An illicit drug user was defined as a person who, at the time of diagnosis, regularly used heroin or cocaine, or was receiving methadone replacement therapy. A homeless person was defined as a person who frequently used day-care, night-care, residential, or other facilities for homeless persons.

All notified TB cases among illicit drug users with a registered address in Rotterdam and homeless persons residing in Rotterdam who were diagnosed between January 1, 1993, and December 31, 2005, were included in the study. To compare demographic and disease-related characteristics, all TB cases not belonging to these risk groups, and notified in the same period, were also studied. Information about the patients was extracted from the TB register of the Department of Tuberculosis Control and from individual patient records.

A targeted mobile TB screening program among illicit drug users and homeless persons in Rotterdam was reintroduced in May 2002, visiting an incomplete number of services and facilities in that year. Since January 1, 2003, all known facilities—that is, four methadone-dispensing centers, six day- or night-care facilities (some of them also having safe drug consumption rooms), three residential homes, and the street prostitution zone—were visited twice per year with an MDXU. The screening of the 14 facilities was performed during 12 working days per half-year, with an average of 102 persons per day, ranging between 36 and 319. The screening results were obtained from the Client Information System of the Department of Tuberculosis Control. The yield of the screening was cross-checked with individual patient records.

Since 1993, all M. tuberculosis isolates in the Netherlands have been subject to standardized insertion sequence (IS)6110-based restriction fragment length polymorphism (RFLP) typing, so-called DNA fingerprinting (7). Clusters are defined as groups of patients having isolates with fully identical RFLP patterns or, if strains harbor fewer than five IS6110 copies, with identical subtyping by use of the polymorphic GC-rich sequence probe (8). The mycobacterial DNA fingerprints of TB cases among illicit drug users and homeless persons were compared with those of other cases with the same risk profile in previous years. If the mycobacterial DNA clustered with a preceding case within a 2-year period, the disease was considered to be due to recent transmission related to illicit drug use or homelessness (9). Because of this 2-year time lag, only cases in the period 1995 to 2005 were included in the analysis of recent transmission.

We analyzed data with SPSS version 12 (SPSS, Inc., Chicago, IL). We calculated χ² trends for the proportion of TB cases among illicit drug users and homeless persons of the total caseload and the proportion of recently infected TB cases in these risk groups for three periods, that is, a first period from calendar years 1993 through 1996 and 1995 through 1996, respectively, with mobile screening, a second period from calendar years 1997 through 2001 without screening, and a third period from calendar years 2002 through 2005 with the MDXU screening program. For comparison of the proportion of smear-positive TB cases and cases found through active case finding among illicit drug users and homeless persons, the third period started in May 2002. We calculated the odds ratio of demographic and disease-related characteristics of TB cases among illicit drug users and homeless persons. Multivariate logistic regression was performed, considering factors significant in univariate analysis.

During the 13 years of study, 90 TB cases were notified as illicit drug users, 37 were homeless persons, and 79 were both illicit drugs user and homeless. These 206 cases represented 11.4% of the total TB caseload of 1,811 in Rotterdam (Table 1). The annual number of TB cases among illicit drug users and homeless persons doubled from an average of 12 in the years 1993 until 1998 to 23 in 2001. After the introduction of the mobile screening program the annual number and the proportion fell between 2002 and 2005 from 24 to 11 and 16.4 to 8.5% (p < 0.05), respectively. The annual notification rate decreased from 533 per 100,000 persons in 2002 to 244 per 100,000 persons in 2005, based on an estimated number of 4,500 illicit drug users and homeless persons in Rotterdam.

