Case fatality ratio estimates for the 2013–2016 West African Ebola epidemic: application of Boosted Regression Trees for imputation

Purpose: The 2013–2016 West African Ebola epidemic has been the largest to date with more than 11,000 deaths in the affected countries. The data collected have provided more insight than ever before into the case fatality ratio (CFR) and how it varies with age and other characteristics. However, the accuracy and precision of naïve CFR are limited because 44% of survival outcomes were unreported. We impute unreported outcomes and combine the imputed and reported outcomes to provide improved CFR estimates. Methods & Materials: We imputed for unreported survival outcomes (i.e. recovery or death) using a Boosted Regression Tree (BRT) model and calculated the average out-of-sample performance of the simplified model used for imputation. Results: The average out-of-sample performance measures were as follows: sensitivity = 64·3% (95% CI: 53·8-73·0), specificity = 64·3% (95% CI: 53·8-73·0), percentage correctly classified = 64·2% (95% CI: 54·4-73·1) and area under the ROC curve = 70·2% (95% CI: 59·3-78·1). The CFR estimates obtained with imputation for the 2013-2016 West African epidemic were 66·5% (95% CI 61·8%-71·1%) overall and 68·9% (95% CI 62·1%-74·5%), 65·7% (95% CI 61·4%-69·5%) and 61·4% (95% CI 55·9%-67·3%) for Sierra Leone, Guinea, and Liberia, respectively. Conclusion: We achieved appreciable out-of-sample performance and CFR estimates with imputation were an improvement to CFR estimates obtained without imputation. Updated CFR estimates including imputed outcomes can be used to inform public health contingency planning for future Ebola epidemics. Improved baseline CFR estimates are essential for allocating resources and evaluating the effectiveness of future inventions.