NOTICE: THIS MATERIAL MAY BE PROTECTED BY COPYRIGHT LAW (TITLE 17 U.S.

CODE)
Call #:

Borrower: EOS

Location:

MSU ILLiad TN: 378011

*378011*

E-Mail

Lending String:
*MZF,GKJ,IDU,UIW,KHN,ICS,UUO,TWT,CSU,MTG,WOS
,VBC,PBU,CCT,CC#

MSU Copy Charge: $0.00

Your Max Cost: 0.00

Journal Title: Medecine et maladies infectieuses

Volume: 44 Issue: 9
Month/Year: 201491Pages: 412-416

EASTERN OREGON UNIVERSITY

PIERCE LIBRARY-ILL
ONE UNIVERSITY BLVD
LA GRANDE OR 97850-2899

Article Author: Lefebvre,A.

Article Title: Case fatality rates of Ebola virus diseases:
A meta-analysis of World Health Organization data.
ILL Number: 138015408

Odyssey:
E-mail: illoans@eou.edu

*138015408*

Resource Sharing/Doc Delivery

PO Box 173320
140 Renne Library
Bozeman MT 59717-3320

PHONE (406) 994-3161

EMAIL bozemanill@montana.edu

Within 7 days of receipt, please use this form to contact us for any problems with the item(s) received.
OCLC# _____________________________________ OTHER ILL# __________________________________________

Missing pages, page numbers _______________________________________________________________________

Margins cut off, page numbers ______________________________________________________________________

Unreadable, page numbers _________________________________________________________________________

Other ___________________________________________________________________________________________

Please return this page to Montana State University via the fax number listed above, call us (phone number above), or e-mail (listed above).
We will take care of the problem as soon as possible. We apologize for any inconvenience.
Montana State University

Disponible en ligne sur

ScienceDirect
www.sciencedirect.com
Mdecine et maladies infectieuses 44 (2014) 412416

Original article

Case fatality rates of Ebola virus diseases: A meta-analysis of World Health

Organization data
Taux de ltalit due aux maladies virus bola : mta-analyse des donnes de lOrganisation
mondiale de la sant
A. Lefebvre , C. Fiet , C. Belpois-Duchamp , M. Tiv , K. Astruc , L.S. Aho Gll
Service dpidmiologie et dhygine hospitalires, hpital denfants, CHU, 14, rue Paul-Gaffarel, BP 77908, 21079 Dijon cedex, France
Received 15 April 2014; received in revised form 20 June 2014; accepted 6 August 2014
Available online 2 September 2014

Abstract
Objective. Our objective was to estimate the case fatality rates of Zaire, Sudan, and Bundibugyo Ebola species, responsible for sometimes-lethal
hemorrhagic fevers.
Methods. We performed a meta-analysis of World Health Organization data on outbreaks of infections due to theses species.
Results. Twenty outbreaks, including the current one, were studied. The estimated case fatality rate was 65.4% (CI 95% [54.6%; 75.5%]) and
varied among the outbreaks. A species effect was identified, with a higher case fatality rate for the Zaire species than for Sudan and Bundibugyo
species. The case fatality rate of the Zaire species tended to decrease with time.
Conclusion. The case fatality rates associated with these 3 species was high. A great variability was observed. It could be explained partly by
a species effect and by the decrease of Zaire species case fatality rate, with time.
2014 Elsevier Masson SAS. All rights reserved.
Keywords: Ebola; Case fatality rate; Meta-analysis

Rsum
Objectif. Estimer le taux de ltalit due aux espces bola Zare, Soudan, et Bundibugyo lorigine de fivre hmorragiques parfois fatales.
Mthodes. Mta-analyses des donnes de lOMS relatives aux diffrents pisodes lis ces virus.
Rsultats. Vingt pisodes, dont 1 actuel, ont t relevs. Leur mta-analyse a dmontr une ltalit leve de 65,4 % (IC 95 % [54,6 % ; 75,5 %]).
Ce taux est trs variable selon les pisodes. Un effet de lespce est retrouv, avec un taux de ltalit plus important pour lespce Zare que pour
les espces Soudan et Bundibugyo. Une tendance la diminution au cours du temps est mise en vidence pour lespce Zare.
Conclusion. Le taux de ltalit associ ces 3 espces est lev. Une variabilit importante est observe. Elle est explique en partie par leffet
de lespce et la diminution de la ltalit au cours du temps pour lespce Zare.
2014 Elsevier Masson SAS. Tous droits rservs.
Mots cls : bola ; Taux de ltalit ; Mta-analyse

