These recent outbreaks emphasize the importance of effective tuberculosis infection control efforts in health care settings. In December 1990, the Centers for Disease Control and Prevention (CDC) issued guidelines for preventing transmission of tuberculosis in health care settings [13]; revised guidelines were issued in October 1994 [14]. A hierarchy of control measures (administrative, engineering, personal respiratory protective equipment [that is, respirator masks]) have been recommended by CDC to prevent nosocomial transmission of tuberculosis [14]. However, few data are available on the efficacy of these guidelines [16], and CDC has recognized the urgent need to assess the efficacy of isolation in preventing the transmission of tuberculosis [21]. The cost of implementing these recommendations is high [22], particularly for engineering requirements (for example, retrofitting rooms for respiratory isolation) and personal respiratory protective equipment. The Occupational Safety and Health Administration (OSHA) has mandated the use of high-efficiency particulate air (HEPA)-filtered respirator masks [23] as the minimum level of respiratory protection in U.S. health care institutions. These HEPA respirator masks are expensive; a disposable HEPA respirator costs more than $5 per mask, which is more than five times the cost of dust-mist respirators. The efficacy of these HEPA respirator masks compared with that of other types of respirators is unknown [23].

Resources for these expensive engineering and personal respiratory protective equipment mandates are especially difficult to obtain at inner-city public hospitals, which often care for many patients with tuberculosis. We studied the efficacy of expanded tuberculosis infection control measures (consisting primarily of administrative controls) that were implemented at Grady Memorial Hospital, Atlanta, Georgia, after documented transmission of drug-susceptible tuberculosis occurred on several wards of the hospital [10]. We evaluated two measures of health care worker occupational exposure to tuberculosis: 1) tuberculosis exposure episodes [that is, the number of patients with acid-fast bacilli smear-positive pulmonary tuberculosis not admitted into respiratory isolation at hospital admission] and 2) tuberculin skin test results of health care workers at our urban public hospital, which has limited resources and each year cares for more than 200 patients with laboratory-confirmed tuberculosis.

Methods

Top Methods Results Discussion Author & Article Info References

Tuberculosis Exposure

We measured two aspects of health care worker exposure to tuberculosis before and during implementation of tuberculosis infection control measures at Grady Memorial Hospital, a public, university-affiliated, 1000-bed inner-city hospital in Atlanta, Georgia. We evaluated the number of tuberculosis exposure episodes between 1 July 1991 and 30 June 1994 and the tuberculin skin test conversion results of health care workers at Grady Memorial Hospital between 1 January 1992 and 30 June 1994. An episode of tuberculosis exposure was considered to occur when a patient not placed in respiratory isolation at hospital admission subsequently had a diagnosis of acid-fast bacilli smear-positive and culture-positive pulmonary tuberculosis (on the basis of isolation of M. tuberculosis from sputum or bronchoalveolar lavage fluid specimens) during that admission or within 2 weeks of discharge. Days of tuberculosis exposure were defined as the cumulative number of days that patients with acid-fast bacilli smear-positive, culture-positive pulmonary tuberculosis were not housed in respiratory isolation rooms. We began to prospectively follow tuberculosis exposure episodes on 1 March 1992, after an expanded respiratory isolation policy was implemented. We retrospectively determined the number of tuberculosis exposure episodes that occurred between 1 July 1991 and 29 February 1992 by reviewing laboratory records and patient charts.

Acid-Fast Bacilli Smears and Cultures

The hospital's clinical microbiology laboratory did acid-fast bacilli smears using a concentrated method and fluorochrome staining [24]; respiratory specimens were decontaminated by standard procedures [25]. Acid-fast bacilli cultures were done by inoculating specimens onto Middlebrook 7H11 agar (Becton Dickinson, Cockeysville, Maryland) and into radiometric broth media (Bactec, Becton Dickinson, Sparks, Maryland). Mycobacterium isolates were identified as M. tuberculosis complex by nucleic acid probes (Gen-Probe, San Diego, California).

