BackgroundThe authors used continuous quality improvement (CQI) program data to investigate trends in quality of anesthesia care associated with changing staffing patterns in a university hospital.MethodsThe monthly proportion of cases performed by solo attending anesthesiologistsversusattending–resident teams or attending–certified registered nurse anesthetist (CRNA) teams was used to measure staffing patterns. Anesthesia team productivity was measured as mean monthly surgical anesthesia hours billed per attending anesthesiologist per clinical day. Supervisory ratios (concurrency) were measured as mean monthly number of cases supervised concurrently by attending anesthesiologists. Quality of anesthesia care was measured as monthly rates of critical incidents, patient injury, escalation of care, operational inefficiencies, and human errors per 10,000 cases. Trends in quality at increasing productivity and concurrency levels from 1992 to 1997 were analyzed by the one‐sided Jonckheere‐Terpstra test.ResultsProductivity was positively correlated with concurrency (r = 0.838;P< 0.001). Productivity levels ranged from 10 to 17 h per anesthesiologist per clinical day. Concurrency ranged from 1.6 to 2.2 cases per attending anesthesiologist. At higher productivity and concurrency levels, solo anesthesiologists conducted a smaller percentage of cases, and the proportion of cases with CRNA team members increased. The patient injury rate decreased with increased productivity levels (P= 0.002), whereas the critical incident rate increased (P= 0.001). Changes in operational inefficiency, escalation of care, and human error rates were not statistically significant (P= 0.072, 0.345, 0.320, respectively).ConclusionsMost aspects of quality of anesthesia care were apparently not effected by changing anesthesia team composition or increased productivity and concurrency. Only team performance was measured; the role of individuals (attending anesthesiologist, resident, or CRNA) in quality of care was not directly measured. Further research is needed to explain lower patient injury rates and increases in critical incident reporting at higher concurrency and productivity levels.