A new method to estimate factors associated with tuberculosis transmission using whole genome sequencing and other additional metadata
Anne N. Shapiro, ChuanChin Huang, Meredith B. Brooks, Samantha Malatesta, Leonid Lecca, Mercedes C. Becerra, Roger I. Calderon, Carmen C. Contreras, Judith Jimenez, Rosa Yataco, Zibiao Zhang, Megan B. Murray, Helen E. Jenkins, Laura F. WhiteAbstract
Background
Understanding tuberculosis (TB) transmission dynamics is necessary to interrupt disease spread. We developed a model to estimate adjusted odds ratios (ORs) for factors associated with genetic relatedness, as proxy for transmission.
Methods
We build upon an existing iterative model that modifies genetically linked tuberculosis case data to better represent true transmission links. We incorporate bootstrapped logistic regression to calculate adjusted ORs with confidence intervals that account for correlation from individuals present across multiple transmission pairs. We assess model performance with simulation studies and apply the method to cohort data from Lima, Peru.
Results
Iterative algorithm estimates resembled those from logistic regression but had larger confidence intervals, reflecting the data correlation adjustment. Transmission pairs where at least one member was >34 years had decreased transmission odds. Pairs with at least one incarcerated or male member had increased adjusted transmission odds.
Conclusions
We produce adjusted ORs accounting for the correlation of pairwise genetic relatedness data. These ORs are an accurate proxy for the association between covariates and transmission and further our understanding of factors associated with tuberculosis transmission.