pros in clinical care

Post on 15-Nov-2014

172 Views

Category:

Health & Medicine

3 Downloads

Preview:

Click to see full reader

DESCRIPTION

 

TRANSCRIPT

PROS IN CLINICAL CARE

Paul K. Crane, MD MPH

Associate Professor, Department of Medicine, School of MedicineAdjunct Associate Professor, Department of Health Services, School of Public HealthUniversity of Washington

pcrane@uw.edu

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression PROMIS 2 network / Clinical Practice

Subcommittee Future forecast

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression PROMIS 2 network / Clinical Practice

Subcommittee Future forecast

My background

UW med school (1997) Internal medicine internship (UW)

and residency (Barnes-Jewish) 1997-2000

Hospitalist and Health Behavior Research fellow 2000-01 (Washington U, St. Louis)

General Internal Medicine Fellow, MPH, UW (2001-03)

Interest in measurement

Decided as a Health Behavior fellow to study diabetes, depression, and health related quality of life

Determined that to know anything about that topic, had to know something about how to measure HRQL Lots of diabetes-specific HRQL scales, all of

which made claims that I did not understand And I thought I knew how to read the medical

literature!

Faries et al. paper

A paper on the Hamilton Depression Rating Scale looked at responsiveness of each item, which they defined as differences between placebo and active treatments with their drug

Certainly didn’t like that way of defining things

But it’s the item, not the scale! Faries D et al. The responsiveness of the Hamilton Depression Rating Scale. Journal of Psychiatric Research 2000; 34: 3-10.

GIM Fellowship: Psychometrics

Worked with Gerald van Belle during fellowship

We taught each other modern psychometrics Hambleton et al. (1991): Fundamentals of Item

Response Theory Embretson and Reise (2000): Item Response

Theory for Psychologists McDonald (1999): Test Theory: a Unified

Treatment Would add to this list:

Wainer et al. (2007): Testlet Response Theory and its Applications.

Fellowship psychometrics research - 1

K08 proposal, “Improving cognitive tests with modern psychometrics” – Alzheimer’s disease specific 3-year K award

Worked with item-level cognitive and PRO data Dan Mungas at UC Davis PROMIS I application on self-reported cognition

Not discussed – “cognition is not important to HRQL”

Became aware of a second UW PROMIS I proposal from Dagmar Amtmann

Psychometrics Research – 2 Dagmar’s project was funded, and

she was happy to have me involved as I wished during my K award.

So, I went.

PROMIS 1 Large project (7 U01 projects, 1 Statistical Coordinating

Center, each with an NIH Project Officer) One Danish physician (Jakob Bjorner) involved with one of

the projects No other practicing clinicians with modern psychometrics

expertise No projects really integrated with clinical care I felt the need to speak up! Network structure; first set of domains built by the Network

Depression; Anxiety; Alcohol abuse; Anger; Physical function; Fatigue impact / experience; Social role performance /satisfaction; Pain interference / quality / behavior

Pilkonis PA et al. Item banks for measuring emotional distress from the Patient-Reported Outcomes Measurement Information System (PROMIS®): depression, anxiety, and anger. Assessment 2011; 18: 263-283.

Some PROMIS strengths

Extraordinary amount of attention, with reasonable choices made

Item QC led to items that are relatively easy to read, simple to interpret, similarly worded, etc. Domain-to-domain look and feel is consistent

Scores across the domains look similar Direction based on name of domain Scaled so 50 is mean of US, +/- 10 is 1 SD

Attention to intellectual property Increasingly important consideration Proprietary items = $, risk for lawyer involvement

PROMIS Product: Short forms Brief group of items “Developed … based on simulations

of CAT results, item information, and item content” Candidate items identified based on

psychometric characteristics, and then reviewed by content experts (Pilkonis 2011)

PROMIS product: CAT

Efficient and brilliant use of computers You already know the CAT algorithm: pick a

number (“binary search”) Need an item bank, a scoring algorithm,

and a stopping rule Stopping rule can be composite

Result: precise-enough scores for a domain after a very few items (like 4 or 5)

If IRT assumptions are appropriate, it’s very slick! Assumption: all items are equally useful

PROMIS 2

1 coordinating center becomes 3 (technology center, statistical center, network center

12 PROMIS projects Structure different: no network

projects (or we’re all network projects)

And one of the 12 PROMIS projects was ours! We’ll come back to this in a bit.

