Screening for tuberculosis in an urban shelter for homeless in Switzerland: a prospective study

Homelessness, a global and mostly urban phenomenon, encompasses a spectrum of social situations ranging from inadequate accommodation to the complete lack of housing [1]. In the European Union, it affects an estimated 410′000 people each night and 4.1 million annually [2]. There is no data currently available about the situation in Switzerland.

Homelessness exposes to harsh physical and social environments with a negative impact on health. In Europe, mortality in homeless persons is seven times higher than in the general population and morbidity, both somatic and psychiatric, is also significantly increased [3, 4]. It is associated with a substantial risk of communicable diseases, notably tuberculosis (TB) [5, 6]. In Western Europe, TB prevalence in homeless persons is between 1 and 2% for active and up to 45% for latent infection [5, 7‐9]. This prevalence, considerably higher than in the general population, underlines the public health importance of developing control programs targeting this group. To enhance TB control program effectiveness, guidelines recommend using active decentralized (ie. outside healthcare settings) case-finding strategies in cooperation with community stakeholders and local social services [10, 11]. While screening and treatment is cost effective (US$ 32′000 per QALY) compared to passive strategies among groups with a TB prevalence of 250/100′000 or higher in Europe, low awareness of TB, limited participation in programs, and frequent drop out along diagnostic and therapeutic procedures frequently limit effectiveness of control programs [5, 12, 13]. Various screening modalities have been tested and there is still debate over the optimal strategy [14]. Although there is no gold standard methods for screening active tuberculosis in underserved populations, active case-finding strategies usually rely on chest-X-ray and sputum analysis, sometimes combined with symptoms score. In registry-based studies, chest X-ray provided a limited sensitivity of 42% (95% confidence interval: 28–58%) [15].

In Switzerland, active TB incidence is 7.4/100′000 and the annual number of new active TB cases has been decreasing regularly until 2007, then increasing again slightly, mainly because of an increase in cases among foreign-born subjects accounting for 73% of cases in 2014 [16]. In the 1999–2002 period, vulnerable groups of population such as asylum seekers, refugees, undocumented immigrants and homeless persons accounted for 39% of cases in Geneva [17]. In recent years, we observed a change in populations accommodated in shelters with a growing proportion of vulnerable migrants from medium - to high TB prevalence countries, which correlates with findings in other European cities [5]. Unlike Germany where screening for active TB is mandatory for homeless persons living in shelters, Swiss policies restrict systematic screening to asylum seekers. Since January 2006, all asylum seekers undergo a health questionnaire upon entry and active TB prevalence is currently estimated in this group at 124/100′0000 [18]. Before 2006, screening included systematic chest X-ray and tuberculin skin testing. A comparative study showed that the questionnaire was less sensitive but more specific than systematic chest x-ray, with a similar yield in terms of number of cases started on treatment at 90 days, a lower proportion of screenees requiring additional investigations but a longer time-lag before treatment initiation (mean of 6 vs. 25 days) [18]. Screening for latent tuberculosis infection is not considered a priority in Switzerland for this population, because of its high mobility, rendering completion of preventive treatment unlikely, or at least difficult to monitor.

In 2013, we diagnosed two cases of late presentation of active TB in homeless adults living in city shelters in Geneva with several employees working in the shelter showing seroconversion. Moreover, the time-lag between the onset of symptoms and the diagnosis sharply compromised the contact tracing because of the high turnover rate of users in the shelters. There was thus a need to document the local active TB epidemiology in this setting to guide local health policies and strategies.

We designed a program with the social services of Geneva, Switzerland, with the goal of testing a systematic screening program for active TB using the screening tool applied to asylum seekers. Study objectives were to assess 1/ program acceptability in homeless subjects and 2/TB prevalence in this population.

We designed and managed this program in accordance with European recommendations supporting decentralization and intersectoral collaboration [20]. It provided an opportunity for active TB screening for a large majority of homeless persons accommodated in shelters in Geneva, with high participation and retention rates. During the study and in the post-study surveillance, no case of active TB was diagnosed in the target population.

