Rates of tuberculosis are higher in some states than in others, and they are also higher in large urban centers than in nonurban settings; most cases of tuberculosis in urban centers occur among minority groups [12]. Assessment of the occupational risk for tuberculosis among employees of urban health care institutions is therefore complicated by the sociodemographic diversity of the employees, their risks for exposure to tuberculosis in the community, and the variable degree to which they are exposed to patients who may have tuberculosis.

We did this study to clarify the relative roles of occupational and community exposure to tuberculosis among our hospital employees by determining the prevalence and incidence of positive tuberculin skin tests among our employees and by assessing their risk factors for a positive tuberculin skin test. We also sought to determine whether we could identify groups at particular risk by using employees' job descriptions or departmental groupings.

Methods

Top Methods Results Discussion Author & Article Info References

Patients and Definitions

We obtained tuberculosis screening data for employees of Barnes Hospital from the hospital's employee health department. Barnes Hospital is a 1000-bed, tertiary care referral center in St. Louis, Missouri. The annual incidence of tuberculosis for the State of Missouri is approximately 5 cases per 100 000 persons; the annual incidence of tuberculosis for the City of St. Louis is approximately 11 cases per 100 000 persons. During the study period, 33 patients with pulmonary tuberculosis were admitted to Barnes Hospital. Barnes Hospital employs approximately 6000 persons and hires 1500 to 2000 persons annually.

During the early part of the study, the employee health department's policy was to do tuberculin skin testing at the time of employment and then annually for employees judged to have patient exposure; testing was also done after known exposures to patients with tuberculosis. Since January 1990, it has been the policy of the employee health department to annually test all employees whose duties might require them to be in patient care areas, regardless of degree of exposure to patients.

A positive tuberculin skin test was defined as at least 10 mm of cutaneous induration 48 to 72 hours after intradermal injection of 5 tuberculin units of purified protein derivative (Aplisol, Parke-Davis, Morris Plains, New Jersey) during the study period or as a previously documented positive tuberculin skin test. Employees with a previously documented positive tuberculin skin test were not retested. Skin-test conversion was defined as a positive tuberculin skin test preceded by at least one negative tuberculin skin test. Information about bacille Calmette–Guérin (BCG) vaccination history was unavailable in the database and was not used in the interpretation of skin-test results. Employees with a newly discovered positive tuberculin skin test were evaluated by a physician and were offered preventive isoniazid therapy according to the recommendations of the Centers for Disease Control and Prevention (CDC) and the American Thoracic Society [15].

From January 1989 to July 1991, tuberculin skin-test results were merged by employee number with personnel files that contained demographic information and job descriptions. We classified all 584 job descriptions as having frequent, limited, or no direct contact with patients. Employees who had frequent contact with patients included physicians, nurses, nursing technicians, and respiratory therapists. Employees who had limited contact included activity therapists, chaplains, dietitians, admitting office interviewers, phlebotomists, social workers, and transporters. Employees who had no direct contact with patients included accountants, administrators, cashiers, unit clerks, medical records personnel, custodians, laboratory personnel, laundry workers, and pharmacists.

Employees were also categorized according to departmental groupings: nursing, physicians, support, clinical, and "other." Support personnel included housekeepers, cafeteria workers, and transporters. Respiratory therapists, physical therapists, dietitians, and clinical social workers comprised the clinical group. Employees who did not fall into any group were classified as "other."

Socioeconomic data for employees' postal zones of residence were derived from 1990 U.S. Census Bureau information for the St. Louis Metropolitan Statistical Area and fit to postal zones by the Urban Information Center of the University of Missouri in St. Louis. Socioeconomic variables investigated for each postal zone included average household size, median household income, the proportion of residents below the poverty level, and the proportion of unemployed residents. Tuberculosis case rates by postal zone were derived from case reports from 1988 to 1992 from the Missouri and Illinois State Bureaus of Tuberculosis Control; 1990 U.S. Census Bureau information was used for population denominators.