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression PROMIS 2 network / Clinical Practice

Subcommittee Future forecast

Heidi Crane, MD MPH

UW undergrad, UW med school, Barnes-Jewish Medicine Residency, UW ID fellowship

K23 on body morphology disorder among people with HIV

Self-reported body morphology changes Tablets in the waiting room for people with

long waits ahead of them Other PRO domains on the assessment

Crane HM et al. Routine collection of patient-reported outcomes in an HIV clinic setting: the first 100 patients. Current HIV Research 2007; 5: 109-118)

Chart reviews for same-day visits

Depression: not identified by providers Substance use: not identified by providers Poor adherence: not identified by providers

And alarmingly high prevalence of patients who told the computer they were having problems, whose providers documented “No problems with adherence,” “Perfect adherence,” “Taking all meds”

Reviewed these findings with clinic leadership Imperative to measure these things and

make sure providers have access to the findings at the point of care

Integrating into routine care: not trivial Instead of patients with long anticipated wait

times for research protocol, change to all patients Except not all patients; super frequent fliers for wound

care excluded So who?

Offset the clinical day, so patients scheduled to interact with tablet 20 minutes before provider scheduled in the room Front desk implications, rooming staff / vitals

implications, … In short: a clinical change with an impact on

patient flow Importance of Clinic Leadership buy-in essential Other elements of case on succeeding slides

Source: http://www.improvingchroniccare.org/index.php?p=Chronic+Care+Model&s=124

Extension of Chronic Care Model

clinical information system, delivery system design, decision support “Listening to the patient’s voice in a

systematic standardized way” Delivering data to providers using 21st

century informatics tools Ultimate goals: Tailored, personalized,

evidence-based recommendations for clinical actions

Patient-provider relationship

Devote time during the clinic session to elements both patients and providers deem important

Clarify patient concerns Patients more honest to CASI than they are to a

provider, less social desirability bias. More likely to report to CASI poor adherence, substance abuse, depression, risk behavior than to provider Fredericksen R et al. Integrating a web-based, patient-

administered assessment into primary care for HIV-infected adults. Journal of AIDS and HIV Research

Figure 1A. Common situation in routine clinical care

Figure 1B. Situation with valid adherence measurement incorporated into clinical care

Poor adherence

Adherence not assessed

System not aware No

intervention

Continued poor adherence

Poor HIV outcomes

Structural barriers Provider barriers Patient barriers to assessment

Patient factors *substance abuse *mental illness *other

Poor adherence

Patient factors *substance abuse *mental illness *other

Adherence assessed

Intervention

Poor adherence

Good adherence

Better HIV outcomes

Adherence assessed

System aware

CNICS

Madison Clinic part of UW Centers for AIDS Research (CFAR)

CFARs banded together to form CNICS, the CFAR Network of Integrated Clinical Systems

Initial partnership with the 1919 Clinic at University of Alabama at Birmingham

http://www.positivethebook.com/

PROs at UAB

Paper free clinic Never a feedback form on paper Touch screens provider room with feedback; monitors

in exam rooms Aspects of PROs at every visit (ROS) Patient flow is different – patients in a physical

“circuit” around the clinic Patient flow at Madison was much more waiting room

-> vitals -> back to waiting room -> exam room Addressed the “circuit” with “ticket numbers” so

could pick up the PRO Assessment where they left off Addressed these differences with personal visits

PROs in CNICS

Extension to Fenway (Boston), UC San Diego (large Latino population; Spanish essential), others (UCSF, Hopkins, UNC, Case Western) Different EMR systems, different leadership /

clinic cultures, different patient groups, different patient flow

All PRO collection is local!

