We performed a retrospective cohort study in asylum seekers from Eritrea and Somalia who arrived in the Netherlands from January 1, 2013, through December 31, 2017. Because Eritrea was not yet an independent country before 1991, persons from Eritrea who were born before 1991 were actually born in Ethiopia. Because our data sources reported on country of birth (instead of nationality), we combined the groups of asylum seekers from Eritrea whose country of birth was listed as Eritrea or Ethiopia (3.4% persons in this combined group were born in Ethiopia).

We obtained data from the Netherlands’ Immigration and Naturalization Service (IND) on asylum seekers from Eritrea and Somalia arriving in the Netherlands during the study period. IND provided data on numbers of asylum seekers and reason of request (e.g., first applications, repeated applications, family reunifications, and invited asylum seekers) by country of birth, month and year of arrival, sex, and age group. We excluded repeated applications because these are usually from asylum seekers who reapply after a failed first application without leaving the country in between applications. The IND does not provide information on duration of stay in the Netherlands of asylum seekers after registration (i.e., no linked data of asylum seekers who left the country or died are available). Because asylum seekers from Eritrea and Somalia are usually granted asylum in the Netherlands and therefore stay in the country (11), we assumed that the follow-up period of each person in the study population lasted either until the end of the study period (December 31, 2017) or until the event (TB diagnosis) occurred.

We obtained data from the Netherlands’ National TB Register (NTR) on TB patients from Eritrea and Somalia who arrived in the Netherlands during the study period (i.e., persons diagnosed with active TB through December 31, 2017). NTR contains detailed information on patient demographics (including date of arrival in the Netherlands, age, sex, and country of birth), diagnostic and disease characteristics, and treatment outcome. NTR also has information on whether patients belong to specific risk groups, such as asylum seekers or immigrants. Because asylum seekers who have onset of TB after obtaining asylum are registered as immigrants in NTR, we included asylum seekers and immigrants from NTR as long as they had arrived on or after January 1, 2013. We excluded TB patients with unknown date of arrival in the Netherlands (n = 41).

We defined cases of prevalent TB as active TB in patients who were registered in NTR as being found through entry screening, independent of the time that had passed since arrival. We defined cases of incident TB as active TB in patients who were not found through entry screening, including patients who self-reported symptoms and patients found through follow-up screening or contact tracing. Patients with pulmonary and extrapulmonary TB were included. Because entry screening is conducted by using chest radiograph, mainly pulmonary and some forms of intrathoracic TB (e.g., pleural TB) are detected by screening. TB diagnosis follows the criteria set by European Union member states (i.e., cases are confirmed if M. tuberculosis is found in patient specimens; cases are deemed probable or possible if bacteriologic or clinical criteria are met) (12).

We merged both datasets (the cohort of asylum seekers from Eritrea and Somalia who arrived during the study period and the TB patients registered in NTR) by month and year of arrival, country of birth, sex, and age. We calculated the total follow-up time (in person-months) for cases by subtracting the date of arrival from the date of diagnosis, and for noncases by subtracting the estimated date of arrival (set at the 15th day of the registered month of arrival) from December 31, 2017.

We analyzed data by using the statistics software package Stata/SE 15.1 (https://www.stata.com). We described and compared characteristics of (prevalent and incident) cases and noncases and calculated incidence rates by number of years after arrival. Because we were interested in the risk for TB over time in persons with different follow-up periods, we performed survival analysis with Cox proportional hazards regression. We calculated cumulative incidences and analyzed the effect of country of birth, calendar year of entry, age, and sex on the risk for active TB.

Our study showed that asylum seekers in the Netherlands from Eritrea and Somalia have a high risk for TB: 0.3% had TB upon arrival, and ≈3% had onset of TB in the first 5 years after arrival. Although incidence rates gradually declined, they were still 10–50 times higher than the overall TB incidence in the Netherlands. Furthermore, our study provides additional insight in specific risk groups for active TB: adult (mainly those 18–35 years old) asylum seekers from Eritrea and Somalia were at higher risk for TB compared with those <18 years of age, as were men and boys from Eritrea compared with women and girls from Eritrea.

These results are consistent with a study conducted in the Netherlands in 2004, which showed that the incidence of pulmonary TB remained high in immigrants from high-incidence countries at least a decade after arrival in the Netherlands (13). A study in Denmark showed that in the 1990s, the annual incidence of TB in immigrants from Somalia decreased only gradually during the first 7 years of residence, from 2,000 to 700 cases/100,000 population (14). Although these studies provide data on changes in incidence over time, we used survival analysis methods to analyze and depict the risks of a cohort of newly arrived asylum seekers, which enabled us to take into account the different follow-up periods for each person in the study.