Table 2 compares demographic and disease-related characteristics of illicit drug users and homeless persons with TB and cases without this risk profile in Rotterdam between 1993 and 2005. In the multivariate analysis illicit drug users and homeless persons with TB were more often in the age group between 40 and 59 years, born in the Netherlands, and coinfected with human immunodeficiency virus (HIV). They more often had pulmonary disease with positive sputum or bronchoalveolar fluid smears for acid-fast bacilli. The proportion of smear-positive TB cases among illicit drug users and homeless persons was 55.3% (26 of 47), 58.0% (51 of 88), and 47.9% (34 of 71) during the three periods. The proportion of smear positivity during the MDXU screening program decreased compared with the nonscreening years, although the decrease was not statistically significant (p = 0.11). The M. tuberculosis strains of illicit drug users and homeless persons clustered more frequently. TB cases among illicit drug users and homeless persons were more often identified through active case finding, such as contact investigation and screening.

During the 3 years and 8 months of the MDXU screening program, 8,559 chest X-rays were taken of 3,248 individuals (Table 3). The total yield of the screening was 28 TB cases, of which 12 cases were smear positive and 27 cases were culture confirmed. This translated into a prevalence rate of 327 cases per 100,000 radiographs. In the same period, another 7 TB cases among illicit drug users or homeless persons were identified through contact investigation and 7 via the TB screening program in prisons, resulting in 42 of a total of 71 TB cases (59.2%) among these risk groups found through active case finding during the time of the MDXU screening program. In the first 4 years of the study period, 15 of a total of 47 TB cases (31.9%) were found through active case finding, of which 9 cases were found via mobile screening of these risk groups, 2 cases through contact tracing, and 4 cases via the TB screening program in prisons. During the nonscreening period, 26 of a total of 88 cases (29.5%) were actively traced, of which 17 cases were found through contact investigation and 9 via the TB screening program in prisons. The difference in the proportion of cases identified through active case finding during the MDXU screening program and during the nonscreening years was highly significant (p < 0.001).

During the study period 1993–2005, initial hospitalization was required for two-thirds of the TB cases among illicit drug users and homeless persons, because of the high proportion of infectious cases. Fourteen patients (6.8%) were detained in a designated hospital for a mean period of 10 weeks (3–26 wk), because they did not comply with isolation procedures during the infectious period. Other aspects of case holding are addressed in the discussion. TB cases among illicit drug users and homeless persons had a significant lower treatment completion rate than did cases without the risk profile (79.1 vs. 86.8%, unadjusted odds ratio, 0.6 [95% confidence interval, 0.4–0.8]; p < 0.05). However, 25 of the 28 cases (89.2%) identified through the mobile radiographic screening program completed treatment.

Table 4 gives cluster information of the mycobacterial strains of the culture-confirmed TB cases among illicit drug users or homeless persons in Rotterdam over the period 1993–2005. In 31 cases TB was caused by mycobacteria with an unique DNA pattern in the Netherlands, of which 9 were the first case of a cluster; 3 of these clusters (clusters 13, 19, and 30) also included other illicit drug users or homeless persons with TB in Rotterdam. In the remaining 156 TB cases (83.4%) the mycobacteria belonged to 27 different clusters. Four mycobacterial strains (clusters 1, 13, 17, and 19) were the causative microorganism in 62.8% (117 of 187) of the culture-confirmed cases with an RFLP.

In 1997–2002, more than 80% of the illicit drug users or homeless persons with TB were infected with one of the M. tuberculosis strains prevalent among these risk groups within 2 years before diagnosis (Figure 1). The proportion of prevalent strains declined from 82% in 2002 at the start of the screening program to 45% in 2005 (p < 0.05).

Figure 1. Annual proportion of tuberculosis cases among illicit drug users and homeless persons in Rotterdam with identical restriction fragment length polymorphism with cases in these risk groups 2 years before diagnosis. Fisher exact test for trend first (screening) period (1995–1996): p = 0.22; χ² test for trend second (nonscreening) period (1997–2001): p = 0.77; χ² test for trend third (MDXU screening) period (2002–2005): p = 0.03.