1. Introduction

Corresponding author.
E-mail address: Ludwig.aho@chu-dijon.fr (L.S. Aho Gll).

http://dx.doi.org/10.1016/j.medmal.2014.08.005
0399-077X/ 2014 Elsevier Masson SAS. All rights reserved.

The Ebola virus disease is a hemorrhagic fever due to a virus

of the same name, of the loviridae family, and of the Filovirus
genus [1]. There are 5 species of the virus: Bundibugyo, Zaire,
Sudan, Reston, and Ta Forest (previously called Ebola Cote
dIvoire) [2,3]. The 3 first species cause epidemics in Africa and

A. Lefebvre et al. / Mdecine et maladies infectieuses 44 (2014) 412416

413

Table 1
Number of cases and deaths, case fatality rates, and species among outbreaks of hemorrhagic fevers due to Ebola virus in Africa between 1976 and 2014 (WHO data
[3,8]).
Nombre de cas, de dcs, taux de ltalit et espce en cause dans les pisodes de vres hmorragiques virus bola en Afrique entre 1976 et 2014 (donnes OMS
[3,8]).
Country

Year

Species of virus

Number of cases

Number of death

Case fatality rate (%)

Sudan
Democratic Republic of Congo
Democratic Republic of Congo
Sudan
Gabon
Democratic Republic of Congo
South Africa
Uganda
Congo
Gabon
Sudan
Congo
Uganda
Democratic Republic of Congo
Democratic Republic of Congo
Uganda
Democratic Republic of Congo
Uganda
Uganda
Liberia-Guinea-Sierra Leone (ongoing)

1976
1976
1977
1979
1994
1995
1996
2000
2001
2001
2004
2005
2007
2007
2008
2011
2012
2012
2012
2014

Sudan
Zaire
Zaire
Sudan
Zaire
Zaire
Zaire
Sudan
Zaire
Zaire
Sudan
Zaire
Bundibugyo
Zaire
Zaire
Sudan
Bundibugyo
Sudan
Sudan
Zaire

284
318
1
34
52
315
1
425
59
65
17
12
149
264
32
1
57
7
24
453

151
280
1
22
31
254
1
224
44
53
7
10
37
187
14
1
29
4
17
245

53
88
100
65
60
81
100
53
75
82
41
83
25
71
44
100
51
57
71
54

severe hemorrhagic diseases, with high case fatality rates. No

human death due to the 2 other species has ever been reported.
The Reston species causes death in primates, but is less virulent
than the Zaire and Sudan species [4].
The World Health Organization (WHO) has reported several outbreaks of Ebola virus infections since its discovery in
1976 [3,5]. An outbreak is ongoing in Guinea, Liberia, and in
Sierra Leone [5,6]. The case definition, often difficult during
these epidemics, makes it difficult to estimate the case fatality
rate [7,8]. We had for objective to estimate the case fatality rate
due to Ebola virus Bundibugyo, Zaire, and Sudan species, with a
meta-analysis of the WHO data on the various epidemics related
to the virus.
2. Methods
A meta-analysis of case fatality WHO data was performed
using a Freeman-Tukey (double arcsine) transformation [9] so
as to stabilize the variance of proportions. The number of deaths
was compared to the number of suspected or confirmed cases, to
determiner the case fatality rate for every outbreak [10]. A random effect model was used according to DerSimonian-Lairds
method in case of heterogeneousness [11]. Sensitivity analyses were performed by removing epidemics with a low number
of cases (< 10) and the ongoing epidemic. A cumulative metaanalysis was used to assess the stabilization of the case fatality
rate with time. Sub-group analyses were performed by species.
Finally, the species effect and the time effect were investigated by using univariate and multivariate linear meta-regression
models for changed proportions. A linear trend was investigated globally for the time effect then for every viral species
after having tested the hypothesis of linearity by using fractional

