Is dengue and malaria co-infection more severe than single infections? A retrospective matched-pair study in French Guiana.

Abstract

Dengue and malaria are two major arthropod-borne infections in tropical areas, but dual infections were only described for the first time in 2005. Reports of these concomitant infections are scarce and there is no evidence of more severe clinical and biological pictures than single infections.
To compare co-infections to dengue alone and malaria alone, a retrospective matched-pair study was conducted between 2004 and 2010 among patients admitted in the emergency department of Cayenne hospital, French Guiana.
104 dengue and malaria co-infection cases were identified during the study period and 208 individuals were matched in two comparison groups: dengue alone and malaria alone. In bivariate analysis, co-infection clinical picture was more severe than separated infections, in particular using the severe malaria WHO criteria. In multivariate analysis, independent factors associated with co-infection versus dengue were: masculine gender, CRP level > 50 mg/L, thrombocytopaenia < 50 109/L, and low haematocrit <36% and independent factors significantly associated with co-infections versus malaria were red cells transfusion, low haematocrit < 36%, thrombocytopaenia < 50 109/L and low Plasmodium parasitic load < 0.001%.
In the present study, dengue and malaria co-infection clinical picture seems to be more severe than single infections in French Guiana, with a greater risk of deep thrombocytopaenia and anaemia.

Full text

Page 1

RESEARCHOpen Access
Is dengue and malaria co-infection more severe
than single infections? A retrospective
matched-pair study in French Guiana
Loïc Epelboin1,2,3*, Matthieu Hanf1,2, Philippe Dussart4, Sihem Ouar-Epelboin1, Félix Djossou2,5,
Mathieu Nacher1,2and Bernard Carme1,2
Abstract
Background: Dengue and malaria are two major arthropod-borne infections in tropical areas, but dual infections
were only described for the first time in 2005. Reports of these concomitant infections are scarce and there is no
evidence of more severe clinical and biological pictures than single infections.
Methods: To compare co-infections to dengue alone and malaria alone, a retrospective matched-pair study was
conducted between 2004 and 2010 among patients admitted in the emergency department of Cayenne hospital,
French Guiana.
Results: 104 dengue and malaria co-infection cases were identified during the study period and 208 individuals
were matched in two comparison groups: dengue alone and malaria alone. In bivariate analysis, co-infection clinical
picture was more severe than separated infections, in particular using the severe malaria WHO criteria. In
multivariate analysis, independent factors associated with co-infection versus dengue were: masculine gender, CRP
level>50 mg/L, thrombocytopaenia<50 109/L, and low haematocrit <36% and independent factors significantly
associated with co-infections versus malaria were red cells transfusion, low haematocrit<36%,
thrombocytopaenia<50 109/L and low Plasmodium parasitic load<0.001%.
Conclusions: In the present study, dengue and malaria co-infection clinical picture seems to be more severe than
single infections in French Guiana, with a greater risk of deep thrombocytopaenia and anaemia.
Keywords: Dengue, Malaria, French Guiana, Thrombocytopaenia, Case–control studies
Background
Dengue fever and malaria are the most common
arthropod-borne diseases in humans and represent
major public health problems. Dengue virus (family
Flaviridae, genus Flavivirus) and Plasmodium para-
sites are widespread in American and Asian tropical
regions and their endemic areas overlap extensively.
Nevertheless, reports of malaria and dengue dual in-
fection are scarce. Since the first case reported in
2005 [1], only case-reports and two descriptive studies
have been published. They have been reported with
Plasmodium falciparum and/or Plasmodium vivax in
India and Pakistan [2-5], Southeast Asia [6,7], French
Guiana [8] and Brazil [9]. This phenomenon seems to be
uncommon. In a study performed in Thailand among 194
patients with dengue, no co-infection with malaria was
found [10], but in French Guiana, a retrospective study
performed in 2004–2005 on 1,723 consecutive febrile emer-
gency patients found 17 co-infections, including six acute
concurrent infections (e.g. 1% of dengue and 4% of malaria
cases) [8]. The influence of co-infections on severity is not
straightforward, therefore, the aim of this study was to
differentiate clinical and biological picture of co-infections
from infections alone and determine whether patients
infected by both malaria and dengue (MD) were more
severe than either infection alone (respectively M and D).
* Correspondence: epelboincrh@hotmail.fr
1CIC-EC Antilles Guyane CIE 802 Inserm, Centre Hospitalier Andrée Rosemon,
Cayenne, French Guiana
2Research team EPaT EA 3593, University of French West Indies and French
Guiana, Cayenne, French Guiana
Full list of author information is available at the end of the article
© 2012 Epelboin et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Epelboin et al. Malaria Journal 2012, 11:142
http://www.malariajournal.com/content/11/1/142

