Presentation is loading. Please wait.

Presentation is loading. Please wait.

Evaluating Policies in Cardiovascular Medicine

Similar presentations


Presentation on theme: "Evaluating Policies in Cardiovascular Medicine"— Presentation transcript:

1 Evaluating Policies in Cardiovascular Medicine
Jason H. Wasfy, M.D. M.Phil. Director of Quality and Outcomes Research, Mass General Heart Center Assistant Professor, Harvard Medical School February 20, 2017

2 Disclosures Grant Support, National Institutes of Health (KL2)
Salary, Massachusetts General Physicians Organization Member, New England Comparative Effectiveness Public Advisory Council U.S. Department of Health and Human Services (LAN committee on bundled payments)

3 Financial penalties Passage of the Affordable Care Act created the Hospital Readmission Reduction Program (HRRP) Introduced the prospect of financial penalties based on performance on readmissions for AMI, CHF, pneumonia during 3 year period Uncertain effects – specific concerns regarding equity, competing priorities, vulnerable populations

4 Background 30 day hospital readmissions after AMI, CHF, pneumonia decreased faster after passage of the Affordable Care Act Zuckerman et al NEJM 2016

5 Background We know that the overall readmission rate declined after after passage of the law We do not know individual effects on higher and lower performing hospitals We sought to determine if the acceleration in improvement in readmission rates was greater in lowest-performing hospitals compared with higher-performing hospitals

6 Methods Post intervention

7 Methods Post intervention

8 Methods Difference in differences
Quasi-experimental approach meant to minimize confounding and history bias

9 Methods Queried MedPAR files, which contain Medicare beneficiaries’ inpatient claims data Identified patients discharged alive 2000-November with AMI, CHF, pneumonia, and identified those readmitted to acute care facilities within 30 days Identified comorbidities and hospital characteristics

10 Methods Estimated risk-standardized 30-day all-cause readmission rates (RSRRs) for each hospital (for all three conditions) Fitting a logistic regression to the 30-day all-cause readmissions as a function of patients’ age, sex, and comorbidities We also calculated a combined-condition RSRR, reflecting readmissions for any of the three conditions

11 Methods We identified information regarding hospital penalties from CMS, and classified hospital performance based on initial penalty: Lowest performance (greater than or equal to 99% maximum penalty) Low performance (greater than or equal to 50%, less than 99% maximum penalty) Average performance (less than 50% maximum penalty, but not zero) Highest performance (zero penalty)

12 Methods Pre-post analysis stratified by performance group
Piecewise linear model with change point defined as passage of the law (March 2010)

13 Results 2868 hospitals met inclusion criteria
866 (30.1%) in highest performance group (no penalty) 1261 (44.0%) in average performance group (low penalty) 483 (16.8%) in low performance group (high penalty) 258 (9.0%) in lowest performance group (maximum penalty)

14 Results Given the size of the dataset, patient and hospital characteristics were different in every performance group In particular, highest performance hospitals served more white patients than lowest performance hospitals (87.9% vs. 85.2%, p <0.001) Highest performance hospitals were more often rural than lowest performance hospitals (29.9% vs. 25.2%, p<0.001) Highest performance hospitals served a lesser proportion of dual-eligibles than lowest performance hospitals (17.0% vs. 20.3%, p <0.001)

15 Results

16 Results After controlling for pre-law trends, we found that an additional 67.6, 74.8, 85.4, and 95.1 readmissions per discharges were averted per year in the highest, average, low, and lowest performance groups respectively

17 Results – Acute Myocardial Infarction

18 Results – Congestive Heart Failure

19 Results - Pneumonia

20 Differences in readmissions

21 Results – Sensitivity Analyses
Our main results were robust in sensitivity analyses: Including hospital volume Including other hospital characteristics Changing the “pre-law” period to 15 months Changing the definition of performance group Observed readmission rates instead of RSRRs One-step full hierarchical model

22 Discussion Hospitals with lower performance experienced greater acceleration of performance after the law was passed The rate of decrease in RSRRs for penalized conditions was one-third greater for the lowest-performing hospitals than for the highest performing hospitals

23 Discussion A concerning potential outcome of the HRRP was that low-performing hospitals would not be able to improve Our analysis is reassuring with respect to that concern Our analysis is consistent with previously reported findings that HRRP penalties are applied disproportionally to hospitals that serve vulnerable patients

24 Discussion Although this may be a form of regression to the mean (lower performing hospitals have more room for improvement) we present strong evidence that the passage of the law caused this change in hospital performance patterns Our analysis minimizes confounding bias, a common threat to validity in policy analyses Pre-post analysis with the pre-law period serving as a control

25 Limitations Analysis of billing codes
We do not know that declines in RSRRs are related to genuine improvements in the quality of care We cannot detect effects on observation readmissions We cannot adjust for all confounding by hospital characteristics

26 Conclusions Hospitals with the lowest performance had greater acceleration of improvement in RSRRs for all three initial penalty conditions (AMI, CHF, pneumonia)

27 jwasfy@mgh.harvard.edu @jasonwasfy
Thanks! @jasonwasfy


Download ppt "Evaluating Policies in Cardiovascular Medicine"

Similar presentations


Ads by Google