We found a high (87.3%) acceptance rate for the screening procedure, particularly among young males who had been homeless for a limited period of time. Moreover, retention along the diagnostic procedures was higher than expected. This suggests that homeless persons in our area are concerned about health issues and respond positively to opportunities in contrast to the perception of this group having low inclination for preventive health measures [21]. Cooperation with social services, which also provide essential resources such as food, may have enhanced trust towards our program. Incentives may provide a motivational effect and several authors described how appropriate use of incentives enhanced participation to TB screening or treatment programs [22‐24]. In Frankfurt, implementing a voluntary X-ray screening program with the help of local social services had a positive impact on participation and coverage [5]. Other studies showed how cooperation between health and social sectors helps to promote health opportunities for underserved groups of population [8, 25]. As opposed to other infectious diseases, involving peer educators does not seem effective in increasing homeless persons participation screening campaigns for active TB. A study in London comparing shelters with and without peers providing health information to residents found no difference in chest x-ray attendance [26]. Yet, we believe that our efforts to increase awareness regarding TB by direct information and involving social workers contributed to the high participation rate. Our findings encourage the implementation of health promoting and preventive activities targeting homeless persons such as influenza immunization or smoking cessation programs [13, 22, 27].

The null TB prevalence confirms the variability of risk among homeless groups. A meta-analysis in Western countries found that prevalence was highly heterogeneous depending on population characteristics, setting and screening strategy [7]. Several reasons may explain our findings: a) a measurement bias due to the lack of sensitivity of the test. The threshold value of 10 for the test was defined and tested by Schneeberger et al. with an acceptable value for sensitivity and specificity, in a population which presently has many epidemiological similarities with our study population [18]. Interestingly, as in our homeless population, 4.1% of asylum seekers in Switzerland tested positive to the questionnaire for a TB prevalence of 0.12%. Therefore modifying the threshold value would not seem appropriate; b) a selection bias by excluding homeless persons not registering for accommodation cannot be excluded. As mentioned above, the surveillance system in place did not detect cases occurring in homeless during and up to 3 months after the study which makes this unlikely; c) the high mobility of participants may have caused acute cases to be declared outside Geneva Canton and thus to not be recorded by the local surveillance system; and d) finally, we postulate that the risk of TB in homeless persons in Geneva differs from that prevailing in other settings because of a different demographic and social profile. Our cohort mostly included young migrants as opposed to older socially excluded autochthonous homeless screened in previous studies [5, 28]. Indeed, in the context of enhanced international movements and restrictive immigration policies in Western Europe, migrants represent a growing share of homeless persons in Western Europe [2, 5]. The profile of homeless people in Geneva reflects this trend with 78% being highly mobile foreigners without residency permit and limited access to care [29]. This finding is unexepected considering that migrants from medium and high-risk countries for TB now represent the population with the highest incidence of active TB in Europe. Further and longer observation periods will allow us to determine if the demographic changes occuring among the homeless population is associated with an increased incidence of active TB. This diversity supports the need to carefully assess the local social and health context and its constant evolution in regards to changes in internationality mobility pattern to define tailored rather than “one-fits-all” strategies.

In other European cities, active case-finding programs based on outreach activities led to a decline in number of cases and case clusters, demonstrating the public health value of such strategies among high risk homeless persons [28, 30]. In the current Geneva context, these results suggest that reinforcing TB health education in shelters, awareness of social workers about symptoms suggestive of TB, facilitating access to care irrespective of legal or insurance status, maintaining close cooperation between social services and reference health centers and close monitoring of the number of homeless among subjects treated for active TB at the reference center may be a reasonable alternative to a systematic screening strategy. Yet, given the constant evolution in migrant populations in Europe, close TB surveillance remains of major importance to rapidly respond to a change in local epidemiology.

Some groups have suggested screening for latent tuberculosis infection (LTBI) and proposed treatment for LTBI when indicated in homeless subjects [24]. In our area however, the homeless population is a very mobile group with a low probability of attending a TB clinic for 4 to 9 months for LTBI. Therefore, focus in our study was on active TB, not on LTBI, and the strategy followed that chosen by the Swiss Federal Office for Public Health for asylum seekers restricting screening to active cases.

Our study has several limitations. First, homelessness encompasses different groups with distinct housing, social and health needs [4]. Therefore, we cannot extend our findings to all people with unmet housing needs in Geneva. Moreover, homeless women were underrepresented and more data are needed to assess their TB risk. Second, we used a questionnaire which although promoted by our national health services, has not been formally prospectively evaluated in a comparative study. In spite of our efforts to identify false negative cases, we cannot exclude that some early clinically active TB cases with negative score at time of screening were missed given their rather short stay in the shelters because of the high mobility of participants. Finally, this study describes point prevalence upon admission in shelters and thus cannot reflect the longitudinal risk of TB in this population.