Page 2

Methods
Study location
French Guiana is a French Overseas territory located on
the north-eastern coast of South America. About 90% of
its surface of 84,000 km2is Amazonian rain forest; the
remaining 10% in the north is a coastal plain where 90%
of the 215,000 inhabitants live and Cayenne and sur-
roundings contain almost 50% of the population in 2009
[11]. Malaria and dengue fever (DF) represent two major
public health concerns in French Guiana. Malaria is en-
demic and the annual number of cases ranges from
3,200 to 4,700 [12]. Until 2006, P. vivax represented 50%
of annual cases. The current proportion of P. vivax mal-
aria is 75%, as in the rest of the Americas [12-14]. Since
the first cases of DF were reported in French Guiana in
1943, an increase in the number of DF cases and DF out-
breaks and the emergence of dengue hemorrhagic fever
(DHF) have been observed [15]. All four dengue virus
serotypes circulate in French Guiana. The last two mains
epidemics occurred in 2006 and 2009, and dengue is cur-
rently endemic. Until 2005, dengue outbreaks were ex-
clusively described on the coast. Since 2006, outbreaks of
DF have been reported in interior villages where malaria
is endemic [16].
Study population
A matched retrospective study was conducted comparing
patients infected with concurrent malaria and dengue to
patients with either infection alone. The study popula-
tion included all patients admitted in the emergency de-
partment of Cayenne hospital, between June 2004 and
February 2010. The diagnosis of dengue and malaria co-
infection was made on the basis of concomitant bio-
logical diagnosis of dengue and malaria within seven
days in patients with a compatible clinical picture. Two
control groups were constituted: the group M with posi-
tive biological diagnosis for malaria and negative for den-
gue, according to the criteria defined in the next
paragraph, and the contrary for the group D. Control
cases were matched on the date of biological diagnosis
of infection.
Case definitions were based on compatible clinical his-
tory and biological diagnosis. Malaria diagnosis relied on
the identification of haematozoa on a thin blood film and/
or on a thick blood film stained with Giemsa (group MD
and M). The screening sensitivity was?6 plasmodia/μL
(1/1,000 leukocytes). The asexual parasite load (PL) was
classified in five classes: class 5: >1.25%; class 4: 0.125 to
1.25%; class 3 : 0.0125% to 0.125%; class 2: 0.00125 to
0.0125%; and class 1: ≤0.00125. Malaria rapid diagnosis
tests were not systematically performed on the study
period. Due to the evolution of the techniques between
2004 and 2010, the laboratory diagnosis of dengue relied
on different methods. Direct diagnosis was based on virus
isolation, genome detection by Reverse Transcriptase-
Polymerase Chain Reaction (RT-PCR) or NS1 antigen de-
tection introduced in 2006 in French Guiana. Indirect
diagnosis was based on detection of specific anti-dengue
IgM and/or IgA antibodies in patients’ sera [17]. When
NS1 antigen detection was available, RT-PCR, which
allows serotype identification, was not systematically
performed.
Concerning dengue definition (groups MD and D),
cases were separated in two groups: “confirmed acute
dengue cases” (CADC) were defined by direct biological
diagnosis (NS1 antigen and/or RT-PCR and/or virus iso-
lation), IgM seroconversion (early serum sample negative
for IgM but convalescent sample positive) or IgA anti-
bodies detection. “Likely dengue cases” (LDC) were
defined by IgM antibodies detection. Indeed, IgM appear
between the 3rdand 5thday of fever but can persist for
over three months and IgA appear concomitantly with
IgM but does not persist longer than five to six weeks
[18,19]. There was no discrimination between dengue
primary infection and secondary infection, e.g., further
infection(s) by dengue of a different serotype.
Covariates included, data collection and statistical analysis
Patients’ data, including socio-epidemiologic data, previ-
ous medical history, clinical symptoms, and biological
results, were obtained from the computerized medical
charts. Data were analysed using R version 2.10.0Wand
the EpicalcWpackage.
The continuous variables of interest were categorized
following the laboratory cut-off values, or published
values. They generally were dichotomized because of the
small sample size.
Two analyses were performed comparing separately
MD to D and MD to M. For categorical variables, a
matched bivariate analysis using the Wald test was per-
formed to identify factors associated with co-infections.
Statistical significance was set at p <0.05. Variables with
a p-value<0.2 in bivariate analyses were entered into
multivariate model to identify the factors independently
associated with dengue-malaria co-infections. As bivari-
ate analysis were made in an exploratory way and used
as a selection criterion for inclusion in the final multi-
variate model (p<0.20), we did not judge as a necessity
to adjust bivariate p-values for multiple comparisons.
Thus, bivariate p-values near to 0.05 must be relativized.
To obtain more powerful models, and because of the
missing data inherent to retrospective studies, variables
obtained from anamnesis and clinical examination and
variables with more than 5% of missing data were
excluded from the model. Thus, conditional multivariate
backward stepwise logistic regression estimated the
adjusted odds ratio (OR) and the confidence intervals
linked to co-infections.
Epelboin et al. Malaria Journal 2012, 11:142
http://www.malariajournal.com/content/11/1/142
Page 2 of 8

