Previous reports showed that in California during the early 1990s, operative mortality rates for esophageal, pancreatic, and hepatic cancers were inversely related to hospital volume. It is unknown whether this information has affected referral patterns or operative mortality rates.
Data were analyzed for the 10 years that followed the period covered in the initial studies to determine if: (a) the operative mortality rates had decreased; and (b) a greater proportion of patients with esophageal, pancreatic, and hepatic cancers were treated at high-volume centers.
Hospital discharge data were obtained for 8901 patients who had resections for cancer of the esophagus, 2404 patients; pancreas, 5294 patients; and liver, 1203 patients in California between 1995 and 2004. Logistic regression models were used to calculate adjusted mortality rates at high- and low-volume centers by year. The data were compared with the published results for California during the years 1990–1994.
Operative mortality rates decreased for esophageal, pancreatic, and hepatic resections during the more recent 10 years. Concomitantly, the proportion of patients treated at high-volume centers increased, as did the number of high-volume centers. There was a substantial increase in the proportion of esophagectomies performed in high-volume hospitals, while the overall number of esophagectomies dropped by 22%. For the other 2 operations, total volume and the volume in high-volume hospitals increased greatly, and the volume in low-volume hospitals was about the same during the 3 periods. The mortality rates decreased at all levels of the volume range. Finally, the performance from one period to the next in individual hospitals was mostly similar, but an occasional outlier was also noted.
More resections for esophageal, pancreatic, and hepatic cancer were performed at high volume centers, but mortality rates decreased for all hospital categories. The data suggest that modern hospitals act as complex adaptive systems, whose outputs are determined from the interactions between internal agents and are resistant to analysis by isolating and studying the individual contributions. It is tempting to attribute the desirable changes in these data (eg, more operations being done in high volume centers and better mortality rates at all levels) as consequences of pressures over the past few decades on hospitals to assume greater responsibility for their quality of care and to become more integrated internally.
Thus, many factors appear to influence the volume-outcome relationships, and the identity and individual contributions of these influences may be immune to reductionist analysis. There is substantial evidence that high volume should be part of high quality for these complex operations. Nevertheless, measuring outcomes directly, rather than concentrating on their correlates, may be a more reliable index of hospital performance.