Our PROMIS 2 proposal CNICS infrastructure now with 30,000 PRO

Assessments from 8 CFARs around the country Extensive harmonized clinical data Uniquely situated in clinical care for PLWH Aim 1: PROMIS domains in clinical care

Ask patients which domains they think are most important

Focus on groups of items patients see, not the whole bank Simulated CAT, PROMIS short form As of April 2012, 809 studies set up in Assessment

Center; only 2 administered an entire bank

Other parts: 2 new domains; RCT on adherence; active involvement with the Network

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression

PROMIS 2 network / Clinical Practice Subcommittee

Future forecast

Depression domain

Simulated CAT: 5 items for severe, moderate, and mild depression

Compared with short form content

Short Form CAT: Mild CAT: Moderate, Severe Sad Sad Sad Unhappy Unhappy Unhappy Depressed Depressed Depressed Helpless Helpless Hopeless Worthless Like a failure Nothing to look forward to Discouraged about future Discouraged about future Disappointed in self

Analyses of depression domain Qualitative analyses

97 PLWH in 4 cities, stratified by depression severity “Repetitive,” “Redundant,” “Mas o menos lo mismo”

(both the Short Form and each of the simulated CATs)

“What would a provider need to know to take great care of a person with HIV? SUICIDALITY

Providers: Distinct preference for PHQ-9 content Patients: Distinct preference for PHQ-9 content

Quantitative analyses 1299 PLWH in 4 cities PROMIS and PHQ-9 work fine, nothing to distinguish

either one

WSCD (“What Should CNICS Do”)?

• Administer PHQ-9, score using PROMIS item parameters Gibbons LE et al. Migrating from a legacy

fixed-format measure to CAT administration: calibrating the PHQ-9 to the PROMIS depression measures. Qual Life Res 2011; 1349-1357.

Best of both worlds Content providers and patients want,

scores on PROMIS metric, brief enough

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression PROMIS 2 network / Clinical

Practice Subcommittee Future forecast

PROMIS Clinical Practice Subcommittee

Growing demand for PROMIS scales from clinicians Groundswell of understanding of need for PROMIS

focus on this issue “Sorry, I have to go, there goes my group, and I am

their leader!” Big initiatives to date: two papers (Broderick et al.

2013; Jensen et al. to be re-submitted) and EPIC Steering Committee vote: Work with EPIC to

ensure PROMIS content included Extensive discussion of which domains and

calibrations Initial build: short forms, scored using total scores

(not IRT scoring) Hopefully CATs in next build

PROMIS 2 Network

Large number of additional domains being developed (sexual function, self efficacy, substance use, GI symptoms)

Bigger efforts in pediatric settings (Cincinnati Children’s, CHOP, UNC)

Other initiatives: cancer, instantaneous assessment, broaden physical functioning to address ceiling, …

Less cohesive than PROMIS 1 (by design) in terms of projects

PROMIS standards document, domain framework, PROMIS at NIH Clinical Center, PROMIS in National Children’s Study, PROMIS in DoD care settings, ….

3 Coordinating Centers transitioning to ??? In future PROMIS, NIH Toolbox, NeuroQOL

RFA-CA-13-008 The purpose … is to support the creation of a research resource

infrastructure for the administration of research investigations using person-centered health outcomes … the Person-Centered Outcomes Research Resource (PCORR).

