We conducted a population-based molecular epidemiological study of tuberculosis (TB) in Madrid, Spain (2002 to 2004), to define transmission patterns and factors associated with clustering. We particularly focused on examining how the increase in TB cases among immigrants in recent years (2.8% in 1997 to 1999 to 36.2% during the current study) was modifying transmission patterns. Mycobacterium tuberculosis isolates obtained from patients living in nine districts of Madrid (1,459,232 inhabitants) were genotyped. The TB case rate among foreign-born people was three to four times that of Spanish-born people, and the median time from arrival to the onset of treatment was 22.4 months. During the study period, 227 (36.3%) patients were grouped in 64 clusters, and 115 (50.7%) of them were in 21 clusters with mixed Spanish-born and foreign-born patients. Three of the 21 mixed clusters accounted for 21.1% of clustered patients. Twenty-two of 38 (57.9%) immigrants in mixed clusters were infected with TB strains that had already been identified in the native population in 1997 to 1999, including the three most prevalent strains. Factors identified as independent predictors of clustering were homelessness (odds ratio [OR], 2.3; 95% confidence interval [95% CI], 1.2 to 4.5; P ‫؍‬ 0.011) and to be born in Spain (OR, 1.8; 95% CI, 1.2 to 2.6; P ‫؍‬ 0.002). The results indicated that (i) TB transmission was higher in Spanish-born people, associated mainly with homelessness, (ii) that foreign-born people were much less likely to be clustered, suggesting a higher percentage of infection before arriving in Spain, and (iii) that an extensive transmission between Spanish-and foreign-born populations, caused mainly by autochthonous strains, was taking place in Madrid.

In recent years, the number of cases of tuberculosis (TB) among immigrants in Spain has increased markedly, and led to this analysis of the recent transmission patterns of TB in the immigrant population in Madrid. The countries from which the highest number of immigrant cases have been reported were Ecuador (21%), Romania (16%), Morocco (12%), Peru (11%) and Bolivia (9%). Fifty-one per cent of the cases were from South America. In a multicentre study (2004-2006), IS6110 restriction fragment length polymorphism and spoligotyping were used to genotype the Mycobacterium tuberculosis isolates from 632 immigrant cases from 47 countries. A total of 183 cases (29%) were grouped into 59 clusters, which are markers of potential transmission events. Most of the clusters (81%) included patients living in different healthcare districts, and 54% of the clusters were multinational. When a sample of 478 autochthonous cases was included, 53% of the clusters involving immigrants also included autochthonous cases. This study revealed marked transmission permeability among nationalities and between the immigrant and the autochthonous populations.

Tuberculosis cases infected by the same Mycobacterium tuberculosis (MTB) strain are considered to be clustered and involved in a transmission chain. Large clusters are assumed to represent active transmission chains in a population. In the present study, we focused on the analysis of large clusters defined by IS6110-restriction fragment length polymorphism (RFLP) typing in the immigrant population in Madrid. We identified 12 large clusters (involving 43% of the isolates) comprising 4-23 representatives. We proposed a gradient of epidemiological certainty for these large clusters. For a cluster to be considered robust and a good indicator of recent transmission, the MTB strain involved should not have been identified in a geographically and epidemiologically unrelated population and the cluster had to be re-confirmed by another highly discriminative molecular marker (MIRU-VNTR). The clusters that we discovered were classified into three categories: high, intermediate and low expected epidemiological value. In the largest cluster in the study (cluster M6; 23 representatives), failures by both criteria were identified: the representative seven-band RFLP pattern was also the most prevalent in the unrelated population (25 cases) and the cluster was fully split by MIRU-15, suggesting a lack of epidemiological value. The RFLP pattern representative of this cluster was also identified in 64 isolates from five countries in the Latin American genotype database, and again proved to be heterogeneous according to the MIRU-15 analysis. Specific analysis of large clusters, combined with the application of criteria for evaluating their robustness, could help identify uninformative clusters and target epidemiological resources towards those clusters with higher expected epidemiological value.