Page 3

Ethical considerations
The retrospective use of anonymous patient files on the
site of patient care is authorized by the French National
Commission on Informatics and Liberties. All the data
collected retrospectively were anonymized in a standar-
dized case report form and in the database.
Results
Cases description
Between June 2004 and February 2010, 104 patients sat-
isfied the criteria for MD (Figure 1). Consequently, 208
patients were matched in each comparison group.
Among the 104 MD patients, 75 (72.1%) were men and
11 (10.6%) were 15 years-old or under. The mean age
was 33.8 years (range: 6 months to 83 years). Forty-one
(39.4%) were considered as CADC and 63 (60.6%) as
LDC versus 150 (72.1%) and 58 (27.9%; p <0.001) in the
group D, respectively (Table 1). The dengue virus sero-
type could be identified for only 10 (9.6%) of co-infected
patients (DENV-1: 3 (30%); DENV-2: 2 (20%) and
DENV-3: 5 (50%)) and 91 in the control group (DENV-1:
25 (27.5%); DENV-2: 25 (27.5%); DENV-3: 28 (30.8%) and
DENV-4: 13 (14.2%)). No significant difference was found
in the proportion of P. vivax between the MD group
and M group: P. vivax (76.7% vs. 68.1% respectively),
P. falciparum (20.4% vs. 28%) and association of P. vivax
and P. falciparum (2.9% vs. 3.9%). Species identification
was not possible in two patients because they had received
anti-malarial treatment after positive rapid diagnostic
testing in a health centre. A low PL (class 1 and 2) tended
to be more frequent in the MD group than M group
(p 0.08) (Table 2).
Comparison of co-infection with dengue
Clinical and biological pictures of co-infection cases were
different from single infections and bivariate compari-
sons showed more differences between MD and D than
between MD and M (Table 3). MD patients were more
often adult men. A quarter of them reported having re-
cently visited the forest (military, forest workers or gold
miners) versus 3% in the D group. Patients from MD
group resided more frequently far from the coast than D
patients and had a history of malaria and recent malaria
attacks (<3 months). The duration of fever was longer in
MD patients and they were hospitalized more frequently
than D patients, but hospitalization was not longer.
More patients required a transfusion in the MD group.
The clinical presentation in MD patients was generally
more severe, with more fever above 40°C, tachycardia,
initial hypotension, nausea, vomiting and dehydration
0
2
4
6
8
10
12
0
100
200
300
400
500
600
Monthly number of dengue and malaria co-infection cases

