Africa carries the largest global burden of sepsis, yet international sepsis treatment recommendations and evidence on vasoactive support are driven largely by data from high-income settings. The authors sought to characterize real-world patterns of vasopressor use among adults admitted with sepsis in East Africa and to identify clinical factors associated with vasopressor administration.
This work is a retrospective cohort analysis pooling patient-level data from two prior observational studies conducted at three hospitals: one in Uganda and two in Rwanda. No interventional procedures were performed as part of this analysis.
Included were adults hospitalized with sepsis defined in the source as suspected infection plus at least two quick Sequential Organ Failure Assessment (qSOFA) criteria. The pooled cohort comprised 471 patients. Patient-level demographics reported in the source: median age 51 years (interquartile range [IQR] 36–65), 263 (57%) were female, and 121 (26%) were living with HIV.
The analysis used routinely collected clinical data from the contributing retrospective studies. Variables examined included vital signs at sepsis diagnosis (respiratory rate, blood pressure, oxygen saturation, Glasgow Coma Scale), site of infection (including abdominal infection), IV fluid volumes received, Universal Vital Assessment (UVA) score (a mortality risk score used in the source), and hospital site (Uganda versus Rwanda). The primary exposure of interest was receipt of any vasopressor during hospitalization.
Primary descriptive outcomes reported were the proportion of patients receiving vasopressors and in-hospital mortality. In the pooled cohort, 113 of 471 patients (24%) received a vasopressor. Overall in-hospital death occurred in 139 patients (30%). The authors performed univariable logistic regression to identify predictors of vasopressor receipt and a multivariable model to identify independent predictors. Exploratory machine learning analyses were also used to replicate findings.
In univariable logistic regression, several clinical features at the time of sepsis diagnosis were associated with subsequent vasopressor administration. These included:
These univariable associations indicate that sicker patients across multiple physiologic domains were more likely to be started on vasopressors in these hospitals.
When adjusted for covariates in multivariable logistic regression, the principal independent predictor of vasopressor receipt reported in the source was a higher UVA mortality risk score. The adjusted odds ratio for UVA score predicting vasopressor use was 1.25 (95% CI, 1.14–1.27; p < 0.001). The source also notes that hospitalization in Uganda was predictive of vasopressor administration in the cohort.
The authors conducted exploratory machine learning analyses intended to validate and replicate the relationships observed in regression models. Those exploratory analyses reportedly reproduced the main predictive findings, supporting the association between higher UVA score and vasopressor receipt. The source does not provide further methodological detail in the abstract regarding the specific machine learning methods, hyperparameters, or performance metrics.
Among the 471 patients analyzed, 113 (24%) received vasopressors and 139 (30%) died during hospitalization. The source abstracts these aggregate outcomes but does not provide additional adjusted mortality analyses or causal inference linking vasopressor use to outcomes within the abstract text.
The pooled retrospective analysis found that higher UVA mortality risk scores and hospitalization in Uganda predicted vasopressor administration among adults with sepsis in East Africa. The authors emphasize that guidelines for sepsis management, including vasopressor indications, largely reflect evidence from high-income countries and that data from African settings are limited. They conclude that further research is needed to clarify the appropriate role, optimal timing, and specific indications for vasopressors in the management of sepsis in African hospitals.
As a retrospective, observational description, the study cannot establish causality between clinical features and vasopressor use or between vasopressor use and outcomes. The abstract reports pooled data and key associations but does not include full methodological detail, such as timing of vasopressor initiation, specific agents used, dosing, or centrally adjudicated indications; these details were not reported in the abstract. The authors call for additional research to address these knowledge gaps and to inform context-appropriate sepsis guidance for low- and middle-income settings.
In this multicenter East African cohort of adults with sepsis, nearly one-quarter received vasopressors and nearly one-third died. A higher UVA mortality risk score was the strongest independent predictor of vasopressor use, and care location (Uganda) was associated with higher likelihood of vasopressor administration. The findings underscore the need for prospective and contextualized research on vasopressor timing, indications, and outcomes in African sepsis care.