Limitations of Using Pediatric Respiratory Illness Readmissions to Compare Hospital Performance
BACKGROUND: Adult hospital readmission rates can reliably identify meaningful variation in hospital performance; however, pediatric condition-specific readmission rates are limited by low patient volumes.
OBJECTIVE: To determine if a National Quality Forum (NQF)-endorsed measure for pediatric lower respiratory illness (LRI) 30-day readmission rates can meaningfully identify high- and low-performing hospitals.
DESIGN: Observational, retrospective cohort analysis. We applied the pediatric LRI measure and several variations to evaluate their ability to detect performance differences.
SETTING: Administrative claims from all hospital admissions in California (2012-2014). PATIENTS: Children (age <18 years) with LRI (primary diagnosis: bronchiolitis, influenza, or pneumonia; or LRI as a secondary diagnosis with a primary diagnosis of respiratory failure, sepsis, bacteremia, or asthma).
MEASUREMENTS: Thirty-day hospital readmission rates and costs. Hierarchical regression models adjusted for age, gender, and chronic conditions were used.
RESULTS: Across all California hospitals admitting children (n = 239), using respiratory readmission rates, no outlier hospitals were identified with (1) the NQF-endorsed metric, (2) inclusion of primary asthma or secondary asthma exacerbation diagnoses, or (3) inclusion of 30-day emergency revisits. By including admissions for asthma, adding emergency revisits, and merging 3 years of data, we identified 9 outlier hospitals (2 high-performers, 7 low-performers). There was no association of hospital readmission rates with costs.
CONCLUSIONS: Using a nationally-endorsed quality measure of inpatient pediatric care, we were unable to identify meaningful variation in hospital performance without broadening the metric definition and merging multiple years of data. Utilizers of pediatric-quality measures should consider modifying metrics to better evaluate the quality of pediatric care at low-volume hospitals.