The PCORR will be expected to support the use and enhancement of the following four measurement information systems, currently funded as separate NIH programs: Patient Reported Outcomes Measurement Information System® (PROMIS®); the NIH Toolbox for Assessment of Neurological and Behavioral Function (NIH

Toolbox); the Quality of Life (QOL) Outcomes in Neurological Disorders (Neuro-QOL; and The Adult Sickle Cell Quality of Life Measurement Information System

(ASCQ-Mef). The main goal for the PCORR is to provide an integrated platform for

automated use of the four measurement information systems. This platform must be compatible with various modes of information

collection (including web/mobile-based entry, non-digital paper source data, and others).

The PCORR platform must also be designed to allow resource users (i.e., external researchers and clinicians unaffiliated with the resource) to access and use any of the four systems together or in isolation and tailor use to meet the specific study needs…

PCORI funding

Not surprising PCORI is interested in PROs Their default is that patients are the

experts Atlanta meeting last fall on

integrating PROs in EHRs PROMIS measurement RFP now Additional input on PCORI priorities

sought Additional funding initiatives likely

Network-ness

Outline

How did I come to this topic? Intro to PROMIS

How did my wife come to this topic? Intro to CNICS

PROMIS 2 research on depression PROMIS 2 network / Clinical Practice

Subcommittee Future forecast

“Prediction is very difficult, especially about the future” (N Bohr / Y Berra)

Growing demand for PROs in clinical care ACOs, PCORI, Quality measures, IOM,

Meaningful use…. Technological issues much less of a

barrier Ubiquitous tablets and iPhones Initiatives such as PROMIS developing a lot of

content Where should this head?

42

University of Utah Orthopedics

PROMIS colleague Nan Rothrock got me connected with Orthopedics faculty at U of Utah

Discussion with Dr. Darrel Brodke Chair interested in PRO collection x

years ODI, NDI, SF-36, PROMIS PF, EQ5D Floor effects of ODI and NDI Data warehouse, also scores at point

of care

Score interpretability

Clinicians are not innumerate people! mmHg, HCT, chemistries, creatinine,

saturations, SNPs, omics, MRI physics… How do we get used to all these

numbers and different scales? The old fashioned way – we use them!

Black box of score production Modern psychometrics, confirmatory

factor analysis, item response theory, graded response model, polytomous data, computerized adaptive testing, etc. There’s a whole science in there! But there is a whole science in producing the

creatinine value we use clinically too, and I don’t know what’s in that black box

I don’t think one needs expertise inside the black box to use the output from the black box to take great care of patients

Future will be longitudinal Power will come from integrating

PRO data alongside other clinical data

Conceptualize PRO collection as an extension of history taking Quantified history?

Not the end of the discussion but a launching point Launch from a deeper place than “So

how’s your depression been doing?”

Imagine this visit…

-18 -12 -6 010

30

50

70

90

0

50

100

Time (months)

De

pre

ssio

n s

core

(m

ea

n 5

0,

SD

1

0)

SS

RI

do

se (

mg

)

Next things to tackle

One size fits all Great place to start Heterogeneity across patients within a clinical

setting Primary care may be the hardest

Critical need to value patients’ time – can’t possibly collect everything on everyone

Measurement prior to visits vs. web-based More and more people are connected Critically important to reach those who are not!

Lessons learned

Patient care is local Stakeholder buy-in is critical for clinical change

to survive Patients appreciate being asked The only way to study the data is to have the

data Old questions of whether what we do makes a

difference 21st century technology to address those questions

Personalized care is not just omics, it’s got a person at the center Patient-provider relationship at the center of the

Chronic Care Model makes a ton of sense!

Too many great colleagues to thank

Thanks for the invitation! Thanks to Joan Broderick and other PROMIS Clinical

Practice Subcommittee authors on the eGEMS paper, which caught your eye and led to my talk today

UW/Madison Clinic Colleagues UW PROMIS Colleagues CNICS investigators Dagmar Amtmann and PROMIS 1 Gerald van Belle, Dan Mungas, Eric Larson, Ed

Wagner for mentorship My local shop: Laura Gibbons, Shubhabrata

Mukherjee, Elizabeth Sanders Funding from NIH Patients

top related