Active and passive surveillance of yellow fever vaccine 17D or 17DD-associated serious adverse events: systematic review.

PURPOSE: To identify the rate of serious adverse events attributable to yellow fever vaccination with 17D and 17DD strains reported in active and passive surveillance data.

METHODS: We conducted a systematic review of published literature on adverse events associated with yellow fever. We searched 9 electronic databases for peer reviewed and grey literature in all languages. There were no restrictions on date of publication. Reference lists of key studies were also reviewed to identify additional studies.

PRINCIPAL RESULTS: We identified 66 relevant studies: 24 used active, 17 a combination of passive and active (15 of which were pharmacovigilance databases), and 25 passive surveillance. ACTIVE SURVEILLANCE: A total of 2,660,929 patients in general populations were followed for adverse events after vaccination, heavily weighted (97.7%) by one large Brazilian study. There were no observed cases of viscerotropic or neurotropic disease, one of anaphylaxis and 26 cases of urticaria (hypersensitivity). We also identified four studies of infants and children (n=2199), four studies of women (n=1334), and one study of 174 HIV+, and no serious adverse events were observed. PHARMACOVIGILANCE DATABASES: 10 of the 15 databases contributed data to this review, with 107,621,154 patients, heavily weighted (94%) by the Brazilian database. The estimates for Australia were low at 0/210,656 for "severe neurological disease" and 1/210,656 for YEL-AVD, and also low for Brazil with 9 hypersensitivity events, 0.23 anaphylactic shock events, 0.84 neurologic syndrome events and 0.19 viscerotropic events cases/million doses. The five analyses of partly overlapping periods for the US VAERS database provided an estimate of 6.6 YEL-AVD and YEL-AND cases per million, and estimates between 11.1 and 15.6 of overall "serious adverse events" per million. The estimates for the UK were higher at 34 "serious adverse events" and also for Switzerland with 14.6 "neurologic events" and 40 "serious events not neurological"/million doses. PASSIVE SURVEILLANCE: Six studies of campaigns in general populations included 94,500,528 individuals, very heavily weighted (99%) by the Brazilian data, and providing an estimate of 0.51 serious AEFIs/million doses. Five retrospective reviews of hospital or clinic records included 60,698 individuals, and no serious AEFIs were proven. The data are heavily weighted (96%) by the data from the Hospital for Tropical Diseases, London. Two studies included 35,723 children, four studies included 138 pregnant women, six studies included 191 HIV+ patients, and there was one review of patients who were HIV+, and no serious AEFIs were proven.

MAJOR CONCLUSIONS: The databases in each country used different definitions, protocols, surveillance mechanisms for the initial identification and reporting of cases, and strategies for the clinical and laboratory follow up of cases. The pharmacovigilance databases provide three sets of estimates: a low estimate from the Brazilian and Australian data, a medium estimate from the US VAERS data, and a higher estimate from the UK and Swiss data. The estimates from the active surveillance data are lower (and strongly influenced by the Brazilian data) and the estimates from the passive surveillance studies are also lower (strongly influenced by the London Hospital for Tropical Diseases data from the early 1950s). Sophisticated pathology, histopathology and tests such as PCR amplicon sequencing are needed to prove that serious adverse events were actually caused by the yellow fever vaccine, and the availability of such diagnostic capability is strongly biased towards recent reports from developed countries. Despite these variations in the estimation of serious harm, overall the 17D and 17DD yellow fever vaccine has proven to be a very safe vaccine and is highly effective against an illness with high potential mortality rates.