Monthly number of dengue and malaria cases

Dengue and malaria co-infection Dengue Malaria
Figure 1 Monthly cases of malaria, dengue fever (curves) and dengue and malaria co-infection (columns) in the Emergency
department of Cayenne Hospital, French Guiana, between June 2004 and February 2010. Source : Institut de Veille Sanitaire - CIRE
Antilles-Guyane, Reference National Laboratory of Arboviruses and Parasitology–Mycology Unit, Cayenne Hospital, French Guiana.
Epelboin et al. Malaria Journal 2012, 11:142
http://www.malariajournal.com/content/11/1/142
Page 3 of 8

Page 4

than D patients. Furthermore, cases which fulfilled one
or more of the WHO clinical and/or biological criteria
for severe falciparum malaria [20] were more frequent in
the MD group than in the D group (p<0.001). Increased
C - reactive protein (CRP), especially>50 mg/L, was sig-
nificantly associated with MD co-infections (p <0.001)
relative to dengue alone. Retro-orbital pain, skin rash
and ENT symptoms were significantly associated with D.
Anaemia (p 0.02), severe thrombocytopaenia (p <0.001),
and elevated bilirubin (p <0.001) were more frequent in
MD patients. CADC diagnoses were significantly more
frequent in the D group than in the MD group. Anaemia,
severe thrombocytopaenia, male gender, high CRP level
and LDC diagnosis were significantly associated with co-
infection in multivariate analysis (Table 4).
Comparison of co-infection with malaria
No significant difference between MD and M was found
in terms of gender, age, place of residence and forest-
related activities (Table 2). A history of malaria was more
frequent in the M group. The fever duration was longer
in MD patients but not hospitalization. They received
significantly more transfusions (p 0.02). Low blood pres-
sure, signs of shock, pallor were significantly associated
with the MD group. Anaemia and severe thrombocyto-
paenia were also significantly more frequent in the MD
group. Cases which fulfilled one or more of the WHO
criteria for severe falciparum malaria [20] were more fre-
quent in the MD group than in the M group (p 0.007).
Anaemia, severe thrombocytopaenia, low parasitaemia,
and a high number of blood transfusions were independ-
ently associated with co-infections in multivariate ana-
lysis (Table 4).
Discussion
The unexceptional nature of the association of dengue
and malaria is confirmed in French Guiana. In regions
Table 1 Laboratory diagnosis of dengue infection in the
