E341: External Validation of Nomograms Predicting 30-Day Mortality in Malignant Bowel Obstruction Patients

Tools predicting short-term prognosis and benefits of surgery in malignant bowel obstruction (MBO) patients are valuable for end-of-life clinical decisions. In 2012, Henry et al.¹ developed two nomograms for MBO, one predicting 30-day mortality, and another identifying optimal surgical candidates. This study aims to externally validate these nomograms in a distinct patient cohort.

Methods:

A retrospective chart review was conducted to identify patients aged >18 with stage 4 MBO presenting between 2008-2018. Patients’ 30-day survival status was obtained.

With Henry et al.'s short-term prognosis nomogram, total score from 5 factors (ascites, carcinomatosis, complete small bowel obstruction (SBO), leukocytosis, and hypoalbuminemia) was generated to predict 30-day mortality, irrespective of treatment approach.

Using the surgical benefit nomogram, total score from 4 factors (carcinomatosis, leukocytosis, normal albumin, non-gynecologic cancer) was used to predict 30-day mortality in complete SBO patients who received surgery.

Association between 30-day mortality and groups based on total scores was analyzed using chi-square analysis with a significance threshold of p< 0.05. Nomogram performance was assessed based on discrimination quantified using Harrell’s concordance index (C-statistic).

Results:

In the short-term prognosis validation cohort (n=312), 49.04% of MBO patients died within 30 days. 62.7% of patients exhibited 2-3 of 5 risk factors. Consistent with Henry et al.¹, higher total score categories were associated with a higher incidence of 30-day mortality (χ²=18.75, p=0.002). The short-term prognosis nomogram showed fair discriminative ability with a significant C-statistic of 0.62 (95% CI: 0.56-0.68) in this cohort.

In the surgical benefit validation cohort (n=172) of patients with complete SBO, 55 (31.98%) had surgery. 80% of surgical patients had 2-3 of 4 risk factors. Unlike Henry et al.¹, no significant association was found between total nomogram score and 30-day mortality in complete SBO patients who had surgery (χ²=2.28, p=0.68). The nomogram displayed poor discrimination with a non-significant C-statistic of 0.57 (95% CI: 0.42-0.73).

Conclusions:

Findings support predictive utility of Henry et al.'s short-term prognosis nomogram in MBO patients. However, the surgical benefit nomogram's performance in this cohort of complete SBO patients remains inconclusive, indicating potential limitations in its application. Future directions involve development of a discriminative nomogram for surgical MBO patients using this large patient cohort.