Effectiveness of a Nosocomial Tuberculosis Control Program at an Urban Teaching Hospital

Effectiveness of a Nosocomial Tuberculosis Control Program at an Urban Teaching Hospital* Philip A LoBue, MD; and Antonino Catanzaro, MD

Study objective: To assess the effectiveness of a nosocomial tuberculosis (TB) program at an urban teaching hospital. Design: Retrospective review. Setting: An urban teaching hospital, the University of California, San Diego Medical Center (UCSD), which cares for 25 to 30 culture-proven pulmonary TB cases (>80% of which are smear-positive) pet· year. Study population: Health-care workers. Measurements: (I) Purified protein derivative (PPD) conversion rates. (2) Cases of active TB among health-care workers. (3) Compliance rates with isoniazid (INH) preventive therapy. Results: The UCSD program was evaluated for the years 1993 to 1995. The PPD conversion rate among established employees was 0.6%. Of 556 employees who had an exposure, 494 (88.8%) were compliant with follow-up. Three hundred thirty-seven were skin-tested (the other 157 already had a known PPD > 10 mm). Only 2 of 337 (0.6%) converted. One case of active TB, unrelated to any documented hospital exposure, was discovered in 3 years among approximately 5,000 employees per year (follow-up for convertors, 18 to 54 months). Only 48.4% of eligible employees completed at least 6 months of INH preventive therapy. Conclusions: UCSD's TB control measures appear to be effective in the prevention of nosocomial transmission of TB. Despite poor compliance with INH preventive therapy, cases of active TB (CHEST 1998; 113:1184-89) among health-care workers wet·e rare. Key words: isoni azid prophylaxis; nosocomial infection ; puri fied protein derivative; screening; tuberculosis Abbreviations: AFB = acid-fast bacilli; CDC=Cente rs for Disease Control and Preve ntion; I NH = si oni azid; MDR =multidru g-resistant; PPD = p urifi ed protein de rivative; SH EA = Society for Healtbcare Epide miology of America; UCS D = Univers ity of Californi a, San Diego

n the e rapreceding effective treatment for tuberI culosis (TB ), health-care workers in the United States were at significant risk for infection and disease.t With the advent of effective che motherapy and the subsequent decline in incidence ofTB in the general population beginning in the ,1950s, the risk ofTB to health-care workers a lsofell.Z From the mid 1980s through the early 1990s, however, the re was a res urgence in TB in the United States.3 This coincided with two other phenomena: the HIV epidemic and the e mergence of multidrug-resistant (MDR ) strains of TB. Not surprisingly the re were several subsequent reports of hospital outbreaks of MDR *F ro m th e Divis ion of Pulmonary and Critical Care Medicine, Unive rsity of California, San Diego Manuscript received July 24, 1997; revision accepted O ctober 31, 1997. Reprint requests: Philip A. LoBue, MD, Division of Pulnwnary and Critical Care Medicine, Uniw rsity of Califomia, San Diego Medical Center, Mail Code 8374, 200 W Arbor Dr, San Diego, CA 92103 1184

TB among patients and hospital workers, many of whom were HIV-seropositive.4 ' 5 '6 This resulted in renewed interest in measures designed to prevent both TB infection and development of active disease in health-care worke rs. Subsequently, the Centers for Disease Control and Prevention (CDC ) formulated guidelines for prevention of transmission of TB in health-care facilities based on a three-tier hierarchy.7 The first level of hierarchy is the use of administrative controls. These are designed to p revent the exposure of uninfected persons to persons with active, infectious TB . They include implementation of measures to ensure rapid identification, diagnostic evaluation, and isolation of persons likely to have active TB . Administrative controls also should provide for education and training of health-care workers and screening of health-care workers for tuberculous infection and disease. The second level of hierarchy involves the use of engineering controls to prevent Clinical Investigations

