(CCSP064) DOES THE METHOD OF CALCULATING THE 30-DAY READMISSION RATE AFTER HOSPITALIZATION FOR HEART FAILURE MATTER? DATA FROM THE VANCOUVER COASTAL ACUTE HEART FAILURE (VOCAL-AHF)

Background: The 30-day readmission rate after a heart failure (HF) hospitalization is widely used to evaluate and compare healthcare quality and system performance. However, there are different methodological approaches which may influence estimated rates, and no single accepted approach. The combined impact of these methods is unknown. We calculated 30-day HF readmission rates of hospitalized patients using different published approaches to sampling and measurement.

METHODS AND RESULTS: We included 1,849 patients discharged following unplanned hospitalization with a primary diagnosis of HF between 2016 to 2018 from the VancOuver CoastAL Acute Heart Failure (VOCAL-AHF) registry. We combined five distinct methodological factors to create 64 unique definitions and associated HF readmission rates. The following factors were defined (Table 1): ICD-10 codes (broad vs narrow), index admission selection (single first-in-year vs single random-sampling vs multiple no-blanking vs multiple blanking), 30-day survival (survived cases at discharge vs at 30-day), follow-up period start (discharge date vs day following discharge), and annual reference period (calendar vs fiscal). The readmission rates were averaged over 3-years to minimize random and secular variability. Multiple linear regression was used to determine the impact of different factors on estimated readmission rates.

The calculated 30-day readmission rate for HF varied more than twofold depending solely on the methodological approach (6.4% to 15.0%, 8.6% absolute difference, 134% relative difference). The rates were highest when including all consecutive index admissions (11.1% to 15.0%), and lowest including only one index admission per patient per year (6.4% to 11.4%, Figure 1). The regression model ranked variables affecting readmission rate as: index selection method, reference period, ICD-10 codes, 30-day survival, and index day (p < 0.001 in all). When including multiple index admissions per year and compared to blanking the 30 days post-discharge, not blanking was associated with 2.3% higher readmission rates (95% CI (2.1, 2.4). By contrast, selecting a single admission per year with a first-in-year approach lowered readmission rates by 1.4% (-1.6, -1.3), while random-sampling admissions lowered estimates further by 5.2% (-5.4, -5.1). The remaining factors had smaller associations with the estimated rates: fiscal vs calendar year -0.8% (-1.0, -0.7), narrow vs broad ICD-10 code definition 0.3% (0.2, 0.4), denominator measured at 30 days 0.2% (0.2, 0.3), and first day post-discharge as the start of the 30-day follow-up period 0.2% (0.1, 0.3).

Conclusion: Our findings have important implications for policy and reporting. Transparent and consistent methods are needed to calculate 30-day readmission rates to ensure reproducible and comparable reporting.