Does the Pittsburgh Severity Score predict outcome in esophageal perforation?

ABSTRACT

Esophageal perforation is an uncommon and challenging surgical emergency associated with high rates of morbidity and mortality. At present, no consensus exists on optimal management of the condition. The Pittsburgh Severity Score (PSS) is a tool intended to stratify perforation severity and guide treatment. However, there is a paucity of literature examining the validity of the score or its application in a UK population. This study aims to validate the PSS and explore its use in stratifying patients with esophageal perforation into distinct subgroups with differential outcomes in an independent UK study population.All patients treated for esophageal perforation at Queen Elizabeth Hospital, Birmingham between September 2003 and October 2017 were included in this study. Cases were identified using a combination of ICD-10 and OPCS informatics search codes and prospective case collection. Data relating to the clinical presentation, diagnosis, management, and outcome of cases were recorded using a preformed data collection form. PSS predictive performance was assessed against five outcomes: rates of post-perforation and post-operative complications, in-hospital mortality, length of intensive care (ICU/HDU) stay, and total length of hospital stay.A total of 87 cases were identified, consisting of 48 (55%) iatrogenic perforations, 24 (28%) cases of spontaneous (Boerhaave's) perforation, and 15 perforations due to other etiologies (17%). Operative management was favored in this series, with 47% of all perforations being treated surgically. Overall in-hospital mortality was 13%, coupled with a median length of hospital stay of 24 days (interquartile range [IQR] 12-49), of which a median of 2 days was spent in intensive care facilities (IQR 0-14). A total of 46% of patients developed post-perforation complications, with 59% of the operatively managed cohort developing complications post-operatively.The PSS was not found to be significantly predictive of post-perforation complications (area under the ROC curve [AUROC] 0.62, p = 0.053) or in-hospital mortality (AUROC 0.69, p = 0.057) for the cohort as a whole. However, a subgroup analysis found the accuracy of the PSS to vary considerably by etiology, being significantly predictive of post-perforation complications within the subgroup of Boerhaave's perforations (AUROC 0.86, p = 0.004).In conclusion, we found that the PSS has some utility in stratifying esophageal perforation severity and predicting specific patient outcomes. However, it appears to be of more value when applied to the subgroup of patients with Boerhaave's perforations.