Download Figure | Download Powerpoint

[More] [Minimize]

Figure 2 shows details of cluster 17, the largest TB cluster in the Netherlands with 150 cases in December 31, 2005. Initially it contained only TB cases with the same ethnic background and all residing in Rotterdam (13 cases in 1996–1997). One of them was an illicit drug user who was diagnosed at the end of 1996 with extensive disease. In the years 1998–2001 another 76 cases were added to this cluster, among them 34 cases from Rotterdam with a drug addiction or homeless status (46%) and also documented secondary cases among family members, and medical and social staff involved in the care for these persons. The number of cases among illicit drug users and homeless persons in this cluster declined during the screening years 2002–2005, with only one case during the last 2 years. The growth of cluster 17 continued in 2004 and 2005 due to pub-related transmission among migrants with the same ethnic background as the initial cases of the cluster.

Figure 2. Epidemic curve of tuberculosis cases in cluster 17 in the Netherlands by year, residential area, and risk factors. Column patterns: solid, illicit drug user or homeless person in Rotterdam; hatched, resident of Rotterdam, not an illicit drug user or homeless; open, patient residing elsewhere in the Netherlands.

Download Figure | Download Powerpoint

[More] [Minimize]

Outbreak reports of TB among residents of homeless shelters in large cities in Europe and the United States have cited high TB incidence rates ranging from 250 to more than 1,000 per 100,000 persons (9–14). In one review homeless TB cases represented between 6.1 and 6.7% of all cases between 1994 and 2003 in the United States (15). At the height of the outbreak in Rotterdam, one of six TB cases (16.4%) used illicit drugs or was homeless at the time of diagnosis. We have combined these two risk factors because they were strongly interrelated, although illicit drug use was more dominant.

Homelessness and drug addiction are important risk factors for TB because transmission is facilitated by late presentation, common to marginalized care avoiders, crowding, and poor ventilation of shelters or safe drug consumption rooms. In addition, interrelated factors such as HIV infection, malnutrition, and alcohol and drug abuse increase the risk of progression to active disease once infected (16–20). In our study cohort, HIV coinfection was three times more common among illicit drug users and homeless persons with TB than among cases without these risk factors.

Illicit drug users or homeless persons with TB in Rotterdam were more often born in the Netherlands than were cases without these risk factors, and DNA fingerprinting showed that the majority of cases among these risk groups clustered. This underpins the notion that the increase in the number of illicit drug users and homeless persons with TB was due to intensive and prolonged transmission in Rotterdam.

Illicit drug users or homeless persons with TB more often had disease located in the lungs, which more frequently had reached an infectious stage with positive sputum or bronchoalveolar lavage fluid smears after direct microscopy. They represented 19.3% (111 of 576) of all infectious cases in Rotterdam during the study period. DNA fingerprinting analysis, including all cases from the general population, showed that apart from other illicit drug users or homeless persons, also workers in close contact with the risk groups, such as shelter staff, social and health care workers and volunteers, as well as family members, had developed TB caused by M. tuberculosis strains prevalent among these risk groups.

Systematic chest radiographic screening for TB, especially with M(D)XUs, has been reported in a limited number of outbreaks of TB among homeless persons (21–23). In the Netherlands, MDXUs were already used for several other screening activities, so they could be easily used for the screening of illicit drug users and homeless persons. We considered screening with an MDXU the most appropriate intervention for active case finding because of the urgent public health situation and previous findings that asymptomatic infectious cases may persist undiagnosed for a long time in the targeted community (6). In addition, symptomatic cases may report late because of limited access to care and to patient delay (24). Alternative interventions to control TB among homeless persons, such as symptom screening, spot sputum screening, treatment of latent TB infections in HIV-infected populations, and environmental control activities have been applied (9, 10, 25, 26), although the success of these approaches has been questioned (20).

The screening resulted in a prevalence rate of 741 per 100,000 radiographs in the first year, but declined after some years of systematic screening, suggesting that the epidemic was being controlled. The total yield was 327 TB cases per 100,000 radiographs, which is high compared with the outcomes of other TB screening programs in the Netherlands, such as entry screening of immigrants (111 per 100,000 radiographs) or prisoners (72 per 100,000 radiographs) (5). Our findings were similar to the result of radiographic screening of opiate users in Amsterdam in the 1990s of the last century (400 per 100,000 radiographs) (11). Radiographic screening detects only active intrathoracic TB cases. Additional interventions, such as screening for latent TB infections, might be of use in a phase of a controlled epidemic to identify cases at risk and select candidates for preventive treatment, although low completion rates have been reported among homeless persons (27).

