Dataset Information

Case Fatality Ratio Estimates for the 2013-2016 West African Ebola Epidemic: Application of Boosted Regression Trees for Imputation.

ABSTRACT:

Background

The 2013-2016 West African Ebola epidemic has been the largest to date with >11 000 deaths in the affected countries. The data collected have provided more insight into the case fatality ratio (CFR) and how it varies with age and other characteristics. However, the accuracy and precision of the naive CFR remain limited because 44% of survival outcomes were unreported.

Methods

Using a boosted regression tree model, we imputed survival outcomes (ie, survival or death) when unreported, corrected for model imperfection to estimate the CFR without imputation, with imputation, and adjusted with imputation. The method allowed us to further identify and explore relevant clinical and demographic predictors of the CFR.

Results

The out-of-sample performance (95% confidence interval [CI]) of our model was good: sensitivity, 69.7% (52.5-75.6%); specificity, 69.8% (54.1-75.6%); percentage correctly classified, 69.9% (53.7-75.5%); and area under the receiver operating characteristic curve, 76.0% (56.8-82.1%). The adjusted CFR estimates (95% CI) for the 2013-2016 West African epidemic were 82.8% (45.6-85.6%) overall and 89.1% (40.8-91.6%), 65.6% (61.3-69.6%), and 79.2% (45.4-84.1%) for Sierra Leone, Guinea, and Liberia, respectively. We found that district, hospitalisation status, age, case classification, and quarter (date of case reporting aggregated at three-month intervals) explained 93.6% of the variance in the naive CFR.

Conclusions

The adjusted CFR estimates improved the naive CFR estimates obtained without imputation and were more representative. Used in conjunction with other resources, adjusted estimates will inform public health contingency planning for future Ebola epidemics, and help better allocate resources and evaluate the effectiveness of future inventions.

SUBMITTER: Forna A

PROVIDER: S-EPMC7286386 | biostudies-literature | 2020 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Case Fatality Ratio Estimates for the 2013-2016 West African Ebola Epidemic: Application of Boosted Regression Trees for Imputation.

Forna Alpha A Nouvellet Pierre P Dorigatti Ilaria I Donnelly Christl A CA

Clinical infectious diseases : an official publication of the Infectious Diseases Society of America 20200601 12

<h4>Background</h4>The 2013-2016 West African Ebola epidemic has been the largest to date with >11 000 deaths in the affected countries. The data collected have provided more insight into the case fatality ratio (CFR) and how it varies with age and other characteristics. However, the accuracy and precision of the naive CFR remain limited because 44% of survival outcomes were unreported.<h4>Methods</h4>Using a boosted regression tree model, we imputed survival outcomes (ie, survival or death) whe ...[more]

PMID: 31328221

Similar Datasets

Project description:BackgroundHealthcare and other front-line workers are at particular risk of infection with Ebola virus (EBOV). Despite the large-scale deployment of international responders, few cases of Ebola virus disease have been diagnosed in this group. Since asymptomatic or pauci-symptomatic infection has been described, it is plausible that infections have occurred in healthcare workers but have escaped being diagnosed. We aimed to assess the prevalence of asymptomatic or pauci-symptomatic infection, and of exposure events, among returned responders to the West African Ebola epidemic 2014-2016.Methods and findingsWe used snowball sampling to identify responders who had returned to the UK or Ireland, and used an online consent and questionnaire to determine their exposure to EBOV and their experience of illness. Oral fluid collection devices were sent and returned by post, and samples were tested using an EBOV IgG capture assay that detects IgG to Ebola glycoprotein. Blood was collected from returnees with reactive samples for further testing. Unexposed UK controls were also recruited. In all, 300 individuals consented, of whom 268 (89.3%) returned an oral fluid sample (OFS). The majority had worked in Sierra Leone in clinical, laboratory, research, and other roles. Fifty-three UK controls consented and provided samples using the same method. Of the returnees, 47 (17.5%) reported that they had had a possible EBOV exposure. Based on their free-text descriptions, using a published risk assessment method, we classified 43 (16%) as having had incidents with risk of Ebola transmission, including five intermediate-risk and one high-risk exposure. Of the returnees, 57 (21%) reported a febrile or diarrhoeal illness in West Africa or within 1 mo of return, of whom 40 (70%) were not tested at the time for EBOV infection. Of the 268 OFSs, 266 were unreactive. Two returnees, who did not experience an illness in West Africa or on return, had OFSs that were reactive on the EBOV IgG capture assay, with similar results on plasma. One individual had no further positive test results; the other had a positive result on a double-antigen bridging assay but not on a competitive assay or on an indirect EBOV IgG ELISA. All 53 controls had non-reactive OFSs. While the participants were not a random sample of returnees, the number participating was high.ConclusionsThis is the first study, to our knowledge, of the prevalence of EBOV infection in international responders. More than 99% had clear negative results. Sera from two individuals had discordant results on the different assays; both were negative on the competitive assay, suggesting that prior infection was unlikely. The finding that a significant proportion experienced "near miss" exposure events, and that most of those who experienced symptoms did not get tested for EBOV at the time, suggests a need to review and standardise protocols for the management of possible exposure to EBOV, and for the management of illness, across organisations that deploy staff to outbreaks.

