2019/02/18 by Siddharth Chandra, Julia Christensen · 2 citations
Medicine · Mathematics · #Influenza Virus Research Studies #COVID-19 epidemiological studies #Pandemic #Influenza pandemic #Medicine #Human mortality from H5N1 #Coronavirus disease 2019 (COVID-19) #Virology #2019-20 coronavirus outbreak #Pandemic influenza #Environmental health #Outbreak #Disease #Infectious disease (medical specialty) #Internal medicine
paper · doi:10.1093/aje/kwz044
openalex publication_date 2019/02/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/29
In the article “Reassessing the Global Mortality Burden of the 1918 Influenza Pandemic,” Spreeuwenberg et al. (1) proposed a new estimate of global mortality from the 1918 influenza pandemic of 17.4 million that was markedly lower than the generally accepted estimate of 50 million or more (2–5). This estimate of 17.4 million was influenced in large part by the mortality estimate used for British India (henceforth “India”) of 6.17 million (per Table 2 in that paper). Unfortunately, we believe that this figure for India is a significant underestimation at which the authors arrived via an uncritical approach to using official statistics from India. The authors relied on death rates published in the Statistical Abstract for British India, a government publication that is widely known to severely underreport deaths. For example, the Report of the Population Data Committee by the Department of Education, Health and Lands found that the underregistration of deaths in colonial India in the first half of the 20th century ranged from 35% to 55% depending on the region (6). Taking his cue from these findings, the renowned demographer Kingsley Davis, “a recognized giant in his field” (7), had the following to say in his classic 1951 work The Population of India and Pakistan: The average (reported death) rate during the 1911–20 decade was 34 per thousand (or 31 if the deaths for 1918 are removed), as contrasted to 25 during the next decade and 23 during the 1931–40 decade. These rates are not nearly so high as they should be. The amount of under-registration certainly exceeds 30% at all times, and is probably nearer 50% (8, p. 34). These considerations prompted Davis to eschew the use of official mortality statistics in favor of decennial population head count data from the census when computing estimates of population loss in India from the pandemic, leading to an estimate of influenza deaths in 1918–1919 of 18.5 million (with a possible upward adjustment to 22.6 million). This is 3 times higher than the estimate in the article in question (8, p. 237; see also 9). A second issue relevant to the aforementioned paper that Davis addressed, on which we are able to touch only briefly because of lack of space, is the issue of differences in death rates between India and the West: Nevertheless, with this (under-registration of deaths) defect admitted, the rates are still much higher than those of the West. They are, for example, more than twice the rates of England and Wales during the same decades. If all Indian deaths were registered, the rates would doubtless be three or four times as high as the British rates (8, p. 34). Following Davis’ logic, there is reason to question the claim in the above paper that “[e]ven if most of the countries not included in our sample had excess mortality rates higher than the European (corrected) rates but lower than that of India, the effect on the overall global average would be limited (probably a few million more deaths globally)” (1, p. 2565). A large fraction of the approximately 80% of the global population not included in the calculations of the aforementioned paper lived in colonial Asia and Africa and were subject to living conditions similar to those in India. Therefore, Davis’ statement about the death rates in India being “three to four times as high as the British rates” (8, p. 34) suggests that calculated deaths from the unsampled countries would very likely significantly raise the global mortality estimate from the above paper even further, bringing it into line with earlier estimates. Conflict of interest: none declared.