Understanding g the Impact p of Insurance Reform Methodological Challenges and Solutions Karen Freund MD MPH B t University Boston U i it S School h l off M Medicine di i Boston Medical Center What is our research Question? o Our ultimate question: o “Do “D iimprovements t iin iinsurance coverage improve care outcomes?” o Hypothesis H th i o Insurance reform reduces gaps in insurance o Reduced gaps in care results in care when h needed d d o Needed medical care utilization improves h l h outcomes health Id l D Ideal Data t S Sett tto address dd Longitudinal cohort Information on all insurance changes changes, coverage gaps Health Care Utilization Data Health Outcomes Data Only exists in European countries with single payer systems, t without ith t changes/ h / gaps in i insurance… S Sources off Data D t Consumers Surveys I di id l IInsurance Pl Individual Plan D Data t Utilization Databases Clinical Data from practice/healthcare system What is y your comparison p group? • Cohort: Longitudinal changes • Cross C S Sectional: ti l P Pre// postt reform f • Cross Sectional: Massachusetts vs. Other states • Methods to address confounding g by y temporal trends • Cross Sectional Difference in Differences • Time Series Analysis Consumers Surveys Advantages Can capture coverage across insurance types Women get care across systems of care ((OB/Gyn, y familyy planning, primary care, mental health) h l h status, patient health i reported outcomes Disadvantages Limited utilization Limited ability to assess quality of care or outcomes E Examples l off S Survey D Data t Massachusetts Health Reform Survey MEPS - nationally representative longitudinal – each individual is surveyed 5 times over a 2 ½ year period health status, insurance status for every calendar month, administrative data ((diagnoses g and charges) I Insurance Databases D t b Extensive utilization Exact eligibility data Don’t know what happens pp when patient not enrolled Few systems to link individual patients across multiple insurance systems Cross Sectional Utilization Data Advantages Standardized Collection Comparisons C i over Time and between l locations ti Disadvantages No individual insurance data No N iindividual di id l h health lth status information Example p using g Massachusetts Discharge Data Major gender/ race/ ethnicity di disparities iti iin ““referralf l sensitive iti procedures” These require source of primary care With health insurance reform , would women, and minority women receive a greater increase in procedures? Hanchate A, Mass Health Reform: Poster , Sunday 6/27/2010 2:30 - 4:00 pm Proceedure Rate / 10,0000 populaation Baseline Musculoskeletal Procedure Rates by G d and Gender db by R Race/Ethnicity /Eth i it # procedures/10,000: 2004-2006 100.0 90.00 90 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10 0 10.0 0.0 Female Male White Black Hispanic Hanchate, 2010 % Change in Procedu ure Rate / 10,000 Populattion % Change g ((Annual)) in Musculoskeletal Procedure Rate: Pre- and Post-Reform Comparison 18.0 16.0 14.0 12.0 10 0 10.0 8.0 6.0 4.0 2.0 00 0.0 Female Male Whites Pre Reform Blacks Hispanic Post Reform Hanchate, 2010 D t from Data f Clinical Cli i l C Care Sit Site Advantages Disadvantages Extensive E t i d data t on Cannot C t see care clinical variables, outside of that system clinical li i l status t t t May see patient Can see “switches” in across multiple insurance, but not changes in insurance “gaps” in coverage Example p of use of Clinical Care Data Boston Patient Navigation Research Program 6 Community Health Centers All women with abnormal breast and cervical cancer screening Extensive Clinical Data Large proportion with Medicaid/ uninsured/ Commonwealth care Insurance at each visit H Hypotheses th Uninsured with poorer outcomes Fewer uninsured in post period Those with p private and p public insurance will have the greatest % of timely resolution than those uninsured Commonwealth Care will have % timely resolution similar to public and private and greater than uninsured Breast east Patients at e ts characteristics c a acte st cs pre and post reform Pre Reform % (N=208) Age 18 – 40 6 41 – 64 81 65+ 13 Race/Ethnicity White 23 Black 50 Hispanic 17 Other 9 Insurance Public 38 Private 31 Uninsured 30 C Commonwealth lth C Care 0 Post Reform % (N=718) 9 77 13 29 37 33 1 31 32 27 9 Timely resolution after abnormal b breast t cancer screening, i post insurance su a ce reform eo 80 ** ** 60 % Resolution In 60 days 40 20 0 Private Insurance Public Insurance Commonwealth Care Uninsured What about Gaps p / Switches / Churning? R i Review d data t llooking ki att switches it h Are switches related to outcome of timely resolution? Case: time to resolution 7+ months D t Date I Insurance /Clinical /Cli i l Event E t 6/2007 Medicaid 7/2007 Abnormal Pap Smear 8/2007 Medicaid 9/2007 Commonwealth Care 3/2008 Biopsy – Non-neoplastic finding 3/2008 Commonwealth Care 7/2008 Medicaid Case: time to resolution 43 days Date Insurance / Clinical Event 11/2007 Medicaid 4/2008 Private 4/2008 Abnormal Pap Smear 5/2008 Pi t Private 6/2008 p py – Non-neoplastic p Colposcopy 6/2008 Private 8/2008 Medicaid Summary No one data base will likely allow us t understand to d t d impact i t off health h lth insurance reform Need for multiple sources to assess outcomes data