To the best of our knowledge, this study is the first one carried out in French Guiana that describes and analyzes social and health disparities in specific populations on the basis of their origins and migration status. This study shows that these two cities have large immigrant populations (40.5% of the adult population in Cayenne and 57.8% in Saint-Laurent du Maroni) and that many of these individuals' had been there for long while (60.7% of the immigrants living in Cayenne and 77.5% in Saint-Laurent du Maroni had been living there for more than 10 years). Despite this long duration of residence in French Guiana, a substantial portion of the immigrant population had no stay documents or had a precarious status. An analysis of the population's social and economic conditions shows strong inequalities that follow a gradient according to the individual's legal status with regard to his or her stay. The analyses showed that the health of these populations depends on several migration-related factors, but also on how health is measured. Three key findings are noted. First, in general, of all the subgroups of migrants that were studied, those most vulnerable and with the worst health status were those who were undocumented, regardless of which social and health indicators were considered. Second, when health was measured as perceived health, the analyses showed that the undocumented immigrants and the documented immigrants with a precarious status (with a stay document valid for one year or less) reported poorer perceived health than the native-born. The country of origin and the duration of residence did not change these results very much. These observed associations are only partially explained by the individuals' socioeconomic status. Third, with regard to functional limitations, certain groups of immigrants (recent immigrants and those born in Haiti) reported a more favourable situation than the native-born for a comparable socioeconomic status.
Although the literature on this topic is sparse, several studies and reports suggest it is undocumented immigrants who are the most vulnerable with regard to health [1
]. They suffer from a combination of socioeconomic conditions and working conditions that are precarious or even harmful to their health [19
], and they have difficulty accessing health care. In our study, it was mainly the undocumented immigrants who seemed to be the worst off socioeconomically. The socioeconomic indicators used in this study explain only some the observed differences in health. The remaining differences could be explained by socioeconomic factors that were not taken into account in this study (such as income, working conditions or housing conditions) and by difficulty accessing health care. In French Guiana, as in mainland France, undocumented immigrants can theoretically access health care free of charge through a specific health insurance system called "Aide Médicale État
" (government medical assistance, which is government-run, unlike the usual health insurance system, which is run by Social Security). If, as several reports have shown, there is, in France, a gap between theoretical rights and actual rights to health care (due to the complexity of the system, the difficulty people have in presenting the required administrative documents, the lack of information on the part of administrative personnel, differences in their practices, and so on [15
]), then these difficulties are surely much worse in French Guiana [20
Our results for perceived health are consistent with those of several international and French studies. A systematic review with the objective of examining and comparing self-perceived health among migrants and ethnic minority groups in EU countries showed that most migrants and ethnic minority groups appeared to be disadvantaged in relation to the majority population, even after controlling for age, gender and socioeconomic factors [22
]. A study carried out in mainland France among a sample of more than 20,000 people that was representative of the general population (Enquête décennale santé [Decennial Health Survey]) found that people of foreign origin living in France reported poorer health than the French born in France. It did not find any differences in health between foreign immigrants and those who had been naturalized. As in our study, these populations' poor socioeconomic conditions only partially explained their poorer perceived health [23
]. A study carried out on Mayotte Island, a French overseas territory in the Comoros Archipelago, found that the health of foreigners was less good there as well (and they were found to have more difficulty accessing health care) than that of the French [24
The recent immigrants to French Guiana (≤ 5 years) reported fewer functional limitations than the native-born French. This finding supports the "healthy immigrant effect" hypothesis, according to which migrants represent a selectively healthy group that is not representative of all potential migrants from origin societies [25
]. This hypothesis is also supported by additional analyses in this study suggesting that the migration of sick people (or health care migration) accounts for only a minority of migration movements [28
]. This is not observed for perceived health, which may be due to the cut-off that was chosen. Indeed, other studies suggest that the decline in self-perceived health occurs over a very short period after migration [29
]. In addition, several studies have found relatively better health outcomes for immigrants for indicators such as mortality, chronic conditions and impaired activity than for self-assessed health [26
], which suggests that health selection is stronger for chronic and severe conditions.
