Viral and host determinants of arboviral disease severity in African populations: A systematic review with meta-analysis

Background: Arboviral diseases caused by dengue (DENV), chikungunya (CHIKV), yellow fever (YFV), Rift Valley fever (RVFV), and Zika (ZIKV) viruses are responsible for considerable morbidity and mortality across Africa. Clinical presentations range from self-limiting febrile illness to fatal haemorrhagic syndromes, encephalitis, and chronic arthropathy, yet the factors that govern who develops severe disease remain poorly defined in African settings. We set out to synthesise what is currently known about the viral and host immune determinants of clinical severity for all five arboviruses in African populations. Methods: PubMed/MEDLINE, Scopus, Web of Science, and the WHO Global Index Medicus were searched from database inception to 8 March 2026 for studies reporting clinical severity outcomes or viral or host immune determinants of DENV, CHIKV, YFV, RVFV, or ZIKV infections in African populations or populations of documented African ancestry. Eligibility followed a Population, Exposure, Comparison, Outcome (PECO) framework. Study quality was assessed using the Newcastle-Ottawa Scale (NOS), and the certainty of evidence for meta-analytic outcomes was evaluated using the GRADE framework. The protocol was registered prospectively on PROSPERO (CRD420261334251) on 7 March 2026, before the literature search commenced. The review was conducted in accordance with PRISMA 2020, and the completed checklist is provided as S2 Table. Results: We included 154 studies published between 1978 and 2026, covering 38 African and African-diaspora countries. DENV was the most frequently studied virus (n = 57), followed by RVFV (n = 31), multiple arboviruses (n = 27), CHIKV (n = 15), YFV (n = 15), and ZIKV (n = 9). Random-effects meta-analysis using Freeman-Tukey double arcsine transformation yielded a pooled RVFV case fatality rate of 16.9% (95% CI 11.6 to 22.8; I2 = 88.4%, 19 studies) and a pooled DENV severe disease proportion of 8.5% (95% CI 3.0 to 16.2; I2 = 95.7%, 10 African studies). Given the extreme heterogeneity in both analyses, these pooled estimates should be interpreted as approximate central tendencies of the available data rather than precise epidemiological parameters, and any comparison with reported Asian cohort figures must be made with considerable caution. The certainty of evidence for both pooled outcomes was rated as very low by GRADE criteria. Egger’s test indicated significant funnel plot asymmetry for RVFV (p