Social Contact Patterns Associated with Multidrug-Resistant Tuberculosis

Abstract

Background: Multidrug-resistant tuberculosis (MDR-TB), a mutated strain of Mycobacterium tuberculosis (Mtb) resistant to first line treatment, threatens to derail decades of progress in global TB control. Social contact patterns, defined as the frequency, duration, nature and locations of daily social interactions are known to influence the epidemiology and transmission risk of infectious diseases. The goal of this dissertation was to identify social contact patterns associated with MDR-TB in three domains: 1) exposure to community venues, 2) regular social contacts, and 3) activity spaces. It was hypothesized that greater exposure (duration) to community venues, greater exposure to non-household social contacts, and greater activity space overlap was associated with MDR-TB risk. Methods: A case-control study was designed to address research hypotheses. Study participants were recruited from the areas of Callao and Lima Sur located in to the north and south of Lima, Peru, respectively. Cases were lab-confirmed MDR-TB patients that submitted sputum samples for whole genome sequencing. Genotyping was competed to identify genetic clusters (Mtb strains of ≤5 SNP difference). Controls were health community members with no history of TB residing in case neighborhoods. Participants completed venue-tracing interviews, social contact-tracing interviews and seven days of continuous GPS monitoring. Logistic regression was used to estimate association between social contact patterns and MDR-TB disease. Social network analysis was used to visualize place-affiliation social networks of cases and controls and to identify epidemiologic venue links between cases. Place-affiliation social networks were constructed from place-tracing interviews and modeled using an exponential random graph model (ERGM). Spatial ecology methods were used to construct activity spaces from GPS data and the utilization distribution overlap index (UDOI) was used to estimate the association between activity space overlap and MDR-TB disease. Results: 59 cases and 65 controls completed the place-tracing and contact-tracing interviews and a smaller subset (35 cases and 64 controls) completed GPS data collection. In total, 729 unique community venues and 1,427 social contacts were reported. Cases reported spending more person-time (hours) in healthcare and transportation venues than controls (P<0.05). Person-time in healthcare (Odds Ratio (OR)=1.67, P=0.01), education (OR=1.53, P<0.01), and transportation venues (OR=1.25, P=0.03) was associated with MDR-TB disease. Healthcare, market, and transportation venues were commonly shared among clustered cases. Case networks more densely connected than controls. The final ERGM indicated significant community segregation between cases and controls. Cases reported significantly more contact time with family members while controls reported more contact time with work-related contacts. Cases had more contact time in their homes, in bars and restaurants. Controls reported more contact time in workplace settings, neighborhoods, and churches. Contact time with family members (OR=1.03, CI=1.00-1.05, P=0.02) and persons with TB (OR=1.40, CI=1.17-1.68, P<0.001) was associated with MDR-TB disease, but no specific contact location was implicated. In general, cases reported significantly more prolonged contact time per contact than controls with all contact types and contact locations. Activity spaces of controls (median=35.6km2, IQR=25.1-54) were significantly larger than cases (median=21.3 km2, IQR=17.9-48.6) (P=0.02). Activity space overlap between controls (mean UDOI=0.31, sd=0.47) was greater than overlap between cases (mean UDOI=0.17, sd=0.34); overlap between clustered cases (mean UDOI=0.63, sd=0.67) was highest in all regions. There was a statistically significant association between activity space overlap and genetic similarity of Mtb strains; genetic similarity of Mtb strains increased as spatial overlap increased (P<0.001). The odds of MDR-TB cases being clustered increased by 22% per 0.10 increase in UDOI (OR=1.22, CI=1.09-1.36, P<0.001). Conclusions: Significant segregation between how cases and controls interacted in community venues was identified and spatial overlap between cases and controls was low. This finding suggests the need to focus infection control interventions across the range of congregate community places utilized by high-risk communities, in order to have maximum impact on controlling disease spread. Conversely, infection control interventions in community settings only frequented by healthy or low-risk communities may be less effective. Since MDR-TB cases tended to have fewer contacts, but more prolonged social contacts than controls, this is likely a key factor driving MDR-TB transmission in this population. Case finding may be enhanced by focusing efforts on prolonged contacts occurring in non-household settings. Activity space overlap was strongly associated with genetic clustering of MDR-TB cases; MDR-TB transmission events are likely occurring in concentrated areas of regular case activity. Contact tracing and case finding efforts should focus on areas of concentrated activity and spatial overlap among MDR-TB cases.