Supplementary Text 3: the issue of acquired immunity.
It is reasonable to assume that a significant number of West and Central Africans brought to the US during the slave trade were likely to have acquired immunity or cross-reactive immunity(18) given that YF and other flaviviruses were endemic in their homelands. Nonetheless, several important considerations suggest that this does not explain the data shown above. First, the importation of slaves to the US ended in 1808 and afterwards there was little migration from Africa to the US(9, 10, 23, 24, 27, 34, 36), thereby drastically decreasing the number of non-Caucasians with acquired immunity in the post-1850 epidemics as the average lifespan of non-Caucasians was less than 50 years at the time (17). None of the soldiers in the military data in the 1860s featured above were old enough to have come from Africa via the slave trade(26) and were too old to have maintained protective immunity from maternal antibodies, so the argument would not apply to them. Even the highest estimates of illegal immigration of slaves into the US would not significantly alter the numbers of non-Caucasians in the country, and therefore do not change our conclusion about the low importance of immunity acquired in Africa.
Higher acquired immunity among non-Caucasians could be explained by the substantial immigration of Caucasians into the United States during and after 1850. 19th century medical personnel observed that these new immigrants were more likely to contract YF than people that had been living in the same area for a long period of time (hence YF was commonly referred to as the “Stranger’s disease”(31)) so it was argued that the increased YF mortality for Caucasians was due to the influx of immigrants in areas that later would become YF epidemic sites(21). It was also possible that previous exposure to other flaviviruses could have provided some degree of cross-reactive immunity to Southern United States residents, which would have affected non-Caucasians, the less mobile population, at higher rates(18) (see discussion below).
While we cannot rule out the possibility of higher acquired immunity among individuals of African descent, we posit that multiple clinical infections of YF seemed to be an implicit assumption of these acquired immunity arguments, as a secondary infection recorded during an epidemic could be less severe. The fact that overwhelming majority of the data suggests that multiple clinical infections of YF are seldom, if ever, seen, weakens support for acquired immunity as an explanation for lower case mortality rates among non-Caucasians. Anecdotal reports from Assistant Surgeon Generals and medical practitioners from the 1860s state that they had never seen patients that contracted YF more than once(26), which is consistent with current data (see below). Given that incidence of YF did not seem to be consistently higher among non-Caucasians (Supplementary Table 1), it is unlikely that the rate of immune individuals would be higher for that group. This argument has to be tempered by the fact that incidence cannot account for infections that do not present clinically, which could be substantial(18). While no contemporary data directly address this question, the CDC reported in 2010 that “a second case of clinical yellow fever has never been documented in a human” (http://www.cdc.gov/travel-training/local/HistoryEpidemiologyandVaccination/page24492.html), a claim that is supported by previous assessments(35). The argument could be made that once infected, Caucasians would mount a weaker immune response than non-Caucasians leading to higher case fatality rates in subsequent outbreaks. Current vaccine data, however, indicate outstanding efficacy in all populations. Thus we conclude that the vast majority of 19th century YF patients were naive until they contracted YF during the respective epidemics.