General anesthesia (GA) with endotracheal intubation and monitored anesthesia care (MAC) are the most widely used modalities of sedation for endoscopic retrograde cholangiopancreatography (ERCP). Aim of this study was to determine difference in adverse discharge between patients receiving GA versus MAC for ERCP. We included data of adult patients undergoing ERCP with GA or MAC from January, 2007 to December, 2018 at a tertiary care hospital. The primary outcome was adverse discharge, defined as in-hospital mortality, discharge to a skilled nursing facility or long-term care facility. Unadjusted and adjusted standard logistic regression was performed. To address anesthesiologist preference, which may differ in choosing MAC or GA, we calculated adjusted absolute risk differences (aRD) using provider preference-based instrumental variable analysis. We used a propensity score matching analysis with GA and MAC cases matched 1:1, to further address the possibility of unbalanced confounding between the groups. 18,081 ERCPs were performed with GA or MAC during the study period. 543 were excluded due to incomplete data and 17,538 were included in this study. Of those, 16,238 (92.6%) received MAC and 1,300 (7.4%) received GA. The two groups were similar with respect to mean Age(yrs) ± SD (63.96±17.18 vs 63.34±17.10), mean BMI (27.21±5.98 vs 31.84±9.65). There were no significant differences with respect to ASA class and Charleston comorbidity index. The rates of adverse discharge were 5.8 % in the MAC group compared with 16.2 % in the GA group. GA was associated with significantly increased risk of adverse discharge based on unadjusted and adjusted logistic regression analysis (aRD, [95% confidence interval, CI]) (10.4% [8.3-12.4%] and 2.9 % [1.4-4.3%] respectively). Utilizing provider-related variability in the use of GA in instrumental variable analysis resulted in 8.6% risk increase in adverse discharge in patients who received GA compared with MAC (aRD 8.6%, [95% CI 4.5-12.6%]; p<0.001). This translated into an adjusted number needed to harm of 12 patients to lead to one adverse discharge that can be attributed to receiving GA. Propensity score matching analysis, with 1,248 cases in each group, demonstrated robust results (aRD 5.2% [95% CI 2.6-7.9%]; p<0.001). The effect was modified by a patient’s predicted probability of receiving GA (p-for-interaction=0.01) towards a greater risk in patients with a low probability (aRD 16.7%, [95% CI, 1.9-31.6%]; p=0.027) compared to those with a high probability of receiving GA (aRD 2.9%, [95% CI, 1.1-4.6%]; p=0.001). Results of our data across a wide range of analysis strategies suggests higher risks of adverse discharge in patients who received GA with endotracheal intubation compared with MAC for ERCP.View Large Image Figure ViewerDownload Hi-res image Download (PPT)
Read full abstract