Improving Surgical Research Capacity in Low- and Middle-Income Countries: Can Episodic Data Collection Reliably Estimate Perioperative Mortality?

OBJECTIVE:The aim of this study was to empirically determine the optimal sample size needed to reliably estimate perioperative mortality (POMR) in different contexts. SUMMARY BACKGROUND DATA:POMR is a key metric for measuring the quality and safety of surgical systems and will need to be tracked as surgical care is scaled up globally. Continuous collection of outcomes for all surgical cases is not the standard in high-income countries and may not be necessary in low- and middle-income countries. METHODS:We created simulated datasets to determine the sampling frame needed to reach a given precision. We validated our findings using data collected at Mulago National Referral Hospital in Kampala, Uganda. We used these data to create a tool that can be used to determine the optimal sampling frame for a population based on POMR rate and target POMR improvement goal. RESULTS:Precision improved as the sampling frame increased. However, as POMR increased, lower sampling percentages were needed to achieve a given precision. A total of 357 eligible cases were identified in the Mulago database with an overall POMR rate of 14%. Precision of ±10% was achieved with 34% sampling, and precision of ±25% was obtained at 9% sampling. Using simulated datasets, a tool was created to determine the minimum sample percentage needed to detect a given mortality improvement goal. CONCLUSIONS:Reliably tracking POMR does not require continuous data collection. Data driven sampling strategies can be used to decrease the burden of data collection to track POMR in resource-constrained settings.