co-infected group and in the single dengue infection
group
Dengue laboratory diagnosisCo-infection
n (%)
Dengue
n (%)
Cases
Cell culture virus isolation and/or
RT-PCR*
9 (8.7) 56 (26.9)CADC}
NS1 antigen +/− IgM* 11 (10.6) 84 (40.4)
IgM+IgA* 14 (13.5)9 (4.3)
IgM seroconversion 7 (6.7)1 (0.5)
IgM 63 (60.6)58 (27.9)LDC}
Total
104208
*RT-PCR Reverse transcriptase polymerase chain reaction, IgM Anti dengue
immunoglobulin M, IgA Anti dengue immunoglobulin A.
}CADC Confirmed acute dengue cases, LDC Likely dengue cases.
Table 2 Matched bivariate analysis between co-infected
patients and pure malaria infected patients
Results n (%)Co-infection
(n=104)
Malaria
(n=208)
p**
Males
75 (72.1) 162 (77.9) 0.26
Children (<15 years-old)
11 (10.6) 18 (8.7)0.58
Journey in forest
Inhabitant of the coast¥
26 (25)51 (24.6)0.96
86 (83.5)151 (74.4) 0.11
Long admission (≥2 days)
43 (41.3)72 (34.6) 0.25
Admission
37 (35.6)69 (33.2)0.67
Medical history of malaria
52 (50) 115 (62.8) 0.02}
Time since last malaria
attack<90 days
19 (55.9)49 (51.6) 0.91
Medical history of dengue
3 (3.4)16 (9)0.08}}
Duration of fever > 5 days
48 (48.5)61 (29.9) 0.001}
Red cells transfusion
8 (7.7)3 (1.4)0.01}
Tachycardia (>90 bpm)
79 (79) 165 (82.1) 0.64
Hypotension (<90 mmHg)
10 (10.4) 7 (3.6)0.01}
Fever≥40°C
15 (14.7) 24 (11.9)0.47
Retro-orbital pain
4 (3.9)10 (5) 0.61
Chills
36 (35.6)64 (32.2) 0.7
Hemorrhagic signs
12 (11.5)25 (12.4)0.81
Shock signs
7 (6.8)8 (4)0.16}}
Rash
1 (1) 5 (2.5) 0.4
Nausea and / or vomiting
58 (56.9) 93 (45.8)0.06}}
Neurological disorders
7 (6.8)5 (2.5)0.08}}
Dehydration
7 (7.4) 10 (5)0.37
Mucocutaneous pallor
11 (11.6) 7 (3.5)0.02}
Splenomegaly
9 (12.3)23 (18.3) 0.08}
Mucocutaneous jaundice
9 (8.7) 9 (4.3)0.14}}
ENT symptoms
6 (5.8) 10 (4.8)0.72
Severe “malaria” cases††34 (32,7) 39 (18,7)0,007
Low parasitaemia
(class<3)
20 (19.2) 24 (11.5)0.08}}
Mildly low hemoglobin
(<12 g/dl)
37 (35.6)43 (20.7) 0.05}}
Low hematocrit (<36%)
43 (41.3)49 (23.6)0.002}
Deep thrombocytopenia
(<50 G/L)
23 (22.1) 25 (12)0.02}
Low prothrombin
(<70%)
14 (21.2)22 (15.4)0.33
Hyponatremia (<130)
7 (7.1) 15 (7.4) 0.87
Hypokalemia (<3)
6 (6.1)15 (7.4)0.67
Elevated creatinin
(>111 μmol/L)
4 (4)14 (7) 0.29
Elevated bilirubin
(>50 μmol/L)
16 (16.8)20 (10.5) 0.06
SGPT (>2 N)
10 (10.2)33 (16.7) 0.16}
SGOT >2 N
13 (13.3) 23 (11.6)0.66
Epelboin et al. Malaria Journal 2012, 11:142
http://www.malariajournal.com/content/11/1/142
Page 4 of 8

