Healthcare-associated bacterial outbreaks are difficult to detect with routine phenotypic data, and incidence-based alerts generate false alarms. We present a method to estimate how close two patient isolates are in a transmission chain using antimicrobial resistance profiles. The pairwise transmission proximity (PTR) score combines two ideas: isolates are considered closer when their resistance profiles are more similar, and when that shared profile is uncommon in the sampled population. Practically, this is implemented by counting how many other isolates fall within the same phenotypic distance of each isolate and using these counts to rank pairs; the score is used as a relative proximity measure rather than a calibrated probability. Phenotypic distance is computed from minimum inhibitory concentrations (MICs) as the average absolute difference across drugs, or from binary susceptible/resistant profiles as an equivalent scaled mismatch count. We simulate pathogen divergence, resistance evolution, and patient-to-patient transmissions on evolutionary trees while varying transmission and mutation rates and the number of resistance traits. In closed-population simulations with complete sampling, the PTR ranking tracks time since common ancestry and provides useful discrimination for identifying inter-patient transmission events and for distinguishing inter-from intra-patient pairs, with better performance at moderate rates and larger resistance panels. The method is intended as a high-coverage screening signal to prioritize epidemiologic review or confirmatory genomics; calibration and clinical utility require external validation.
Peron et al. (Thu,) studied this question.