polynomials. The R [12] (meta [13] and metafor packages)

and Stata [14] (fracpoly function) software were used for the
statistical analyses.
3. Results
Nineteen outbreaks were identified between 1976 and 2012
(WHO data [3]), with at least 1 case, and at most 425 cases.
Among them, 2118 cases and 1367 deaths were recorded.
Another recent outbreak concerned, as of June 10, 2014, 453
cases and 245 deaths [15]. The data for these outbreaks is presented in Table 1.
The meta-analysis of the various outbreaks revealed a high
case fatality rate of 65.4% (CI 95% [54.6%; 75.5%]) (Fig. 1).
The heterogeneousness was very important (I2 = 94%). The subgroup analyses revealed a greater mortality for the Ebola Zaire
species (76%, CI 95% [63%; 87%]) than for the Sudan species
(55% CI 95% [50%; 59%]) and Bundibugyo species (37%, CI
95% [14%; 63%]).
The heterogeneousness was low for the Sudan species; it was
high for the Zaire and Bundibugyo species. The estimations were
made with 2 outbreaks only for the Bundibugyo species. The
sensitivity analyses, made by removing epidemics with a low
number of cases (< 10) and ongoing epidemics yielded similar
results.
Using fractional polynomials for the regressions according
to the year did not allow ruling out the hypothesis of linearity.
The multivariate meta-regression model did not allow demonstrating any significant effect of the year of epidemic onset on
the case fatality rate, even if the case fatality rate decreased significantly (P = 0.005) with time in univariate regression model
(Table 2). But it confirmed the species effect. The Zaire species

414

A. Lefebvre et al. / Mdecine et maladies infectieuses 44 (2014) 412416

Outbreak

Death

Species = Sudan
Sudan 1976
Sudan 1979
Uganda 2000
Sudan 2004
Uganda 2011
Uganda 2012
Uganda 2012
Fixed effect model
Random effect model

151
22
224
7
1
4
17

Cases

Proportion

284
34
425
17
1
7
24
792

CI 95% W(fixed) W(random)

0.53 [0.47; 0.59]

0.65 [0.46; 0.80]
0.53 [0.48; 0.58]
0.41 [0.18; 0.67]
1.00 [0.03; 1.00]
0.57 [0.18; 0.90]
0.71 [0.49; 0.87]
0.54 [0.50; 0.58]
0.55 [0.50; 0.59]

11.0%
1.3%
16.5%
0.7%
0.1%
0.3%
0.9%
30.8%
--

6.4%
5.5%
6.5%
4.7%
1.1%
3.4%
5.1%
-32.6%

0.88 [0.84; 0.91]

1.00 [0.03; 1.00]
0.60 [0.45; 0.73]
0.81 [0.76; 0.85]
1.00 [0.03; 1.00]
0.75 [0.62; 0.85]
0.82 [0.70; 0.90]
0.83 [0.52; 0.98]
0.71 [0.65; 0.76]
0.44 [0.26; 0.62]
0.54 [0.49; 0.59]
0.76 [0.73; 0.79]
0.76 [0.63; 0.87]

12.3%
0.1%
2.0%
12.2%
0.1%
2.3%
2.5%
0.5%
10.3%
1.3%
17.6%
61.1%
--

6.4%
1.1%
5.8%
6.4%
1.1%
5.9%
5.9%
4.2%
6.4%
5.4%
6.5%
-55.2%

0.25 [0.18; 0.33]

0.51 [0.37; 0.64]
0.32 [0.25; 0.38]
0.37 [0.14; 0.63]

5.8%
2.2%
8.0%
--

6.3%
5.9%
-12.1%

0.66 [0.63; 0.68]

0.65 [0.55; 0.75]

100%
--

-100%

Heterogeneousness: I-squared=6.4%, squared rate =0.0009, p=0.379

Species = Zaire
Democratic Republic of Congo 1976
Democratic Republic of Congo 1977
Gabon 1994
Democratic Republic of Congo 1995
South Africa 1996
Congo 2001
Gabon 2001
Congo 2005
Democratic Republic of Congo 2007
Democratic Republic of Congo 2008
Guinea-Liberia-Sierra Leone 2014
Fixed effect model
Random effect model

