The impact of migration on tuberculosis epidemiology and control in high-income countries: a review

Surveillance data, and findings from previous meta-analyses (Table 2), have shown that the proportion of migrants with active TB present at the time of migration is relatively small (0.35 %) [18, 19].

Latent TB prevalence figures in migrants are primarily derived from cross-sectional studies where, depending on the specific population tested and the diagnostic tool used, 5-72 % of migrants test positive for LTBI [20‐43]; this is independently associated with increasing age and TB incidence in country-of-origin [41, 42, 44]. One can therefore infer that it is both the cumulative duration of exposure and the TB burden in the source country which determines whether individuals will have LTBI [29, 41, 42, 44, 45]. Migrants with LTBI are coming to lower incidence settings and in the initial years following arrival in the destination country, have a higher risk of LTBI reactivation which decreases slowly over time but remains higher than rates in the host population [12, 46‐48] This higher rate of reactivation in the initial one to two years after migrants arrive likely reflects latent tuberculosis infection which has been acquired in their country of origin shortly before migration although there is also likely to be an ongoing complex interplay, in the destination country, of host (such as age, and comorbidities including diabetes mellitus) and environmental factors (such as nutritional status) which contribute to the observed epidemiology. Understanding the natural history of TB in recently arrived migrants is important when we are considering how best to implement TB control in this population.

Whilst the literature has expanded rapidly in respect of cross-sectional data on LTBI prevalence, longitudinal data on the risk of migrants with diagnosed, untreated, LTBI progressing to active TB disease remain limited partially due to low numbers of chemoprophylaxis-naïve patients (due to recommendations to treat, and not withhold treatment from, individuals identified with latent TB such as in the UK) and because studies to answer this research question need a large sample size and long duration of follow-up. A UK study followed migrants, predominantly from the Indian Subcontinent, and found a TB progression rate of 16.3 % amongst untreated tuberculin skin test (TST) positive patients over a 15 year period following UK arrival [49]. The risk was significantly higher for young migrants (aged 16–19) and for women [49]. However this study did not adjust incidence rates for travel back to countries of origin nor underlying medical comorbidities. A Norwegian study of asylum seekers found a progression rate of 1.1 % over a follow-up period of up to 32 months [50]. By contrast, Marks and colleagues evaluated a large cohort of predominantly Southeast Asian migrants, with no evidence of active TB, over a mean follow-up of 10.3 years post-arrival in Australia and found that 0.12 % with a positive TST progressed to active TB disease per year [51]. Further work is needed to determine both the overall risk of progression from LTBI to active TB disease in migrants as well as the contribution of concomitant medical co-morbidities and demographics such as diabetes mellitus, chronic kidney disease and age. Diabetes mellitus and chronic kidney disease are more common in migrant populations and significantly increase the risk of reactivation from LTBI to active TB [13, 52]. This is likely to result in increasing TB notifications in the foreign-born, migrant populations, particularly as the migrant population ages, and this will, therefore, need to be taken into consideration when developing TB control programmes [13].

An additional issue which is likely to play a part in determining TB epidemiology in migrant populations is the acquisition, and subsequent reactivation, of LTBI following re-exposure during travel back to their countries of origin. Previous work in this area has indicated that travel to high TB burden countries increases the risk of acquiring LTBI with the risk increasing with more prolonged travel and a higher TB burden in the country visited [53]. Although there is a paucity of data on the proportion of TB acquired through travel, published work suggests this could be anywhere between 20 % and 50 % [54, 55]. However further prospective research in this area is needed to more accurately quantify the risk.

