From QI to Research to Community and Back
Ken Wells, UCLA / RAND
QI Research: 1) QI Design and Evaluation
Research used to:
– Enhance the evidence-basis of QI program – Improve QI design and buy-in through (participatory) evaluation feedback – Describe program effects
Methods (Type 1)
Literature review /meta-analysis/ consensus development Intervention tool kits Partnerships with stakeholders Measurement strategies Design: Comparison groups, program assignment options, sampling strategies Real-world practice context
Example: Partners in Care (PIC)
Multi-site study 46 primary care practices (public & private) 181 primary care clinicians 1,356 depressed patients Patient outcomes measured over five years
PIC DESIGN: Clinics Were Randomized
QI-THERAPY
support for psychotherapy
QI-MEDS
support for medication management
Usual Care (UC)
PIC Interventions
Based on chronic disease management literature and provider change theory Practices provide in-kind resources Local teams supervise interventions “Depression nurses” support patient education, assessment, and treatment initiation Local teams support clinician education QI-Meds, trained nurses for follow-up QI, reduced copay for study therapy (CBT)
Probable Depression at One-Year
Source: Miranda et al. (2003)
50%
30%
QI Intervention Usual Care
White
African American
Hispanic
Personal Economic Benefits
Sources: Wells et al. (2000); Schoenbaum et al. (2001); Schoenbaum et al. (under review)
25-47 fewer days of depressiondisability 18-21 more employed days over two years
QI Research: 2) QI as Design Tool
Example: Estimating effects of guidelineconcordant care on outcomes Naturalistic implementation conditions Observational Analysis Methods
– Instrumental variables analysis (as-treated) of treatment effectiveness
Methods (2)
QI program design and study design meet assumptions for strong observational analysis (e.g., instrumental variable) Adequate sample size for as-treated goal Measurement strategy to match goal Real-world study context
PIC As-Treated Analysis: Effect of Appropriate Care on Clinical Status
Source: Schoenbaum et al. (2002)
Percent Clinically Depressed at 6 months
70% No appropriate care
24% Appropriate care
PIC As-Treated Analysis: Effect of Appropriate Care on Employment
Source: Schoenbaum et al. (2002)
Percent Employed at 6 months
53% No appropriate care
72% Appropriate care
QI Research: 3) Planning/Policy
Estimate outcomes and costs of taking QI to scale (MOS, Sturm and Wells, 1995) Adapt QI program for another context or population (Youth Partners in Care)
Target another societal goal (overcoming disparities, PIC 5-year analysis)
Methods (3)
Planning: Appropriate conceptual model, parameters and data sources Adaptation: Suitable population, and revised clinical and system model Societal Aims: Design and measurement modification
– Opportunity is key, not generalizability
Examples of Design Modifications for Youth PIC
Problem
Depressed adolescents have comorbidities that differ from those common in adults. Contacts with care manager are more sporadic than expected.
Solution
Comorbid problems specific to adolescence considered in treatment plan. Care managers use multiple phone contacts and conduct CBT over the phone.
Preliminary Findings for Youth Partners in Care (YPIC)
Sample: 450 depressed youths (13-21) visiting primary care providers Design: Patients randomly assigned to usual care or QI for depression – Outcomes: 10 %-point decline in severe depression at 6 months (CESD) under QI relative to usual care, p=.02
Source: Asarnow et al., in preparation)
Whites
80%
30%
UC
QI-MEDS
QI-THERAPY
Societal Goal: PIC Five-Year Follow-up; Reduction in Disparities
African Americans and Hispanics
80%
30%
Well / In Treatment Still Depressed
UC
QI-MEDS
QI-THERAPY
QI Research: 4) Community Impact
Methods: Long-term partnerships with practices, communities Action research models, mixed methods for evaluation Participatory partnerships to achieve uptake, sustainability Feasible outcome tracking
6/2004
Options for Proposed Project
Enhance evidence-basis for Intervention using chronic disease management literature and provider change theories
– Multi-modal design; social influence theory
Power evaluation for process effects and model expected health outcomes/mortality from parameters in existing efficacy literature Enhance practice/consumer participation in goal setting and evaluation design and analysis to increase uptake and sustainability