2019/02/18 by Peter Spreeuwenberg, Madelon Kroneman, John Paget · 1 citation
Medicine · #Influenza Virus Research Studies #Travel-related health issues #Medicine
paper · doi:10.1093/aje/kwz041
openalex publication_date 2019/02/18 · openalex created_date 2022/05/12 · openalex updated_date 2026/07/29
We thank Drs. Chandra and Christensen for their letter (1) regarding our paper on the mortality burden of the 1918 influenza pandemic (2). We assessed the distribution of excess mortality rates due to the pandemic by country based on observed age-specific population mortality rates and, critically, used knowledge from other influenza pandemics and about expected variation in mortality both between and within countries. The average of this distribution was taken as the excess rate for the world and used to calculate the total worldwide mortality given a world population of 1.92 billion. Incomplete registries are not problematic as long as the methodology was consistent for the 6 years during which the data we used were collected. As long as there was good representativeness, the rate of excess mortality due to the pandemic would be the same for incomplete and complete registries. We did not want to make an assessment of the pandemic burden in India, which in our view is extremely challenging (2); rather, we used India so that we could include a country with an excess at the very high end of the distribution. It is important to note that Chandra and Christensen have misread our Table 2 (1, 2). The difference between the 2 total death counts is not the mortality estimate for British India but the difference between what the total global estimates would be if the British India data were present or absent from the sample of countries used to make the global estimations. We stressed that the validity of the estimation should be studied extensively. Yearly mortality rates from the early 20th century can vary strongly between adjacent years and between countries (because of famine, war, outbreak of infectious diseases, etc.). We tried to control for this by using a short reference period, age-specific mortality rates, and 3 estimation scenarios and by controlling for a few factors. Most importantly, one should use biomedical knowledge and experience with pandemic influenza to validate the findings (see the Web material in our original article (2)). The second point made by Chandra and Christensen is that some parts of the world (colonial Asia and Africa) would have looked more like India than Europe. This needs to be carefully validated; China was hit mildly by the 1918 influenza pandemic (3). Another example of a high-population country in Asia is Indonesia, where estimates of excess mortality range from 402,163 (4) to 4.3 million (5). If the rate is presented as the excess per 100,000 persons, it varies from 1,159 to 12,464. The first rate is based on weekly registry mortality rates (4) for Java (population of 34 million) in the second wave. The second value of 4.3 million, which is several times India’s excess, would imply that 12.5% of the total population of Java had died in 1918–1919. The 1918 pandemic age signature places the highest excess mortality rate in the group of persons 20–49 years of age and the lowest in those 5–19 years of age and 50 years of age or older (6). Van Steenis (7) compared Magelang Indonesia, with Amsterdam, the Netherlands. In Amsterdam, 33% of the deaths were among those younger than 20 years of age; 51.9% were among those 20–50 years of age; and 15.1% were among those older than 50 years of age. In Magelang, the numbers were 53.5%, 32.2%, 14.3%, respectively. This clearly indicates that most of the 1918 excess was among young persons in Magelang, which suggests that other factors played a role in Indonesia. Furthermore, one would expect such a high mortality rate to lead to a significant decline in the population size from which the population would take several years to recover. The population registry (8), however, does not show such a decline. In comparison, in 1773 in Sweden, 5.2% of the total population died (the highest mortality rate found in the Human Mortality Database (www.mortality.org)), which resulted in a population decline that required a recovery period of approximately 4 years. Chandra and Christensen implied a mortality distribution with 2 peaks: a “European” peak with an excess mortality rate of approximately 500 deaths per 100,000 persons and an “Indian” peak of approximately 3,000 deaths per 100,000 persons. For the purposes of an experiment, assuming there were 200 countries and 50 million total deaths, the world rate would be approximately 2,750. This requires 20 “European-like” and 180 “Indian-like” countries. One hundred million deaths would mean that slightly more than 5% of the world population died from the pandemic, which would mean that in most countries, the population would have declined and would have needed several years to recover. This is contrary to what was observed and to how influenza pandemics behave and lead to excess mortality. Conflict of interest: none declared.