Clinical and laboratory predictors of Lassa fever outcome in a dedicated treatment facility in Nigeria: a retrospective, observational cohort study.

BACKGROUND: Lassa fever is a viral haemorrhagic disease endemic to west Africa. No large-scale studies exist from Nigeria, where the Lassa virus (LASV) is most diverse. LASV diversity, coupled with host genetic and environmental factors, might cause differences in disease pathophysiology. Small-scale studies in Nigeria suggest that acute kidney injury is an important clinical feature and might be a determinant of survival. We aimed to establish the demographic, clinical, and laboratory factors associated with mortality in Nigerian patients with Lassa fever, and hypothesised that LASV was the direct cause of intrinsic renal damage for a subset of the patients with Lassa fever. METHODS: We did a retrospective, observational cohort study of consecutive patients in Nigeria with Lassa fever, who tested positive for LASV with RT-PCR, and were treated in Irrua Specialist Teaching Hospital. We did univariate and multivariate statistical analyses, including logistic regression, of all demographic, clinical, and laboratory variables available at presentation to identify the factors associated with patient mortality. FINDINGS: Of 291 patients treated in Irrua Specialist Teaching Hospital between Jan 3, 2011, and Dec 11, 2015, 284 (98%) had known outcomes (died or survived) and seven (2%) were discharged against medical advice. Overall case-fatality rate was 24% (68 of 284 patients), with a 1·4 times increase in mortality risk for each 10 years of age (p=0·00017), reaching 39% (22 of 57) for patients older than 50 years. Of 284 patients, 81 (28%) had acute kidney injury and 104 (37%) had CNS manifestations and thus both were considered important complications of acute Lassa fever in Nigeria. Acute kidney injury was strongly associated with poor outcome (case-fatality rate of 60% [49 of 81 patients]; odds ratio [OR] 15, p<0·00001). Compared with patients without acute kidney injury, those with acute kidney injury had higher incidence of proteinuria (32 [82%] of 39 patients) and haematuria (29 [76%] of 38) and higher mean serum potassium (4·63 [SD 1·04] mmol/L) and lower blood urea nitrogen to creatinine ratio (8·6 for patients without clinical history of fluid loss), suggesting intrinsic renal damage. Normalisation of creatinine concentration was associated with recovery. Elevated serum creatinine (OR 1·3; p=0·046), aspartate aminotransferase (OR 1·5; p=0·075), and potassium (OR 3·6; p=0·0024) were independent predictors of death. INTERPRETATION: Our study presents detailed clinical and laboratory data for Nigerian patients with Lassa fever and provides strong evidence for intrinsic renal dysfunction in acute Lassa fever. Early recognition and treatment of acute kidney injury might significantly reduce mortality. FUNDING: German Research Foundation, German Center for Infection Research, Howard Hughes Medical Institute, US National Institutes of Health, and World Bank.

Transforming Clinical Data into Actionable Prognosis Models: Machine-Learning Framework and Field-Deployable App to Predict Outcome of Ebola Patients.

BACKGROUND: Assessment of the response to the 2014-15 Ebola outbreak indicates the need for innovations in data collection, sharing, and use to improve case detection and treatment. Here we introduce a Machine Learning pipeline for Ebola Virus Disease (EVD) prognosis prediction, which packages the best models into a mobile app to be available in clinical care settings. The pipeline was trained on a public EVD clinical dataset, from 106 patients in Sierra Leone. METHODS/PRINCIPAL FINDINGS: We used a new tool for exploratory analysis, Mirador, to identify the most informative clinical factors that correlate with EVD outcome. The small sample size and high prevalence of missing records were significant challenges. We applied multiple imputation and bootstrap sampling to address missing data and quantify overfitting. We trained several predictors over all combinations of covariates, which resulted in an ensemble of predictors, with and without viral load information, with an area under the receiver operator characteristic curve of 0.8 or more, after correcting for optimistic bias. We ranked the predictors by their F1-score, and those above a set threshold were compiled into a mobile app, Ebola CARE (Computational Assignment of Risk Estimates). CONCLUSIONS/SIGNIFICANCE: This method demonstrates how to address small sample sizes and missing data, while creating predictive models that can be readily deployed to assist treatment in future outbreaks of EVD and other infectious diseases. By generating an ensemble of predictors instead of relying on a single model, we are able to handle situations where patient data is partially available. The prognosis app can be updated as new data become available, and we made all the computational protocols fully documented and open-sourced to encourage timely data sharing, independent validation, and development of better prediction models in outbreak response.