Can workplaces be predictors for recent onset latent tuberculosis in health care workers?

Objective To study the association of workplaces and recent onset latent tuberculosis (LTB) in health care workers (HCW). Methods A case-control study was conducted at Srinagarind Hospital, Khon Kaen University, Thailand. We recruited HCW who had results of tuberculin test within 2 consecutive years from 2001–2008 and also had fixed workplaces (working hours more than 40 hours/week). Cases were subjects with tuberculin conversion, while controls were subjects with negative results of tuberculin test in two consecutive years. Tuberculin conversion was defined if a subject had a negative baseline tuberculin test and a positive tuberculin test in the next consecutive years. Baseline characteristics, workplaces (office, in-patient unit, out-patient unit, intensive care, operating room, and laboratory unit), tuberculosis related variables, and prevention strategies were studied. Multiple logistic regression analysis was used to identify predictors for tuberculin conversion. Results There were 624 subjects who met the criteria and 163 subjects had tuberculin conversion (26.1%; case group). The median age and male/female ratio of both groups were 39 years old and about 1:4. The cases group had higher percentage of subjects who worked at in- and out-patient department (30.7 vs 20.2 and 17.2 vs 12.2, respectively), had history of tuberculosis exposure in the past year (32.1 vs 16.1), and had history of prevention by any method and by surgical mask (49.4 vs 37.0 and 54.3 vs 38.3, respectively). Workings at in- and out-patient unit and history of tuberculosis exposure in the past year were significant predictors for tuberculin conversion (adjusted odds ratio and [95% confidence interval] of 1.99 [1.25–3.17], 1.91 [1.10–3.17], and 2.26 [1.47–4.96], respectively). Summary Workplaces in health care facilities do increase risks of LTB in HCW, particularly in in- and out-patient unit. Policy development regarding tuberculosis infection control programs focused on workplace prevention in health care facilities in Thailand is needed.

Latent tuberculosis (LTB) is the stage of Mycobacterium tuberculosis that is asymptomatic, dormant and non-contagious [1]. A positive for the tuberculin skin test (TST) is the evidence of LTB as a prevalence case. The TST conversion from negative to positive from one year to the next indicates recent onset LTB or an incidence case. This individual is at risk to develop active tuberculosis in the future [2,3] Health care workers (HCW) are considered as high risk for LTB [4]. The prevalence of LTB in HCW is different from country to country [5][6][7]. In addition, its prevalence may vary by places of work. HCW who work in a bronchoscopy room [8] or serve particular patients such as tuberculosis or HIV patients might have a higher risk for LTB. Tuberculosis infection control programs should be emphasized in health care facilities [9]. There are limited numbers of studies in this country assessing the incidence and predictors for recent onset LTB in specific health care facilities. We evaluated the association of various workplaces in health care facilities and recent onset LTB by a case-control study in an endemic area of tuberculosis.

Study population
The study was conducted at Srinagarind Hospital, Khon Kaen University, Thailand. We recruited HCW who had results of TSTs from 2 consecutive years from 2001-2008. Cases were defined as subjects with TST conversion, while controls were subjects without TST conversion. TST was done by using the two-step technique.
We enrolled HCW who worked continuously at least 8 hours in 5 different locations in our hospital including, hospital office, inpatient unit, outpatient unit, critical care unit and operating room. Exclusion criteria that was applied tosubjects of both groups included history of recent or active tuberculosis, suspicion of tuberculosis by previous chest X ray, diabetes mellitus, HIV infection or having received immunosuppressive therapy or steroids. These factors may contribute to a false negative TST due to the suppression of immune system, while people with tuberculosis infection will already have a positive TST.
The TST was given by injection of 0.1 ml of 5 tuberculin units of liquid tuberculin intradermally on the forearm. A subject's forearm was examined independently by two infectious disease control nurses 48-72 hours after the injection. The reaction was seen as an area of induration around the site of the injection. The diameter of the indurated area was measured in millimeters. The average diameter of the two readers was reported. An induration of 10 or more millimeters was considered a positive reaction. If the TST was negative, the TST was repeated within the next three weeks to eliminate the boosting effect. This phenomenon occurs in people who are skin tested many years after becoming infected with M. tuberculosis. An initial TST may be negative, followed by a positive reaction to a TST given up to a year later; this happens because the first TST boosts the immune response.
TST conversion was defined if a subject had a negative baseline TST and a positive TST in the next year. In addition, the TST conversion was also defined if the induration of the second-year TST was more than 10 millimeters and greater than the first-year result that was initially greater than the 10 millimeters. The TST was repeated in 12 months later for those with a negative first-year TST. An annual TST surveillance of health care workers is not a standard practice in Srinagarind Hospital nor most health care facilities in Thailand.

