Developing a tool to predict outcomes in surgical patients with sepsis at university teaching hospitals of Kigali (CHUK) and Butare (CHUB)

Abstract

Background: Sepsis is common in surgical patients, and its presence influences the outcomes in those to undergo surgery. Factors such as advanced age, presence of comorbidities and many other conditions increase mortality in surgical patients with sepsis. There is no single test to diagnose sepsis, but a set of criteria that have kept evolving from 1991 onwards. The current definition of sepsis generated in 2016 introduced the Sequential (Sepsis-related) Organ Failure Assessment (SOFA) score simplified into quick Sequential (Sepsis-related) Organ Failure Assessment score qSOFA score that not only helps to define sepsis but also to identify patients who are likely to die from it. The qSOFA score has been validated in high income countries but some authors advocated for its recalibration. Objectives: The aim of this study was to develop a prognostic tool accurate in predicting outcomes in surgical patients with sepsis who presented at University Teaching Hospital of Kigali (CHUK), University Teaching Hospital of Butare (CHUB) and in other centers with limited resources Methods: This was a prospective cohort study conducted over a period of one year from February 2018 to January 2019. The patients recruited in the first 6 months at CHUK served as the derivation cohort and those recruited in the next 6 months from both CHUK and CHUB served as the validation cohort. We used a pre-established questionnaire for data collection, the data were entered in excel, and analyzed in STATA version 14. Appropriate statistical tests were used for the derivation of the Kigali Surgical Sepsis (KiSS) score and its prognostic accuracy was tested by comparing it with qSOFA score in terms of sensitivity, specificity and their area under receiver operator characteristic (AUROC) curves. Results: A total of 288 patients were recruited with 144 in each cohort. The mean age was 36.5 and median age was 32.6. Males were 117/288 (40.6%) and females were 171/288 (59.4%). The mean LOHS was 22.9 days. The overall intensive care unit (ICU) admission rate was 51.4% and in-hospital mortality rate was 21.7%. Factors associated with hospital mortality were age above 55 years (p = 0.034), presence of comorbidities (p = 0.069), hypotension (p = 0.014), tachycardia (p = 0.061), tachypnea (p = 0.028), decreased level of consciousness (p = 0.021), presence of GIT perforation (p = 0.026) and number of impaired organ function (p = 0.035). A predictive score (KiSS score) consisting of six parameters was derived from these factors and compared to qSOFA score. The sensitivity of KiSS score in predicting mortality was 73% (vs 52% for qSOFA), and the specificity was 97% (vs 87% for qSOFA). The predictive validity for hospital mortality was assessed by Area under Receiver Operator Characteristic (AUROC) curve and it was 0.939 (95% CI, p<0.001) for KiSS and 0.684 (95% CI, p<0.001) for qSOFA. Conclusion: The Kigali Surgical Sepsis (KiSS) score developed from this study was found to be superior to the qSOFA score in predicting hospital mortality. The KiSS score showed an added advantage of stratifying surgical patients to be operated on into those with good prognosis, those with variable prognosis and those with poor prognosis