M L Munang1, C Browne1, J T Evans2, E G Smith3, P M Hawkey4, S B Welch1, H Kaur1, M J Dedicoat1. 1. Birmingham and Solihull TB Service, Heart of England NHS Foundation Trust, Birmingham, UK. 2. Wales Centre for Mycobacteria, Public Health Wales, Penarth, UK. 3. Midlands Regional Centre for Mycobacteriology, Public Health England, Birmingham, UK. 4. Midlands Regional Centre for Mycobacteriology, Public Health England, Birmingham, UK; University of Birmingham Medical School, Birmingham, UK.
Abstract
SETTING: Birmingham, United Kingdom, 2010-2014. OBJECTIVE: To investigate predictors for clustering of tuberculosis (TB) cases and cluster size and to evaluate the impact of cluster investigation using social network data. DESIGN: Retrospective observational cohort study. Prioritised cases linked using 24-locus mycobacterial interspersed repetitive units-variable number of tandem repeats (MIRU-VNTR) were interviewed using a social network approach to find epidemiological links. RESULTS: Of 2055 TB cases notified, 56% could be typed. Clustering was associated with younger age, UK birth, Black Caribbean ethnicity, social risk factors, pulmonary TB and negative human immunodeficiency virus status. Only UK birth and presence of more than one social risk factor were associated with larger cluster size, while drug resistance was associated with smaller cluster size. Social network data from 139/431 clustered cases found new epidemiological links in 11/19 clusters with ⩾5 members (undirected median network density 0.09, interquartile range 0.05-0.4). Ninety-eight additional contacts were assessed, with one case of active TB and 24 with latent tuberculous infection diagnosed. CONCLUSION: A social network approach increased knowledge of likely transmission events, but few additional TB cases were diagnosed. Obtaining social network data for all typed and untyped TB cases may improve contact tracing and reduce unexpected transmission detected from molecular data.
SETTING: Birmingham, United Kingdom, 2010-2014. OBJECTIVE: To investigate predictors for clustering of tuberculosis (TB) cases and cluster size and to evaluate the impact of cluster investigation using social network data. DESIGN: Retrospective observational cohort study. Prioritised cases linked using 24-locus mycobacterial interspersed repetitive units-variable number of tandem repeats (MIRU-VNTR) were interviewed using a social network approach to find epidemiological links. RESULTS: Of 2055 TB cases notified, 56% could be typed. Clustering was associated with younger age, UK birth, Black Caribbean ethnicity, social risk factors, pulmonary TB and negative human immunodeficiency virus status. Only UK birth and presence of more than one social risk factor were associated with larger cluster size, while drug resistance was associated with smaller cluster size. Social network data from 139/431 clustered cases found new epidemiological links in 11/19 clusters with ⩾5 members (undirected median network density 0.09, interquartile range 0.05-0.4). Ninety-eight additional contacts were assessed, with one case of active TB and 24 with latent tuberculous infection diagnosed. CONCLUSION: A social network approach increased knowledge of likely transmission events, but few additional TB cases were diagnosed. Obtaining social network data for all typed and untyped TB cases may improve contact tracing and reduce unexpected transmission detected from molecular data.