In order to distinguish reactivation of remotely acquired LTBI from TB transmission, authors have genotyped DNA isolated from M. tuberculosis cases in foreign-born individuals. Here, individuals with a TB isolate with a clustered DNA pattern that matches at least one other case in the cohort is attributed to recent transmission whereas cases with unique, DNA patterns are thought to arise due to the reactivation of LTBI [56, 57]. In a large meta-analysis across a range of TB burden settings, foreign-born individuals were significantly less likely to have a clustered isolate as compared to local-born individuals (25 % versus 45.8 % clustered respectively) [57]. Moreover, there is little evidence that the foreign-born population transmit TB to the local born population [58]. These data, in conjunction with TB surveillance data, appear to highlight the importance of reactivating LTBI amongst migrants in determining TB epidemiology in high-income countries and, therefore, the need to have appropriate control measures in place.

Most OECD high-income countries screen migrants for active TB although the specifics of how this is performed vary significantly across countries. Most active TB screening programs in Western Europe are performed on or soon after arrival with a chest radiograph (CXR). Other countries such as Canada, the US, Australia, New Zealand and recently the UK, screen for active TB with a CXR prior to arrival. If the CXR is abnormal a sputum smear and culture are performed. Those found to have active TB are treated prior to arrival and granted permission to enter the country if cultures are negative at the end of treatment. Those found to have an abnormal CXR but negative sputum cultures or those with prior treated TB are followed after arrival in a post-landing surveillance program [65‐70].

The criteria for which migrant groups are screened is also highly variable [67, 69]. Alvarez and colleagues found that countries differed in which migrants were selected for screening basing their decision on a number of factors including: type of migrant (e.g. refugee, students, workers), duration of stay, intended occupation or TB burden in country of origin [67]. In a large survey of OECD countries, the authors identified heterogeneity in source country incidence thresholds for screening although, in general, migrants arriving from high TB burden settings were preferentially selected for screening [69]. The reasons for this remain unclear but it may reflect a lack of evidence in this area.

Whilst screening for active TB is frequently undertaken, there is a lack of trial data for its effectiveness as a public health intervention. Policy decisions therefore rely on national evaluations and cross-sectional observational data.

Pre-arrival screening, in country of origin, is part of the immigration process of several high-income countries – including the US, Australia, Canada and the UK. Aldridge and colleagues undertook a detailed systematic review and meta-analysis on the yields of active TB through pre-arrival screening. They found that the overall yield for culture positive active TB was 0.22 % (219 cases/100,000) [71]; the yield for culture positive TB increased with increasing TB prevalence in country of origin [71] suggesting that setting an incidence threshold for pre-arrival screening of migrants may be needed to ensure cost-effective use of resources. Klinkenberg et al. reviewed the yields for active TB at different stages of the immigration process and found that the yields for pre-arrival, as compared to post-arrival, screening were higher (1.21 % versus 0.31 % respectively) although these data were based entirely on non-EU studies [19]. A recently published analysis of the US pre-arrival screening programme found that 4032 cases of culture positive TB were diagnosed amongst 1,561,460 migrants screened (yield 0.26 %) resulting in a reduction in the number of TB cases diagnosed amongst migrants within one year of arrival in the US [43]. The UK pre-arrival screening programme yield for active TB has steadily increased from 0.05 % in 2006 to 0.16 % in 2014 which most likely reflects the use of sputum cultures [72, 73].

Post-arrival screening for active TB involves chest radiography once the migrant has arrived in the host country. Two systematic reviews have evaluated the outcomes and yields for active TB in high-income countries [18, 19]. Arshad et al. reviewed 22 studies comprising 2,620,739 migrants and found that the overall yield for post-arrival screening was 0.35 % with higher yields in migrants from Africa and Asia [18]. Klinkenberg and colleagues reviewed 40 studies and found that the yields for post-arrival screening, which were lower than that for pre-arrival screening, ranged from 0.20 % to 0.36 % depending on the specific setting in which screening was undertaken [19].

Previous work has shown that the current models of active TB screening have weaknesses including: individuals not completing the screening processes, limited yields for active disease and an inability to identify active TB occurring through LTBI reactivation [13].

Whilst most countries offer some form of screening for active TB, screening for LTBI is much less commonly performed [69]. High-income countries that screen for LTBI usually undertake this post-arrival with a tuberculin skin test (TST) or an interferon gamma release assay (IGRA) [69, 74].

