Is Patient Activation Associated with Future Glycemic Control and Utilization Outcomes for Adults with Diabetes? Advancing Know ledge to Im prove H ealth David M M. Mosen Mosen, PhD PhD, MPH Carol L. Remmers, PhD, MPH Judy Hibbard, Hibbard PhD © 2010, Kaiser Permanente Center for Health Research B k Background d • Patient Activation refers to an individual’s: • Skills and abilities to manage their own health • Ability to engage health providers in shared decisiondecision making practices • Increased levels of patient activation may improve patient outcomes • Some evidence to suggest that Patient Activation Measure (PAM) is associated with improved outcomes • Further work needed to understand association with future glycemic control and utilization measures © 2010, Kaiser Permanente Center for Health Research Preliminary Results: Logistic and OLS Regression ResultsAssociation of PAM1 with Significant g Non Disease-Specific p Outcome Measures: DM Population* PAM Stage Use of > 1 SM Resources OR Stage 3-41 1Reference 95% CI 1.48* 1.15-1.91* No Missed Medications in Past 7 Days OR 95% CI 1.51* 1.15-1.98* SM Behavior Index Beta Coeff PCS Beta Coeff MCS Beta Coeff +0.29*** +0.09** +0.08* group = PAM Stage 1-2. * p < .05 ** p < .01 001 *** p < .001 *Mosen DM, Schmittdiel J, Hibbard J, Sobel D, Remmers C, Bellows J. Is Patient Activation Association with Outcomes of Care for Adults with Chronic Conditions? J Ambulatory Care Manage. Vol 30, No 1, pgs 21-29. © 2010, Kaiser Permanente Center for Health Research St d Obj Study Objectives ti • Among adults with Diabetes Mellitus (DM), identify the association of PAM with: • Glycemic control • All-cause inpatient admissions • DM-specific DM ifi inpatient i ti t admissions d i i • Acute myocardial infarction (AMI)specific inpatient admissions © 2010, Kaiser Permanente Center for Health Research 2004 CMI Self-care/ Sh d Decision-making Shared D i i ki Survey S • Survey assessed several domains across five chronic condition cohorts: • • • • • • • Patient activation System support for shared decision-making Medication adherence Use of self-management services Performance of self-management behaviors Functional status Patient satisfaction • Information intended to be used for quality improvement purposes • Seven of eight KP Regions participated • Group Model Health Maintenance Organization (HMO) serves 8.7 million members across United States • Survey funded by KP’s Care Management Institute (CMI) • Development process guided by experts in the field (e.g., Judy Hibb d K Hibbard, Kate t L Lorig, i R Russ Gl Glasgow)) © 2010, Kaiser Permanente Center for Health Research Overview of KP’s Care M Management t Institute I tit t (CMI) Mission: • Improve quality and efficacy of care delivered to members with specific health conditions, including asthma and di b t diabetes Goals: • Ensure that effective care processes are uniformly and consistently delivered to KP members • Improve knowledge inside and outside of KP about effective health-care delivery, including new approaches to physician and staff learning • Advance understanding of how patient-provider interactions impact quality of care and outcomes • C Collect ll t h health-related lth l t d quality lit off life lif (HRQOL) iinformation f ti on patients with chronic conditions © 2010, Kaiser Permanente Center for Health Research M th d Methods • Survey of 1 1,180 180 persons with diabetes completed in fall 2004 / early winter 2005 • Mode of administration = mail/telephone survey • Survey sample identified using CMI diabetes inclusion criteria: • 1/96 through 12/03 • • • • > 1 DM drug dispensing > 1 DM-related hospital admission > 2 DM DM-related l t d outpatient t ti t visits i it in i 24-month 24 th period i d HgbA1c value > 8% • 1/01 through 1/03, either test result: • Fasting blood sugar > 125 mg/dl • Random blood sugar > 200 mg/dl • Survey data linked with 2006 administrative data • Glycemic control and utilization measures obtained from CMI’s CMI s Clinical Outcomes, Reporting and Evaluation (CORE) Database • Measures routinely used for performance reporting © 2010, Kaiser Permanente Center for Health Research S Survey Process P Fl Flow Initial sample N = 3,000 Contact rate N = 2,250 (75.0%) Response rate N = 1,433 (63.7%) Initial Analytic sample N = 1,306 (91.1%) Maintained continuous health plan coverage 36 post-surveyy months p N = 1,180 (90.4%) © 2010, Kaiser Permanente Center for Health Research Patient Activation to Perform Self-Management Behaviors (Independent Variable) • PAM • Developed by Judy Hibbard (University of Oregon) • 22 items total • 4-item scale (1= ( disagree strongly, 4 = agree strongly)) • Scoring algorithm applied • 0 = lowest activation • 100 = highest activation • Created using Rasch methodology • 1-dimensional interval-level Guttman-like scale © 2010, Kaiser Permanente Center for Health Research Four Stages of PAM and E Example l Items It from f Each E h Stage St • Stage St 11: N Nott yett believing b li i active ti role l iis important i t t • Ex. Question: Taking an active role in my own health care is the most important factor in determining my health and ability to function • Score < 47.0 • Stage 2: Lacking confidence and knowledge to take action • Ex. Question: I am confident that I can tell my health care provider concerns I have even when he or she does not ask • Score > 47.1 and < 55.1 • Stage g 3: Beginning g g to take action • Ex. Question: I have made the changes in my lifestyle (such as to diet and exercise routines) that are recommended for my health condition • Score > 55.2 and < 67.0 • Stage 4: Maintaining behaviors over time • Ex Question: I am confident that I can maintain lifestyle changes in my diet and exercise routines even during times of stress • Score > 67.1 © 2010, Kaiser Permanente Center for Health Research Utilization Outcome Measures • 2006 and 2007 Glycemic Control Measures • HgA1c < 8% (vs. > 8%) • 2006 and 2007 Utilization Measures All-cause cause inpatient admissions ((> 1 vs. • All none) • DM-specific admissions (> 1 vs. vs none) • AMI-specific admissions (> 1 vs. none) © 2010, Kaiser Permanente Center for Health Research P di t Variables Predictor V i bl • Age A (continuous) ( ti ) • Gender (male vs. female) • Race/ethnicity • White vs. African-American, Hispanic-non-white, Asian-American, Other/unknown • Educational attainment • LLess th than hi highh school h l vs. hi highh school h l graduate, d t some college/technical ll /t h i l school or higher, unknown • Self-reported health status • Fair/poor/very poor vs. vs good/very good/excellent • Comorbidity status • None vs. 1 or more comorbidities • Geographic location • California region vs. non-California region • 2006 Insulin or oral hypoglycemic dispensings • 0 vs. vs 11-5, 5 66-10, 10 11+ © 2010, Kaiser Permanente Center for Health Research A l i Analysis • Descriptive statistics • Bivariate analysis • Logistic regression models constructed to assess independent association of PAM with outcome measures for significant associations in bivariate analysis (p < .05) 05) • predictors included in final models had significant relationship with outcome measures (p < .10) 10) © 2010, Kaiser Permanente Center for Health Research D Descriptive i ti Statistics St ti ti Study Population Characteristics Mean Age +/- S.D. Male (%) Total Sample (N = 1,180) 61.7 +/- 9.4 55.3 Race/Ethnicity (%) White African-American Hispanic non-white A i A i Asian-American Other/unknown 51.4 15.7 10.2 10 8 10.8 11.9 Educational Attainment (%) Less than high school High school graduate Some college or higher Unknown 8.0 19.3 60.8 11.9 Comorbidity Status > 1 comorbidities (%) 41.6 Self-reported health status Fair/Poor/Very Poor (%) Good/Very Good/Excellent (%) 34.7 65.3 © 2010, Kaiser Permanente Center for Health Research Descriptive Statistics: Insulin Dispensings, PAM and Outcome Measures 2006 insulin or oral hypoglycemics dispensings (%) 0 1-5 6-10 11+ Mean PAM Measure +/- S.D. (0=lowest, 100=highest) % Stage 1 % Stage 2 % Stage 3 % Stage 4 23.2 29.8 27.5 19 5 19.5 57.1 +/- 9.9 11.7 44.4 26.5 