Sample Size Calculation for Binary Outcomes

We performed sample size calculations recommended for binary outcomes [18 (link)], using the ‘pmsampsize’ module in Stata software version 16 and 17 (StataCorp) [19 ]. For these calculations, we used an expected R² of 0.072 from a recent study with a similar model [20 (link)], expected outcome incidences ranging from 1.7% for ambulatory surgery [9 (link)] and 8.0% for inpatient surgery [20 (link)], and the maximum candidate predictor parameters of 32. The minimum sample size ranged from 3838 to 3881 participants with 66 to 308 events depending on the incidence values used for the expected outcomes (Supplementary Table 5).