Apart from active case-finding activities, completion of treatment is essential to control TB and prevent new cases (28). High completion rates can be achieved by directly observed therapy, in which patients take their medication under supervision (28, 29). In Rotterdam, all illicit drug users and homeless persons with TB have a directly observed therapy indication, and as a result of the screening program additional human and other resources were needed to maximize case holding. Incentives such as public transport tickets, priority accommodation in shelters, voluntary admission to specialized TB hospitals, or assistance in applying for temporary residence permits also contributed to the high treatment completion rate. The use of detention as a last resort to isolate noncompliant infectious patients and complete treatment has been successfully applied in other programs to control tuberculosis (30).

DNA fingerprint studies, most of them based on RFLP, combined with conventional epidemiologic information, have increased our knowledge of TB transmission (10, 31–34). Community studies demonstrated unidentified outbreaks of TB and underlined the importance of ongoing transmission, also in low-prevalence countries (33, 34). Molecular studies have pointed out the limitations of standard contact-tracing procedures to identify or prevent new cases and have showed the contribution of DNA fingerprinting to surveillance and the evaluation of TB control programs (33–35). For adequate interpretation and comparison, molecular studies should ideally involve a high proportion of cases in a population, be conducted in conjunction with conventional epidemiologic investigations, provide information about patient characteristics, and report the time period of the study (36). In the Netherlands, a comprehensive database of DNA fingerprints of nearly all culture-confirmed TB cases over more than 10 years is available (37–39). Together with detailed epidemiologic information from the patient records the present study fulfills the above-cited criteria.

DNA fingerprinting has been used to document clonal relationships in shelter-associated outbreaks over long time periods (9, 40, 41). In our study, the four most prevalent mycobacterial strains continued to reoccur in the illicit drug–using or homeless population for 8 to 13 years. It is unlikely that these cases were due to a single point source, because 54% (data not shown) of the illicit drug–using or homeless cases in these clusters were infectious. Ongoing transmission from multiple sources among illicit drug users or homeless persons sustained the outbreak for many years. The growth of one of these clusters clearly illustrated the epidemic spread both inside and outside these risk groups. DNA fingerprint analysis also revealed that isoniazid-resistant strains were introduced in the homeless community in Rotterdam, leading to a secondary case and creating the potential for new and more complicated outbreaks, as documented elsewhere (42).

The proportions of TB cases among illicit drug users and homeless persons with prevalent strains varied between 80 and 90% at the height of the epidemic in Rotterdam and were higher than the reported proportions in Denver (49%), Los Angeles (53%), San Francisco (60%), New York (60%), or Budapest (70%) (9, 10, 13, 31, 43). One of these studies also measured the effect of an intervention on the proportion of prevalent strains, and showed a decrease from 49 to 14% after a mandatory screening program using symptom screening and tuberculin skin testing was implemented (9). During the mobile radiographic screening program in Rotterdam, which was not mandatory but used an opting-out strategy and strong persuasion of clients to comply with screening, the proportion of TB cases among illicit drug users and homeless persons with prevalent strains declined from 82% in 2002 to 45% in 2005.

After the introduction of systematic and targeted TB screening among illicit drug users and homeless persons with mobile digital X-ray units in Rotterdam, the annual number of notified TB cases among these risk groups decreased and transmission fell. This study demonstrates that DNA fingerprinting can be a useful tool to evaluate the impact of a TB screening program. Although the yield declined after some years of systematic screening, we advocate that screening of illicit drug users and homeless persons in Rotterdam should be continued to prevent a resurgence of TB as was experienced after discontinuing a screening program at the end of the last century.