After adjustment for the socioeconomic conditions, the people born in Haiti reported fewer functional limitations than the native-born French. This subgroup of immigrants had the worst socioeconomic indicators (47.2% of the people born in Haiti had no or only a primary education, 60.7% reported that they did not have enough to live on, and only 34.5% were working). Moreover the proportion of recent immigrants (≤5 years) among immigrants born in Haiti (14.2%) was not different from the one among immigrants from other countries (15.5%). Therefore, three hypotheses could explain this paradoxical finding. One is that of cultural differences in reporting functional limitations, although it hardly seems plausible (this hypothesis will be detailed below). Another is that of selection bias due, in this case, to the return of migrants in poor health to their country of origin, which seems even more unlikely, given the overall situation in Haiti. A third hypothesis seems the most probable: that of greater migration selection among migrants from Haiti, the poorest country in the Americas [34
]. A recent study carried out in Spain found that "[f]oreign immigrants from poor countries reported the worst socio-economic conditions, but relatively good health" [33
]. Other studies suggest that long distance migration may be associated with a stronger selection effect [7
]. It may be that Haitians in better physical health are the ones more likely to move to French Guiana because they are able to manage the difficulties and stress associated with immigrating. The undocumented immigrants from Surinam had poor health indicators, regardless of which health indicator was used or which adjustments were made. These people have special attributes: all of them were living in Saint-Laurent du Maroni (a town on the border with Surinam), and their median duration of residence in French Guiana was 16 years (as opposed to 6 years for the other undocumented immigrants). Thus, a number of hypotheses can be proposed to explain their particularly poor health status: the circumstances of their immigration to French Guiana (fleeing from the civil war between 1986 and 1992 and economic hardships in Surinam), the geographical proximity of Saint-Laurent du Maroni (which limits the possibility of positive immigration-selection bias), and the many years spent underground.
Limitations and strengths of this study
This study has a certain number of strengths: a sampling method ensuring that the final sample would be representative, a high participation rate, and the inclusion of several migration variables.
Several limitations should be discussed. First, this was a cross-sectional study, and no definite conclusions can be drawn regarding causality. Second, this survey was conducted among people over the age of 18 years who had been living or were intending to live in French Guiana for at least 6 months and who were residing in single-family dwellings. It therefore excluded people living collectively, people with no fixed address, and transient migrants. Third, we did not have a means of measuring the representativeness of the subgroup consisting of undocumented immigrants, since they are, by definition, undocumented in the national statistics. On the other hand, the sampling procedure (the stratification and sampling intervals used) and the large proportion of this population in the two survey cities make it unlikely that we under- or overrepresented the neighborhoods inhabited by undocumented immigrants. Lastly, a few words need to be said about the choice of indicators. The three health indicators of the MEHM had the advantage of being widely used in epidemiological surveys, and their reliability had been evaluated in a European population [12
]. However, they have not been validated in the populations of French overseas departments (especially in French Guiana). Moreover, questions remained about their interindividual comparability, since health perceptions vary according to health norms and people's aspirations, who are influenced by their social and cultural environment [36
]. Of the three health indicators used, a self-reported chronic disease is the most prone to differential reporting bias between social groups [38
]. In this study, homemakers and individuals with little schooling reported poor perceived health and functional limitations more often, but these associations were not found for the indicator 'chronic disease'. Several analyses have reported a trend toward chronic diseases in population groups in the lowest education and income brackets being underreported [38
]. This can be explained by less medical information, which is due to less use of the health-care system. In addition, it is questionable whether the concept of chronic disease is clearly understood by all sociocultural groups. Perceived health is the mostly widely used indicator, and numerous studies have shown associations with mortality [36
], morbidity and the use of the health-care system [42
], regardless of the ethnic group [44
]. However, a few studies found that this indicator tended, once again, to underestimate social health inequalities [38
]. As for the indicator 'functional limitations', its transcultural validity has not been investigated, but several studies that have examined this indicator between different ethnic groups suggest that information biases are weak [47
The choice of migration variable has its limitations, too. The main one is that the groupings that were made (to construct the six subgroups based on migration status) mask very different sociocultural situations and migration paths. For instance, the subgroup consisting of people born in French Guiana was actually quite heterogeneous (among the main ethnic groups that make up the population in French Guiana are the Creoles, the Bushinenge and Amerindians). Furthermore, when constructing these six groups, we took into account the individual's status on the day of the survey. This categorization did not take into account how long the person had had that status, for some statuses are not stable. Immigrants with a temporary stay document can have their renewal request turned down and quickly become undocumented. In contrast, some of the interviewees may have very recently regularized their status.