Predicting in-hospital mortality risk for perforated peptic ulcer surgery: the PPUMS scoring system and the benefit of laparoscopic surgery: a population-based study.

Abstract

The major treatment for perforated peptic ulcers (PPU) is surgery. It remains unclear which patient may not get benefit from surgery due to comorbidity. This study aimed to generate a scoring system by predicting mortality for patients with PPU who received non-operative management (NOM) and surgical treatment. We extracted the admission data of adult (≥ 18years) patients with PPU disease from the NHIRD database. We randomly divided patients into 80% model derivation and 20% validation cohorts. Multivariate analysis with a logistic regression model was applied to generate the scoring system, PPUMS. We then apply the scoring system to the validation group. The PPUMS score ranged from 0 to 8 points, composite with age (< 45: 0 points, 45-65: 1 point, 65-80: 2 points, > 80: 3 points), and five comorbidities (congestive heart failure, severe liver disease, renal disease, history of malignancy, and obesity: 1 point each). The areas under ROC curve were 0.785 and 0.787 in the derivation and validation groups. The in-hospital mortality rates in the derivation group were 0.6% (0 points), 3.4% (1 point), 9.0% (2 points), 19.0% (3 points), 30.2% (4 points), and 45.9% when PPUMS > 4 point. Patients with PPUMS > 4 had a similar in-hospital mortality risk between the surgery group [laparotomy: odds ratio (OR) = 0.729, p = 0.320, laparoscopy: OR = 0.772, p = 0.697] and the non-surgery group. We identified similar results in the validation group. PPUMS scoring system effectively predicts in-hospital mortality for perforated peptic ulcer patients. It factors in age and specific comorbidities is highly predictive and well-calibrated with a reliable AUC of 0.785-0.787. Surgery, no matter laparotomy or laparoscope, significantly reduced mortality for scores < = 4. However, patients with a score > 4 did not show this difference, calling for tailored approaches to treatment based on risk assessment. Further prospective validation is suggested.