17.4 2006 Outcome Measures (%) HgA1c < 8% All-cause admissions DM-specific admissions AMI-specific admissions 68.5 21.4 19.2 1.4 2007 Utilization (%) HgA1c < 8% All-cause admissions DM-specific DM specific admissions AMI-specific admissions 69.3 19.8 19.5 2.0 © 2010, Kaiser Permanente Center for Health Research Bivariate Results: Association of PAM with 2006 Outcome Measures PAM Stage HgA1c < 8 ((%)) > 1 All-cause > 1 DM Admissions ((%)) Admissions ((%)) > 1 AMI Admissions ((%)) Stage 1-2 64.5 25.2 21.9 1.8 Stage 3-4 72.9 17.8 17.2 1.1 p-value .004 .004 .05 NS © 2010, Kaiser Permanente Center for Health Research Bivariate Results: Association of PAM with 2007 Outcome Measures PAM Stage HgA1c < 8 ((%)) > 1 All-Cause > 1 DM Admissions ((%)) Admissions ((%)) > 1 AMI Admissions ((%)) Stage 1-2 68.8 18.8 19.5 3.0 Stage 3-4 71.3 21.8 20.7 0.9 p-value NS NS NS .03 © 2010, Kaiser Permanente Center for Health Research Logistic Regression Results: g Association of PAM with Significant 2006 and 2007 Outcome Measures PAM Stage 20061 HgA1c < 8% OR 95% CI 20062 > 1 Hosp. Admissions OR 95% CI 20063 > 1 DM Admissions OR 95% CI 20074 > 1 AMI Admissions OR 95% CI Stage 1-2 1.00 NA 1.00 NA 1.00 NA 1.00 NA Stage 3-4 1.50 1.16 -1.94 0.58 0.43-0.78 0.66 0.49 – 0.90 0.32 0.12 – 0.89 1 Model adjusted for age, race/ethnicity, and geographic location adjusted for age, gender, race/ethnicity, HgA1c value (2006), and health status 3 Model adjusted for age, gender, race/ethnicity, HgA1c value (2006), geographic location, and health status 4 Model adjusted for geographic location and health status 2 Model © 2010, Kaiser Permanente Center for Health Research C l i Conclusions • Glycemic Control • Higher levels of PAM (stage 3-4) independently associated with higher HgA1c control in 2006 • Utilization outcome measures • Higher levels of PAM (stage 3-4) independently associated with: • Lower all-cause hospital admissions in 2006 • Lower DM-specific hospital admissions in 2006 • Lower AMI-specific AMI specific hospital admissions in 2007 © 2010, Kaiser Permanente Center for Health Research Li it ti Limitations • Results cannot be generalized beyond ggroup-model p HMO settingg • Self-reported medication adherence • PAM only measured during one time period © 2010, Kaiser Permanente Center for Health Research I li ti Implications ffor IInterventions t ti • Implications for interventions • Further efforts are needed to develop and implement interventions to increase PAM scores • Determine whether movement on PAM scores results in improvement in quality of care/quality of life measures © 2010, Kaiser Permanente Center for Health Research Candidate Intervention: KPNW – S b itt d R01 to Submitted t AHRQ • Use the PAM tool and structured health coaching and motivational interviewing (MI) among adults with diabetes • Health interventionist would train clinical teams how to effectively use PAM • Health interventionist would use MI and follow up with patients over 6-month period • Methods • 12 clinical teams randomized (6 intervention, 6 usual care) • Target population: • Persons with diabetes and poor HgA1c control (> 8%) • Total sample size: 600-900; 300-450 each group • Primary outcomes: • PAM: baseline, 6 and 12 months • Secondary outcomes (assessed baseline and 12 months) • • • • Health-related quality of life (measured by health utilities index [HUI]) Clinical indicators (glycemic control, blood pressure, lipids, weight) Medication adherence Utilization (emergency department and/or hospital admissions) © 2010, Kaiser Permanente Center for Health Research A Appendix: di Demographics D hi andd PAM Demographic characteristic Gender (%) Male Female Age (%) 18-44 45 64 45-64 65-74 75+ Race/Ethnicity (%) White African-American Asian-American Hispanic, non-white Education (%) Less than High School High School Graduate Some College/Tech College Grad+ High PAM (Stage 3-4) p value NS 40.6 42.1 NS 36.4 42 7 42.7 41.1 38.7 NS 42.1 38.6 43.5 39.0 < 0.001 32.3 36.7 38 9 38.9 49.2 © 2010, Kaiser Permanente Center for Health Research