To the Editor:
Sandberg and colleagues address the view expressed elsewhere that estimates of excess deaths after hurricane Maria were “all over the place”.1 Although we mostly agree with their findings, we want to address three elements of their article: (1) the discussion of our results,2 (2) mortality risk within the displacement model, and (3) the use of vital registration data.
Our estimate of excess deaths applied to the period September to December2 (1139 excess deaths; 95% CI: 1006, 1272), not September to October as described by Sandberg et al. Our estimate is close to the census-based model for September to February (1191 excess deaths, 95% CI: 836, 1544).1,3 Estimates produced using vital statistics and nondisplacement counterfactuals, and their reanalyses are all similar1,4 so are not “all over the place,” as also explained by Sandberg et al.
For the displacement models, the authors indicate that to estimate excess deaths we need “… a counterfactual estimate of what mortality would have been had the disaster not occurred.”1 The incorporation of displacement of persons from Puerto Rico to elsewhere due to Hurricane Maria ought to account for migration selectivity. Our main concern is that the displacement models assume a higher mortality hazard for those displaced,4,5 suggesting that displaced persons would have higher mortality than persons who remained in Puerto Rico. Models where the mortality hazard is similar for both populations result in an estimated 2322 excess deaths. Estimates that account for migrant selectivity and assume that displaced persons have a lower mortality hazard would result in estimates of less than 2322 excess deaths. Within displacement models, migration selectivity would imply lower mortality hazard for those displaced. Selectivity favoring lower mortality hazard is well-documented for Puerto Rican migration dynamics and for infant mortality,6 although admittedly in contexts preceding hurricane Maria.
Regarding the use of vital registration data, the Puerto Rico Vital Statistics System captures virtually every death that occurs in Puerto Rico7 and this data should be used for future analysis of this kind, as has been the convention.1
Alexis R. Santos-Lozada
Pennsylvania State University, University Park, PA, firstname.lastname@example.org
Jeffrey T. Howard
University of Texas at San Antonio, San Antonio, TX
1. Sandberg J, Santos-Burgoa C, Roess A, et al. All over the place?: differences in and consistency of excess mortality estimates in Puerto Rico after Hurricane Maria. Epidemiology. 2019;30:549–552.
2. Santos-Lozada AR, Howard JT. Use of death counts from vital statistics to calculate excess deaths in Puerto Rico following Hurricane Maria. JAMA. 2018;320:1491–1493.
3. Santos-Burgoa C, Sandberg J, Suárez E, et al. Differential and persistent risk of excess mortality from Hurricane Maria in Puerto Rico: a time-series analysis. Lancet Planet Health. 2018;2:e478–e488. doi:10.1016/S2542-5196(18)30209-2
4. Howard JT, Santos-Lozada AR. Mortality risk within counterfactual models: Puerto Rico and Hurricane Maria. Lancet Planet Health. 2019;3:e207–e208.
6. Landale NS, Gorman BK, Oropesa RS. Selective migration and infant mortality among Puerto Ricans. Matern Child Health J. 2006;10:351–360.
7. Rodriguez-Ayuso I, Marazzi-Santiago M. Nuevas Estadisticas de Mortalidad. 2010.San Juan:Puerto Rico Institute of Statistics.