Transmission of tuberculosis and predictors of large clusters within three years in an urban setting in Tokyo, Japan: a population-based molecular epidemiological study

Abstract

ObjectiveMolecular epidemiology is a promising tool for understanding tuberculosis transmission dynamics but has not been sufficiently utilised in Asian countries including Japan. The aim of this study was to estimate the proportion of TB cases attributable to recent transmission and to identify risk factors of genotype clustering and the development of large clusters within 3 years in an urban setting in Japan.Design and settingLong-term cross-sectional observational study combining the characteristics of patients with culture-positive TB notified in Shinjuku City, Tokyo (2002–2013), with genotype data ofMycobacterium tuberculosis.Primary outcome measureGenotype clustering rate and association between genotype clustering status and explanatory variables.ResultsAmong 1025 cases, 515 were localised within 113 genotype clusters. The overall clustering rate was 39.2%. Significantly higher rates were found in patients aged <40 years (adjusted odds ratio (aOR)=1.73, 95% CI 1.23 to 2.44), native Japanese individuals (aOR=3.90, 95% CI 2.27 to 6.72), full-time workers (aOR=1.63, 95% CI 1.17 to 2.27), part-time/daily workers (aOR=2.20, 95% CI 1.35 to 3.58), individuals receiving public assistance (aOR=1.81, 95% CI 1.15 to 2.84) and homeless people (aOR=1.63, 95% CI 1.02 to 2.62). A significant predictor of large genotype clusters within 3 years was a registration interval ≤2 months between the first two cases in a cluster.ConclusionOur results indicated that a large proportion of patients with culture-positive TB were involved in the recent TB transmission chain. Foreign-born persons still have a limited impact on transmission in the Japanese urban setting. Intensified public health interventions, including the active case finding, need to focus on individuals with socioeconomic risk factors that are significantly associated with tuberculosis transmission and clusters with shorter registration intervals between the first two cases.