Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Feb 23;11(2):e0005356.
doi: 10.1371/journal.pntd.0005356. eCollection 2017 Feb.

Predicting Ebola infection: A malaria-sensitive triage score for Ebola virus disease

Mary-Anne Hartley  1   2 , Alyssa Young  1 , Anh-Minh Tran  1 , Harry Henry Okoni-Williams  1 , Mohamed Suma  1 , Brooke Mancuso  1 , Ahmed Al-Dikhari  1 , Mohamed Faouzi  3
Affiliations

Predicting Ebola infection: A malaria-sensitive triage score for Ebola virus disease

Mary-Anne Hartley et al. PLoS Negl Trop Dis. .

Abstract

Background: The non-specific symptoms of Ebola Virus Disease (EVD) pose a major problem to triage and isolation efforts at Ebola Treatment Centres (ETCs). Under the current triage protocol, half the patients allocated to high-risk "probable" wards were EVD(-): a misclassification speculated to predispose nosocomial EVD infection. A better understanding of the statistical relevance of individual triage symptoms is essential in resource-poor settings where rapid, laboratory-confirmed diagnostics are often unavailable.

Methods/principal findings: This retrospective cohort study analyses the clinical characteristics of 566 patients admitted to the GOAL-Mathaska ETC in Sierra Leone. The diagnostic potential of each characteristic was assessed by multivariate analysis and incorporated into a statistically weighted predictive score, designed to detect EVD as well as discriminate malaria. Of the 566 patients, 28% were EVD(+) and 35% were malaria(+). Malaria was 2-fold more common in EVD(-) patients (p<0.05), and thus an important differential diagnosis. Univariate analyses comparing EVD(+) vs. EVD(-) and EVD(+)/malaria(-) vs. EVD(-)/malaria(+) cohorts revealed 7 characteristics with the highest odds for EVD infection, namely: reported sick-contact, conjunctivitis, diarrhoea, referral-time of 4-9 days, pyrexia, dysphagia and haemorrhage. Oppositely, myalgia was more predictive of EVD(-) or EVD(-)/malaria(+). Including these 8 characteristics in a triage score, we obtained an 89% ability to discriminate EVD(+) from either EVD(-) or EVD(-)/malaria(+).

