Surviving paediatric sepsis in Tanzania: a prospective cohort study to identify risk factors

Abstract

Abstract Background In 2015, there were 5·9 million deaths in children aged under 5 years worldwide: mortality is highest in sub-Saharan Africa (81 per 1000 livebirths) and sepsis is an important cause of these deaths. Paediatric sepsis is preventable and treatable, yet remains a serious and life-threatening condition. Prompt recognition and early treatment can improve survival; however, the understanding of how to identify and manage paediatric sepsis in sub-Saharan Africa is poor because of a lack of regional data. These data are crucial for the identification of high-risk patients, development of triage systems, and the reduction of barriers to care. In this study, we aim to identify mortality risk factors for paediatric sepsis patients at a national referral hospital in Tanzania. Methods We conducted an exploratory analysis of prospective cohort pilot data. We included children aged between 28 days and 14 years in the emergency medicine department at Muhimbili National Hospital in Dar es Salaam with sepsis (defined as ≥2 clinical criteria for systemic inflammatory response syndrome). The primary outcome was in-hospital mortality and secondary outcomes included mortality in the emergency department and length of stay. We used t -tests and Wilcoxon rank-sum, Kruskal Wallis, χ 2 , and Fisher's exact tests for data analysis. Findings Of the 2232 children screened between July 1 and September 30, 2017, 433 were eligible, 405 were enrolled, and 402 were followed to discharge or death. Median age was 25·2 months (IQR 11·4–63·5) and prevalence of malaria and HIV was 9·1% (28/307) and 1·7% (n=7), respectively. 247 (61·0%) patients were referred from outside health facilities and 116/244 (47·5%) had received antibiotics before arrival. Mortality was 14·2% (n=57): 89·5% (n=51) died on a hospital ward and 10·5% (n=6) died in the emergency department. Non-survivors were younger than survivors (median age 16 and 27 months, respectively, p=0·002), and more likely to have respiratory insufficiency (54 (94·7%) vs 293 (84·9%), p=0·046) or altered mental status (43 (75·4%) vs 171 (49·6%), p Interpretation We identified several risk factors for mortality, but only that of referral status can be modified. Most children—and a disproportionately high number of non-survivors—were assessed at other health facilities before coming to Muhimbili, suggesting that there could be an opportunity for improved sepsis identification and earlier intervention. Next steps include development of a clinical illness severity score to allow risk stratification of patients and also investigation of barriers to care. Data collection is ongoing. Funding University of California, San Francisco, Department of Pediatrics Clinical-Translational Pilot.