nosocomial spread of TB. Institution of adequate ventilation and air cleaning via filtration or ultraviolet radiation are the basis of engineering controls. The final level of hierarchy is the use of personal respiratmy protection (eg, masks, etc) in situations where there is a known high risk for TB infection to the health-care worker which is unavoidable (eg, isolation room with known TB patient, bronchoscopy, etc). We have had a nosocomial TB control program at the University of California, San Diego (UCSD) Medical Center since 1971. UCSD Medical Center is an urban teaching hospital in a county with a relatively high TB case rate (15.4 to 17.2/100,000 for the years 1993 to 1995) (E. Haas, MPH; Tuberculosis Control Clinic, San Diego County Department of Health Services; personal communication; October, 1996). The current policies and procedures have been in place since November 1, 1992. Therefore, we analyzed the effectiveness of our program by examining purified protein derivative (PPD) conversion rates and follow-up of exposures to active cases over a recent 3-year period, 1993 to 1995.

MATERIALS AND METHODS Number of Employees and Cases of TB at UCSD

The number, race, ethnicity, and count1y of origin of new and established employees and volunteers for the years 1993, 1994, and 1995 were dete rmined from payroll and staffing records maintained by the medical center. The medical center patient database was searched for all patients with a positive sputum culture for Mycobacterium tuberculosis for the years 1993 to 1995. It was also determined if the patient was sm ear-positive or smear-negative at the time of hospitali zation. The charts of all active TB cases were reviewed to determine the delay in diagnosis. TB Screening Policy

UCSD Medical Center has a writte n policy that stipulates infection control practices regarding TB. The current policy described below has been in effect since \Tovember 1, 1992, and was not changed clUJing th e period of this report. All employees and voluntee rs at UCSD Medical Ceuter are required to undergo screening for TB prior to beginning work and subsequently at yearly inte1vals. Physicians are subject to the same requirements as all other personnel. New employees are not permitted to begin work until they have completed th e screening process. If an established employee has not complied within 1 month of notification , the e mployee's supe1visor is contacted. Monthly warning letters are sent to the employee and supe1visor th ereafter. Failure to comply within 6 months of notification can lead to te rmination of ernployment. All new and established employees without a prior documented reactive P1>D (10 mm or greater [5 mm is used as the cutoff for the PPD for HIV seropositive employees] ) at UCSD (or an outside facility, for new employees) must have a PPD skin test. Five tuberculin units of PPD anti'gen (Tubersol; Connaught; S\\~ftwater , Penn) are

administered intradennally by the Mantoux method by an experi enced t echnician. The e mployee is instructed to return in 48 to 72 hours for a reading. The amount of induration is measured and recorded. A convertor is defin ed as one whose PPD reaction is at least 10 mm in induration and bas increased at least 6 mm within a 2-year period. (This was the definition per the State of California/California Tuberculosis Controllers Association guidelines in 1993. Although newer Ame1ican Thoracic Society/CDC guidelines were published in late 1994, UCSD did not change its definitions. ) A reactor is defin ed as anyone with a PPD reaction of 10 mm or greate r who does not meet th e criteria of a convertor. If the initial PPD for a new employee produces less than 10 mm of induration, a s ceond PPD (booster) is required '~thin 10 days of the original. All new employees and volunteers \\~th a PJ>D of 10 mm or greate r are required to have a chest radiograph. This is also true of all established employees unless th ey are in one of two categories. Those who have completed at l aest 6 months of isoniazid (INH ) therapy or whose significantly reactive PPD has been documented to be present for least 5 years are considered low-risk. They are given a questionnaire which inquires about symptoms compatible with TB. A questionnaire is also given to any personn el (new or established ) in lieu of a chest radiograph if there is acontraindication to radiography (eg, pregnancy). If they answer yes to any qu estion, a c hes t radiograph is required. All chest radiographs are read by a radiologist and then examined by the pulmonmy physician in charge of the employee TB control clinic. Any employee with a radiograph possibly compatible with active TB undergoes furth er evaluation, which includes, at minimum, obtaining sputum samples for acid-fast bacilli (AFB ) smear and culture. These employees are not allowed to begin or continue working until they have been cleared by this clinic. Isolation Procedures, Determination of Employee Exposure, and Follow-Up of Exposu·res

