A scoring system for the prognosis and treatment of malignant bowel obstruction

Abstract

Background: Malignant bowel obstruction is a common result of end-stage abdominal cancer that is a treatment dilemma for many physicians. Little has been reported predicting outcomes or determining the role of surgical intervention. We sought to review our experience with surgical and nonsurgical management of malignant bowel obstruction to identify predictors of 30-day mortality and of who would most likely benefit from surgical intervention.

Methods: A chart review of 523 patients treated between 2000 and 2007 with malignant bowel obstruction were evaluated for factors present at admission to determine return to oral intake, 30-day mortality, and overall survival. Propensity score matching was used to homogenize patients treated with and without surgery to identify those who would benefit most from operative intervention.

Results: Radiographic evidence of large bowel obstruction was predictive of return to oral intake. Hypoalbuminemia and radiographic evidence of ascites or carcinomatosis were all predictive of increased 30-day mortality and overall survival. A nomogram of 5 identified risk factors correlated with increased 30-day mortality independent of therapy. Patients with large bowel or partial small bowel obstruction benefited most from surgery. A second nomogram was created from 4 identified risk factors that revealed which patients with complete small bowel obstruction might benefit from surgery.

Conclusion: Two nomograms were created that may guide decisions in the care of patients with malignant bowel obstruction. These nomograms are able to predict 30-day mortality and who may benefit from surgery for small bowel obstruction.

Fig 1

Kaplan–Meier overall survival curves of surgical versus nonsurgical therapy for the entire cohort of 523 patients (P < .001).

Fig 2

(A) Nomogram to estimate 30-day mortality for patients presenting with malignant bowel obstructions independent of therapy. One point was assigned for each of the 5 variables. (B) The distribution of 523 patients with malignant bowel obstruction for each number of risk factors contributing to 30-day mortality. The blue portion of the bar represents the number of patients alive at 30 days and the red represents the number of patients who were dead at 30 days. The percentage above each bar is the percentage dead at 30 days. (Color version of figure is available online.)

Fig 3

Propensity score matching based upon the characteristics of age, gender, the presence of carcinomatosis, the presence of complete small bowel obstruction, the presence of ascites on imaging, leukocytosis, hypoalbuminemia, and cancer diagnosis. (A) Representative of the patients from the entire cohort who had data for each factor (n = 395), which showed very little overlap between the 2 groups. (B) Representation of the population (n = 226) created after propensity scoring was conducted to produce more comparable groups. (Color version of figure is available online.)

Fig 4

(A) Nomogram to estimate surgical benefit in patients with complete small bowel obstruction. One point was assigned for each of the 4 variables. (B) Thrity-day mortality based on the computer model created from the survival characteristics for surgery and nonsurgery patients based on the number of risk factors present. As the number of factors increase, the mortality from surgery increases, whereas an inverse correlation is seen for nonsurgical therapy. (Color version of figure is available online.)

Fig 5

An algorithm purposed for use of both scoring systems in patients with malignant bowel obstruction. (Color version of figure is available online.)