Academic leaders do not agree on the level of supervision attending physicians should entrust to surgery residents when performing operations in teaching centers. Some favor granting more autonomy to residents to better develop the clinical skills necessary to enter independent practice while others are concerned about the impacts to patient safety, care quality, and service effectiveness. We use health records from the Department of Veterans Affairs (VA), the largest healthcare training platform in the US, to evaluate VA’s Office of Academic Affiliations policies governing the supervision of residents in teaching surgery encounters. We focus on whether attending physicians had scrubbed or not scrubbed for surgery teaching encounters where residents served as the primary surgeon. Attending physicians not scrubbed implies a lower level of supervision when they can be in the room and supervising but unavailable to do any of the surgery directly with immediacy. Secondary data came from a national, systematic 1:8 sample of n=862,425 teaching encounters where residents were listed as primary surgeon at 122 VA medical centers from July 1, 2004, through September 30, 2019. Estimates from the sample reveal that under VA’s policies, attending physicians had scrubbed in 85% [AS] and not scrubbed in 15% [ANS] of all teaching surgery encounters. Two prior studies focused on the 15% of ANS and found using a propensity score match to AS encounters that there was no change in patient outcomes had these not-scrubbed attending physicians scrubbed. This suggests VA surgeons made the correct choice in those teaching encounters when residents were granted more autonomy. Our study focused on the 85% of encounters to assess how patient outcomes would change had attendings not scrubbed. Propensity score matching was not an option because there were too few ANS cases to match to AS encounters. We thus used a complex regression analyses called control functions to estimate the association between attending scrubbed status on patient outcomes. Estimates were tested for confounding biases, robustness to different regression models, stability over time, and validated using moderator and secondary factors analyses. We found that for every 1,000 surgery patients from among the 733,997 AS encounters, had the scrubbed attending not scrubbed an estimated 2.03 [0.04, 4.26] deaths would have been added to the reported 14.54 deaths per 1000 patients during the 30 days post-surgery; 6.46 [1.64, 11.38] cases with complications would have been added to the reported 90.52 cases per 1000 patients, and 10.93 [7.03, 15.05] cases with 30-day readmissions would have been added to the reported 67.80 readmissions per 1000 patients. Over the 15 study years, scrubbed surgery attending physicians may have averted an estimated 13,700 deaths, 43,600 cases with complications, and 73,800 readmissions. The study affirms VA’s strict supervision policies governing surgery residents assigned as the primary surgeon to surgery patient encounters. Also discussed are advantages to estimate policy outcomes using control function analyses on health records. Validated metrics that quantify resident supervision are recommended to better inform attending physicians how to supervise residents that optimizes both patient safety and resident education outcomes.