Outcome in patients with open abdomen treatment for peritonitis: a multidomain approach outperforms single domain predictions

Abstract

Numerous patient-related clinical parameters and treatment-specific variables have been identified as causing or contributing to the severity of peritonitis. We postulated that a combination of clinical and surgical markers and scoring systems would outperform each of these predictors in isolation. To investigate this hypothesis, we developed a multivariable model to examine whether survival outcome can reliably be predicted in peritonitis patients treated with open abdomen. This single-center retrospective analysis used univariable and multivariable logistic regression modeling in combination with repeated random sub-sampling validation to examine the predictive capabilities of domain-specific predictors (i.e., demography, physiology, surgery). We analyzed data of 1,351 consecutive adult patients (55.7% male) who underwent open abdominal surgery in the study period (January 1998 to December 2018). Core variables included demographics, clinical scores, surgical indices and indicators of organ dysfunction, peritonitis index, incision type, fascia closure, wound healing, and fascial dehiscence. Postoperative complications were also added when available. A multidomain peritonitis prediction model (MPPM) was constructed to bridge the mortality predictions from individual domains (demographic, physiological and surgical). The MPPM is based on data of n = 597 patients, features high predictive capabilities (area under the receiver operating curve: 0.87 (0.85 to 0.90, 95% CI)) and is well calibrated. The surgical predictor "skin closure" was found to be the most important predictor of survival in our cohort, closely followed by the two physiological predictors SAPS-II and MPI. Marginal effects plots highlight the effect of individual outcomes on the prediction of survival outcome in patients undergoing staged laparotomies for treatment of peritonitis. Although most single indices exhibited moderate performance, we observed that the predictive performance was markedly increased when an integrative prediction model was applied. Our proposed MPPM integrative prediction model may outperform the predictive power of current models.

Keywords: Decision support; MPI score; Mortality; Open abdomen; Peritonitis; SAPS-II score.

Fig. 1

Calibration plots of predicted mortality versus observed mortality. Calibration plots of predicted mortality versus observed mortality using demographic predictors (age and sex of the patients; panel A), physiological predictors (SAPS-II and MPI scores; panel B), surgical predictors (wound healing disorders and skin closure; panel C). Panel D illustrates the calibration of the multidomain peritonitis prediction model, which includes the predictors from all three domains. The diagonal red lines denote a 1:1 relationship between predicted and observed mortality

Fig. 2

Marginal effects plots of the multidomain peritonitis prediction model. Shaded bands and error bars denote the 95% confidence interval. A, B demographic predictors, C-E physiological predictors and F-G surgical predictors. Only one predictor is varied in each panel while the other predictors are held constant: here, the predictor-specific predictions are adjusted for a 66 year old male patient with SAPS-II and MPI scores of 46 and 21, respectively, 21 days at ICU with no wound healing disorders and successful skin closure. Note that changing these adjustment values would result only in a vertical shift the outcome predictions – the shape of the curves as well as the prediction differences between categories would remain the same

Fig. 3

Diagnostic performance of single predictor models, domain-specific models and the multidomain peritonitis prediction model in predicting the survival outcome in patients with open abdomen treatment for peritonitis. A repeated random sub-sampling validation was used to compute distributions of quantitative indicators (balanced accuracy, log diagnostic odds ratio, negative predictive value, positive predictive value, sensitivity and specificity). Box plots illustrate the median and interquartile ranges of these distribution. Capitalized predictors denote logistic regression models including all predictors of a particular domain, i.e., the model DEMOGRAPHICS includes the age of the patient and sex as predictors