We acknowledge several potential limitations to this
study. Data for this study were abstracted from occupational
health clinic employee medical records and limited by the
quality of the records. Due to missing data of interest, we
excluded nearly 8% of medical charts. It is likely that some
HCWs sought and received care outside of the hospital at
private providers or hospitals and did not report their case
of TB to the occupational health nurse and therefore, the incidence
of TB in this sample may be underestimated. Similarly,
a large proportion (74%) of HCWs did not have their
HIV status documented in their medical charts suggesting
that the HIV prevalence in this population may be underestimated.
Many HCWs (15%), especially clinical staff, worked
in more than one ward during the study period and this
made it difficult to pinpoint where TB exposure may have
occurred. However, our analysis included a multilevel multivariate
model controlling for multiple wards to address this
concern. It should be noted that the comparison of TB incidence
rates among HCWs in this sample with the general
population of KwaZulu-Natal is a crude comparison and
does not account for potential confounding.