Statistical Analysis

Risk factor analyses were done using univariate and multivariate statistical methods (UNIX-SAS, SAS Institute, Cary, North Carolina). Univariate comparisons of class variables were done using the Fisher exact test. Continuous variables were assessed using the Wilcoxon rank-sum test. For class variables with multiple levels, analysis of variance was used to test for differences among the distributions. Multivariate analyses were done by stepwise logistic regression; the Bonferroni correction for multiple comparisons was used to determine which variables to include in the final regression model. Because of particular interest in their importance, patient contact variables were included in the models despite their statistical insignificance.

Results

Top Methods Results Discussion Author & Article Info References

Of the 6070 active employees for whom we had tuberculosis screening data, 684 (11.3% [CI, 10.4% to 12.1%]) had a positive tuberculin skin test during the study period. Complete personal demographic data and postal zone socioeconomic and tuberculosis case rate data for postal zones of residence were available for 5475 employees (90%), including 627 (92%) of those who had a positive tuberculin skin test. The characteristics of the employees who were screened and the univariate statistics for groups that had positive and negative tuberculin skin tests are shown in Table 1. Although less than half of the employees (2444 of 5475; 44%) were older than 35 years of age, these employees accounted for most of those with positive tuberculin skin tests (463 of 627; 74%). They had a prevalence of 18.9% (CI, 17.3% to 20.5%) compared with a prevalence of 5.4% (CI, 4.6% to 6.2%) in those younger than 35 years of age. Blacks composed 40% of the work force but 56% (353 of 627) of employees with positive tuberculin skin tests. The rate of positive tuberculin skin tests among blacks was 16% (353 of 2202 [CI, 14.5% to 17.5%]); the rate among non-Hispanic whites was 7.4% (237 of 3198 [CI, 6.5% to 8.3%]). The rates of positive tuberculin skin tests were highest among Asians (28 of 54; 52% [CI, 38% to 65%]) and Hispanics (9 of 21; 43% [CI, 21% to 65%]), although the absolute number of persons with positive tuberculin skin tests in these groups was small. Among the categories of patient contact, the highest rate of positive tuberculin skin tests (327 of 2414; 13.5% [CI, 12.1% to 14.9%]) was seen in employees without patient contact, and, among the departmental groupings, the highest rate was seen in support personnel (231 of 1692; 13.6% [CI, 12.0% to 15.2%]). Employees with positive tuberculin skin tests lived in postal zones that had significantly lower median household incomes and higher rates of poverty, unemployment, and tuberculosis. Household size was marginally higher among those with positive tuberculin skin tests.

By multivariate analysis Table 2, age; black race, Asian race, and Hispanic ethnicity; and the percentage of persons below the poverty level within the employee's postal zone of residence were significantly and independently associated with a positive tuberculin skin test. There was a statistically insignificant trend toward an increased risk for a positive tuberculin skin test among employees who had frequent contact with patients. None of the other variables in the univariate analysis Table 1 was independently associated with an increased risk for a positive tuberculin skin test. The relation of race and age to a positive tuberculin skin test is shown in Figure 1. Compared with non-Hispanic whites, blacks had higher rates of positive tuberculin skin tests across all age groups.

View larger version (28K): [in this window] [in a new window]   Figure 1. Relation of positive tuberculin skin tests to age and race. White bars represent whites; striped bars represent blacks. Vertical bars indicate the 95% CIs.

Among 2154 nursing department employees (10.2% [CI, 8.9% to 11.5%]), a group that had relatively uniform contact with patients, 219 had positive tuberculin skin tests. Of the 1615 (75%) nursing department employees who were white, 118 (7.3% [CI, 6.0% to 8.6%]) had positive tuberculin skin tests. Eighty-seven of the 519 (24%) blacks in this department (16.8% [CI, 13.6% to 20.0%]) had positive tuberculin skin tests. Eight of 13 (62% [CI, 34% to 89%]) Asian employees and 6 of 7 (86% [CI, 58% to 100%] Hispanic employees in this department had positive tuberculin skin tests. By multivariate analysis, age (odds ratio, 2.16 per decade [CI, 2.01 to 2.33]; P < 0.0001); black race (odds ratio, 1.5 [CI, 1.08 to 2.08]; P = 0.014); Asian race (odds ratio, 26.2 [CI, 7.95 to 86.3]; P < 0.0001), and Hispanic ethnicity (odds ratio, 62.2 [CI, 6.77 to 572]; P < 0.0002) remained independently associated with a positive tuberculin skin test. Adding the proportion of persons below the poverty level in the employee's postal zone of residence to the model negated the statistical association of black race with a positive tuberculin skin test, but Asian race and Hispanic ethnicity remained significantly associated (data not shown).