When a patient is identified as a TB case or suspect, th e patient is placed in an appropriate isolation room (negative pressure, more than 6 air changes/h). Negative pressure for each isolation room is verified on a quarterly basis. The number of air changes is ve rified at the time the room is set up as an isolation room, but not routinely th ereafter. Isolation rooms are not equipped \~th ultraviolet lighting. A sign is placed on th e door (which is kept closed all the time except when th e patient, an employee, or a visitor ente rs or exits) indicating that airborne isolation precautions are in effect. All individuals entering the room are required to wear a protective mask (from 1993 to 1994; Submicron Molded Surgical Mask; 3l'v1 ; St. Paul , Minn; in 199.5; Particulate Filter Respirator 95; Tecnol Inc; Fort Worth , Tex). Fit testing is not done. If it is necessmy for the patie nt to leave the room, he is required to wear a protective mask until he returns. The patient remains in isolation until he has been d etermined to be noninfectious, usually by having three respiratmy specimens that are AFB smear-negative on 3 consecutive clays, or until discharge. In gene ral a significant exposure is considered to have occurred if an employee has unprotected face-to-face contact with a patie nt (regardless of duration or frequency of visits ) who is later cletermin ecl to have AFB smear-positive pulmon
1185

Table !-Employees with PPD of at Least 10 mm (Reactor or Convertor}: Distribution by Country of Origin and Race/Ethnicity Demographic Cou ntry of origin United States Other country Unknown Hace/ethnicity for United States-born Asian

African-American Hispanic White Native American Other Unknown

Total No. of Employees (%)

No. of Emp loyees with PPD 10 mm or Greater (%)

12,225 (79. 1%) 3,106 (20.1 %) 124 (0.8%)

147 (18.4%) 645 (80 6%) 8 (1.0%)

929 (7.6%) 1,112 (9.1 %) 1,014 (8.3%) 8,423 (68.9%) 73 (0.6%) 123 (1.0%) .538 (4.4% )

0 (0.0%) 38 (26.3%) 31 (21.1%) 70 (47.4 %) 0 (00%) 8 (5.3 %) 0(0.0%)

<0.001*

< 0.001*

another PPD is required at 10 wee ks after exposure. If either of these PPD tests is 10 mm or greater, a chest radiograph is requirecl. Those with prior PPD of 10 mm or greater are screenecl with a questionnaire inquiring about potential symptoms compatible with TB 10 weeks after the initial exposure. 1f the e mployee answers yes to any question , a chest radiograph is required. TNH Preventive Therapy

All PPD convertors, regardless of age, ancl all employees with a PPD of 10 mm or greater who are < 35 years of age are offerecl an appointment for evaluation for INH preventive therapy. lf the physician determines that preventive therapy is inclicated, a 6- to 12-month course of isoniazid is offered to the employee free of charge·' (We offer our employees the opti on of continuing isoniazid for 12 months because th ere are data showing that this is more efficacious). At the start of lNH therapy, employees receive an instruction sheet detailing signs ancl symptoms of possible drug toxicity. They are told to discontinue th e medication immediately and contact the e mpl oyee clinic for an appointme nt with a physician should any of these signs or symptoms occur. They also receive a monthly questionnaire for signs or symptoms of drug toxicity. If they respond affirmatively to any question, they are immediately instructed to discontinue the medication until evaluated by the physician. As th ere is only one part-time physician available for th e skin testin g program , follow-up with the physician is not clone routinely. Monthly transaminase levels are obtained in patients at high risk for hepatitis as determined by American Thoracic Society guidelines." They are monitorecl by th e skin testing technicians, who immecliately alert th e physician if any abnormaliti es occur. Co mplian ce with INI-1 therapy was evaluated by tracking prescription refills. Employees who are eligible for IN I-I receive a prescription for 1 month of medication with 11 refills. An employee was considered compliant if' he or she picked up at least 5 refills (6 months total) within a 12-month period.

RES ULTS

Number of Cases of TB and Delay in Diagnosis There were 85 cases of sputum culture-positive TB at UCSD Medical Center between 1993 and 1186

p Value

1995. Of these, 69 (81.2%) were AFB smear-positive. Seventy-one patients with TB were diagnosed upon admission. In 14 patients who were not initially placed in isolation, the diagnosis was delayed a mean of 4.6 days and a median of 2 days (SD , 5.4 days; range, 1to 18 days). Eleven of these 14 (78.6%) had a positive AFB smear from a respiratory specimen (although several had a positive smear from bronchoscopy only).

