Does Managed Care Affect Quality

Does Managed Care Affect Quality? Appropriateness, Referral Patterns and Outcomes of Carotid Endarterectomy Ethan A. Halm, MD, MPH Depts of Medicine and Health Policy Mount Sinai School of Medicine, NY Funding: AHRQ, CMS, and RWJF Background: Carotid Endarterectomy (CEA) • Carotid endarterectomy (CEA), surgery to prevent stroke, is a common vascular procedure in Medicare patients • Costly procedure with upfront risks of death/stroke due to CEA • High rates of overuse: in 1980s, 32% of CEAs were considered inappropriate CEA as a Model for Comparing Fee-For-Service (FFS) to Managed Care (MC) • Managed care plans have financial and quality incentives to prevent overuse and selectively refer patients (Pt) to high quality providers • CEA valuable model for comparing FFS to MC: – CEA is almost always elective – Subject to prior authorization by MC plans – Rich evidence-base of RCTs and national guidelines to guide appropriate Pt selection and provider performance benchmarks Study Aims • Compare appropriateness of CEA in Fee-ForService (FFS) v. Managed Care (MC) • Compare referral patterns between FFS v. MC – Are MC Pts more likely to be operated on by high volume providers or low complication rate providers? • Compare clinical outcomes: FFS v. MC New York Carotid Artery Surgery (NYCAS) Study Population • Population-based, retrospective cohort study of all CEAs in Medicare patients in NY State ’98-99 • Partnered with Medicare and NY Medicare QIO • Research RNs abstracted data on SES factors, indications for CEA, comorbidities, outcomes • Total of 9588 cases – FFS (8691) v. MC (897) – 166 hospitals – 488 surgeons Determining Appropriateness of CEA • RAND methodology: multidisciplinary, national expert panel rated appropriateness of 1557 indications for CEA • Appropriateness rating for each clinical scenario: – Inappropriate: risks > benefits – Uncertain: benefits = risks – Appropriate: benefits > risks 1 Provider Volume, Performance, and Clinical Outcomes • Surgeon and hospital volume (quintiles) • Performance: Low complication rate provider – Based on RCTs and national guidelines – Death/stroke rate for asymptomatic cases of <3% – Surgeon or hospital with at least 25 cases Analysis Plan • Univariate tests for differences in FFS v.MC in patients factors, indications, appropriateness, referral patterns, and 30 day outcomes • Multiple logistic regression (with GEE) to assess differences in risk-adjusted rates of death/stroke – Model A: adjusted for published CEA-specific risk factor model – Model B: adjusted for differences in FFS v. MC cases • Outcomes: Death, stroke, MI within 30 days – Reviewed medical records of index CEA and all readmissions within 30 days of surgery – Complications confirmed by MD over-reading CEA Patients in FFS v. MC FFS Patients Age* Male White* CEA: Symptomatic CEA: Asymptomatic High comorbidity CAD *P<.05 No Difference in Rates of Appropriateness: FFS v. MC 100 90 80 70 60 50 40 30 20 10 0 86.9 88.6 MC 897 74.1* Percent 8691 74.6 56 93 28 72 11 62 57 90* 27 73 10 60 FFS MC 4.4 Appropriate 3 8.6 8.4 Uncertain Inappropriate P=.78 Comparing Referral Patterns: Surgeon Characteristics FFS (%) Lowest quint. volume* Highest quint. volume* Low complication rate* Vascular surgeon General surgeon ≤10 yrs since med school* *P<.001 Comparing Referral Patterns: Hospital Characteristics FFS (%) Lowest quint. volume Highest quint. volume* Low complication rate Teaching hospital MSA > 1 million* *P<.001 MC (%) 23* 13* 55* 52 35 11* MC (%) 20 13* 64 38 88* 20 20 63 53 34 8 20 20 62 39 64 2 No Difference in 30 Day Outcomes of CEA: FFS v. MC Complication Rate (%) 5 4 3 2 1 0 1.2 1 3.4 3.7 3 3.3 No Difference in Risk-Adjusted Rates of Death/Stroke: FFS v. MC • 30 Day Risk-Adjusted Odds of Death/Stroke • Model A: OR=0.97; (0.69-1.37) – Adjusted for CEA-specific risk model (stroke, CAD, stenosis, anes, repair type) • Model B: OR=0.93 (0.65-1.31) – Adjusted for imbalances in FFS v. MC incl: age, race, stroke, HTN, COPD, DM, provider volume, MSA size 4.2 4.3 1.1 1.3 FFS MC th St ro ke tro ke e st ro k D ea on -fa ta l ea th /S P > .50 for all N Limitations • One state: NY is 14% of Medicare Pts • MC competition in NY in 1998-99 was moderate: similar to average state in US • Only Pts who underwent surgery: can not comment on influence of MC on eligible Pts who did not have CEA • No Pt zip code so couldn’t measure proximity to high volume/low complication rate provider – Used MSA size as proxy for access, most CEAs in cities of >1 million, and more MC Pts in big cities D M I Conclusions: FFS v. MC • FFS and MC Pts were similar in indications for CEA and most key patient attributes • No difference in rates of appropriateness or clinical outcomes • MC Pts less likely operated on by high volume surgeon or in high volume hospitals • MC Pts less likely to have CEA from low complication rate surgeon Implications: FFS v. MC • Disconnect between theoretical plan financial and quality incentives and plan behavior • Medicare MC plans did not appear to manage care, at least regarding CEA • Do not know if plans tried and failed, or never tried at all – Actually, very few really tried Acknowledgments Co-investigators: Mark Chassin, Stan Tuhrim, Ed Hannan, Jason Wang, Mary Rojas In order of heroics: Patricia Formisano, Virginia Chan, Hugh Dai, Matt Press In alphabetical order: Anna Arreglado, Virginia Chan, Camille Cohen, Wilfredo Gaerlan, Edwin Gravereaux, Chiaki Nakazono, Ying Qiu, Diane Thomas, Aymara Triana, and George Wang Funders/Collaborators: AHRQ, CMS, RWJ, IPRO National Expert Panel 3 4