Introduction

In December 2013, local transmission of chikungunya virus (CHIKV) was identified in the Caribbean for the first time. A large outbreak ensued, which hit the French Caribbean Islands of Guadeloupe (1,434 km²) and Martinique (1,128 km²) in a single wave in 2014. That year the population sizes of the two territories were 403,750 and 381,326 inhabitants, respectively. After the outbreak terminated in January 2015, it was estimated that approximately 308,000 people had developed clinically overt chikungunya, i.e. approximately 40% of the population [1]. However, an accurate estimation of the attack rate of this outbreak, based on laboratory-confirmed cases, is lacking. There is no evidence that CHIKV had circulated in the Caribbean for at least 200 years and the whole population was immunologically naïve to CHIKV before 2014 [2]. This provided an opportunity to estimate the outbreak attack rate by measuring the CHIKV seroprevalence rate in the immediate aftermath of the outbreak in a population sample reproducing the age and sex structure of the two island populations. We did so in a population-based cohort of HIV-infected patients, based on the assumption that the distribution of the living places of HIV-infected patients matched that of the general population in the two islands where most people live in urban and peri-urban economic areas. We also estimated the proportion of asymptomatic CHIKV infections.

Methods

Results

A total of 388 patients were enrolled but the result of CHIKV serology could not be obtained for 11. Therefore, the study population consisted of 377 patients (198 in Martinique and 179 in Guadeloupe). Patients’ characteristics are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Main characteristics of the 377 participants.

https://doi.org/10.1371/journal.pntd.0009267.t001

CHIKV serology was positive in 230 patients, which translates in a CHIKV seroprevalence rate of 61.0% [95% CI 55.9–66.0]. The CHIKV seroprevalence rates were almost identical in Guadeloupe and Martinique. Seroprevalence rates were not significantly different in men and women, while there was a trend towards lower seroprevalence in youngest adults and higher seroprevalence in the oldest (see Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Seroprevalence of CHIKV in patients in Martinique and Guadeloupe according to territory, gender, and age (N = 377).

https://doi.org/10.1371/journal.pntd.0009267.t002

Among the 230 patients who were seropositive for CHIKV, 153 recalled having had symptoms consistent with chikungunya in 2014, while 77 did not, which accounts for a proportion of asymptomatic cases of 33.5% [95% CI 27.4–39.6]. Symptoms most frequently reported by patients who were seropositive for CHIKV included arthralgia (94.1%), fever (73.2%), myalgia (53.6%), headache (45.8%), and skin rash (26.1%) (S1 Data). We also compared the distribution of these symptoms by levels of CD4 lymphocyte counts (below versus above 200/mm³) and of HIV viral loads (below vs above 50 HIV-RNA copies/mL) in both seropositive and seronegative patients (S2 Table).

Discussion

The two main findings of this study are that the estimation of CHIKV seroprevalence was approximately 60% in both islands and a third of cases were asymptomatic. Both figures are higher than previously reported. Regarding seroprevalence, our figures are higher than those observed in volunteer blood donors at the end of the outbreak in the same area (48% in Guadeloupe and 41% in Martinique) [4] and at the end of the outbreak in the Reunion island in 2005 (38.2%) [5–6] or in Mayotte islands (37,2%) [7] but are very close to those observed after the outbreak in Haiti (57.9%) [8]. It is highly likely that seroprevalence was underestimated in blood donors, probably because recruitment of donors was biased towards subjects less exposed to CHIKV. Lower seroprevalence rates observed in Reunion Island probably results from the fact that this territory is both larger and less uniformly populated than Guadeloupe and Martinique islands. At any rate, our results may not result from a lack of specificity of the assay we used, since the specificity of the Euroimmun assay was demonstrated to be 95% or more for IgG [9]. The relatively high rate of asymptomatic cases we observed in our study (33%, vs 20–25% in prior outbreaks [7]) may have resulted from a recall bias that could have led people who had presented with minor, paucisymptomatic forms of the disease to have forgotten these manifestations, although this hypothesis appears relatively unlikely as patients were interrogated soon after the end of the outbreak. Moreover, when considering that 19.7% of noninfected patients reported having presented with symptoms consistent with chikungunya, the actual rate of asymptomatic forms of the disease might even be higher than 33%.

A genuine strength of our study is that we did our best to minimize any selection bias by analyzing a study population that matched the age and sex distribution of the two island populations.

On the other hand, we acknowledge that our study has some limitations. As already mentioned, a recall bias may have led to underestimate the rate of symptomatic forms of chikungunya. Besides, since the most common symptoms of chikungunya are nonspecific and may be observed in other arboviral or flu-like diseases, some patients may have misreported such symptoms as chikungunya manifestations. Likewise, the fact that our study population was made of HIV-infected people could have introduced a bias. Although there is no data suggesting that HIV infection might alter host susceptibility to or clinical manifestations of CHIKV infection [10], one cannot exclude that exposure to CHIKV might have been different in PLWHIV as compared to the general island populations. On the one hand, the distribution of the living places of PLWHIV in the two islands does not differ from that of the general population. On the other hand, we acknowledge that socio-economic characteristics of PLWHIV differ from those of the general population in several aspects. For instance, the proportion of immigrants is higher in PLWHIV than in the general population (17.0% vs 2.4% and 38.2% vs 4.5% in Martinique and Guadeloupe, respectively) (Nadis data) [11]. Likewise, in Guadeloupe the proportions of persons that are professionally active and highly educated are lower in PLWHIV than in the general population (29.2% vs 40.0% and 11.2% vs 20.0%, respectively), while this is not true in Martinique (42.6% vs 41.9% and 19.8% vs 21.7%, respectively) (Nadis data) [11]. Whether these differences may have been detrimental to the representativeness of our study population is uncertain, especially as it was shown in both islands that all urban and rural districts have been equally affected by the chikungunya outbreak in 2014. As a result, it is unlikely that socioeconomic differences may have had an impact on seroprevalence estimates. The very close seroprevalence rates observed in Martinique (60.1%) and in Guadeloupe (62.0%) may be regarded as one more argument against a differential impact of socioeconomic factors on the spread of CHIKV infection within these territories.

The high seroprevalence rate that we consistently found in both Caribbean islands probably explains the abrupt and persistent cessation CHIKV circulation in these territories. It also provides useful information for modelling the risk of development of a new outbreak in these territories. Given the long-term persistence and protective efficacy of anti-CHIKV antibodies [12], at least one generation replacement and the re-introduction of CHIKV will condition the reemergence of chikungunya in these territories [13].

Supporting information

Acknowledgments

The authors are grateful to the research assistants at the Centre d’Investigation Clinique Antilles who helped collect data and deliver information to the participants (Isabelle CALMONT, Véronique PELONDE-ERIMÉE), to Benoît TRESSIERES for his support to data management, and to the nurses of our departments (Francelise COCLES, Elisabeth DILOU, Marie-Paule FERDINAND, Isabelle JEANVILLE, Marie-Agnès LAURETTA, Astrid ORNEM, Sylvie TURLEPIN). They are also indebted to the medical doctors who followed the participants (Drs Anne-Sophie BATALLA, Julien CUZIAT, Isabelle FABRE, Patrick HOCHEDEZ, Isabelle LAMAURY, Rachida OUISSA, Mathilde PIRCHER, Vincent RONIN, Benoit ROZÉ, Kinda SCHEPERS) and to the patients who accepted to participate in this study.