The TB prevalence and incidence rates in asylum seekers from Eritrea and Somalia in our study were much higher than the WHO-estimated TB incidence in Eritrea (74 cases/100,000 population) and in Somalia (270 cases/100,000 population) (8). A plausible explanation for this finding is the additional risk for infection while traveling to Europe, where overcrowding and unsanitary conditions are common along travel routes, on top of the baseline infection risk in the country of birth (15,16). Walker et al. (17) found molecular and epidemiologic evidence for this in their study of a cluster of multidrug-resistant M. tuberculosis infections among patients arriving in Europe from the Horn of Africa. A second explanation could be an increased risk for TB because of vitamin D deficiency, malnutrition, and stress (15). These conditions are common in asylum seekers during the often stressful asylum application procedures and during the first years of settlement in the new country. Third, transmission within ethnic groups in the new country of residence can also contribute to the higher TB rates found in asylum seekers. Occasional outbreaks have been reported in ethnic groups, including recently arrived asylum seekers (18). Finally, the WHO figures could be an underestimation of actual TB rates (18–20).

Our findings support the recommendations for LTBI screening of asylum seekers from high TB-incidence countries (4). In our study, most TB cases in asylum seekers from Eritrea and Somalia occurred after initial radiologic screening for active TB, and only a few cases were identified by radiologic follow-up screening (partly because of the low coverage of follow-up screening). LTBI screening and treatment can prevent active TB, including extrapulmonary forms. Furthermore, because LTBI might reactivate to active disease many years after infection, LTBI screening also has the potential to prevent TB in asylum seekers many years and even decades after arrival in the new country. Implementing an LTBI screening program in asylum seekers is not easy. A survey conducted by the WHO Regional Office for Europe and the European Respiratory Society showed that 53% of countries in Europe performed systematic LTBI screening in refugee populations (21). A study conducted in 11 selected countries of Europe indicated that these countries had very different methods and policies for migrant TB or LTBI screening (22). Systematic reviews have demonstrated that limited information is available on the yield and effectiveness of migrant LTBI screening (22,23). Furthermore, cost-effectiveness of LTBI screening as predicted in mathematical models is highly setting-specific, with best results achieved if restricted to migrants from high-incidence countries (24).

Our study had several strengths. We were able to combine national comprehensive data on immigration and TB notification; thus, we could calculate incidence rates for specific groups of asylum seekers and show trends over time after arrival. Moreover, by using a Cox proportional hazards regression model, we were able to assess the effect of risk factors such as sex and age on the risk for TB.

Our study also had some limitations. First, no data were available on the actual follow-up period of persons in the study population who did not have TB diagnosed. Although most asylum applicants from Eritrea and Somalia are granted asylum and will therefore stay in the Netherlands (11), some might have moved to another country (or died) before the end of our study period, meaning that we possibly underestimated the TB incidence rates in asylum seekers from Eritrea and Somalia. On the other hand, whereas we included only asylum seekers (not immigrants) from the IND database, we did not differentiate between asylum seekers and immigrants among the TB patients included in our study. The reason for this was that asylum seekers who obtained a residence permit before having their TB diagnosed are registered in NTR as immigrants, even though most arrived in the Netherlands as asylum seekers. Very few asylum seekers from Eritrea and Somalia come to the Netherlands as immigrants (11,25), so we expect that this limitation has not led to a substantial overestimation of TB incidences in our study. Second, no information on travel routes was available for our study population, and no distinction could be made between asylum seekers who had undertaken the often long and stressful journey by land and water and those who came directly by airplane, for example, for the purpose of family reunification. Future studies should take these differences into account. Moreover, we only analyzed data on asylum seekers from 2 countries with high TB incidence countries (Eritrea and Somalia), so the results might not reflect trends in onset of active TB in asylum seekers from other high-incidence countries. However, because asylum seekers from other countries in Africa often share the same hazardous journey, their risk for TB is probably similarly elevated. An investigation into whether the risks for active TB after arrival in the Netherlands differ between asylum seekers and other migrants from high-incidence countries (e.g., persons migrating to the Netherlands because of work or study) would be warranted. We recommend additional studies of longer follow-up periods to enable a more extensive analysis of trends in TB incidence rates, molecular studies differentiating disease caused by in-country transmission or reactivation of premigration acquired infection, and studies evaluating the effectiveness and impact of LTBI screening programs in asylum seekers.

In conclusion, our study results clearly show that asylum seekers from Eritrea and Somalia remain at high risk for active TB for at least the first 5 years after arrival in the Netherlands. This finding underscores the need for an LTBI screening and treatment program for high-risk groups. LTBI screening and preventive treatment will also accelerate TB control and contribute toward the elimination of TB.