Date of Award

January 2011

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Gerald H. Friedland

Subject Area(s)

Medicine, Public health, South African studies


Healthcare-associated tuberculosis (TB) transmission occurs in resource-limited health facilities, putting patients and healthcare workers (HCWs) at risk. TB infection control (IC) can prevent such transmission but is inadequately implemented. We attempted to better characterize HCW TB IC implementation. We hypothesized that TB IC implementation would correlate in the manner predicted by the Information-Motivation-Behavioral skills (IMB) behavioral model. We conducted direct observations of TB IC behavior and staff questionnaires of TB IC IMB and behavior at two district hospitals (Church of Scotland Hospital, COSH, and Charles Johnson Memorial Hospital, CJM) in rural South Africa. Direct observations were conducted on 10-14 consecutive working days. Observed TB IC practices varied greatly by department (natural ventilation: 50.7-97.0%, respirator use: 5.0-100.0%). Questionnaires were completed by 123 HCWs at COSH and 75 at CJM. Information levels were generally high at both hospitals. Motivation responses were generally appropriate, though 29.4% would not be bothered "very much" by catching TB, and 22.8% thought TB IC was not worth the effort. Behavioral skills assessment indicated that HCWs found the majority of TB IC procedures to be easy to perform, though respondents highlighted several discrete tasks as being relatively difficult, especially those relating to personal HIV testing and relocation to low-risk departments if HIV-positive. When in high-risk TB areas of the hospital, more than half of respondents claimed to "always" wear a respirator (54.3%), instruct patients on cough hygiene (63.0%) and ensure effective natural ventilation (67.4%). Most (74.0%) knew their HIV status (81.0% at COSH, 63.8% at CJM, p=0.012). Correlations were noted between self-reported TB IC implementation and several IMB variables, particularly those related to social support. A social support Motivational sub-scale correlated with self-reported respirator use (p=0.002), cough hygiene instruction (p=0.001), and natural ventilation (p=0.006). A global model was created to compare IMB variables to aggregated self-reported TB IC behaviors. The only significant global scale variable was Motivation as a covariate of Behavioral skills (p<0.000). IMB models were created for self-reported respirator use, cough hygiene instruction to patients, natural ventilation implementation, and knowledge of personal HIV status. The respirator IMB model performed much better than the others. Information did not vary significantly with other variables in any of the models. Results suggest that rather than focusing on improving staff Information, efforts to increase TB IC implementation should focus on HCW Motivation and Behavioral skills development. TB IC implementation in this study compared favorably to other reports from the developing world. Social support, especially that of colleagues and supervisors, is an important element in ensuring better TB IC implementation, which is crucial to preventing healthcare-associated transmission. Though individual models require refinement, IMB modeling offers a promising avenue for further research and guiding interventions.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.