Background
Uncontrolled urbanization, human mobility and changes in ecosystems have led to a 30-fold increase of dengue incidence in the last 50 years with geographic expansion to new countries [
1]. The America, South-East Asia and western Pacific regions are the most affected [
2]. Although dengue is known to circulate in Africa since the nineteenth century, and despite that up to 30 African countries have identified cases, the clinical impact and epidemiology of dengue in this part of the world remains poorly characterized [
3]. Current estimates suggest that sub-Saharan Africa carries 16% of the annual worldwide burden, but dengue is often not recognized, and hence under-reported because of its non-specific clinical presentation leading to presumptive diagnosis of malaria. Lack of awareness among clinicians, limited diagnostic capacities and weak surveillance systems may also contribute to underestimation of dengue. A limited number of dengue-infected patients will progress to severe disease which has more specific characteristics and is easier to identify. Although the incidence of dengue across Africa is high, severe dengue has been reported infrequently [
3]. To our knowledge, no study has been conducted in Africa evaluating the association between race and dengue presentation and outcome.
Since 2010, dengue outbreaks have been identified in Tanzania [
4‐
6]. We describe the clinical features and outcome of dengue during the first documented outbreak in Dar es Salaam that occurred in 2013–14 in both native and expatriate populations of different races.
Methods
Study design and setting
Patients were recruited between December 2013 and July 2014 in four public clinics and one private clinic in Kinondoni, the most populated (1.8 million inhabitants) District of Dar es Salaam, the largest city and economic center of Tanzania.
Study participants
Patients recruited in public and private clinics
A prospective cohort study to document the etiologies of fever was performed between December 2013 and July 2014 at the emergency departments of Mwananyamala Regional Hospital and connected health care facilities (Sinza Hospital, Magomeni Health Centre, Tandale Health Centre). All consecutive adult (age ≥ 18 years) patients presenting with fever (tympanic temperature ≥ 38.0 °C) lasting for 7 days or less at the emergency departments were prospectively screened for inclusion in the study. Simultaneously, consecutive adult patients who presented for medical care in the IST private clinic were screened for dengue in case of clinical suspicion with symptoms lasting for 7 days or less by the medical doctor in charge. Patients diagnosed with dengue (definition below) were included in the present study. The clinical outcome was assessed by a visit or a call 7 days after inclusion in the study.
Dengue
Dengue was defined as a positive NS1 antigen or IgM detection with rapid diagnostic test (SD BIOLINE Dengue Duo®) and/or a positive PCR (Fast-track DIAGNOSTICS tropical fever core®).
Secondary dengue infection was defined as evidenced of previous dengue infection as determined by anti-dengue virus IgG detection with rapid diagnostic test during the acute phase (≤5 days of symptoms) of the disease as previously defined [
7,
8].
Race
Patients were categorized into two groups according to their self-defined race. Those with African ancestry were in the group of patients with black race (black patients from Tanzania and other sub-Saharan African countries) and those without African ancestry were in the group with non-black race (non-black patients from Europe, US, Australia, Asia, Middle East, South Africa).
Study procedures
Data collection
Demographic characteristics, race, comorbidities and symptoms and signs were collected at inclusion using electronic or paper case report forms. Blood pressure was measured with an automated device (Omron® M6). The socio-economic status of black patients was categorized based on indicators of education and wealth. Non-black patients were all expatriates and were considered as having a high socio-economic status.
GPS localization of patients’ home was recorded. Data were entered directly into an open data kit in a personal digital assistant with real-time error, range and consistency checks [
9].
Laboratory investigations
Rapid diagnostic tests for dengue (SD BIOLINE Dengue Duo®) and malaria (ICT Malaria P.f.®) were systematically performed in patients on site on the day of the interview. Participants enrolled in the public clinics were systematically screened for HIV in accordance with the national algorithm (rapid test, Alere Determine™ HIV-1/2, and for confirmation, a second rapid test, Trinity Biotech Uni-gold™ Recombigen® HIV-1/2). Real-time multiplex PCR (Fast-track DIAGNOSTICS tropical fever core®) targeting dengue virus and Plasmodium was performed in all patients recruited in the public hospitals. Real-time multiplex PCR (Fast-track DIAGNOSTICS Dengue differentiation®) for the detection of dengue virus type 1, 2, 3 and 4 was done in all patients with a positive rapid diagnostic test for dengue in the public and private clinics. RT-PCR analyses were performed at the virology laboratory of the University Hospital of Geneva in Switzerland. Genotyping of dengue virus was performed at the arbovirus and imported viral disease laboratory, National Centre of Microbiology, Madrid, Spain, in a subgroup of randomly selected cases; 4 patients per week (2 in the public and 2 in the private clinic) during the dengue outbreak. A partial and complete envelope gene sequence was obtained using previously described protocols for amplification and sequencing for dengue serotype 2 (DENV-2) [
10].
