Can we estimate crisis death tolls by subtracting total population estimates? A critical review and appraisal

Background: In the absence of high-quality data, the death tolls of epidemics, conflicts, and disasters are often estimated using ad hoc methods. One understudied class of methods, which we term the growth rate discontinuity method (GRDM), estimates death tolls by projecting pre-crisis and post-crisis total population estimates using crude growth rates and then subtracting the results. Despite, or perhaps due to, its simplicity, this method is the source of prominent death toll estimates for the Black Death, the 1918 influenza pandemic, the Great Chinese Famine, and the Rwandan genocide, among others. Objective: In this article, we review the influence and validity of GRDM and its applications. Methods: Using statistical simulation and comparison with better-validated demographic methods, we assess the accuracy, precision, and biases of this method for estimating mortality in absolute and relative terms. Results: We find that existing GRDM estimates often misestimate death tolls by large, unpredictable margins. Simulations suggest this is because GRDM requires precision in its inputs to an extent rarely possible in the contexts of interest. Conclusions: If there is sufficient data to specify GRDM well, it is probably possible to also use a more reliable method; if there is not sufficient data, GRDM estimates are so sensitive to their assumptions that they cannot be considered informative. Contribution: These findings question the empirical suitability of existing demographic and econometric work that has used GRDM to analyse mortality crises. They also underscore the need for improved data collection in crisis settings and the utility of qualitative methods in contexts where quantification using better-validated methods is not possible.