Transmission classification model to determine place and time of infection of tuberculosis cases in an urban area.

We conducted a population-based study in the Rotterdam region of The Netherlands to determine the place and time of infection of tuberculosis (TB) cases using conventional epidemiological and genotyping information. In particular, we focused on the extent of misclassification if genotyping was not combined with epidemiological information. Cases were divided into those with a unique mycobacterial DNA fingerprint, a clustering fingerprint, and an unknown fingerprint. We developed transmission classification trees for each category to determine whether patients were infected in a foreign country or recently (<or=2 years) or remotely (>2 years) infected in The Netherlands. Of all TB cases during the 12-year study period, 38% were infected in a foreign country, 36% resulted from recent transmission in The Netherlands, and 18% resulted from remote infection in The Netherlands, while in the remaining cases (9%) either the time or place of infection could not be determined. The conventional epidemiological data suggested that at least 29% of clustered cases were not part of recent chains of transmission. Cases with unknown fingerprints, almost all culture negative, relatively frequently had confirmed epidemiological links with a recent pulmonary TB case in The Netherlands and were more often identified by contact tracing. Our findings highlight the idea that genotyping should be combined with conventional epidemiological investigation to establish the place and time of infection of TB cases as accurately as possible. A standardized way of classifying TB into recently, remotely, and foreign-acquired disease provides indicators for surveillance and TB control program performance that can be used to decide on interventions and allocation of resources.