N. Lunardi1,2, M. Thornton1,2, B. L. Zarzaur3, S. Agarwal4, M. Berger5, S. Sharath6,7,8, P. Kougias6,8, S. Bhat1, T. H. Pham1,2, C. J. Balentine3,9,10 1University Of Texas Southwestern Medical Center, Department Of Surgery, Dallas, TX, USA 2VA North Texas Health Care System, Department Of Surgery, Dallas, TX, USA 3University Of Wisconsin, Department Of Surgery, Madison, WI, USA 4Duke University Medical Center, Department Of Surgery, Durham, NC, USA 5Duke University Medical Center, Department Of Anesthesiology, Durham, NC, USA 6SUNY Downstate Health Sciences Center, Department Of Surgery, Brooklyn, NY, USA 7SUNY Downstate Health Sciences Center, Department Of Epidemiology & Biostatistics, Brooklyn, NY, USA 8New York Harbor Health Care System, Department Of Surgery, Brooklyn, NY, USA 9Wisconsin Surgical Outcomes Research Program, Madison, WI, USA 10William S. Middleton VA, Department Of Surgery, Madison, WI, USA
Introduction: The popularity of nonoperative management for acute appendicitis is based on the untested assumption that it offers a lower risk alternative to surgery in patients who are at high risk for morbidity and mortality with appendectomy. We hypothesized that patients with appendicitis who were high-risk for morbidity/mortality with appendectomy would also be high risk for complications following nonoperative management, potentially reducing the benefits of a nonoperative approach.
Methods: This is a retrospective cohort study of patients with acute, uncomplicated appendicitis in the 2004 to 2017 National Inpatient Sample. We used a logistic regression model to predict the risk of morbidity or death following laparoscopic appendectomy and used this model to predict the risk of morbidity or death if patients managed nonoperatively had been treated with surgery. We defined patients as high-risk if their predicted morbidity or mortality risk was ≥2 standard deviations above the mean. We used inverse probability weighting of the propensity score to compare outcomes of nonoperative management versus laparoscopic appendectomy for the high-risk patients. For sensitivity analysis, we used the robustness of inference to replacement (RIR) to quantify the percentage of the estimated difference that would need to be attributed to an unmeasured confounder to change the results.
Results: The sample included 17,776 patients classified as high-risk for morbidity or mortality following appendectomy. The median age was 70 years (IQR 60-79) and 31% were managed nonoperatively. Compared to surgery, nonoperative management in high-risk patients was associated with an 11% decrease in complications (95% CI 10%-12%), but a 3% increase in mortality (95% CI 2%-4%), a $9,924 increase in hospital costs (95% CI $8,688-$11,160), a 3 day increase in length of stay (95% CI 3-4) and a 6% greater likelihood of discharge to skilled nursing facilities (95% CI 5%-8%). The RIR for the difference in morbidity indicated that 224 additional nonoperative patients would need to have experienced a complication for there to be no significant difference compared to surgery. This suggests that the observed difference is robust to a strong unmeasured confounder. Similar results were observed for differences in mortality, length of stay, and costs.
Conclusion: Nonoperative management of acute appendicitis in high-risk patients may reduce morbidity, but could increase mortality, duration of hospitalization, and costs while leading to greater loss of independence and non-home discharge. Surgeons should exercise caution when considering nonoperative management in these vulnerable patients.