlob Health. 2017;5(4):e428–38.

16. Key ICT indicators for developed and developing countries and the world(totals and penetration rates) [https://www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx]

17. GSMA: Mobile industry driving growth and enabling digital inclusion. 2018.18. Curtis V, Schmidt W, Luby S, Florez R, Touré O, Biran A. Hygiene: new hopes,

new horizons. Lancet Infect Dis. 2011;11(4):312–21.19. Michie S, Prestwich A. Are interventions theory-based? Development of a

theory coding scheme. Health Psychol. 2010;29(1):1.20. Inauen J, Mosler H-J. Developing and testing theory-based and evidence-

based interventions to promote switching to arsenic-safe wells inBangladesh. J Health Psychol. 2014;19(12):1483–98.

21. Contzen N, Meili IH, Mosler HJ. Changing handwashing behaviour insouthern Ethiopia: a longitudinal study on infrastructural and commitmentinterventions. Soc Sci Med. 2015;124:103–14.

22. Taylor N, Conner M, Lawton R. The impact of theory on the effectiveness ofworksite physical activity interventions: a meta-analysis and meta-regression.Health Psychol Rev. 2012;6(1):33–73.

23. Webb TL, Joseph J, Yardley L, Michie S. Using the internet to promotehealth behavior change: a systematic review and meta-analysis of theimpact of theoretical basis, use of behavior change techniques, and modeof delivery on efficacy. J Med Internet Res. 2010;12(1):e4.

24. Dreibelbis R, Kroeger A, Hossain K, Venkatesh M, Ram PK. Behavior change withoutbehavior change communication: nudging handwashing among primary schoolstudents in Bangladesh. Int J Environ Res Public Health. 2016;13(1):129.

25. Dreibelbis R, Winch PJ, Leontsini E, Hulland KR, Ram PK, Unicomb L,Luby SP. The integrated behavioural model for water, sanitation, andhygiene: a systematic review of behavioural models and a frameworkfor designing and evaluating behaviour change interventions ininfrastructure-restricted settings. BMC public health. 2013;13(1):1015.

26. Carpenter CJ. A meta-analysis of the effectiveness of health belief modelvariables in predicting behavior. Health Commun. 2010;25(8):661–9.

27. Mosler H-J. A systematic approach to behavior change interventions for thewater and sanitation sector in developing countries: a conceptual model, areview, and a guideline. Int J Environ Health Res. 2012;22(5):431–49.

28. Ajzen I. From intentions to actions: A theory of planned behavior: Springer; 1985.29. Rogers RW. A protection motivation theory of fear appeals and attitude

change1. J Psychol. 1975;91(1):93–114.30. Contzen N, Mosler HJ. Identifying the psychological determinants of

handwashing: Results from two cross-sectional questionnaire studies in Haitiand Ethiopia. Am J Infect Control. 2015;43(8):826–32.

31. Curtis VA, Danquah LO, Aunger RV. Planned, motivated and habitual hygienebehaviour: an eleven country review. Health Educ Res. 2009;24(4):655–73.

32. Weil AA, Khan AI, Chowdhury F, LaRocque RC, Faruque A, Ryan ET,Calderwood SB, Qadri F, Harris JB. Clinical outcomes in household contactsof patients with cholera in Bangladesh. Clin Infect Dis. 2009;49(10):1473–9.

33. George CM, Ahmed S, Talukder KA, Azmi IJ, Perin J, Sack RB, Sack DA, Stine OC,Oldja L, Shahnaij M, et al. Shigella Infections in Household Contacts of PediatricShigellosis Patients in Rural Bangladesh. Emerg Infect Dis. 2015;21(11):2006–13.

34. Black RE, Merson MH, Rowe B, Taylor PR, Abdul Alim AR, Gross RJ, Sack DA.Enterotoxigenic Escherichia coli diarrhoea: acquired immunity andtransmission in an endemic area. Bull World Health Organ. 1981;59(2):263–8.

35. Burrowes V, Perin J, Monira S, Sack DA, Rashid MU, Mahamud T, Rahman Z,Mustafiz M, Bhuyian SI, Begum F, et al. Risk Factors for HouseholdTransmission of Vibrio cholerae in Dhaka, Bangladesh (CHoBI7 Trial). Am JTrop Med Hyg. 2017;96(6):1382–7.

36. Figueroa ME, Kincaid DL, Sobsey N, Clasen T. Social, cultural and behavioralcorrelates of household water treatment and storage. Household watertreatment and safe storage; 2007.

