A majority of the respondents were found to score “good” on TBIC knowledge and practice. In addition to the good overall knowledge about TBIC, the respondents also demonstrated good knowledge about the need for an infection prevention (IP) committee to implement TBIC activities, window opening, patient isolation, and patient education. Furthermore, most of the respondents saw the need to prioritize TB patients to minimize hospital stay and the need for healthcare worker screening to quickly detect TB infection. These positive findings may be attributable to the dissemination of TBIC guidelines, trainings, and supportive supervisions by the national and regional TB control programs as well as the support from nongovernmental organizations working on TBIC.
The lower knowledge level regarding respirators and the use of fans may be attributable to unavailability of this equipment in the respective health facilities. Those who use respirators and fans probably received them from organizations working in partnership with the government on TBIC.
The proportion of TBIC knowledge questions answered correctly for all participating health professionals was 70%. On this measure, health workers’ knowledge was higher than that reported from the United States (median 55%) . Eighty-six percent of the respondents in our study knew that educating patients on cough etiquette is important. This finding is in line with a report from the United States . Knowledge levels in our study that are similar to or higher than those reported by healthcare workers in the United States could be explained by the high TB patient load in the study hospitals: The frequent exposure respondents in our study have to TB cases provides frequent opportunities to learn about TB treatment and prevention.
Of the health professionals in our study, 74.4% scored “good” knowledge in TBIC. This was better than rates reported from Russia (“overall scores were low”)  and the Philippines (misunderstanding of TB infection transmission despite high magnitude of TB in the general population) . On the other hand, proportions of healthcare workers with “good” knowledge from Iraq (95.5%) , Saudi Arabia (81.8%), and Bangkok (85%) are higher than ours ,. Explaining this knowledge difference is generally difficult; however, variations in the rates of “good” knowledge across countries could be attributed to the different settings and methodologies used by different studies.
TBIC knowledge was not found in our study to have any association with years of service, unlike the findings from Iraq . Only 18.8% of the respondents were found trained on TBIC. This is contradictory to the recommendation on giving training on TBIC before the initial assignment of each health worker to any unit in the health facilities .
Training was found to have a statistically significant association with TBIC knowledge--OR 3.631 and 95% CI (1.496, 8.813). This was not the case in studies from the United States (on medical residents)  and Australia (on new graduate nurses) . We found no significant differences in knowledge by category of healthcare workers, whereas in a study of Russian health workers physicians scored significantly higher than nurses and laboratory staff . In our study a higher proportion of the respondents (91.4%) trained on TBIC responded as having contact with TB patients/suspected cases, while a study in South Africa found the opposite .
Our findings also revealed that only 34.2% of the healthcare professionals felt that a respirator should be worn to prevent TB infection transmission; this is far below what is reported from the United States (65% of those with no TB patient contact and 88% of those with patient contact) . The unavailability of respirators in our context is the most likely reason for the reportedly low levels of this practice.
While training on TBIC was significantly associated with window opening and getting tested for TB when exposed to TB, it was not found to be a predictor of the other specific practice questions or of overall TBIC practice. This finding is contradictory to the report from Australia . The lack of association between training and practice may be due to an emphasis on the theoretical aspects of training rather than skill-based components. Similarly, though we found statistically significant differences in window opening practices between those trained on TBIC and those not, OR 1.9 and 95% CI (1.02, 3.75), there were no significant differences in other standard IP practices (educating patients on cough etiquette, giving priority to suspected cases of TB) by training. This indicates that the basic minimum practice for healthcare facilities caring for patients suspected of air born diseases (including TB) is not in line with recommendations ,. The Ethiopian TBIC guidelines clearly state that healthcare providers should triage and fast-track TB suspected cases, educate on cough etiquette, improve the cross ventilation of the room by opening windows, adjust seating arrangements, isolate TB suspected cases in the waiting area and wards, provide ambulatory management, conduct routine TB screening, follow up on TBIC activity implementation by the IP committee, and use respirators and fans to minimize TB infection .
Overall, 63.3% of the respondents scored “good” practice, where the median percent of TBIC practice questions correctly answered was 50%. This is much better than what is reported from Iraq (38.2%) . However, a study from Bangkok revealed that all of the IP committee members and 85% of the hospital personnel attempted to implement TBIC measures .
Concerning specific practices, 21.8%, 32.3%, 44.2%, 40.8%, 30.2%, and 39.4% of the health professionals working in the outpatient department (OPD), TB/HIV clinic, medical ward, other wards, laboratory, and pharmacy, respectively, prioritize TB patients. This is much lower than what is observed from Bangkok . Mask usage is reported by only 21.1% of the respondents, which is much lower than that reported from Bangkok  and South Africa . The difference in these findings may be attributed to the unavailability of the masks. Window opening is practiced by the majority (64.9%) of respondents; this is much better than findings from the Philippines (39%) . Nurses, pharmacy health professionals, and laboratory professionals are found to be better than physicians and health officers in giving priority for TB patients; however, it is not statistically significant, unlike the finding in Saudi Arabia .
In contrast to the high scores under each knowledge question, a majority of the health professionals did not demonstrate appropriate TBIC practice; only 21% use masks, 33.5% prioritize TB patients, and 39.9% use a fan (ventilator). The discrepancy between healthcare workers’ knowledge on the one hand and practice on the other is probably attributed to shortage or unavailability of supplies like fans (ventilators) and respirators (masks). Hence, health care workers have not had opportunities to put their knowledge into practice in an effective way.
Working in the wards, pharmacy, and laboratory were significant predictors of good practice compared to working in the outpatient department. This could probably be due to exposure of suspected or confirmed cases of TB patients coming to each of these units. TB suspects or cases come to these units for inpatient management, to get a sputum examination done, and to collect drugs for respiratory tract infections with documents indicating TB. These documents (patient chart, sputum laboratory request paper, and prescription papers) bring the issue of TB infection transmission to the attention of the healthcare workers in these units and alert them to be vigilant in IP. On the other hand, patients come to the outpatient clinics with numerous complaints, and only after a full clinical evaluation will suspicion of TB come to mind. By then, it is already too late to implement good TBIC measures.
The findings of this research were in line with what one may expect under each TBIC practice question: Health professionals with good TBIC knowledge were more likely to implement the TBIC practices compared to those with poor knowledge, OR 10.6 and 95% CI (5.8, 19.3).
In this study, though univariate regression showed mulitple predictor variables, AOR revealed that training is the only statistically significant determinant of TBIC knowledge. Similarly, multivariate logistic regression revealed that good TBIC knowledge is a strong determinant for good TBIC practice.
Limitations of the study
There is limitated reference material available to compare our study with other studies done in similar settings.
Observing practices may produce more accurate results than asking about practices in a questionnaire, but this study did not include observation as a data collection method.