This is the first exhaustive epidemiological study in Reunion Island. Our strength was to include patients from both mandatory declarations and microbiological database, reducing bias in incidence estimation (6.4% of missing declaration in our study) and giving precise description of demographical, clinical and microbiological patients characteristics, that could not be reported in previous studies uniquely based on declarations [6, 7]. The rate of missing declarations was low. Since 2016, every microbiological result positive for tuberculosis in universitary hospital laboratory is referred to local tuberculosis center to ensure patient treatment, optimize screening around the cases and lower missing declarations.
We confirm low endemicity in Reunion Island with incidence of 6.23 per 100,000 habitants per year in the 2014–2018 period, closed to recent public health data [7] and to national statistics [10]. However, risk is higher for migrant population: 47% of patients were born outside Reunion Island including 32.1% from Madagascar with increasing incidence in this population over the study period. Our estimated incidence in migrant population must be taken with caution as it is extrapolated from 2013 INSEE data but it is coherent with higher incidences in surrounding islands in 2018: 11.5/100000 habitants per year in Mayotte, 13/100000 habitants per year in Mauritius, 35/100000 habitants per year in Comoros and 233/100000 habitants per year in Madagascar [1, 8]. As in other parts of France, the Office Français de l’Immigration et de l’Intégration (OFII) oversees tuberculosis screening in recently arrived migrants, representing 7.9% of tuberculosis cases in France every year [11, 12]. In our study, median delay between arrival in Reunion Island and tuberculosis diagnosis is 30 months. It is corroborant with a maximum risk of tuberculosis in the 4 first years of arrival in United Kingdom in Aldridge et al. cohort [13]. It suggests that chest X-ray should be repeated in the first years of arrival to improve tuberculosis screening on the territory.
Our work is also the first to describe antituberculosis drug resistance in Reunion Island. Previous studies [6, 7] reported declared resistance at diagnosis but without any phenotypic susceptibility testing after culture results and without any rapid genotypic resistance testing before 2016. In our study, antimicrobial susceptibility data were available in 216 out of 265 patients; the first limit was absence of phenotypic testing results in some patients: those with negative cultures and those whose culture was performed in private laboratories; the second limit was absence of rapid genotypic testing results before 2016 in our university laboratory.
Antimicrobial resistance was low in Reunion Island on the study period. MDR tuberculosis represented 1.4% of the tuberculosis cases whereas national reference center and WHO estimate it represents 1.9% of the tuberculosis in France in 2018 and 3 to 5% of the tuberculosis in the world respectively [1, 14]. There was no XDR tuberculosis in our cohort. However, 14.4% of the Reunion Island patients had resistance for at least one first-line antituberculosis drugs, compared to 13.3% in France. Among them, 5.6% had at least isoniazid resistance versus 2.3% in France, and 1.8% had isoniazid monoresistance versus 2.5% in France [14]. This resistance was not associated with a poorer outcome in our patients, but this analysis was limited by small number of patients. Frequent difficulties to treat patients with resistance and/or toxicity with first line antituberculosis drugs suggest developing rapid second-line antituberculosis susceptibility testing [15] in our laboratory to reduce delay for adequate treatment.
We looked for several potential risk factors in antituberculosis drugs resistance.
First, we described mostly primary resistance (29 out of 31 patients). Previous tuberculosis and previous treatment for tuberculosis was higher in patients with resistance than patients with susceptible strains (respectively 12.9% vs. 5.9%) but difference was not significant in these few patients.
Secondly, we considered clinical characteristics. HIV prevalence was low in our cohort. Multivariate analysis showed a higher proportion of women in group with antimicrobial resistance in contrast with previous literature in France or other countries [14]. Besides, diabetes mellitus was more frequent in patients with antituberculosis drugs resistance in univariate analysis, but with no significant differences in multivariate analysis. This corroborates some studies in other specific areas [16,17,18]. Physiopathological explanations may include delayed time to sputum conversion in diabetic patients leading to secondary resistance, lower rifampicin plasmatic concentrations in diabetic patients and/or chronic inflammation and immunosuppression in diabetic patients making them more susceptible to resistant strains [19]. Since diabetes mellitus is particularly prevalent in Reunion Island [20], it suggests a specific attention to diabetic patients with tuberculosis, who represent 17.3% of our cohort.
Thirdly, we analyzed social characteristics. No significant difference was found when considering health insurance and homelessness between patients with susceptible or resistant tuberculosis. Direct observed therapy was more frequently used in patients with resistance but it did not reach significance. Lost to follow-up was low in our study (20.8% versus 35.2% in France [10], probably due to insularity.
At least, we compared demographic data in patients with or without resistance. Resistance frequency was lower in patients who were born or had lived in Madagascar than patients from Reunion Island. Little is known about antimicrobial resistance in Madagascar, because of extreme poverty and healthcare system difficulties. Between 2012 and 2017, a nationwide program tried to evaluate drug susceptibility in patients with tuberculosis relapse and/or close contacts of MDR tuberculosis. In these high-risk patients, only 4.5% of patients had MDR tuberculosis and none of them had XDR tuberculosis. However, authors estimated that only two thirds of Malagasy patients were diagnosed and that only 1% of them were tested for genotypic drug resistance [21]. Our study is one of the few which systematically tested phenotypic and genotypic anti tuberculosis drug resistance in Malagasy people, reporting a very low frequency of resistance (10.3%). Considering migrants from Comoros or Mayotte islands, there was no available data about tuberculosis resistance in the literature so far. Our work reported no resistance in 10 patients from Mayotte and resistance in 2 out of 8 patients from Comoros but we could not draw any conclusion from these small subgroups. Since Mayotte became a French department in 2009, migratory flows from Comoros islands intensified as well as growing numbers of migrants coming from Central and Austral Africa, with some concerns about potential impact on tuberculosis epidemiology in the area.