Applying Frameworks for Implementation Research to RealReal-World Examples: Improving Care For Depression Lisa Rubenstein, Rubenstein MD, MD MSPH 6/28/09 Academy Health, Chicago This Talk Purpose of implementation/QI research Moving from efficacy to routine care Depression p example p Other examples (in brief) Next talks will explain how the pathway works (theory) and how to learn on the wayy ((evaluation)) Implementation p Research is about Impact We want patients or community populations p p to experience p better health We achieve this by understanding what should be done (evidence), & figuring out how to achieve it in the real world Successful Systems y Embed Health Services Researchers Eg., VA, Kaiser, Group Health, Mayo Clinic, Harvard Community Health Plan, Geisinger, Rocky Mountain, National Health Service (Britain) Health services researchers help systems function as learning organizations Purpose of implementation science/QI research is to support this process But HSR Must Deliver Deliver… Success requires theory & methods – Partnership, teamwork – Understanding pathways into practice – Linking methods to real world conditions A Guiding G idi Framework Efficacy and Effectiveness What actions, maneuvers, processes p patient p health in “laboratories” improve run by researchers? – Efficacy research Which of these efficacious actions work (are effective), effective) or can be made to work work, in real world settings run by managers? – Effectiveness research Effectiveness Versus QI In effectiveness, the intervention aims generalizable models to create g In QI, the goal is to make the intervention work within a site or sites In truth, it is a spectrum Efficacy Routine Care Effectiveness Quality Improvement Efficacy Routine Care Effectiveness Quality Improvement Clinical ProblemSolving VA HSR&D Center for the Study of Of HealthCare Provider Behavior Efficacy Routine Care Problems, Adverse Outcomes Effectiveness Quality Improvement Clinical ProblemSolving VA HSR&D Center for the Study of Of HealthCare Provider Behavior Efficacy Basic Science, Epidemiology Routine Care Problems, Adverse Outcomes Effectiveness Quality Improvement Clinical ProblemSolving VA HSR&D Center for the Study of Of HealthCare Provider Behavior Depression: an Elephant in the Living g Room… Patient Factor: Depressed people are poor consumers… Provider & Organizational g Factors No systematic method of detecting depression p or assessing g symptoms y p Primary care doctors not trained Mental health specialists and primary care not linked/collaborative Pi Primary care practice ti structure t t nott adequate for depression 1990 – 2000: Effectiveness Phase 1 What didn’t work in randomized trials – Depression screening in primary care – Depression education – Computer decision support in primary care What did work – Practice redesign for collaborative care – Katon: ResearcherResearcher-delivered care management, patient selfself-management support active collaboration with MHS support, Effectiveness Randomized Trials Stage 1: Katon, Group Health (1996) Intervention 6 mos 88 Usual Care 6 mos 57 % Satisfaction 93 75 % Reduced depression p scores by 50% 74 44 % Approp. Antidepressants Effectiveness Randomized Trials Stage 2: Less Researcher Control Researchers did not directly touch the p practices However, intervention was still prepre-set – Tools and protocols designed by researchers – Researchers trained champions – Researchers evaluated Effectiveness Randomized Trials Ph Phase 2 2: AHRQ P Partners t iin Care C Study Int Cont Mean mental health41** 39 related QOL (MCS-12) % with symptoms of 42*** 52 major depression % still working 84 90* Rubenstein, Health _________________________ Services Research Research, 1999 *p<.05; **p<.005; ***p<.001 (intervention); Wells, JAMA, 2000 (evaluation) Additional Effectiveness Studies of Collaborative Care Meta-analysis of 36 randomized trials Metashows effectiveness (Gilbody,2006) Meta--analysis of over 10 randomized Meta trials shows costcost-effectiveness (Gilbody,2006) Collaborative care is a “Best Practice” However, still not incorporated into routine care 1995-2005: Q 1995QI Trials Phase 1 Hands--off CQI for depression: negative Hands St t d CQI with Structured ith access to t experts: t (Goldberg H,1998, J of QI; Solberg L,2001, J of Effective Clin Practice) – IHI Breakthrough Series inconclusive M, 2006, Psych Services) (Meredith – EvidenceEvidence-Based QI Kaiser/VA Kaiser/VA---Collaborative Collaborative care tools, structured QI, senior leader guidance: id Modest M d t + (Rubenstein LV, 2006, JGIM; Parker L, 2007, Qual Health Res) Our theory: QI teams needed more technical support Implementation and Spread: d QI Phase h 2 VA QUERI EBQI for depression 2001 – Translating Initiatives for Depression into Effective Solutions (TIDES) Eventually a series (WAVES, COVES, PONDS, HiTIDES, RIPPLES, AQUADEP, ReTIDES) – Chaney, Kirchner, Yano, Fortney, Owen, Williams, Liu, Ritchie EBQI: PDSA Cycles of Research-Based Approaches Plan the •Improvement •Data D t collection ll ti Do the •Improvement •Data collection •Analysis •Hold gains Plan D Do Act Study •Interpret Data 23 Researcher Role In EBQI EBQI: Q Clinical Management g as Decision Decision--Makers Senior leader expert panel to set g priorities p design Local sites PDSA to adapt the model Researchers support TIDES Work Group p Structure: Chronic Care Model Workgroups supported implementation – Informatics – Care management and patient self self-management support – Senior leaders – Provider education/decision support – Collaboration Felker BL,, 2006,, Prim Care Companion J Clin Psychiatry Research/Clinical / Partnership Majority of >100 project leaders were / g , not researchers clinical/managers, Majority of hours spent were by researchers Liu CF, CF 2009, 2009 Health Services Research Theories Underlying y g TIDES Quality improvement (Demming, IHI) Social Marketing (Andreasen) (Luck, (Luck Implementation Science, after July 2009) Innovations (Rogers) Re--Engineering (Goldratt) Re A ti research Action h (Kirchner, Parker, Yano et al) TIDES Primary Careand Clinic Sites 2006 VISN MAP of TIDES ReTIDES CBOC (4,906 PC Patients) VAMC VAMC (5,470 PC Patients) (12,963 PC Patients) VAMC (3,836 PC Patients) CBOC (7,604 P PC Patients) 9 New VAMC’s (90,000 PC Patients) CBOC (5 856 PC Patients) (5,856 2 New VAMC’s (40,000 PC Patients) CBOC (10,122 PC Patients) 2 New VAMC’s (40,000 PC Patients) 2 new VAMC’s (90,000 PC Patients) VAMC (5 355 PC Patients) (5,355 CBOC (12,329 PC Patients) CBOC (7,700 PC Patients) ReTIDES Spread Intervention Sites Control Sites TIDES Spread Results In 2005, 2005 spread TIDES to 50 sites in 4 VA multistate regions (some spontaneous) In 2006, TIDES/BHL became an official VA Central Office priority (Primary Care/Mental Health Integration Initiative) In 2008, 2008 TIDES/BHL became VA policy – “Handoff” to clinical care (Post, Katz, Agarwal) g ) – Minimal continued support by research team Triangulation: g Example p #1,, VA Patient Centered Medical Home Began as one pilot site 1989 – Evidence review for design Researcher local and national primary care evaluation Research/manager assistance teams National Directive 1994 – 40% implemented by 1996 (early adopters/early majority) Mandated 1998 Example p #2: VA Electronic Medical Record Developed in four pilot sites 1990 1994 – Extensive clinician & health services researcher involvement Merged CPRS system 1994 – 30 % adoption 1996 Mandated adoption 1998 Example p #3: Smoking g Cessation Randomized trials showed effectiveness g cessation clinics of meds,, smoking EBQI for depression resulted in emphasis on smoking cessation clinics – Low impact on primary care population in randomized trial (Yano, Sherman) Revised intervention emphasizing telephone support underway Efficacy Routine Care Effectiveness Quality Improvement