280
1
31
254
1
44
53
10
187
14
245

318
1
52
315
1
59
65
12
264
32
453
1572

Heterogeneousness I-squared =93.2%, squared rate =0.1084, p<0.0001

Species = Bundibugyo
Uganda 2007
Democratic Republic of Congo 2012
Fixed effect model
Random effect model

37
29

149
57
206

Heterogeneousness: I-squared =91.8%, squared rate =0.1342, p=0.0005

Fixed effect model

Random effect model

2,570

Heterogeneousness: I-squared =94.4%, squared rate =0.1397, p<0.0001

0,2

0,4

0,6

0,8

Fig. 1. Meta-analysis of case fatality rate due to Ebola (Sudan, Zaire and Bundibugyo) as of June 10, 2014: meta-analysis (global and by species).
Mta-analyse du taux ltalit due aux virus bola (espces Soudan, Zare, Bundibugyo) au 10 juin 2014 : mta-analyse (globale et par espce).
Table 2
Meta-regressions according to species and year.
Mta-rgressions selon lespce et lanne.
Variable

Coefficient

Freemann and Tukay transformation of proportion

2.4

CI 95%

Univariate analysis
Species
Sudan (reference) (n = 7)
Zaire (n = 11)
Bundibugyo (n = 2)
Year

0.306
0.419
0.010

[0.028; 0.641]
[0.915; 0.089]
[0.025; 0.005]

0.073
0.107
0.005

Multivariate analysis
Species
Sudan (reference) (n = 7)
Zaire (n = 11)
Bundibugyo (n = 2)
Year

0.360
0.284
0.009

[0.082; 0.637]
[0.717; 0.150]
[0.018; 0.002]

0.011
0.200
0.109

2.2
2.0

1.8
1.6

1980

seemed to have a higher case fatality rate than the Sudan species
(coefficient: 0.360; P = 0.011 in multivariate analysis), contrary
to the Bundibugyo species for which it was lower but nonsignificantly. The case fatality rate tended to decrease with time
for the Zaire species (coefficient: 0.02; P < 0.001) (Fig. 2),
but no trend was observed for the Sudan species (coefficient:
0.001; P = 0.235). Residuals of meta-regressions did not reveal
any trend (for example curvilinear) and seemed randomly distributed. The cumulative meta-analysis revealed a stabilization
of the estimated case fatality rate from 2008 on (Fig. 3).

1990

2000

2010

Year

Fig. 2. Meta-regression according to year of case fatality rate during outbreaks

of Zaire species Ebola virus hemorrhagic fever, between 1976 and 2014.
Mta-rgression selon lanne du taux de ltalit dans les pisodes de vres
hmorragiques virus bola, espce Zare, entre 1976 et 2014.

4. Discussion
Twenty outbreaks of Ebola virus hemorrhagic fever were
reported in Africa, one of which is ongoing in West Africa. Our
meta-analysis proved a high case fatality rate (65%, CI 95%
[55%; 75%]) for the 3 studied species.
Our meta-analysis had some limitations. The important variability of case fatality rates (from 25% to 88% not taking into
account single cases) led to a great heterogeneousness among

A. Lefebvre et al. / Mdecine et maladies infectieuses 44 (2014) 412416

415

Proportion

Outbreaks

CI 95%

Added Sudan 1976 (k=1)

Added Democratic Republic of Congo 1976 (k=2)
Added Democratic Republic of Congo 1977 (k=3)
Added Sudan 1979 (k=4)
Added Gabon 1994 (k=5)
Added Democratic Republic of Congo 1995 (k=6)
Added South Africa 1996 (k=7)
Added Uganda 2000 (k=8)
Added Congo 2001 (k=9)
Added Gabon 2001 (k=10)
Added Sudan 2004 (k=11)
Added Congo 2005 (k=12)
Added Uganda 2007 (k=13)
Added Democratic Republic of Congo 2007 (k=14)
Added Democratic Republic of Congo 2008 (k=15)
Added Uganda 2011 (k=16)
Added Democratic Republic of Congo 2012 (k=17)
Added Uganda 2012 (k=18)
Added Uganda 2012 (k=19)
Added Guinea-Liberia-Sierra Leone 2014 (k=20)

