Variation in Pediatric Readmission Rates across Hospitals

Transcription

1 Variation in Pediatric Readmission Rates across Hospitals Jay G. Berry, MD, MPH, 1,6 Alan M. Zaslavsky,! PhD, 2 Ashish K. Jha, MD,MPH, 3 Shanna Shulman, PhD, 1 David Klein, MS, 1 Sara L. Toomey, MD, MPhil, MPH, MSc, 1,6 Matt Hall, PhD, 4 Vincent K. Chiang, MD, 5,6 Katie Alijewicz, BS, 1 William Kaplan, BS, 1 Mark A. Schuster, MD PhD 1,6 1 Division of General Pediatrics, Boston Children s Hospital 2 Department of Health Care Policy, Harvard Medical School, Boston, MA 3 Department of Health Policy and Management, Harvard School of Public Health, Boston, MA 4 Child Healthcare Corporation of America, Shawnee Mission, KS 5 Department of Emergency Medicine, Boston Children s\ Hospital 6 Department of Pediatrics, Harvard Medical School, Boston, MA

2 This project was funded by the Agency for Healthcare Research and Quality (U18 HS020513) There are no conflicts of interest to report. 2

3 Hospital Readmission Admission Discharge Preventable Readmission Unpreventable Readmission

4 Readmission Rate Variation Across Hospitals Unwarranted variation is an indicator of inconsistent quality of care Some variation may be legitimate or desirable Some variation may be explained by differences in case-mix 4

5 Readmission Rate Variation in Adult Patients Heart Failure, Myocardial Infarction and Pneumonia Mean 30-day readmission rates = 18-20% Rates vary 73-93% across hospitals With adjustment for case-mix differences With removal elective readmissions 5

6 Readmission Rate Variation in Pediatric Patients 30-day readmission after all-cause admissions Mean readmission rate = 5-8% Adjusted rates vary by 13% across hospitals 30-day readmission after sickle cell crisis Mean readmission rate = 17% Unadjusted rates vary by 420% across hospitals 6

7 Readmission Variation Across Hospitals in Children Gaps in knowledge Do readmission rates vary meaningfully across hospitals for other conditions? What are the rates for the best-performing hospitals? How many readmissions might be avoided if all hospitals had the best rates? 7

8 Study Objectives To determine whether pediatric readmission rates vary meaningfully across hospitals To determine how many readmissions may be avoided if all hospitals achieve the rates of the best-performing hospitals 8

9 Methods Retrospective cohort analysis Patients 18 years old Discharged between 7/2009 and 6/ children s hospitals in the NACHRI CaseMix Dataset ~ 50% of U.S. children s hospitals ~ 20% of U.S. acute care hospitalizations for children 9

10 Index Admission Exclusions Healthy newborn deliveries Chemotherapy admissions Patients that left against medical advice Patients that transferred to another hospital Patients that died during the hospitalization 10

11 Index Admissions All-cause Admission for any reason Condition-specific 10 conditions with the highest readmission frequency All Patient Refined Diagnostic Related Groups (APRDRG) Groupings of ICD-9-CM codes Used to identify the reason for admission 11

12 Hospital Readmission within 30 days All-cause Readmission for any reason Unplanned Elective readmissions removed To the same hospital Readmissions to a different hospital were not measureable within the dataset 12

13 Case-Mix Adjustment Age at admission <1 year, 1-4, 5-12, and years Feudtner s Complex Chronic Conditions Uses ICD9 codes to identify children with the conditions Correlate with risk of readmission Reason for index admission Ranked each admission by readmission rate Grouped into ten categories 13

14 Readmission Rate Variation Across Hospitals Hierarchical regression models Random effect for hospital Fixed effects for case-mix adjusters Test for significance Covariance test of the hospital random effect Statistical significance, p<.05 14

15 Study Cohort No. of index admissions 535,281 Mean age 6 years Complex chronic condition 33% Non-Hispanic White 49% Public or no health insurance 52% 15