Tuberculin Skin Testing

Tuberculin skin testing for health care workers was done at the hospital's Occupational Health Services Clinic using the Mantoux method; a 0.1-mL (5 tuberculin units) solution of purified protein derivative (Aplisol, Parke-Davis, Morris Plains, New Jersey, or Tubersol, Connaught, Swiftwater, Pennsylvania) was placed intradermally on the volar surface of the forearm and was read 48 to 72 hours later by a member of the Occupational Health Services staff (nurse, physician, or physician assistant). Self-reporting of results by health care workers was not permitted. A positive tuberculin skin test result was defined as an induration of at least 10 mm. A baseline tuberculin skin test was done when the health care worker began working at the hospital; results of this baseline testing were excluded from the analysis. Two-step tuberculin skin testing [14] for new health care workers was not done during the study period. Tuberculin skin testing has been mandatory for all Grady Memorial Hospital employees since 1976. Extremely high compliance (almost 100%) has been assured by the requirement of documentation of a recent tuberculin skin test result (or documentation of a previous positive tuberculin skin test result) before issuance and renewal of the hospital's identification badge, which must be worn by all health care workers. A tuberculin skin test conversion was defined as a documented positive test result following a documented negative result of a test done by the Occupational Health Services staff. We included all health care workers whose tuberculin skin tests were done by the Grady Occupational Health Services during the study period (that is, all Grady Memorial Hospital employees, Grady-based university physician faculty, and housestaff and medical and allied health students who were tested at the Grady Occupational Health Services). We did not include results of tuberculin skin testing of housestaff and students who rotate through Grady Memorial Hospital but who were tested at other sites (for example, at other Emory University-affiliated hospitals).

Control Measures

Expanded tuberculosis infection control measures were implemented at Grady Memorial Hospital after documented transmission of tuberculosis occurred on several wards in late 1991 and early 1992 [10]. The following are the control measures and the dates they were implemented.

1. Administrative controls. These controls consisted primarily of an expanded respiratory isolation policy that was implemented on 1 March 1992. The previous isolation policy required isolation of patients known or suspected to have tuberculosis; also, under this policy, respiratory isolation was discontinued after 2 weeks of antituberculous therapy. The new expanded respiratory isolation policy included mandatory isolation of all patients with active tuberculosis, those with tuberculosis that was established in the differential diagnosis (or when an acid-fast bacilli sputum smear and culture were ordered), and those with HIV infection (or patients at high risk for HIV infection if serologic status was unknown) who had an abnormal chest radiograph. Isolation could be discontinued only after three consecutive negative acid-fast bacilli sputum smears were obtained or when the patient was discharged from the hospital. Other new administrative controls included increased surveillance by the Epidemiology/Infection Control Department to ensure that all patients with acid-fast bacilli sputum smear and culture specimens received by the clinical microbiology laboratory were in respiratory isolation (1 March 1992); the hiring of a nurse epidemiologist to serve as tuberculosis infection control coordinator (1 July 1992); and expanded health care worker education about tuberculosis (1 March 1992), which accompanied implementation of the expanded respiratory isolation policy. This education was especially targeted to nursing staff, house staff, and attending physicians. The frequency of routine mandatory tuberculin skin testing was increased from yearly to every 6 months and was expanded to cover nonemployee health care workers (such as attending physicians, housestaff, and medical students) (July 1992).

2. Interim engineering controls. These controls were implemented by 1 March 1992, pending renovation of the hospital, and consisted of converting 90 rooms that had no recirculated air to negative-pressure rooms (which could be used as respiratory isolation rooms) by the addition of a window fan.

3. Personal respiratory protection equipment. On 1 June 1992, a new mask (3M 1812 submicron mask, 3M Health Care, St. Paul, Minnesota), which provides more than 75% filtration efficiency of 0.3-micron particles and which is similar to a particulate respirator-type mask, was implemented for use by all health care workers when they entered a respiratory isolation room. Standard surgical masks had previously been used at the hospital.

Review of Medical Records and Evaluation of Isolation Rooms

We reviewed the medical records of all patients whose respiratory specimen (sputum or bronchoalveolar lavage fluid) was acid-fast bacilli smear-positive and culture-positive for M. tuberculosis during the study period to assess admission dates and to determine whether the patient had been in respiratory isolation during their admission or admissions, when respiratory isolation was initiated, and HIV serologic status.