Project description:BackgroundTo date, epidemiological studies at the index site of the 2013-16 west African Ebola outbreak in Meliandou, Guinea, have been restricted in their scope. We aimed to determine the occurrence of previously undocumented Ebola virus disease (EVD) cases and infections, and to reconstruct transmission events.MethodsThis cross-sectional seroprevalence survey of the adult population of Meliandou used a highly specific oral fluid test and detailed interviews of all households in the village and key informants. Each household was interviewed, with all members prompted to describe the events of the outbreak, any illness within the household, and possible contact with suspected cases. Information for deceased individuals was provided by relatives living in the same household. Symptoms were based on Ebola virus Makona variant EVD case definitions (focusing on fever, vomiting, and diarrhoea). For antibody testing, we used an Ebola virus glycoprotein IgG capture enzyme immunoassay developed from a previously validated assay. A maximum exposure level was assigned to every participant using a predetermined scale. We used a generalised linear model (logit function) to estimate odds ratios for the association of sociodemographic variables and exposure level with Ebola virus infection. We adjusted estimates for age and maximum exposure, as appropriate.FindingsBetween June 22, and July 9, 2017, we enrolled 237 participants from 27 households in Meliandou. Two households refused to participate and one was absent. All adults in participating households who were present for the interview provided an oral fluid swab for testing, of which 224 were suitable for analysis. In addition to the 11 EVD deaths described previously, on the basis of clinical description and oral fluid testing, we found two probable EVD deaths and eight previously unrecognised anti-Ebola virus IgG-positive survivors, including one who had mild symptoms and one who was asymptomatic, resulting in a case fatality of 55·6% (95% CI 30·8-78·5) for adults. Health-care work (adjusted odds ratio 6·64, 1·54-28·56; p=0·001) and level of exposure (odds ratio adjusted for linear trend across five levels 2·79, 1·59-4·883; p<0·0001) were independent risk factors for infection.InterpretationEbola virus infection was more widespread in this spillover population than previously recognised (21 vs 11 cases). We show the first serological evidence of survivors in this population (eight anti-Ebola virus IgG seropositive) and report a case fatality lower than previously reported (55·6% vs 100% in adults). These data show the high community coverage achievable by using a non-invasive test and, by accurately documenting the beginnings of the west African Ebola virus outbreak, reveal important insight into transmission dynamics and risk factors that underpin Ebola virus spillover events.FundingUS Food and Drug Administration, Wellcome Trust, and German Research Council.

Dataset Information

Case Fatality Ratio Estimates for the 2013-2016 West African Ebola Epidemic: Application of Boosted Regression Trees for Imputation.

Background

Methods

Results

Conclusions

Publications

Case Fatality Ratio Estimates for the 2013-2016 West African Ebola Epidemic: Application of Boosted Regression Trees for Imputation.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets