August's Paper of the Month aims to systematically review and compare risk prediction algorithms for prolonged postoperative ileus (PPOI) following colorectal surgery.
Risk prediction algorithms for prolonged postoperative ileus: A systematic review
Liu GXH, Milne T, Xu W, Varghese C, Keane C, O’Grady G, et al.
Colorectal Dis. 2024; 26: 1101–1113.
What is known about the subject?
Prolonged postoperative ileus (PPOI) affects 10%-30% of patients after gastrointestinal surgery. It is associated with nausea and vomiting, obstipation, abdominal distention, dehydration, electrolyte disturbances and sepsis. This differs from physiological postoperative ileus, which occurs in response to surgical stress and is typically self-limiting. PPOI could persist for many days and is associated with a significant healthcare burden. Nevertheless, a notable degree of heterogeneity is apparent in the literature with regard to the definition of PPOI. The risk factors for PPOI have been the subject of extensive studies, although the findings have not been entirely consistent. Male sex, age, cardiac comorbidities, previous abdominal surgery and laparotomy or ostomy creation are among the most cited.
Risk prediction algorithms would enable clinicians to quickly identify patients at high risk for PPOI and inform appropriate management and prophylaxis.
What the study adds?
The aim of this study was to systematically review and compare risk prediction algorithms for PPOI following colorectal surgery.
A total of 11 studies (n = 87 549 patients), published between 2011 and 2022, were included. All risk prediction algorithms assessed PPOI following elective laparoscopic procedures, most of which were colorectal resections. There was significant heterogeneity between how risk prediction algorithms were developed and tested, including differences in PPOI definition, assessment of model performance and the frequency and rigor of statistical validation. Most risk prediction algorithms generated a numeric score based on the number of PPOI risk factors using a point-based system, which was then used to categorize patients into PPOI risk categories. Overall PPOI risk prediction algorithms demonstrated moderate predictive value: AUCs ranged from 0.68 to 0.79.
Implications for colorectal practice
Prolonged postoperative ileus (PPOI) poses a significant burden on patients and the healthcare system. Previous studies have attempted to predict PPOI using risk prediction algorithms. An accurate risk prediction algorithm for PPOI would fulfil a critical role in guiding effective resource allocation for patients after surgery. While the ideal risk prediction algorithm would demonstrate excellent sensitivity, specificity and positive and negative predictive values, models which are unidirectionally predictive would also be of immense clinical benefit. However, none of the 11 studies identified in the review fulfilled these criteria.
As previously stated, a common limitation observed in PPOI studies is the absence of a widely accepted definition for PPOI, which restricts both the generalisability of findings and the ability to compare different risk prediction algorithms. This study corroborates these concerns. In particular, the analysis revealed significant discrepancies in the definition of PPOI across studies. These included differences in whether a diagnosis required clinical, radiological, or procedural (e.g., NGT insertion) parameters, the specific characteristics of each criterion, or the point of departure (POD) after which PPOI could be defined.
It is therefore recommended that future investigators build on existing data by conducting prospective multicentre studies using a standardised definition of PPOI and transparent and consistent model performance assessment.