Login to Access Video or Poster Abstract: MP37-17
Sources of Funding: none

Introduction

Adrenalectomy is performed to treat functional pathology and to remove tumors of malignant concern. The NSQIP Risk Calculator predicts both length of stay and 30 day complications following a given surgery by inputting a set of risk factors. Our objective is to assess whether this tool accurately predicts complications following adrenalectomy thereby allowing accurate counselling regarding surgical risk.

Methods

A retrospective review was performed for all adrenalectomies at our institution between 2008-2016. 126 patients underwent adrenalectomy without concurrent resections. Predicted risk for NSQIP complications was calculated for each patient. The mean predicted and observed risks (%) at 30 days across all patients within each category were determined, and these were compared with a two-sided one-sample t-test.

Results

Of 126 adrenalectomies, 111 were laparoscopic and 15 were open. Mean age was 55 and 66% of patients were female. For laparoscopic adrenalectomy, 5 (4%) complications were observed including 3 (2%) Clavien III or greater. All observed complication rates were significantly different than predicted, except for surgical site infection. (Table) Rates of severe, any, and cardiac complications, as well as venous thromboembolism, renal failure, and death were less than expected, while rates of pneumonia, urinary tract infection, readmission, return to operating room, and discharge to rehabilitation were greater than expected. Mean observed length of stay (LOS) was significantly less than predicted (1.44 vs. 2.11 days, p<0.001). In the open adrenalectomy subgroup, there were no observed complications with observed mean LOS greater than predicted (6.67 vs. 5.27, p<0.05) without a higher readmission rate (6.67 vs. 5.83%, p=0.064).

Conclusions

Laparoscopic adrenalectomy has a low complication rate, and most patients leave the hospital in 1 to 2 days. Most of the NSQIP calculator's predictions were significantly different than observed; some complications were overestimated and others underestimated. Overall, absolute complication rates seem comparable when considered for clinical significance with a larger sample size requisite to better delineate which complications are most accurately predicted by the calculator.

Funding

none