Page 5

where these infections are transmitted in close proximity,
the classical concept that malaria occurs in rural areas
and dengue in urban areas may thus also be contradicted
by facts in many countries and simultaneous infections
may result from the overlap of the mosquito biotopes
[16].
This study presented some minor biases. There were a
higher number of cases based on the IgM detection in
the MD group than in the D group. Co-infected patients
with LDC were tested separately, so the association be-
tween MD and thrombocytopaenia, anaemia and fre-
quent transfusions persisted in bivariate analysis, but not
in multivariate analysis. However, when testing separately
all patients with CADC diagnosis, no significant differ-
ence was found in bivariate and multivariate analysis
which is probably due to the loss of power. The separ-
ation in two groups, LDC and CADC is arbitrary since
the decision to perform the direct diagnosis relied on
non verifiable information provided by patients on fever
duration, a relatively unreliable answer given the fre-
quent linguistic difficulties in FG. Studying confirmed
acute cases alone was questionable, and the authors
decided to study together likely and confirmed acute
cases because it allowed a larger sample. Furthermore,
associating likely cases would have minimized differences
between the group MD and malaria alone because “real”
associations would have been mixed with isolated mal-
aria cases. It appears that co-infected patients consulted
significantly later than the other groups, which may
explains the predominance of IgM-diagnosed cases.
However, almost all previous studies on dengue and mal-
aria co-infections relied on IgM diagnosis [1,2,4-8,21].
Another hypothesis to explain the relatively high number
of LDC in the study group is that malaria attack could
have been triggered by dengue infection, especially as
there is a majority of P. vivax infection, possibly relapses,
which are coherent with the high frequency of malaria
medical history, especially in the last three months in the
Table 2 Matched bivariate analysis between co-infected
patients and pure malaria infected patients (Continued)
CRP <5
2 (2)1 (0.5) 0.22
CRP 5 à 50
29 (29)78 (38)-
CRP>50
69 (69)126 (61.5) -
* p calculated with Wald test in matched bivariate analysis.
}Significant p-values<0.05.}}
p-values<0.2 included in the matched multivariate model.
bpm Beats per minute.
¥Inhabitants of the coast are defined as people living in Cayenne, Rémire-
Montjoly, Matoury Macouria, Kourou, Irakoubo or Mana.; Other people
declared to live in Cacao, Roura, Montsinnéry-Tonnégrande, Régina, Saül,
St-Elie, Saint Laurent du Maroni, Apatou, Maripasoula, Papaïchton, St George
de l’Oyapock, Camopi or Trois-Sauts. 4 people came from France, 1 from the
French Caribbean and 1 from French Polynesia.
††WHO 2000 criteria for severe P. falciparum malaria
ENT Ear, Nose and Throat, SGPT Serum Glutamopyruvate Transferase, SGOT
Serum Glutamooxaloacetate Transférase, CRP C-reactive protein.
Table 3 Matched bivariate analysis between co-infected
patients and pure dengue infected patients
Results n (%)Co-infection
(n=104)N (%)
Males
Dengue
(n=208)N (%)
p*
75 (72.1) 119 (57.2)0.01}
Children (<15 years-old)
11 (10.6)49 (23.6) 0.008}
Journey in forest
Living on the coast¥
26 (25)7 (3.4)
<0.001}
86 (83.5) 183 (92.9)0.025}
Long admission (≥2 days)
43 (41.3)71 (34.1) 0.23
Admission
37 (35.6)45 (21.6) 0.01}
Medical history of malaria
52 (50)21 (10.2)
<0.001}
Time since last malaria
attack<90 days
15 (44.1) 0 (0)-
Medical history of dengue
3 (3.4) 5 (2.7)0.64
Duration of fever > 5 days
48 (48.5) 33 (16.3)
<0.001}
Red cells transfusion
8 (7.7) 2 (1)0.03}
Tachycardia (>90 bpm)
79 (79) 128 (64.6) 0.03}
Hypotension (<90 mmHg)
10 (10.4) 3 (1.8)0.02}
Fever≥40°C
15 (14.7)10 (4.9)0.006}
Retro-orbital pain
4 (3.9)26 (12.6) 0.01}
Chills
Hemorrhagic signs†
Shock signs{
36 (35.6)37 (18)0.002}
12 (11.5) 32 (15.5) 0.31
7 (6.8) 5 (2.4)0.06}}
Rash
1 (1) 30 (14.6) 0.007}
Neurological disorders
7 (6.8)8 (3.9) 0.28
Nausea and / or vomiting
58 (56.9)94 (45.6)0.07}}
Dehydration
7 (7.4) 4 (1.9)0.04}
Mucocutaneous pallor
11 (11.6) 3 (1.5) 0.003}
Splenomegaly
9 (12.3)3 (3.1) 0.04}
Jaundice
9 (8.7) 6 (2.9)0.03}
ENT symptoms
Severe “malaria” cases††
6 (5.8) 42 (20.2)0.002}
34 (32,7)33 (15,9)
<0,001
Confirmed acute
dengue cases
41 (39.4) 150 (72.1)
<0.001}
Mildly low hemoglobin
(<12 g/dl)
37 (35.6) 40 (19.2) 0.002}
Low hematocrit (<36%)
43 (41.3)34 (16.4)
<0.001}
Deep thrombocytopenia
(<50 G/L)
23 (22.1)6 (2.9)
<0.001}
Low prothrombin (<70%)
14 (21.2)19 (13.9) 0.18}}
Hyponatremia (<130)
7 (7.1) 5 (2.5)0.08}}
Hypokalemia (<3)
6 (6.1)3 (1.5) 0.04}
Elevated creatinin
(>111 μmol/L)
4 (4)4 (2) 0.3
Elevated bilirubin
(>50 μmol/L)
16 (16.8) 3 (1.6)
<0.001}
SGPT (>2 N)
10 (10.2) 32 (16)0.2
SGOT >2 N
13 (13.3)41 (20.5) 0.16}}
CRP <5
2 (2)73 (35.3)
<0.001}
Epelboin et al. Malaria Journal 2012, 11:142
http://www.malariajournal.com/content/11/1/142
Page 5 of 8