Implementation of the Patient Centered Medical Home and Quality of - PowerPoint PPT Presentation

Implementation of the Patient Centered Medical Home and Quality of Life in Patients with Multimorbidity Linnaea Schuttner, MD VA Puget Sound Healthcare System 1

MULTIMORBID PATIENTS ARE COMMON, VULNERABLE  Multimorbidity: > 2 chronic diseases in > 2 body systems National prevalence (~55-83%) • 6,012,186  High-cost, high-risk population Primary care patients  Different interaction with primary care o More in-person visits o Competing demands from illness, acute needs 5,095,532 (84.8%)  How to measure impact of care? Multimorbid in the VA o Not applicable for many quality metrics, diverse o Universal, patient-centered outcomes 2 Marengoni et al., Aging Research, 2011

AIMS Was implementation of the VA’s medical home model associated with health-related quality of life for multimorbid patients? Was this effect moderated by illness severity? Or by primary care visit number (i.e. “dose” of care)? 3

ASSESSING CARE FOR MM PATIENTS Measure Measure Implementation exposure to HRQoL for of the VA PCMH patients PCMH 2010 2011 2012 2013 2014 4

RETROSPECTIVE COHORT STUDY Inclusion criteria, from administrative data • 6,012,186 > 2 chronic conditions in > 2 body systems • Primary care in ‘13 -14 Primary care empaneled, 2012 • 27,813 Responded to patient experience survey 2013/14 • Survey Excluded: respondents 60 Non-Veterans 4,585 Non-MM 521 Not enrolled in PC 22,095 (95.4%) 677 Missing covariates MM patients included 5

EXPOSURE: PCMH IMPLEMENTATION  Patient Aligned Care Team (PACT) model is the VA’s version of the PCMH  PI 2 : The PACT Implementation Progress Index o Clinic-level measure of PCMH implementation o Composed of survey and administrative data o 8 domains (access, continuity, care coordination, comprehensiveness, self-management support, communication, shared decision making, staffing) o Higher scores associated with clinical quality, patient satisfaction, lower staff burnout  Total score divided into 5 categories (lowest to highest implementation) for our study 6 Nelson et al., JAMA Int Med 2014, 2017

OUTCOMES: HEALTH-RELATED QUALITY OF LIFE (HRQOL)  Patient experiences survey (PCMH-CAHPS) of outpatients o Stratified random sample of outpatients with encounters in past 1 month  Average response rate 45.4% in 2014  Includes validated measure of HRQoL: Short Form-12 (SF-12) o Raw scores transformed with validated algorithm o Physical and Mental component scores (PCS, MCS), 0-100 o MCID 2.2 points for PCS, 2.0 points for MCS 7 Ware et al., 1996 Samsa et al., 1999

STATISTICAL ANALYSIS Covariates Generalized estimating equation models • Patient characteristics Clinic Factors Clustered by clinic • Exchangeable correlation, identity link • Age MDs / 10k patients Outcomes as marginal means • Sex Hospital/community Test of trend by ANOVA • Race Rural or urban • Missing responses imputed with Marital status Census Division Location modified estimation regression Co-pay exemption • Survey weighting for non-response bias Household income (by county) Education *Quarter of survey response also included 8 Spiro et al., 2004

ADDITIONAL ANALYSES • Domains of PACT • Comparing 3 levels of implementation ❖ Lowest (< 25%) v. Average (25-75%) v. Highest (> 75%) implementation • Effect moderation as interactions Hospitalizations (2012) = Proxy for illness severity, differences in care • Total primary care clinic visits (2012) = Dose-relationship • 9

PATIENT DEMOGRAPHICS (2012) Overall Mean (SD) * Lowest Highest Implementation Implementation N = 22,095 N = 1,879 N = 2,075 Age, y 68.4 (11.1) 67.5 (11.2) 69.1 (10.9) Male, No. (%) 21,189 (96) 1,796 (96) 2,018 (97) Non-Hispanic white, No. (%) 18,345 (83) 1,481 (79) 1,784 (86) Total chronic diagnoses 4.4 (1.7) 4.2 (1.6) 4.3 (1.7) Primary care visits 4.6 (4.4) 4.6 (4.6) 4.8 (4.5) Mental health visits 2.9 (10.0) 2.8 (8.8) 2.4 (7.5) Hospitalizations 0.09 (0.41) 0.08 (0.36) 0.09 (0.41) 10 Mean, SD except where indicated