During the study period, 3106 employees had at least two skin tests, and 29 of them had skin-test conversion (0.93% [CI, 0.60% to 1.3%]). The characteristics of the employees who had at least two skin tests and the univariate statistics comparing those who consistently had negative tuberculin skin tests with those who had skin-test conversions are shown in Table 3. Nonwhite race was associated with skin-test conversion; the conversion rate (18 of 1254; 1.4% [CI, 0.73% to 2.07%]) was more than twice that of whites (11 of 1852; 0.59% [CI, 0.24% to 0.94%]). Among those who had contact with patients, the conversion rate was lowest in employees classified as having frequent patient contact. Among the departmental groupings, the conversion rate in support personnel had the greatest statistical significance.

Among all of the variables considered, the characteristics of the employees' postal zones of residence had the strongest statistical association with skin-test conversion. Compared with employees who had consistently negative tuberculin skin tests, employees who had skin-test conversions lived in postal zones with lower household incomes and higher rates of poverty, unemployment, and tuberculosis. By multivariate analysis, only the proportion of persons below the poverty level within the employee's postal zone of residence (odds ratio, 1.39 per 10% [CI, 1.09 to 1.78]; P = 0.0075) had a statistically significant association with skin-test conversion.

Discussion

Top Methods Results Discussion Author & Article Info References

We have shown that in the employee population of an urban, midwestern hospital, the variables most strongly associated with a positive tuberculin skin test were older age, minority group status, and the percentage of poor persons in the employee's postal zone of residence. No independent risk could be associated with degree of patient contact or with departments grouped according to similar job tasks, although there was a statistically insignificant trend toward increased risk among those with frequent direct exposure to patients. The relations between age and race and a positive tuberculin skin test persisted when the analysis was restricted to nursing department employees, a group with relatively uniform patient contact, further indicating that employee demographics played an important role in the acquisition of tuberculosis. The only factor independently associated with tuberculin skin-test conversion was the proportion of persons living below the poverty level in the employee's postal zone of residence. Our findings support those of earlier studies of hospital employees, which show an increased risk for positive tuberculin skin tests among minority groups and no apparent relation between a positive tuberculin skin test and degree of patient exposure [16, 17]. As did our study, one of these earlier studies [17] found an association between tuberculin skin-test conversion and an employee's residence in an area with lower socioeconomic status.

Although we found a significant association between race and both a positive tuberculin skin test and skin-test conversion, the precise nature of this association remains unclear. The analysis of racial characteristics is significantly confounded by socioeconomic, cultural, geographic, and biological factors that are difficult to separate. It has been shown in other populations that among white and black persons who have similar exposure to persons with active tuberculosis, blacks have an increased risk for skin-test conversion, which possibly reflects a greater susceptibility to mycobacterial infection [18, 19]. However, these data have been challenged [20]. Although we cannot rule out the possibility that the higher rate of positive tuberculin skin tests among nonwhite employees in this study reflects an increased susceptibility to tuberculous infection, it is more likely that this association reflects a greater likelihood of exposure to tuberculosis in the community.