Conclusions/significance: This study proposes a highly predictive and easy-to-use triage tool, which stratifies the risk of EVD infection with 89% discriminative power for both EVD(-) and EVD(-)/malaria(+) differential diagnoses. Improved triage could preserve resources by identifying those in need of more specific differential diagnostics as well as bolster infection prevention/control measures by better compartmentalizing the risk of nosocomial infection.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1
Fig 1. Demographic and epidemiological characteristics of EVD infection.
(A) Number of patients according to EVD or malaria test result. (B) Gender distribution of EVD infection. (C) Malaria prevalence among EVD(-) and EVD(+) cohorts. (D) Average age of EVD(-) and EVD(+) cohorts. (E) Probability of testing EVD(+), EVD(-)/malaria(+) or EVD(-)/malaria(-) according to age. (F) Geographical distribution of EVD and malaria prevalence among admissions at the GOAL-Mathaska ETC, Sierra Leone†. (G) Number of admissions and EVD prevalence according to distance of the referred patient from the ETC†. † Representing 525/552 patients, for which EVD status and geographical location is known. *: p<0.05, **: p<0.005, ***: p<0.001, ns: not significant, ETC: Ebola Treatment Centre.
Fig 2
Fig 2. Accuracy of current triage methods.
(A) Number of EVD(+) and EVD(-) patients triaged into the low-risk “suspect” and high-risk “probable” wards using the WHO triage protocol [7]. (B) Number of days spent in the ETC according to the probability of being diagnosed as either EVD(+) (red) or EVD(-) with malaria (green) or with neither EVD nor malaria (blue). (C) The sensitivity and specificity of predicting EVD(+) patients in our cohort using the scoring system of Levine et al. [11]. The area under the receiver-operator characteristic (ROC) curve represents the discriminative power of the score. (D) Percentage of EVD(+) and EVD(-) patients in our cohort classified in the various risk categories as proposed by the scoring system of Levine et al. [11].
Fig 3
Fig 3. Prevalence of the clinical signs and symptoms recorded at triage.
(A) Prevalence of triage symptoms for EVD(+) and EVD(-) cohorts ranked according to the prevalence in EVD(+). Rankings from 1–16 are listed above each bar: black for EVD(+) and grey for EVD(-). (B) Differences in symptom prevalence between EVD(+) and EVD(-) cohorts. Positive values are more prevalent in EVD(+) cases. Negative values are more prevalent in EVD(-) cases. (C) Differences in symptom prevalence between EVD(+)only patients and malaria(+)only patients. Positive values are more prevalent in EVD(+)only cases. Negative values are more prevalent in malaria(+)only cases. EVD(+)only: EVD(+)/malaria(-); Malaria(+)only: EVD(-)/malaria(+)
Fig 4
Fig 4. Impact of EVD on referral sensitivity.
(A) Mean referral time (days since symptom onset at triage) for EVD(+) and EVD(-) cohorts. (B) Fitted relationship between referral time and outcome using fractional polynomial analysis [25]. (C) Mean referral time for EVD(+) and EVD(-) patients according to age categorisation. (D) Mean referral time for EVD(-) and EVD(+) patients over the entire time course of the study (December 2014 to October 2015). *: p<0.05, **: p<0.005, ***: p<0.001, ns: not significant, ETC: Ebola Treatment Centre.
Fig 5
Fig 5. Derivation of a malaria-sensitive triage scoring system for EVD.
The sensitivity and specificity of predicting EVD(+) patients in our cohort using the triage scoring system developed from the multivariate analysis of groups comparing (A) EVD(+) vs. EVD(-) and (C) EVD(+)/malaria(-) vs. EVD(-)/malaria(+). The area under the receiver-operator characteristic (ROC) curve represents the discriminative power of each score. (B) Sensitivity (green) and specificity (blue) according to the 22 score points. Prevalence of EVD(+) and EVD(-) patients are displayed as bar graphs and risk category cut-offs are shown as vertical lines. (D) Probability of being diagnosed as either EVD(+)only (red), Malaria(+)only (green) or double-negative (blue) according to the 22 points of the triage score. (E) Percentage of EVD(+) and EVD(-) patients classified in each risk category. (F) Percentage of the cohort captured in each risk category.
Fig 6
Fig 6. Scorecard to extrapolate the Ebola infection risk at triage.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Weyer J, Grobbelaar A and Blumberg L, Ebola virus disease: history, epidemiology and outbreaks. Curr Infect Dis Rep, 2015. 17(5): p. 480 10.1007/s11908-015-0480-y - DOI - PubMed
    1. Centers for Disease Control and Prevention, Ebola Virus Disease Distribution Map, 17 Feb 2016.
    1. Bolkan HA, Bash-Taqi DA, Samai M, Gerdin M and von Schreeb J, Ebola and indirect effects on health service function in Sierra Leone. PLoS Curr, 2014. 6. - PMC - PubMed
    1. Awah PK, Boock AU and Kum KA, Ebola Virus Diseases in Africa: a commentary on its history, local and global context. Pan Afr Med J, 2015. 22 Suppl 1: p. 18. - PMC - PubMed
    1. WHO Ebola Response Team, After Ebola in West Africa—Unpredictable Risks, Preventable Epidemics. N Engl J Med, 2016. 375(6): p. 587–96. 10.1056/NEJMsr1513109 - DOI - PubMed

Publication types

Grants and funding

This study was funded by the Department for International Development (DfID) (https://www.gov.uk/government/organisations/department-for-international-development, Grant number: 04890) via the humanitarian aid organisation, GOAL Global (https://www.goalglobal.org/). The funder had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to the data in the study and had final responsibility for the decision to submit for publication.
-