Data collection
We recorded each subject's data on an infectious control unit chart including the baseline characteristics such as age, gender, and duration of employment, working unit, duration of employment in years, presence of bacillus Calmette-Guerin (BCG) scar, previous history of tuberculosis in family members or colleagues, previous history of tuberculosis exposure at workplace in the past year, previous history of previous TST, and previous history of using surgical, N95 or hepa masks. The frequency of mask use was defined as either, using at all times or occasionally while working.

Data analysis
Baseline and clinical characteristics of cases and controls were compared using descriptive statistics. Wilcoxon rank-sum or Students t-test and Fisher's exact tests or Chisquare test were applied to compare the differences in numbers and proportions between the two groups.
Univariate logistic regression analyses were applied to calculate the crude odds ratios of individual variables for the development of TST conversion. All variables with p values < 0.25 in univariate analysis were included in subsequent multivariate logistic regression analyses. All variables with p < 0.10 were retained in the final model by the backward elimination technique. Analytical results were presented as crude odds ratios (OR), adjusted OR, and 95% confidence intervals (CI).
The goodness-of-fit of the final model was evaluated using Hosmer-Lemeshow statistics [10]. To evaluate the discriminatory power or accuracy of the model, c statistics or area under the receiver operating characteristic curves were examined [11]. All data analyses were performed with SAS software version 8.2.

Results
In 2001, there were 3,075 health care workers in Srinagarind Hospital and 35 subjects were excluded due to history of recent or active tuberculosis. Of the remainder, 871 subjects or 28.3% had TSTs in two consecutive years. In total, 624 subjects or 71.6%, were eligible, while 247 subjects, 28.4%, were excluded due to variable or rotating workplaces (Figure 1). There were 163 subjects, 26.1%, who had TST conversion and became the cases group.
The baseline characteristics and variables related to tuberculosis of the case and control groups are presented in Table 1. The median age of both groups was 39 years-old. Male subjects accounted for about one-fourth of the cases group. The case and control groups were significantly different in working units, history of tuberculosis exposure in the past year, and history of prevention by any method or by surgical masks.
Almost half of the 291 subjects or 48.7%, worked in the office unit, while 143 (22.9%), 84 (13.5%), 27 (4.3%), 52 (8.3%), and 27 (8.3%) worked at inpatient, outpatient, intensive care, operating room, and laboratory units. Of the subjects who worked in the inpatient, outpatient, intensive care, and operating room units 34.6%, were nurses, 34.6% were nurse assistants and 30.8% were ward staff, while subjects from the laboratory units were all lab technicians. There were three working units that had TST conversion rates of more than 30% including the inpatient, outpatient, and laboratory units (Figure 2).
Univariate analyses showed that factors significantly associated with having TST conversion were the working unit, history of tuberculosis exposure in the past year, and history of prevention by any method or by surgical masks (Table 2). There were only two factors that remained in the final model predictive of having TST conversion; the history of tuberculosis exposure in the past year and the working unit (

Discussion
The prevalence of TST conversion in HCW was about onefourth of the total tested subjects. A history of tuberculosis exposure in the past year and working in either the inpatient or outpatient units were predictive for being TST conversion. This is the first study that indicated that workplaces were significantly associated with TST conversion in an endemic area of tuberculosis. These particular working areas, inpatient or outpatient units, should be monitored closely in tuberculosis control programs of health care facilities.
Srinagarind Hospital, a university hospital, is a tertiary care and teaching hospital with 800 beds, located in the Northeastern part of Thailand. The World Health Organization (WHO) report indicates an incidence rate of new and relapsed tuberculosis in 2006 was 89/100,000 population [12]. Similar to other ASEAN countries, Thailand is Flow chart showed enrollment process Figure 1 Flow chart showed enrollment process.
included in the list of twenty-two tuberculosis high-burden countries by the WHO.
The results of this study indicated that working in health care facilities, particularly the in-and out-patient units, is significantly associated with recent onset LTB. Working in either place increased the risk of TST conversion of 99% and 91%, by the adjusted odds ratio (Table 2). These risks are statistically significant for the working unit in multivariate analysis. Both workplaces have been ignored regarding tuberculosis prevention programs in most health care facilities. Not surprisingly, the critical care unit and operating rooms were not associated with new tuberculosis infection. Both places had negative air pressure systems which could prevent spreading of tuberculosis. A previous report from Brazil [13] did not find the workplace a significant factor in TST conversion.
A history of tuberculosis exposure in the workplace increased the risk of TST conversion about 2.3 times. Even though a history of prevention, particularly using surgical masks in TST conversion subjects, as shown by univariate analysis (Table 2), was, unfortunately, not an effective prevention. Previous reports showed that wearing surgical masks did not prevent tuberculosis infection. Masks were designed to stop droplet nuclei from being spread into the air by person wearing them [14]. In the University hospital, HCWs were wearing surgical or other masks only when they were caring for a suspected tuberculosis patient.  Percentage of subjects with tuberculin skin test (TST) con-version by working units Figure 2 Percentage of subjects with tuberculin skin test (TST) conversion by working units.