Ebola Virus Disease mathematical models and epidemiological parameters: a systematic review and meta-analysis

Abstract

SummaryBackgroundEbola Virus Disease (EVD) poses a recurring risk to human health. Modelling can provide key insights informing epidemic response, hence synthesising current evidence about EVD epidemiology and models is critical to prepare for future outbreaks.MethodsWe conducted a systematic review (PROSPERO CRD42023393345) and meta-analysis of EVD transmission models and parameters characterising EVD transmission, evolution, natural history, severity, risk factors and seroprevalence published prior to 7th July 2023 from PubMed and Web of Science. Two people screened each abstract and full text. Papers were extracted using a bespoke Access database, 10% were double extracted. Meta-analyses were conducted to synthesise information where possible.FindingsWe extracted 1,280 parameters and 295 models from 522 papers. Basic reproduction number estimates were highly variable (central estimates between 0.1 and 12.0 for high quality assessment scores), as were effective reproduction numbers, likely reflecting spatiotemporal variability in interventions. Pooled random effect estimates were 15.4 days (95% Confidence Interval (CI) 13.2-17.5) for the serial interval, 8.5 (95% CI 7.7-9.2) for the incubation period, 9.3 (95% CI 8.5-10.1) for the symptom-onset-to-death delay and 13.0 (95% CI 10.4-15.7) for symptom-onset-to-recovery. Common effect estimates were similar albeit with narrower CIs. Case fatality ratio estimates were generally high but highly variable (from 0 to 100%), which could reflect heterogeneity in underlying risk factors such as age and caring responsibilities.InterpretationWhile a significant body of literature exists on EVD models and epidemiological parameter estimates, many of these studies focus on the West African Ebola epidemic and are primarily associated with Zaire Ebola virus. This leaves a critical gap in our knowledge regarding other Ebola virus species and outbreak contexts.FundingUKRI, NIHR, Academy of Medical Sciences, Wellcome, UK Department for Business, Energy, and Industrial Strategy, BHF, Diabetes UK, Schmidt Foundation, Community Jameel, Royal Society, and Imperial College London.Research in ContextEvidence before this studyWe searched Web of Science and PubMed up to 7th July 2023 using the search terms: Ebola, epidemiology, outbreaks, models, transmissibility, severity, delays, risk factors, mutation rates and seroprevalence. We identified 179 reviews or overviews of different aspects of Ebola virus disease (EVD) transmission, of which we explored 11 that had “systematic” or “meta” in the title plus one included by expert recommendation. Five reviews focused on case fatality ratios, with estimates ranging between 34-42% for the Bundibugyo Ebola virus species, 53-69% for the Sudan species, 31.6-100% for the Zaire species, and pooled estimates ranging between 28-65% from reviews not specifying the species. Three reviews estimated seroprevalence to be between 3.3-8% depending on the setting and time. Three reviews investigated risk factors and found that caring for a case in the community and participation in traditional funeral rites are strongly associated with acquiring disease. Two reviews reported the incubation period to be 6.3 days for the Bundibugyo species, a range of 3.35-14 days for the Sudan species, and a range of 9-11.4 days across studies on the Zaire species. We found one review considering each of the following: basic reproduction number (1.34– 2.7 for Sudan species and 1.8 for Zaire species), serial interval (15-15.3 days for Zaire species), latent period (11.75 days for a combination of Zaire and unspecified species), and secondary attack rates (12.5%, species unspecified). Two reviews consider transmission models, identifying that it is difficult to accurately model the impact of time-dependent changing factors without high quality data, and data are often missing, complicating proper parameterisation of the underlying transmission mechanisms. One specific review looked at the Sudan EVD in response to the outbreak in Uganda in 2023, which highlighted the lack of vaccines and treatment available for this species.Added value of this studyWe provide a comprehensive summary of all available peer reviewed literature of transmission models and the variables needed to parameterise them across all EVD species and outbreaks. Our study synthesises all available analyses until 2023 and additionally considers attack rates, overdispersion and mutation rates. We give updated pooled random effects meta-analyses of incubation periods, serial intervals, symptom onset to death and symptom onset to recovery and, where possible, provide species-specific estimates in the Supplementary Material. We also provide ranges for the basic reproduction number and case fatality ratios without running meta-analyses because these are very setting dependent. We identify that most evidence (92%) is for the Zaire species and highlight that there are knowledge gaps for other species, which should be explored in the future. All our data is held within a bespoke open-source R package to enable others to use this information easily during their model building and updates.Implications of all the available evidencePrevious outbreaks of infectious pathogens, including the 2013-2016 West African EVD epidemic, emphasise the usefulness of computational modelling in assessing epidemic dynamics and the impact of mitigation strategies. Our study provides an updated and broader overview of all the necessary information for designing and parameterising mathematical models for use in future outbreaks of EVD, including a centralised database for other researchers to use and contribute data to.