37. Amin N, Pickering AJ, Ram PK, Unicomb L, Najnin N, Homaira N, Ashraf S,Abedin J, Islam MS, Luby SP. Microbiological evaluation of the efficacy ofsoapy water to clean hands: a randomized, non-inferiority field trial. Am JTrop Med Hyg. 2014;91(2):415–23.

38. George CM, Monira S, Sack DA, Rashid MU, Saif-Ur-Rahman KM, Mahmud T,Rahman Z, Mustafiz M, Bhuyian SI, Winch PJ, et al. Randomized ControlledTrial of Hospital-Based Hygiene and Water Treatment Intervention (CHoBI7)to Reduce Cholera. Emerg Infect Dis. 2016;22(2):233–41.

39. The Safe Water System: Free Chlorine Testing [http://www.cdc.gov/safewater/chlorine-residual-testing.html]

40. George CM, Jung DS, Saif-Ur-Rahman KM, Monira S, Sack DA, Rashid MU,Mahmud T, Mustafiz M, Rahman Z, Bhuyian SI, et al. Sustained Uptake of aHospital-Based Handwashing with Soap and Water Treatment Intervention(Cholera-Hospital-Based Intervention for 7 Days [CHoBI7]): A RandomizedControlled Trial. Am J Trop Med Hyg. 2016;94(2):428–36.

41. Devine J, Karver J, Coombes Y, Chase C, Hernandez O. Behavioraldeterminants of handwashing with soap among mothers and caretakers:emergent learning from Senegal and Peru. Washington, DC: Learning note,Water and Sanitation Program; 2012.

42. Porzig-Drummond R, Stevenson R, Case T, Oaten M. Can the emotion ofdisgust be harnessed to promote hand hygiene? Experimental and field-based tests. Soc Sci Med. 2009;68(6):1006–12.

43. Aunger R, Schmidt W-P, Ranpura A, Coombes Y, Maina PM, Matiko CN,Curtis V. Three kinds of psychological determinants for hand-washingbehaviour in Kenya. Soc Sci Med. 2010;70(3):383–91.

44. George CM, Biswas S, Jung D, Perin J, Parvin T, Monira S, Saif-Ur-RahmanKM, Rashid MU, Bhuyian SI, Thomas ED, et al. Psychosocial factors mediating

George et al. BMC Public Health (2019) 19:1028 Page 17 of 18

the effect of the CHoBI7 intervention on handwashing with soap: Arandomized controlled trial. Health Educ Behav. 2017;4(4):613–25.

45. Etikan I, Musa SA, Alkassim RS. Comparison of convenience sampling andpurposive sampling. Am J Theor Appl Stat. 2016;5(1):1–4.

46. Crowdsourcing citizen feedback on district development in Ghana usinginteractive voice response surveys. Making All Voices Count Practice Paper.Brighton: Institute of Development Studies [http://opendocs.ids.ac.uk/opendocs/handle/123456789/12716]

47. Chesterton P. Evaluation of the Meena communication initiative.Kathmandu: UNICEF; 2004. https://www.unicef.org/evaldatabase/files/ROSA_2004_800_Meena_Comm_Initiative.pdf

48. Anis F, White J. The Meena Communicative Initiative in Bangladesh. In:Inclusive Education: Springer; 2017. p. 61–76.

49. Rowntree O. The Mobile Gender Gap Report 2018: GSMA, Retrieved fromhttps://www.gsma.com/mobilefordevelopment/programmes/connected-women/the-mobile-gender-gap-report-2018; 2018.

50. Kabeer N. Between affiliation and autonomy: navigating pathways ofwomen's empowerment and gender justice in rural Bangladesh. DevChang. 2011;42(2):499–528.

51. Rajan R, Raihan A, Alam M, Agarwal S, Ahsan A, Bashir R, Lefevre A, KennedyC, Labrique A. MAMA ‘Aponjon’formative research report; 2013.

52. Aranda-Jan CB, Mohutsiwa-Dibe N, Loukanova S. Systematic review on whatworks, what does not work and why of implementation of mobile health(mHealth) projects in Africa. BMC Public Health. 2014;14:188.

53. Bangladesh. UNESCO.54. Wald DS, Butt S, Bestwick JP. One-way versus two-way text messaging on

improving medication adherence: meta-analysis of randomized trials. Am JMed. 2015;128(10):1139.e1131–5.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

George et al. BMC Public Health (2019) 19:1028 Page 18 of 18