Differences in hospital performance for noncancer vs cancer colorectal surgery

Considerable hospital-to-hospital variations in surgical outcomes have been reported across surgical procedures. However, it is unclear whether hospital quality rankings are consistent for noncancer and cancer operations. We investigated the differences in hospital performance for noncancer and cancer colorectal resections at 52 hospitals participating in the Michigan Surgical Quality Collaborative (MSQC).

Patients undergoing colorectal resections between 2008 and 2012 were identified. Hierarchical risk-adjusted models were used to evaluate hospital level 30-day morbidity, major morbidity, extended length-of-stay (LOS > 75(th) percentile), and mortality outcomes. Hospital performance, as ranked by observed-to-expected ratios, was compared by rank-order changes, interquartile ranges (IQR), and Spearman's correlations.

Of the 19,990 colorectal resections, 7,292 (36.5%) were for cancer. We observed wide variations in all risk-adjusted 30-day outcomes between hospitals, but only weak correlations in cancer and noncancer performance within hospitals. Overall hospital performance in mortality after noncancer and cancer operations was not correlated (Spearman's rho: 0.02). Of the best performing hospitals in mortality after noncancer resections, 69% were reclassified to a worse quartile for cancer operations (median rank-change of 12.5 ranks [IQR 5 to 27]). Similarly, hospital performance in morbidity was only moderately correlated (rho: 0.59; p < 0.001). Of the hospitals with lowest morbidity rates for noncancer resections, 31% were reclassified. We noted a similar lack of relationship in major morbidity and extended LOS.

A hospital's performance ranking in risk-adjusted outcomes after noncancer colorectal resections does not correlate to its performance for cancer-related colorectal resections. Indication for operation should be considered when leveraging risk-adjusted hospital outcomes for quality improvement efforts.