CHANGE IN HRQOL BY LEVEL OF PCMH IMPLEMENTATION (ADJUSTED MARGINAL MEANS) 45.00 Physical HRQoL: P trend < 0.001 43.00 42.3 Low to High: 2.1 MCS/PCS Score 41.00 40.8 40.4 40.3 40.3 39.00 Mental HRQoL 37.00 36.30 36.20 P trend = 0.03 36.00 36.00 35.20 35.00 Low to High: -0.8 33.00 1 2 3 4 5 Low High 11 Implementation quintile of PCMH

PRIOR HOSPITALIZATIONS (2012) MODERATE EFFECT FOR MCS P = 0.01 P = 0.94 6.0 Physical HRQoL 4.0 Change in Score 2.1 2.7 2.0 Mental HRQoL 0.1 0.0 -1.2 -2.0 -4.0 No Hosp No Hosp Hosp No Hosp Hosp 12 No findings for primary care visits

DOMAIN ANALYSES PCS Domain Average PI 2 Top PI 2 P Mean, SE Mean, SE Trend Continuity 0.20 (0.40) 0.64 (0.52) 0.01 Communication 0.47 (0.42) 0.69 (0.46) <0.001 Shared decision-making 0.63 (0.44) 1.06 (0.51)* <0.001 13 No findings for remaining 5 domains

LIMITATIONS  Observational study (Unobserved/hidden bias)  Limited to those surviving over study period / retained in care  Limited to receipt of care in VA only 14

CONCLUSIONS  Greater PCMH implementation associated with higher physical quality of life o Driven by domains of shared decision making, communication, and continuity o No difference when varied threshold to 3+ chronic diseases o Difference of 2.1 ~ minimal clinical difference  Effect for mental quality of life did not meet threshold for minimal clinical difference o Until effect moderated by prior hospitalizations (2.7 points higher) – qualitatively different interaction? o PCMH not same as behavioral health integration, no specific processes for MH in PI 2  No differences in effect from number of visits to primary care (exposure dose) 15

IMPLICATIONS  Implementation of the medical home model is associated with higher HRQoL for MM patients, but may be variation in subgroups by severity of illness, prior utilization  Outcomes in multimorbid patients challenging, but important – vulnerable, diverse, high prevalence 16

ACKNOWLEDGEMENTS Karin Nelson, MD, MSHS • Ashok Reddy, MD, MS • Questions/comments: Ann-Marie Rosland, MD, MS • linnaea.schuttner@va.gov @LSchuttner Edwin Wong, PhD, MA • VA HSR&D Fellowship (David Au, K. Nelson, Ann O’Hare) • Primary Care Analytics Team (PCAT; esp. L. Taylor, I. Curtis, A. Mori, & R. Orlando) • PCAT High-Risk Investigator Network (esp. E. Chang, D. Zulman, & M. Maciejewski) • Disclosures: This work was undertaken as part of the Veterans Administration’s Primary Care Analytics Team (PCAT), with fundi ng provided by the VA Office of Primary Care. Funding for the primary author was through an Advanced Physician Fellowship through the VA Office of Academic Affairs. 17 The views expressed are those of the authors and do not necessarily reflect those of the VA.

QUESTIONS: ADJUSTING FOR DISEASE SEVERITY  Conceptual model  PCMH → total disease burden  Sensitivity analysis including total diseases, CAN o No difference Burden ? PCMH Quality 18

DOMAIN ANALYSIS PCS MCS Domain Average PI 2 Top PI 2 P Average PI 2 Top PI 2 P Mean, SE Mean, SE Trend Mean, SE Mean, SE Trend Continuity 0.20 (0.40) 0.64 (0.52) 0.01 -0.23 (0.31) -0.67 (0.41) <0.01 Communication 0.47 (0.42) 0.69 (0.46) <0.001 -0.19 (0.32) -0.48 (0.39) <0.001 Shared decision- 0.63 (0.44) 1.06 (0.51)* <0.001 -0.43 (0.35) -0.79 (0.43) <0.01 making 19