Our findings should not be interpreted as suggesting that patient contact does not pose a significant risk for acquisition of tuberculosis. Common sense and historical experience dictate otherwise [7, 8, 21]. Our study has several limitations that may account for our inability to find an association between degree of patient contact and a positive tuberculin skin test. First, the retrospective nature of our study may have introduced unrecognized biases. Second, the classification of an employee's degree of patient exposure according to job descriptions may not have always been accurate. Third, some studies have suggested that the purified protein derivative preparation used in this study (Aplisol, Parke-Davis) may give a higher rate of positive results than other preparations [22]. However, unless rates of positive results differed according to age, race, and income levels in the employees' postal zones of residence, this phenomenon would be unlikely to affect our conclusions. Fourth, it is notable that physicians were poorly represented in the employee health data, a selection bias that may have limited our ability to detect the effect of frequent exposure to patients. However, we did not find an association between a positive tuberculin skin test and frequent exposure to patients despite the fact that more than 40% of the employees in our study (most of them nurses) were classified as having such exposure. Fifth, two-step tuberculin skin testing was not done at baseline, and thus it is possible that some of the skin-test conversions were due to the booster phenomenon [16]. However, because a boosted skin-test reaction is considered a true-positive result, this would not affect our findings with respect to variables associated with the prevalence of positive tuberculin skin tests, and factors associated with conversion would be expected to parallel these variables. Finally, the study design limited our ability to detect the risk for occupational exposure. The most appropriate comparison group for such an assessment would be drawn from among age- and socioeconomically matched persons employed in non-health care settings. We did not seek to determine whether an occupational risk exists for acquisition of tuberculosis. We believe it does. Our important finding is that, in an urban institution such as ours, an employee's job description or departmental grouping is less useful than traditional risk factors in identifying those at greatest risk for tuberculosis.

The CDC recommends mandatory tuberculin skin testing at the time of employment for all persons who work in health care facilities [9, 23, 24]. The recommended frequency for periodic screening is based on a risk assessment intended to identify health care workers with the potential for occupational exposure to tuberculosis [9]. The CDC also recommends including health care workers from "risk groups with increased prevalence of tuberculosis ... even if they do not have potential occupational exposure ..." in a periodic skin-testing program [9]. Our findings support this recommendation. Analyses of current data on hospital employees in areas with higher rates of tuberculosis and in institutions with greater tuberculosis case loads than our own would be of interest.

We have shown low rates of tuberculin skin-test conversion among our hospital employees and no cases of active tuberculosis, even though the engineering controls that govern negative-pressure rooms were suboptimal during the study period [25] and high-efficiency particulate air (HEPA) filter masks were not used for personal respiratory protection. Other studies have questioned tuberculosis control strategies that emphasize the use of HEPA filter masks [26, 27]. Allowing the use of minimally less efficient but much less expensive masks awaits approval by the National Institute for Occupational Safety and Health [9].

Finally, data from our urban institution suggest that among some employees, the overall risk for acquiring tuberculosis from exposure in the community may be greater than the risk for acquiring it from exposure to patients. Although some have disputed the necessity of screening hospital employees [28], we believe that screening is worthwhile. Identifying persons infected with Mycobacterium tuberculosis is important because highly effective preventive therapy can be offered to eligible persons [29]. Among employees who work in patient care areas, preventing tuberculosis not only benefits the employee but also averts a potential for nosocomial transmission, which is the thrust of government regulations in this area [30]. Further, some persons with tuberculosis are highly infectious [20, 31], and even minimal exposure to such persons could result in transmission to patients or employees. The consequence of this to immunocompromised persons can be particularly severe. Finally, the risk for transmitting tuberculosis has historically been greatest among close contacts [32]. Thus, an additional potential public health benefit of tuberculosis screening is to prevent tuberculosis in members of the employee's household and in other persons with whom the employee has close contact. For all of these reasons, we advocate a skin-testing policy for urban institutions that is as inclusive as possible rather than one that singles out any particular group. However, screening programs are labor intensive and costly [33], and further study is required to determine an approach to tuberculosis prevention among hospital employees that has acceptable efficacy, risk–benefit ratio, and costs, recognizing that the optimal approach may differ within and among institutions.

Presented in part at the 32nd Interscience Conference on Antimicrobial Agents and Chemotherapy, Anaheim, California, October 1992 (Abstract 1195).