Complete blood count was performed (Horiba Medical ABX Pentra 80 hematology analyzer) at inclusion. Low platelet count was defined as < 100 × 109/L, high hematocrit level as > 45% and low leukocyte count as < 3.5 × 109/L.
Dengue warning signs and severe dengue
Dengue warning signs and severe dengue were defined according to WHO recommendations [
11]. Warning signs included abdominal pain, persistent vomiting (vomiting during two or more consecutive days), clinical fluid accumulation and mucosal bleed. Three of the seven warning signs, namely liver enlargement, lethargy and increase in hematocrit concurrent with rapid decrease in platelet count were not included as these parameters were not routinely collected. Severe dengue was defined as the presence of at least one of the four following criteria: 1) Circulatory compromise or shock defined as narrow pulse pressure ≤ 20 mmHg or low systolic blood pressure < 90 mmHg, 2) Severe hemorrhage defined as gastrointestinal tract bleeding such as hematemesis, melena or rectorrhagia or menorrhagia, 3) altered mentation defined as a Glasgow coma score of 14 or lower, 4) death within 7 days of follow-up. Severe organ impairment was not a criterion of severe dengue as liver and renal functions were not systematically measured. The clinical outcome and the occurrence of warning signs were recorded at inclusion and by a follow-up visit or call at day seven with the exception of blood count values and blood pressure which were measured at inclusion only.
Data analysis
Demographic, comorbidities, clinical and laboratory characteristics as well as outcome of dengue patients of black race were compared to those of non-black raceusing Wilcoxon-Mann-Whitney and chi-square tests. In a sub-group analysis, black patients included in the public clinics were compared to those included in the private clinic to account for potential inter-observer variation.
Each warning sign as well as hematocrit and platelets count were evaluated for their association with black race using univariate logistic regression and multivariate logistic regression including potential confounders, i.e. age, malaria coinfection, secondary dengue and duration of symptoms at inclusion.
The independent association between demographic and comorbidities characteristics of the patients and the occurrence of severe dengue was evaluated by multivariate logistic regression that included potential confounders.
Statistical analyses were performed using Stata software (StataCorp, College Station, TX, USA, version 12) and GraphPad Prism 6. Google Earth was used for Fig.
2.
Discussion
Although all patients were infected with the same dengue virus genotype, black race was independently protective against a severe course of dengue. These results support the hypothesis of protective genetic or environmental (such as protection after previous exposure to the same dengue strain and cross-protection after exposure to other arboviruses or vaccine) host factors among people of African ancestry.
Our results are in line with previous epidemiologic data showing the absence of severe dengue in Haiti between 1994 and 1996 despite high transmission and by data from the 1981 and 1997 dengue epidemics in Cuba showing a lower rate of hemorrhagic manifestations and hospitalization among subpopulations of African ancestry [
12‐
16]. The Cuban studies also showed that non-black individuals were disproportionately susceptible to dengue [
17]. De La Sierra et al. reported a stronger and cross-reactive dengue virus-specific memory CD4
+ T cell proliferation and interferon-gamma release in white people compared to black people in 80 Cuban donors previously infected with dengue [
18]. Blanton et al. also reported an association between African ancestry and a reduced risk of severe dengue in Brazil [
3,
19]. In addition, a genetic study from Brazil identified a strong association between a polymorphism in JAK1 and severe dengue and showed a different distribution of mutations by race consistent with the epidemiologic data [
20]. In Colombia, an epidemiological study also showed that the Afro-Colombians population had a significantly lower risk of getting dengue and its complications, compared with the non-Afro-Colombians population [
21].
In our study, several factors may explain a lower susceptibility to severe disease among the population of black race. Differences in patients’ characteristics, health facility attended and environmental exposure have been identified between black and non-black patients and might have led to some bias in the link between dengue severity and race.
Different patients’ characteristics such as age and socio-economic status were related to race. Black patients were younger and most of them had a low socio-economic status. Older age has been associated with severe dengue in previous reports [
11,
22,
23]. However, our analyses were adjusted for age and age was not a factor associated with a severe course of dengue. Low et al. reported that older age was associated with a lower prevalence of myalgia, arthralgia, headache and mucosal bleeding [
24]. Age difference between patients of black and non-black race could explain part of the differences observed in the clinical presentation of dengue. Low socio-economic status has also been linked to an increased risk of severe dengue [
25]. Blanton et al. addressed both race and socioeconomic factors in a case control study in Brazil and concluded that both ancestry and income are factors associated with severe dengue [
19]. In our study, socio-economic status could be a potential confounder as income and race are closely interrelated in this Tanzanian setting. However, it cannot explain the reduced dengue severity among black patients as they had a lower socio-economic status compared to non-black patients. Hemoglobin and hematocrit values might also be linked to race as they are known to be lower in persons of African race [
26]. As we do not have values outside of the disease episode, we cannot establish with certainty a link of causality between higher hematocrit value in non-black patients and dengue severity.