0.53
0.72
0.81
0.76
0.72
0.74
0.77
0.73
0.73
0.75
0.72
0.73
0.68
0.68
0.66
0.68
0.66
0.66
0.66
0.65

Random effect model

0.65 [0.55; 0.75]

-1

-0.5

0.5

[0.47; 0.59]
[0.34; 0.98]
[0.36; 1.00]
[0.42; 0.99]
[0.45; 0.94]
[0.55; 0.90]
[0.57; 0.93]
[0.54; 0.89]
[0.57; 0.88]
[0.60; 0.87]
[0.58; 0.84]
[0.59; 0.85]
[0.53; 0.82]
[0.55; 0.81]
[0.54; 0.79]
[0.54; 0.80]
[0.54; 0.78]
[0.54; 0.78]
[0.54; 0.77]
[0.55; 0.75]

Fig. 3. Cumulative meta-analyses of case fatality rates of Ebola virus hemorrhagic fever (Zaire, Sudan, and Bundibugyo species) between 1976 and 2014.
Mta-analyse cumulative des proportions de dcs rapports aux cas dans les pisodes de vres hmorragiques virus bola (Zare, Soudan et Bundibugyo) entre
1976 et 2014.

studies (I2 = 94%), taken into account with the random-effect

model. This resulted in a greater estimated CI for case fatality
rates. The heterogeneousness may be due to the viral species,
with a greater case fatality rate for the Zaire species than for
the Sudan species. The Bundibugyo species, discovered more
recently, seemed to be associated to lower case fatality rates,
but only 2 epidemics, including 206 cases, were reported. The
heterogeneousness persisted for Ebola Bundibugyo species and
Zaire species, for which a decrease of case fatality rate was
observed in time. This could be due to a lower virulence of the
species with time, improved medical management, a better nutritional state of patients, improved reporting of probable cases, or
immunization of exposed populations that could increase with
time. The observed heterogeneousness could be partly due to this
evolution in time, but also related to the biological variability of
strains, more or less virulent, depending on the outbreak settings
(country for example) and to probably different case definitions
depending on the epidemic. Thus, this heterogeneousness makes
it difficult to predict case fatality rates related to a given strain
for an early epidemic.
Our meta-analysis relied on available WHO data [3,15]. This
data concerns clinically probable cases, but all the cases were
not confirmed biologically. Furthermore, we did not evaluate the
case definition for each outbreak.
It would have been interesting to take into account the age
and gender effect on the analyses. Indeed, MacNeil et al. [7]
had reported an effect of these 2 factors on the risk of death.
Nevertheless, this was not available in WHO data. Some study
results allowed obtaining aggregated demographic data [7,16],
other studies did not allow obtaining this data systematically
[8,17].

Finally, the exhaustiveness of compilations of infectious outbreaks as well as of number of probable cases in an outbreak
was not guaranteed. Some small epidemics with low virulence
may have been overlooked; but several outbreaks with a single case were included. Some probable cases, less symptomatic,
could also have been overlooked during epidemics in which,
lethal cases were included more exhaustively. Only half of the
patients in reported cases presented with exteriorized hemorrhage in the recent epidemic [18]. In both of these settings, the
meta-analysis probably overestimated the case fatality rate. The
difficulty of access to care could also have decreased the estimated case fatality rate if some patients with probable infection
were not able to consult. Finally, the authors of seroprevalence
studies conducted in Gabon, for example, reported that some
individuals not exposed to epidemics were immunized against
the virus [19]. These individuals had probably been exposed
to the virus in the environment. The viral load to which these
individuals were exposed, lower than when in contact with
an individual presenting with exteriorized hemorrhage, may
have resulted in asymptomatic or paucisymptomatic presentations. Thus, during epidemics, some individuals in contacts with
infected patients may have presented with more or less severe
symptoms.
5. Conclusion
The case fatality rate due to the Ebola virus was high (65%, CI
95% [55%; 75%]) for the Zaire, Sudan, and Bundibugyo species,
but very variable depending on the outbreaks reported by the
WHO. Variability was observed among species, with greater
case fatality rate for the Zaire species than for the Sudan and

