Assessment of abdominoperineal resection rate as a surrogate marker of hospital quality in rectal cancer surgery

Background Rates of abdominoperineal resection (APR) have been suggested as a solitary surrogate marker for comparing overall hospital quality in rectal cancer surgery. This study investigated the value of this marker by examining the associations between hospital APR rates and other quality indicators. Methods Hospital-level correlations between risk-adjusted APR rates for low rectal cancer and six risk-adjusted outcomes and six care processes were performed (such as 30-day mortality, complications, timely treatment). The ability of APR rates to discriminate between hospitals' performance was examined by means of hospital variance results in multilevel regression models and funnel plots. Results A linked population-based data set identified 1703 patients diagnosed in 2007 and 2008 who underwent surgery for rectal cancer. Some 15·9 (95 per cent confidence interval (c.i.) 14·2 to 17·6) per cent of these patients had an APR. Among 707 people with low rectal cancer, 38·2 (34·6 to 41·8) per cent underwent APR. Although risk-adjusted hospital rates of APR for low rectal cancer varied by up to 100 per cent, only one hospital (1 per cent) fell outside funnel plot limits and hospital variance in multilevel models was not very large. Lower hospital rates of APR for low rectal cancer did not correlate significantly with better hospital-level outcomes or process measures, except for recording of pathological stage (r = -0·55, P = 0·019). Patients were significantly more likely to undergo APR for low rectal cancer if they attended a non-tertiary metropolitan hospital (adjusted odds ratio 2·14, 95 per cent c.i. 1·11 to 4·15). Conclusion APR rates do not appear to be a useful surrogate marker of overall hospital performance in rectal cancer surgery. Abdominoperineal resection not a marker of hospital quality