We identified 767 references, of which 282 were duplicated and thus excluded. After reading the titles and abstracts of the 485 studies, 170 were selected for full text reading. For abstract selection, initial agreement between the two main reviewers was 88%. In the 10% sample selected for check by the two additional reviewers, initial agreement was 93%. For full texts, initial agreement between the two main reviewers was 85%. In the 10% sample checked by the two additional reviewers, initial agreement was 80%. Final consensus was obtained in 100% of them.
Sixty-seven studies were included in the present review. Reasons for exclusion of the other 103 studies are displayed in Figure 1. Information on countries of origin and host continent was available in 16 studies, which were included in the meta-analyses.
Sixty-six studies were published in English and one in French(31). Fifteen studies reported active tuberculosis prevalence, 21 reported LTBI prevalence and 31 reported both. The total screened population was 599,072.
Active Tuberculosis Prevalence
Study Characteristics and Population
Among the 46 studies that reported active tuberculosis prevalence, 56.5% were cross-sectional; none were clinical trials (Table 1). Two thirds of these studies were published after 2009, and among them, half completed data collection before 2011. Sixty-three per cent of studies included over 500 people.
The total of individuals screened for active tuberculosis was 537,218, with a single study evaluating 232,738 individuals. Eighty-one per cent of studies were conducted in refugees (n=437,264), 18% in asylum seekers (n=95,283), and 0.9% in both (n=4671).
The mean age, reported by 33% of studies(4,10,32–44), ranged from 18 to 40.7 years. Prevalence by sex was reported by only 17% of the studies.
The average time since arrival of refugees and asylum seekers to the host country to the time of evaluation for tuberculosis was 3.9 months (ranging from 0.7 to 12.8 months) in the 9% of studies(45–48) conveying this information.
Sixty-seven (31/46) per cent of studies concomitantly evaluated the presence of LTBI in their populations. Among these, 57% used the LTBI diagnostic as a prerequisite to investigate the presence of active tuberculosis. In other words, they performed a diagnostic test for LTBI with TST or IGRA to rule out active tuberculosis; if they were positive, a chest X-ray was also performed and if it was indicative of active tuberculosis, bacteriological tests were conducted. Otherwise, TST or IGRA-positive individuals were considered to have LTBI.
Main findings
Active tuberculosis prevalence rates varied from 0 to 11,364 per 100 thousand inhabitants, with 89% of studies reporting values less than 5000 per 100 thousand. Considering studies that reported the prevalence by country of origin, the average prevalence of active tuberculosis was 1331/100 thousand inhabitants (95% CI, 542-2384), with high heterogeneity (I²=98%, Figure 2). The prevalence was higher among refugees from Syria (11,364/100 thousand inhabitants, 95% CI, 3794-24,558), observed in a single study with 44 hospitalised participants. Studies with individuals from Ethiopia, Ghana and Tunisia had results with large confidence intervals, in samples smaller than 100 persons (10,48).
With regards to the host continent, refugees who immigrated to Europe, Asia and America presented a similar average prevalence of 1458, 860 and 1080 per 100 thousand inhabitants, respectively (Figure 3). Europe was the continent that received refugees from the most diverse nationalities, better allowing an assessment of heterogeneity. In the other continents, this individual evaluation was not possible due to the small numbers. Refugees from Eritrea, Ethiopia and Somalia immigrated to Europe(10,48) in the three studies that contained this information, and had a slightly larger tuberculosis prevalence. More information about the meta-analyses data on active tuberculosis can be found on the supplement material (Table S3).
LTBI Prevalence
Study Characteristics and Population
Fifty-four per cent of the 52 studies that reported LTBI prevalence were cohort studies; there were no clinical trials (Table 2). Sixty per cent of studies were published after 2009, and among them half completed data collection before 2011; one study did not provide this information. Thirty-one studies included over 500 participants.
A total of 271,544 individuals were screened for LTBI: 233,688 individuals were refugees (reported by 67% of studies) and 27,960 individuals were asylum seekers (reported by 21% of studies). The remaining were studies including both types of situations.
The mean age, reported by 33% of studies(4,10,31–38,49–55), ranged from 3.5 to 39 years, with the lowest prevalence in a study that included only children (mean age 3.5 years). Only 21% of studies reported the prevalence by gender.
The average time since arrival of refugees and asylum seekers to the host country at the time of evaluation for LTBI was 3.8 months (range: 0.7-12.8 months) among the 13% of studies that reported this information(45–48,56–58).
Eighty-nine per cent of studies performed TST and 77% of those reported the TST cut-off point used to define LTBI: 10 mm was used in 78% of studies. Some studies considered different cut-off points to different populations (children, human immunodeficiency virus (HIV)-infected or BCG-vaccinated individuals).
Main findings
Prevalence of LTBI ranged from 0.4% to 81.5%, with 61% of the studies reporting a prevalence rate higher than 30%.
In the meta-analysis by country of origin, prevalence rates were highly heterogeneous (I2= 99.8%), with an average measure of 37% (95% CI, 23-52) (Figure 4). Refugees from Cuba and Iraq presented the lowest rates, 0.4 and 5% respectively, and from North Korea, the highest rate, 81%, systematically screened when arriving in South Korea(6). Targeted populations and sample sizes varied largely, with the Cuban study(55) evaluating 241 children under 7 years of age (and finding one LTBI case) and large systematic screening for active tuberculosis of Iraq refugees(58,59) applying for visa or recently arrived in the United States of America (USA). Small sample sizes eventually resulted in very wide confidence intervals(10).
In the subgroup analysis by host continent, refugees who immigrated to Europe presented the highest prevalence (41%, 95% CI, 20-65), followed by those who went to the Americas (28%, 95% CI, 18-40) (Figure 5). However, one study in the USA excluded individuals with immunosuppressive conditions and thus had a high risk of false negative results(56). Somali refugees who went to America had a higher prevalence rate (54%) than those who went to Europe (38%), while Iraq refugees who went to the Americas had a higher prevalence rate (14%) than those who went to Asia (2%). Overall, there were very few studies with information on country of origin and host continent. More information about the meta-analyses data on LTBI can be found on the supplement material (Table S4).
Quality of studies and risk of bias
None of the 65 studies fulfilled all quality criteria. Among the 33 cross-sectional and the 32 cohort studies, only 13 and 11 respectively were considered high quality (Figure S2 and S3); 14 and 18 studies were of medium quality; and 6 cohort and 3 cross-sectional studies were considered low quality. Two studies(58,61) were organization reports; it was not possible to perform the quality assessment.
In 85% of studies (n=569,880), routine screening of all the individuals who arrived in the host country was the reason for the enrollment and 9% of studies tested individuals who sought health service with symptoms (n=11,234). Only one study was conducted in refugee camps(60) (n=15,455). Among the 37 studies that informed the diagnostic method for active tuberculosis, 73% confirmed tuberculosis bacteriologically.
Six per cent of studies involved hospitalised populations (n=4). In these studies, overestimation of the prevalence is likely. Three studies (43,44,70) reported only active tuberculosis prevalence, and one described both active and latent tuberculosis (77). None of them entered the meta-analyses.
Finally, only fifteen articles used the 1951 United Nations Convention or a very similar definition of refugee and asylum seeker.