16 30-Day Readmission Rates Following All-Cause Admissions All Patients 8% Complex chronic condition (CCC) Present 14% Absent 5% Age in years % % 1-4 8% <1 8% 16

17 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X Readmission Rate (%) 8 4 X 0 17

18 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X Readmission Rate (%) 8 4 X Statistically significant variation across hospitals, p<

19 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X Readmission Rate (%) 8 4 X 8% Mean Readmission Rate 0 19

20 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X Readmission Rate (%) 8 4 X 8.9% Readmission Rate (+ 1 SD) 0 20

21 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X Readmission Rate (%) 8 4 X 8.9% Readmission Rate (+ 1 SD) 7.1% Readmission Rate ( - 1 SD) 0 21

22 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X 10.5% Readmission Rate (+ 2 SD) Readmission Rate (%) 8 4 X 0 22

23 30-Day Adjusted Readmission Rate Variation Across Hospitals 12 X 10.5% Readmission Rate (+ 2 SD) Readmission Rate (%) 8 4 X 6.2% Readmission Rate ( - 2 SD) 0 23

24 Reduction in Readmission Rate Variation Across Hospitals If all 69 hospitals had readmission rates = 7.1% 12% readmission reduction 5,000 readmissions avoided If all 69 hospitals had readmission rates = 6.2% 22% readmission reduction 9,500 readmissions avoided 24

25 35 30-Day Readmission Rates Condition Specific Admissions 30 Readmission Rate (%) Asthma Bronchiolitis Pneumonia Upp. Resp. Infec. Seizure Gastroenteritis Oth. digest. diag. Craniotomy Sickle cell crisis Anemia/neutropenia 25

26 30-Day Readmission Rates Condition-Specific Admissions Significant variation (p<.05) across hospitals Aside from upper respiratory infection (p = 0.2) Adjusted readmission rates 40-60% higher for hospitals 1 SD above vs. below the mean % higher for hospitals 2 SD above vs. below the mean 26

27 Reduction in Condition-Specific Readmission Rate Variation Across Hospitals If all 69 hospitals had readmission rates 1 SD below the mean 18% readmission reduction 1,820 readmissions avoided If all 69 hospitals had readmission rates 2 SD below the mean 34% readmission reduction 3,330 readmissions avoided 27

28 Main Findings Significant variation in readmission rates exist across a cohort of children s hospitals For all-cause admissions For condition-specific admissions A substantial number of readmissions could be avoided if all hospitals had the best rates Thousands avoided for all-cause and conditionspecific admissions 28

29 Limitations Readmissions to hospitals other than the index admission hospital were not measured Hospital sample did not include non-children s hospitals The degree of preventability of the unplanned readmissions could not be determined 29

30 Implications The degree of readmission rate variability suggests room to improve the quality of care for hospitalized children. We must understand how hospital discharge and outpatient follow-up care might be executed better in hospitals and surrounding areas with lower readmission rates. We should strive to achieve the readmission rates of the best performing hospitals. 30

31 EXTRA SLIDES 31

32 30-Day Readmission Rates Following All-Cause Admissions Hospital Readmission Rates 1 Standard Deviation Above the Mean Below the Mean Adjusted Readmission Rate Ratio 9% 7%

33 30-Day Readmission Rates Following All-Cause Admissions Hospital Readmission Rates 2 Standard Deviations Above the Mean Below the Mean Adjusted Readmission Rate Ratio 10% 6%

34 30-Day Readmission Rates Following All-Cause Admissions Hospital Readmission Rates Maximum Minimum Adjusted Readmission Rate Ratio 11% 5%

35 Variation in Adjusted Readmission Rates Across Hospitals 35 Number of Hospitals Unplanned Adjusted 30-day Readmission Rate 35

36 Variation in Adjusted Readmission Rates Across Hospitals 35 Number of Hospitals Unplanned Adjusted 30-day Readmission Rate 36

37 Thank you 37