A. D. Jalilvand1, M. Al-Mansour1, K. A. Perry1 1Ohio State University,General And Gastrointestinal Surgery,Columbus, OH, USA
Introduction: The ACS-NSQIP Surgical Risk Calculator (ANS-RC) predicts 30-day complication rates for specific surgical procedures. The goal of this study was to assess the accuracy of the ANS-RC for predicting 30-day complication rates in a cohort of patients undergoing laparoscopic paraesophageal hernia repair (LPEHR) in a large academic medical center.
Methods: One hundred seventy-seven patients underwent primary LPEHR between 2011 and 2016 and were included in the study. Using the definitions in the ANS-RC, risk factors and 30-day post-operative complications were obtained for all patients from the electronic medical record. Predicted complication rates were calculated for each patient. Data are presented as incidence (%), mean ± SD, or median (IQ range). Comparisons between predicted and observed complication rates were made using one sample proportion or Wilcoxan paired signed rank tests. A p-value of <0.05 was considered statistically significant.
Results: During the study period, LPEHR was performed for 177 patients with a mean age of 66.2 ± 14.0 years and BMI of 30.2± 6.1 kg/m2. Seventy-three percent (n=156) were female and most patients had an ASA score of 2 (n=47, 26.6%) or 3 (n=122, 68.9%). Compared to the ANS-RC general population, this cohort had higher risks for serious complications (7.0% vs 5.7%), cardiac complication (0.4% vs 0.2%), reoperation (2.3% vs 2.1%), and readmission (6.5% vs 5.2%). Overall, the observed complication rates for any complication (13.6% vs 7.7%, p<0.01), serious complication (11.4% vs 7%, p=0.02), death (1.7% vs 0.3%, p<0.01), reoperation (6.8% vs 2.3%, p<0.01), and readmission (11.3% vs 6.5%, p<0.01) were higher than those predicted by the ANS-RC. The median hospital length of stay (LOS) was significantly shorter than predicted (2 vs 2.5, p<0.01). When stratified for patients with ASA scores of 2 or 3, the calculator more accurately predicted the observed complication rates, although reoperation (p=0.02) for ASA 2, and reoperation (p=0.04), SNF placement (p=0.03) and readmission rates (p=0.04) for ASA 3 were higher than predicted by the ANS-RC. The calculator most accurately predicted complication rates when patients were stratified by age group (<65, 65-79, 80+). Predicted values were lower than observed rates for reoperation in patients <65 (p=0.01) and 65-79 (p<0.01), readmission for patients <65 (p<0.01), and any or serious complications for patients >80 years of age (p=0.01). ANS-RC significantly overestimated LOS for patients <65 (p<0.01) and 65-79 years (p<0.01).
Conclusion: While the ANS-RC provides a useful tool for guiding preoperative discussions, this cohort comprised primarily of elderly patients with significant medical comorbidities had significantly higher than predicted complication rates compared to the general NSQIP population. However, stratifying patients by age and ASA improves the accuracy of the ANS-RC for LPEHR.