For reasons of practicality and cost-effectiveness, most high-income countries attempt to limit the eligible population to refugees or asylum seekers or those individuals arriving from high TB burden settings [69]. However, countries vary considerably in their definition of a high TB burden setting for the purposes of migrant screening [69]; the UK has taken a decision to screen migrants arriving from countries in Sub-Saharan Africa or those countries with a TB incidence above 150 per 100,000 whereas Canada screens at a lower threshold of 30/100,000 but only migrants with increased risk of reactivation. Given the prevalence of comorbidities in the migrant population (such as diabetes mellitus,) which increase the risk of reactivation, these factors will likely need to be taken into account when determining which migrants to screen. At the present time, however, the variation between guidelines could reflect uncertainty about the optimal screening threshold which balances the need to identify the majority of LTBI with cost effectiveness.

Successful screening for LTBI involves a number of key interlinked steps including the accurate identification of migrants, appropriate screening of migrants, initiation of chemoprophylaxis and completion of therapy. However most of the data on LTBI screening outcomes come from cross-sectional studies which have been conducted with the primary aim of calculating the prevalence of LTBI in migrants (estimated at around 25-30 % in young adult migrants from high incidence countries) [20‐43]. However, there is less data on the other elements of the screening pathway – including uptake, and completion, of chemoprophylaxis. A Canadian group recently reviewed the data on LTBI screening effectiveness and found that migrants dropped out at each step of the screening pathway so that overall only 31 % of the cohort completed the programme successfully highlighting the need for research into interventions to optimise the LTBI screening pathway [75]. Programmatically, at a national level, there is little observational data on the impact of LTBI screening on TB notifications in migrants although the recently commenced UK national migrant screening programme will provide this useful data in the next few years [76].

Screening migrants for active TB is widely implemented by high-income countries albeit with different models of care. However there are few studies formally examining the cost-effectiveness of this intervention. Previous studies have come to differing conclusions about the cost-effectiveness of screening migrants for active TB. Schwartzman and colleagues modelled the comparative cost-effectiveness of migrant screening using chest radiography versus tuberculin skin test (versus no screening) and found that in migrants with a high prevalence of infection that chest radiography was the most cost-effective screening modality although the TST strategy prevented the most cases of TB. By contrast, Dasgupta et al. constructed a Markov model informed by empirical data to compare the cost effectiveness of screening migrants with chest radiography pre-arrival followed by TST (if the CXR showed any abnormalities) with screening close contacts of index sputum smear-positive cases (with TST followed by CXR) [77]. The authors found that migrant screening using chest radiography was not cost-effective due to difficulties with operationalising screening [77]. The lack of research in this area highlights the need for further health-economic analyses to objectively assess, and make conclusions about, the cost-effectiveness of screening migrants for active TB.

A larger number of published studies from high-income countries have focused on evaluating the cost-effectiveness of screening migrants for LTBI [78‐86]. These studies, which have evaluated different aspects of LTBI screening including which migrants to screen and how to screen, have generally concluded that LTBI screening of migrants from high burden countries, mainly Asia and Africa is a cost-effective intervention in high-income countries [41, 42, 78‐86].

Methods for diagnosing LTBI have evolved over the last decade with IGRAs increasingly replacing the TST [87]. This is reflected by the several studies which have explored the relative cost-effectiveness of different screening modalities and algorithms for LTBI [41, 42, 79‐84]. These health-economic analyses have, in general, found that IGRA are more cost-effective than TST [41, 42, 79, 81, 84]. However, in the absence of robust longitudinal data as for TST, there remains ongoing debate about the use of IGRA as a screening tool with certain national guidelines instead advocating the use of TST [13].

The scarcity of randomised clinical trials has meant that policy decision are mostly based on of health-economic modelling and observational data. However models must, by definition, make a number of simplifying assumptions and their results are therefore highly dependent on model structure and specific model parameters – even if there is uncertainty around these due to a lack of empirical data.