The engineering department staff and nurse epidemiologist-tuberculosis control coordinator at Grady Memorial Hospital tested all respiratory isolation rooms using a smoke tube method (Drager Air Flow Tester, National Draeger, Inc., Pittsburgh, Pennsylvania), as recommended by CDC [14]. Persons doing the tests noted whether the rooms were under negative, neutral, or positive pressure. Respiratory isolation rooms that were not under negative pressure with respect to the hallway were noted to have "failed" the airflow testing. A single representative isolation room was evaluated with a tracer gas (sulfur hexafluoride) concentration decay test to determine the rate of air change per hour according to published guidelines by the American Society of Testing and Materials [26].

Statistical Analysis

We did statistical analyses of the number of tuberculosis exposure episodes and the number of exposure days per month using the chi-square test and the Wilcoxon rank-sum test, respectively [27]. We evaluated the tuberculin skin test results of health care workers using chi-square analysis for trend and proportions (Mantel extension method) [28]. A P value of less than 0.05 was considered statistically significant.

Results

Top Methods Results Discussion Author & Article Info References

Between July 1991 and June 1994, there were 752 admissions (for 673 patients) to Grady Memorial Hospital during which the patient had a positive culture for M. tuberculosis. The mean age of these patients was 39.3 years. Among these 752 patient admissions, 334 patients (44.4%) were HIV seropositive, 289 (38.4%) were HIV seronegative, and 129 (17.2%) refused or were not offered HIV testing (Table 1). During the 3 years of the study, 461 (61%) of the admissions had respiratory specimens (sputum or bronchoalveolar lavage fluid) that were acid-fast bacilli smear-positive and culture-positive for M. tuberculosis Figure 1, Table 1; these patients were considered to be potentially infectious.

View larger version (18K): [in this window] [in a new window]   Figure 1. Number of admissions of patients with tuberculosis (TB) to Grady Memorial Hospital. The number of admissions of patients with tuberculosis and the number of admissions in which patients with tuberculosis had respiratory (sputum or bronchoalveolar lavage fluid) specimens that were acid-fast bacilli (AFB) smear-positive and culture-positive is shown in 6-month intervals during the 3-year study period.

Tuberculosis Exposure Episodes

The number of tuberculosis exposure episodes (that is, the number of hospitalized patients not placed in respiratory isolation on admission but subsequently having a diagnosis of acid-fast bacilli smear-positive pulmonary tuberculosis during that admission or within 2 weeks of discharge) occurring during two time periods were compared: The time periods were the 8 months before (July 1991 to February 1992) and the 28 months after (March 1992 to June 1994) the implementation of an expanded respiratory isolation policy. The number of tuberculosis exposure episodes decreased markedly after the implementation of the expanded policy. Thirty-five tuberculosis exposure episodes occurred during the 8 months (average, 4.4 episodes per month) before implementation of the expanded policy, compared with 18 episodes during 28 months (average, 0.6 episodes per month) after policy implementation (Figure 2). Under the previous policy, 35 (34%) of 103 potentially infectious patients with tuberculosis were not appropriately isolated, compared with only 18 (5%) of 358 patients with positive smears under the new policy (odds ratio, 9.72; 95% CI, 4.99 to 19.25 [P < 0.001]). The cumulative number of days that potentially infectious patients with tuberculosis were not in respiratory isolation also decreased significantly, from an average of 35.4 days per month to 3.3 days per month after implementation of the expanded respiratory isolation policy (P < 0.001). The number of patients per month with culture-confirmed tuberculosis and a respiratory specimen positive for acid-fast bacilli did not differ between the two time periods (103 patients during 8 months [12.9 per month] in the first period compared with 358 patients during 28 months (12.8 per month) in the second period; (Table 1). Under the previous respiratory isolation policy, 6 patients subsequently diagnosed with acid-fast bacilli smear-positive pulmonary tuberculosis had more than one admission in which they were not appropriately isolated; no patients had multiple tuberculosis exposure episodes after the expanded policy was implemented.

View larger version (17K): [in this window] [in a new window]   Figure 2. Number of tuberculosis exposure episodes per month. The arrow indicates when the new expanded respiratory isolation policy was introduced.

