Factors associated with differences between conventional contact tracing and molecular epidemiology in study of tuberculosis transmission and analysis in the city of Barcelona, Spain.

The aim of this study was to analyze the factors associated with conventional contact tracing (CCT) and molecular epidemiology (ME) methods in assessing tuberculosis (TB) transmission, comparing the populations studied and the epidemiological links established by both methods. Data were obtained from TB case and CCT registries, and ME was performed using IS6110-based restriction fragment length polymorphism (RFLP) analysis and mycobacterial interspersed repetitive unit 12 (MIRU12) typing as a secondary typing method. During two years (2003 and 2004), 892 cases of TB were reported, of which 687 (77%) were confirmed by culture. RFLP analysis was performed with 463 (67.4%) of the 687 isolated strains, and MIRU12 types in 75 strains were evaluated; 280 strains (60.5%) had a unique RFLP pattern, and 183 (39.5%) shared patterns, grouping into 65 clusters. CCT of 613 (68.7%) of 892 cases detected 44 clusters involving 101 patients. The results of both CCT and ME methods yielded 96 clusters involving 255 patients. The household link was the one most frequently identified by CCT (corresponding to 80.7% of the cases clustered by this method), whereas nonhousehold and unknown links were associated with 94.1% of the strains clustered by ME. When both methods were used in 351 cases (39.3%), they showed the same results in 214 cases (61%). Of the remainder, 106 (30.2%) were clustered only by ME, 19 (5.5%) were clustered only by CCT, and 12 (3.4%) were clustered by both methods but into different clusters. Patients with factors potentially associated with social problems were less frequently studied by CCT (P = 0.002), whereas patients of <15 years of age, most with negative cultures, were less frequently studied by ME (P = 0.005). Significant differences in the populations studied by ME versus CCT were observed, possibly explaining the scarce correlation found between the results of these methods. Moreover, ME allowed the detection of nonhousehold contact relationships, whereas CCT was more useful for tracing transmission chains involving patients of <15 years of age. In conclusion, the two methods are complementary, suggesting the need to improve the methodology of contact study protocols.