416

A. Lefebvre et al. / Mdecine et maladies infectieuses 44 (2014) 412416

Bundibugyo species. The Zaire species case fatality rate tended

to decrease with time.
Disclosure of interest
The authors declare that they have no conflicts of interest
concerning this article.
References
[1] Zeller H, Georges-Courbot MC. Les fivres hmorragiques virales. Antibiotiques 2009;8(4):21520.
[2] Li YH, Chen SP. Evolutionary history of Ebola virus. Epidemiol Infect
2014;142(6):113845.
[3] Organisation mondiale de la sant. Maladie virus bola. Genve Suisse:
OMS; 2014 http://www.who.int/mediacentre/factsheets/fs103/fr/
[4] Morikawa S, Saijo M, Kurane I. Current knowledge on lower virulence of Reston Ebola virus [Connaissances actuelles sur la moindre
virulence du virus bola Reston]. Comp Immunol Microbiol Infect Dis
2007;30(56):3918.
[5] Organisation mondiale de la sant. Afrique de lOuest Bulletin
dinformation sur les flambes en cours mise jour 10 juin 2014. Maladie virus bola; 2014 http://www.who.int/csr/don/2014 04 07 ebola/fr/
[6] Baize S, Pannetier D, Oestereich L, Rieger T, Koivogui L, Magassouba
NF, et al. Emergence of Zaire Ebola Virus Disease in Guinea Preliminary
report. N Engl J Med 2014, http://dx.doi.org/10.1056/NEJMoa1404505
[PMID: 24738640].
[7] MacNeil A, Farnon EC, Wamala J, Okware S, Cannon DL, Reed Z, et al.
Proportion of deaths and clinical features in Bundibugyo Ebola virus infection, Uganda. Emerg Infect Dis 2011;16(12):196972.

[8] MacNeil A, Farnon EC, Morgan OW, Gould P, Boehmer TK, Blaney DD,
et al. Filovirus outbreak detection and surveillance: lessons from Bundibugyo. J Infect Dis 2011;204(3):S7617.
[9] Freeman MF, Tukey JW. Transformations related to the angular and the
square root. Ann Math Stat 1950;21:60711.
[10] Elandt-Johnson RC. Definition of rates: some remarks on their use and
misuse. Am J Epidemiol 1975;102(4):26771.
[11] DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clin
Trials 1986;7(3):17788. Epub 1986/09/01.
[12] R Core Team [Internet] R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2014
http://www.R-project.org
[13] Schwarze. Meta: meta-analysis with R Version 3.6-0; 2014.
[14] StataCorp LP. Stata/SE 12.0. Stata Corp. College Station, TX, USA; 2011.
[15] Organisation mondiale de la sant. Afrique de lOuest Bulletin
dinformation sur les flambes en cours mise jour 7 avril 2014. Maladie
virus bola; 2014 http://www.who.int/csr/don/2014 04 07 ebola/fr/
[16] Roddy P, Howard N, Van Kerkhove MD, Lutwama J, Wamala J, Yoti Z, et al.
Clinical manifestations and case management of Ebola haemorrhagic fever
caused by a newly identified virus strain, Bundibugyo, Uganda, 20072008.
PLoS One 2012;7(12):e52986.
[17] Grard G, Biek R, Tamfum J-JM, Fair J, Wolfe N, Formenty P, et al.
Emergence of divergent Zaire ebola virus strains in Democratic Republic of the Congo in 2007 and 2008. J Infect Dis 2011;204(Suppl. 3):
S77684.
[18] Feldmann H. Ebola - A growing threat? N Engl J Med 2014,
http://dx.doi.org/10.1056/NEJMp1405314 [PMID: 24805988].
[19] Nkoghe D, Padilla C, Becquart P, Wauquier N, Moussavou G, Aku JP,
et al. Risk factors for Zaire ebolavirus Specific IgG in rural Gabonese
populations. J Infect Dis 2011;204(3):S76875.