Patients who were seropositive for HIV and who had acid-fast bacilli smear-positive pulmonary tuberculosis were much more likely to be appropriately isolated at admission under the new expanded respiratory isolation policy than were those under the previous respiratory isolation policy. Under the previous policy, 22 of the 33 admissions of HIV-seropositive patients with tuberculosis and positive respiratory acid-fast bacilli smears were associated with exposure episodes compared with 7 of 143 admissions under the new policy (odds ratio, 44.69; CI, 13.9 to 151.8 [P < 0.001]). Exposure episodes more commonly involved HIV-seropositive patients with tuberculosis under the previous policy than they did under the new policy, but the difference did not reach statistical significance (22 of 35 episodes [63%] under the previous policy compared with 7 of 18 episodes (39%) under the new policy; odds ratio, 2.66; CI, 0.71 to 10.16 [P = 0.096]). The proportion of HIV-seropositive patients with tuberculosis in the two time periods did not differ significantly (Table 1).

Tuberculin Skin Testing of Health Care Workers

Tuberculin skin test results for Grady Memorial Hospital health care workers were evaluated over 2.5 years, from 1 January 1992 to 30 June 1994. Tuberculin skin testing was mandatory for all employee health care workers and was done annually before July 1992. After that time, routine mandatory testing was done every 6 months, and the mandatory tuberculin skin testing program was expanded to include all hospital health care workers. The 6-month rates of tuberculin skin test conversion for health care workers are shown in Figure 3. Tuberculin skin test conversion rates decreased steadily from 3.3% (118 tuberculin skin test conversions among 3579 health care workers tested between 1 January 1992 and 30 June 1992), 1.7% (51 conversions among 2975 workers between July 1992 and December 1992), 1.4% (67 conversions among 4715 workers between January 1993 and June 1993), 0.6% (30 conversions among 4775 workers between July 1993 and December 1993), to 0.4% (23 conversions among 5153 workers between January 1994 and June 1994) (P < 0.001).

View larger version (10K): [in this window] [in a new window]   Figure 3. Six-month tuberculin skin test (TST) conversion rates for health care workers at Grady Memorial Hospital. The number of health care workers with a positive test result divided by the total number of health care workers tested during that time is shown above each bar.

Further evaluation of the 23 Grady Memorial Hospital health care workers (among 5153 tested) with tuberculin skin test conversions between January 1994 and June 1994 showed that 10 had direct patient contact in areas where patients with or at risk for tuberculosis receive care (for example, inpatient wards and outpatient clinics); 4 had direct patient contact in areas where infectious patients with tuberculosis were not likely to be housed (for example, neonatal intensive care unit, special care nursery, long-term care facility); and 9 had no direct patient contact (for example, the laundry, cafeteria, administrative offices). This finding suggests that more than half of the tuberculin skin test conversions observed during this time may have been caused by community exposure to tuberculosis or other factors rather than by nosocomial exposure. Among these 23 health care workers, positive tuberculin skin test results were not clustered by work area.

Engineering Controls

Window fans were placed in respiratory isolation rooms between December 1991 and March 1992. Airflow testing of all respiratory isolation rooms by the smoke-tube test method was done seven times (approximately every 3 months) beginning in September 1992. During the study period, respiratory isolation rooms were found not to be under negative pressure an average of 16.5% of the time (range, 6.1% to 21.7%). By the use of a tracer gas (sulfur hexafluoride) concentration decay test, the number of air changes in a single representative respiratory isolation room was determined to be 4.9 per hour.

Discussion

Top Methods Results Discussion Author & Article Info References

Expanded tuberculosis infection control measures were instituted at our hospital after documentation of nosocomial transmission of drug-susceptible tuberculosis caused by health care worker-to-health care worker transmission and patient-to-health care worker transmission [10]. Skin test conversions of health care workers were clustered on two "outbreak" wards of the hospital as well as on "nonoutbreak" general medicine wards that had 1-year conversion rates for nursing personnel of 13.2% [10]. At our hospital, most patients (>85%) with a diagnosis of laboratory-confirmed tuberculosis present without a previous diagnosis of tuberculosis. Our expanded respiratory isolation policy was developed to detect all patients with active pulmonary tuberculosis and to meet CDC guidelines on early identification, isolation, and treatment of persons with active tuberculosis [13, 14]. Our expanded respiratory isolation criteria included isolating all patients with HIV infection who were admitted to the hospital with an abnormal radiograph of the chest because it is clinically difficult to determine whether such patients with respiratory symptoms or signs have tuberculosis rather than another pulmonary infection (such as Pneumocystis carinii pneumonia or fungal or bacterial infections) [17, 19, 29].

