Combining Data from Medicare HOS and Medicare Advantage CAHPS® Academy Health Annual Research Meeting June 30, 2009 Richard D. Hector, MA, MPH, PhD Health Services Advisory Group Goal To demonstrate that linking Medicare Health Outcomes Survey (HOS) and Medicare Advantage Consumer Assessment of Healthcare Providers and Systems (MACAHPS®) survey data produce patient-related outcomes and perception / experience of care information that is actionable, meaningful and relevant ® CAHPS is a registered trademark of the Agency for Healthcare Research and Quality (AHRQ). 2 When one is not enough… • In addition to the usual demographics, the HOS routinely collects patient related outcomes • In addition to the usual demographics, the MA-CAHPS routinely collects “perception / experience of care” • Neither instrument by itself can tell you whether positive perceptions are associated with more desirable patient related outcomes 3 Advantages of HOS Data • Unique identifier for linking data • Outcomes are calculated • Baseline and Follow up data • Items are validated * * * * * * * Physical Component Summary (PCS) Score Mental Component Summary (MCS) Score Falls General Health Chronic Diseases Urinary Incontinence Osteoporosis 4 Advantages of MA-CAHPS Data • Unique identifier for linking data • Standard core questions • Usable comparative information • Items are validated * Number of personal MD visits * Number of clinic visits * Number of specialists seen 5 Advantages of Both • Trained interviewers • Established sampling protocol • Results that can be compared • Large sample size (power) 6 Trade-offs Due to Merging • The merged data set is not representative of HOS or MACAHPS • The merged data set is a convenience sample from Medicare Advantage beneficiaries • Data do not apply at the level of health plan (unless merged on health plan identifier) • The 95% confidence intervals would be wider than HOS or MA-CAHPS 7 HOS vs. MA-CAHPS Analytic Data Set HOS MA-CAHPS Changes in health status of enrollees Patient-centric care Patient reported outcomes Patient evaluated experience with health care Mixed mode (mail/phone) Mixed mode (mail/phone) Random sample Random sample 1,000 sample size per plan 300 completed surveys per plan ≥ 65 years ≥ 65 years Fielded: April – July (current year) Fielded: January – March (previous year) Enrolled for six months Baseline with two-year follow up 8 Data Considerations • Hopefully both surveys capture major life events • Both data sets have the same “merge by” variable – Type and size are the same for the “merge by” variable • Expect fewer observations in resulting (merged) data set – Persons that responded to both surveys 9 Link HOS to MA-CAHPS Data HICNUM Clinic visits MD visits Specialists visits 12345 1 2 0 12346 0 1 1 12347 0 1 0 …….. …. ….. ….. MA-CAHPS Use HIC numbers to merge data sets HICNUM Age Gender GenHlth PCS MCS 12345 72 Female Fair 55 67 12346 66 Male Good 60 61 12347 65 Female Very good 75 80 …….. …. …. …. …. …. HOS 10 Direction of Linking Data HICNUM Age GenHlth Clinic MD visits Specialists 12345 71 Very good 1 2 0 12346 65 Excellent 0 1 1 12347 64 Excellent 0 1 0 …….. …. ….. …. ….. ….. MA-CAHPS Use HIC numbers to merge data sets HICNUM Age Gender GenHlth PCS MCS 12345 72 Female Fair 55 67 12346 66 Male Good 60 61 12347 65 Female Very good 75 80 …….. …. …. …. …. …. HOS 11 Examples for Demonstration • HOS distribution of obesity by race and gender • PCS score by BMI level • PCS score by experience of care (MA-CAHPS) • Health services utilization (MA-CAHPS) by number of chronic diseases by BMI level (both HOS) 12 BMI by Race for Females and Males (HOS) (n= 93,066) Females Actual Data Males 13 Unadjusted PCS Score by Number of Chronic Diseases (HOS) (n= 76,535) Obese beneficiaries have lower PCS scores at any number of chronic diseases Each additional chronic disease reduces mean PCS scores by a constant amount Actual Data 14 PCS Scores (HOS) by MD Experience (MA-CAHPS) (n= 4,216) Percentage Actual Data 15 Low vs. Moderate Utilization Clinic Visits (HOS & MA-CAHPS) (n= 4,216) 1 – 2 visits 3 – 9 visits Similar rates of into Overweight Normal Obese cross cross into moderate moderateutilization utilization (4) (6+) (3) utilization (0) MA-CAHPS HOS Actual Data 16 Summary • HOS and MA-CAHPS populations can overlap • Some beneficiaries respond to both • Highly trained interviewers • Standard interview methods • Merge to take advantage of different features • Due to sample size, these data support advanced statistical methods 17 Technical Queries Health Services Advisory Group For inquiries, please contact the Medicare Health Outcomes Survey (HOS) Information and Technical Support Telephone Line at (888) 880-0077 or E-mail Address (hos@azqio.sdps.org). CMS HOS Website: http://www.hosonline.org 18 Presentation Queries Richard D. Hector, MA, MPH, PhD Health Services Advisory Group Phone: (602) 665 – 6133 Email: rhector@azqio.sdps.org 19 Thank you for your attention Please ask QUESTIONS Please share your COMMENTS 20