The patients were included by two different study teams which might have led to differences in patients’ evaluation. Patients included in the private clinic were mostly non-black (76%) while patients included in the public clinics were all black. However, a different appreciation of the warning signs and dengue severity criteria is unlikely as both study teams were trained to detect severe signs such as mucosal bleed and used the same case report form. Blood pressure was measured with the same device and hematocrit value is not clinician-dependent. Furthermore, when comparing black patients included in the different setting, there was no difference in the prevalence of mucosal bleeding and disease severity.
Black and non-black patients lived in different areas of the city: most non-black patients living in a privileged part of the city while most black patients living in poor and overcrowded wards. However, they were all infected by the same dengue virus strain. This outbreak was not caused by an endemic virus strain that had been circulating in Africa before, but was probably imported from Asia to Tanzania [
4,
5]. Black patients were mostly native of Tanzania while non-black patients were mostly expatriate coming from different parts of the world and thus with different environmental exposures in the past. Therefore, the rate of previous infection by another dengue virus serotype might have been different between black and non-black patients and explain a different dengue severity as secondary dengue infection by another virus serotype is a risk factor for severe disease via antibody-dependent infection enhancement while secondary infection by the same serotype induces protection [
2]. However, the proportion of patients with secondary dengue was not different between patients of black and non-black race. Furthermore, we do not expect protection in the native black population as only dengue serotype 3 has been described in the past in Tanzania [
6,
27]. Exposure or previous vaccination to other flavivirus might have conferred cross-protection against dengue virus. Indeed, black patients native of Tanzania might have been exposed to yellow fever, West Nile or other unrecognized flaviviruses in the past or been vaccinated against yellow fever while Asian patients might have been exposed to or vaccinated against Japanese encephalitis. All flaviviruses are antigenically related and serological cross-reactions between flaviviruses are frequent suggesting that past exposure to a flavivirus might facilitate a secondary booster enhancement effect or cross-protection upon exposure to a different but related virus. In a sero-epidemiologic study, previous exposure to dengue was associated to a reduced severity of yellow fever among military personnel detached in the Ecuadorian Amazonia [
28]. A study performed in mice showed a protective effect against dengue viruses induced by the Japanese encephalitis vaccines [
29]. Cross-protection between the Japanese encephalitis virus and dengue virus was also detected in humans vaccinated with the Japanese encephalitis vaccine [
29]. There has not been any infection with yellow fever in Tanzania for more than 20 years and yellow fever vaccine is not part of the vaccination program in Tanzania. But, most expatriates are vaccinated against yellow fever as recommended for travelers visiting Tanzania. However, there is no evidence that previous yellow fever vaccination has an impact on dengue severity.
Our study has several strengths. Patients were prospectively included during a dengue outbreak and were well characterized. Virus genotyping data allowed ruling out virus characteristics as a factor leading to a difference in disease severity between racial groups. Our study has some limitations. First, in this African setting, race cannot be disentangled from previous exposure to different dengue virus serotypes or other endemic flaviruses as non-black patients were mostly expatriate. Another limitation is the inclusion of patients by two different study team leading in a potential different appreciation of the clinical signs and symptoms of the patients. Third, blood count analysis was only done at inclusion and we could not analyze the rise of hematocrit in the course of the disease.
Acknowledgements
We thank all the patients who accepted to participate and make this study possible. We thank all the clinical officers, nurses and recruiters of Ifakara Health Institute, Mwananyamala Hospital, Sinza Hospital, Magomeni Health Care Center and Tandale Dispensary, who worked with full dedication in this study. We thank the Medical Officers in charge of Mwananyamala Hospital, Sinza Hospital, Magomeni Health Care Center and Tandale Dispensary for their support throughout the study. We are grateful to Lara Turin who managed sample shipment and storage and performed the molecular analyses. In addition, we thank Francisca Molero and Aldo Rojas from ISCIII for technical assistance during sequencing and molecular analysis. We also thank Emilie Pothin for preparing the GPS data figure and Isabella Locatelli for statistical support. We are grateful to Prof. Marcel. Tanner, head of the SwissTPH, for his support and positive input on this research.