After the expanded respiratory isolation policy was implemented, 95% of patients in whom tuberculosis was subsequently diagnosed were appropriately isolated at hospital admission. At the same time, the number of tuberculosis exposure episodes per month dramatically decreased (from 4.4 per month to 0.6 per month), as did the cumulative number of exposure days per month (from 35.4 per month to 3.3 per month). Concomitant with the decrease in tuberculosis exposure episodes, there was a marked decrease in the number and rate of tuberculin skin test conversions among health care workers. During the study period, the conversion rate decreased eightfold, from 3.3% (118 tuberculin skin test conversions among 3579 health care workers tested) between January 1992 and June 1992 to just 0.4% (23 tuberculin skin test conversions among 5153 health care workers tested) in a similar period 2 years later, January 1994 to June 1994.

Our data and those recently reported from follow-up studies by CDC [30, 31] show that nosocomial outbreaks of tuberculosis can be halted and transmission prevented by infection control measures that emphasize administrative controls and do not involve the use of OSHA-mandated HEPA respirator masks. After implementation of the expanded respiratory isolation policy, patients at our hospital were overisolated; only about 1 in 8 (14%) patients placed in respiratory isolation had culture-confirmed tuberculosis [32]. However, we believe that overisolation is necessary to ensure a high efficacy in detecting and isolating patients with tuberculosis at hospital admission.

Although our findings strongly suggest that expanded infection control measures led to the prevention of nosocomial transmission of tuberculosis, our study had several limitations. First, several different control measures (administrative, engineering, and personal respiratory protection) were introduced at about the same time. It is therefore difficult to determine the individual effect of each of these measures. We believe that the administrative controls were most important in preventing nosocomial transmission of tuberculosis because the time course of the decrease in tuberculin skin test conversion rates among health care workers mirrored the decrease in the number of tuberculosis exposure episodes that occurred after the implementation of an expanded respiratory isolation policy and other administrative controls. This is further emphasized by the data on the somewhat limited effectiveness with which engineering controls were implemented during the study period. For example, air-flow testing showed that not infrequently (16.5% of the time), the respiratory isolations rooms were not under negative pressure; in addition, a single representative isolation room that was evaluated by tracer gas study did not have the recommended six air changes per hour [14].

The second limitation of our study is that the comparison periods (the 8 months before and the 28 months after implementation of expanded infection control efforts) were unequal. We prospectively followed tuberculosis exposure episodes after implementation of the expanded respiratory isolation policy and other administrative controls, but exposure episodes in the period before intervention could only be retrospectively determined. We do not believe that this factor biased the conclusions of the study because the retrospective review of the period before intervention may have led to an underestimate of the number of tuberculosis exposure episodes during that time period. We chose an 8-month period before intervention because data for that period were accessible and available for analysis. Availability of information also dictated our use of data on tuberculosis exposure episodes beginning on 1 July 1991 and our use of data on tuberculin skin test conversions of health care workers beginning on 1 January 1992. Tuberculin skin test data collected before January 1992 would be needed to definitely show that the downward trend in the tuberculin skin test conversion rates of health care workers shown after January 1992 did not start before interventions were initiated on 1 March 1992. However, it is highly unlikely that the overall skin test conversion rate of health care workers was decreasing before January 1992, as it was documented that there was an increase in health care worker tuberculin skin test conversion rates on several wards of the hospital that was due to nosocomial transmission of tuberculosis in late 1991 and early 1992 [10].

In summary, expanded infection control measures consisting primarily of administrative controls were effective in preventing nosocomial transmission of tuberculosis to health care workers. Our experience at an institution located in an area with a high incidence of tuberculosis shows that a high index of suspicion and careful screening of patients for tuberculosis is essential and can dramatically reduce both health care worker exposure to tuberculosis and tuberculin skin test conversion rates. Administrative controls appear to be the most important component of a tuberculosis infection control program and should be the first focus of such a program, especially at public hospitals, where resources are most likely to be limited.