General Surgery

AR training improves novice technical performance but effects vary by experience and task

Systematic review of 11 studies (n=347 trainees across 7 specialties) found objective technical improvements with AR in most trials. Consistent reductions in errors were reported and several studies noted faster learning curves and lower cognitive load. An expertise-reversal effect emerged: novices derived the largest benefits, while experienced surgeons saw diminishing returns. Heterogeneous platforms, outcomes, and limited long-term data mean AR is promising but not universally proven. 1

Double negative controls mitigate bias from non-compliance in surgical RCT simulations

Simulation study showed DNC estimators maintained low bias, small MSE, and near-nominal coverage across scenarios with 10–40% crossover. Compared with ITT, PP, and AT (severely biased) and IV (high variability), DNC reduced MSE by >90% versus IV at n=100 with 40% crossover. DiD performed well only when equal-confounding assumptions held; its performance fell when assumptions were violated. For trials with expected non-compliance, DNC methods offer superior accuracy but require appropriate negative controls and assumptions. 2

ESPB yields modest early pain and recovery benefits after median sternotomy, high heterogeneity limits certainty

Meta-analysis of 19 RCTs (N=1344) found modest reductions in 24‑hour pain after median sternotomy (MD −0.65 on 0–10 scale). ESPB was associated with lower opioid consumption and shorter time to extubation (MD −1.28 hours) and ICU stay in sternotomy subgroup. Heterogeneity was substantial (I2 >85%) and prediction intervals often cross null, limiting generalizability across centers. No consistent clinically meaningful benefits were seen in minimally invasive cardiac surgery trials. 3