Inter-institutional variation in risk-adjusted paediatric cardiac surgical outcomes

Abstract

Plans to reduce from four to two the number of hospitals performing paediatric cardiac surgery in Sweden prompted an inter-institutional comparison of risk-adjusted 30-day post-operative mortality. Pre-operative, intra-operative and follow-up data were abstracted from medical records of all patients listed by the hospitals as having undergone ‘complex’ paediatric cardiac surgery in 1992, based on previously established criteria. Surgeon-investigators abstracted clinical data to place all cases meeting these criteria into four preoperative risk categories. Discrepancies were resolved by consensus. Odds ratios were used to compare mortality in three hospitals relative to the fourth hospital, before and after adjusting for risk group distribution. Of 320 admission records submitted by the hospitals, 284 admissions involving 261 patients were considered complex procedures by criteria that included some re-operations but excluded heart transplants. Mortality risks and odds ratios increased in higher-risk groups, indicating the validity of the risk grouping. One-stage procedures or the initial components of multi-stage interventions were performed in 196 patients. Mortality odds ratios unadjusted for pre-operative risk in three other centres relative to the centre with the most patients were 0.72, 0.37 and 0.32, respectively (P=0.2750 by log-likelihood Chi-square). Risk-adjusted mortality odds ratios among the three centres (relative to the baseline hospital and the lowest risk category) were 0.44, 0.17 and 0.30, respectively (P=0.0001). For all 261 patients, unadjusted odds ratios for the three centres were 0.44, 0.27 and 0.39 (P=0.1130), while risk-adjusted odds ratios were 0.24, 0.12 and 0.32, respectively (P=0.0001). In this study, higher institutional volumes of complex procedures were not consistently associated with increased survival. Adjusting for preoperative risk did significantly alter institutional mortality odds ratios. Formal approaches for comparing specific–specific mortality following paediatric cardiac surgery are evolving, and adjusting for risk may enhance the validity of inter-institutional comparisons. Independent review of risk classification and mortality data submitted by hospitals may enhance the consistency of such analyses.