Individual surgeon mortality rates: can outliers be detected? A national utility analysis

Objectives There is controversy on the proposed benefits of publishing mortality rates for individual surgeons. In some procedures, analysis at the level of an individual surgeon may lack statistical power. The aim was to determine the likelihood that variation in surgeon performance will be detected using published outcome data. Design A national analysis surgeon-level mortality rates to calculate the level of power for the reported mortality rate across multiple surgical procedures. Setting The UK from 2010 to 2014. Participants Surgeons who performed colon cancer resection, oesophagectomy or gastrectomy, elective aortic aneurysm repair, hip replacement, bariatric surgery or thyroidectomy. Outcomes The likelihood of detecting an individual with a 30-day, 90-day or in-patient mortality rate of up to 5 times the national mean or median (as available). This was represented using a novel heat-map approach. Results Overall mortality rates for the procedures ranged from 0.07% to 4.5% and mean/median surgeon volume was between 23 and 75 cases. The national median case volume for colorectal (n=55) and upper gastrointestinal (n=23) cancer resections provides around 20% power to detect a mortality rate of 3 times the national median, while, for hip replacement, this is a rate 5 times the national average. At the mortality rates reported for thyroid (0.08%) and bariatric (0.07%) procedures, it is unlikely a surgeon would perform a sufficient number of procedures in his/her entire career to stand a good chance of detecting a mortality rate 5 times the national average. Conclusions At present, surgeons with increased mortality rates are unlikely to be detected. Performance within an expected mortality rate range cannot be considered reliable evidence of acceptable performance. Alternative approaches should focus on commonly occurring meaningful outcome measures, with infrequent events analysed predominately at the hospital level.