Denwgraphics of Employee Population UCSD Medical Center is an urban teaching hospital with 440 beds. For the years 1993, 1994, and 1995, the numbers of new and established employees and volunteers at UCSD Medical Center were 5,784, 5,705, and 5,720, respectively. While the majority of personnel was born in th e United States, a significant percentage of employees was foreignborn (20.1 %) (Table 1). Of United States-born employees, 26.7% were members of a racial or ethnic minority group.

Tuberculin Screening During the three years 1993 to 1995, 15,455 PPD tests were placed and read for the purpose of TB screening among hospital personnel. Compliance with screening for new employees was 100%, as they were not permitted to begin work unless they completed the screening process. Compliance for established employees was approximately 85%. The PPD conversion rate was 0.6% per year for established personnel (Table 2). Among new personnel, 708 were PPD reactors (12.8%). Of these, 678 were detected on the initial PPD (95.6%) and 30 were detected upon booster PPD testing (4.4%). The vast majority of employees and volunteers with a PPD of at least 10 mm (reactors and convertors) were forClinical Investigations

Table 2-Tuberculin Screening: New and Established Personnel

Personnel PPD testing Total tested Convertors Reactors Initial Delayed (booster) Total Questionnaire results Total Positive Negative Chest radiograph and clinical evaluation Total Normal Abnormal, not TB TB inactive TB active

No. of New Employees (%)

No. of Established Employees (%)

5,550 N/A

9,905 59 (0.6%)

678 30 708 (12.8%)

124 N/A 124 (1.3%)

66 1 (1.5%) 65 (98.5%)

2,815 59 (2.1%) 2,756 (97 9%)

948 940 (99.2%) 4(04%) 4 (0.4%) 0 (0.0%)

428 424 (99.1%) 2 (0.5%) 1 (0.2%) 1 (0 2%)

eign-born (80.6%) (Table 1). Of United States-born employees, 52.6% of those with a PPD of 10 mm or more belonged to a racial or ethnic minority. Sixty-six new personnel received a TB symptom questionnaire because it was decided that a chest radiograph should be deferred (Table 2). Only one individual (1.5%) responded positively to any of the questions. TB symptom questionnaires were given to 2,815 returning employees (either the employee was low-risk or it was felt that a chest radiograph should be deferred). Sixty (2.1 %) responded affirmatively to one or more questions. Chest radiographs were performed on 1,376 employees and volunteers. The vast majmity of radiographs (99.1 %) was normal (Table 2). After further clinical evaluation by the employee clinic's pulmonaly physician, one employee (0.1 %) was diagnosed with active TB. The single case ofTB occurred in an established employee who was a PPD convertor. The diagnosis of active TB was made at the time of PPD conversion and therefore INH had not previously been offered to this person. This employee had no known exposure to an active TB case and was identified on routine screening. Exposure Evaluations From 1993 to 1995, 556 personnel were deemed to have had significant exposure to a patient with active TB (Table 3). Four hundred sixty-two of these employees (83.1%) were exposed to a patient with a positive AFB smear, although several of these patients had a positive smear from a bronchoscopy specimen only.

Table 3-Exposure Evaluations Evaluation Method

No. (%)

Total employees exposed Number compliant with follow-up PPD (prior PPD < 10 mm ) Total PPD Conve rtors Reactors Questionnaire (prior PPD 2: 10 mm ) Total Positive Negative Chest radiograph and clinical evaluation Total Normal Abnormal, not TB TB inactive TB active

556 494 (88.8%) 337 2 (0.6%) 2(06%) 157 2 (1.3%) 155 (98 7%) 6* 6 (100.0%) 0 (0.0%) 0 (0.0%) 0(0.0%)

*2 convertors, 2 r eactors , 2 positive questionnaires

The remaining 94 personnel were exposed to smearnegative, culture-positive patients. In these cases, the Tuberculosis Control Officer decided exposure evaluation was warranted based on other factors, such as participation in a high-risk procedure (eg, bronchoscopy), prolonged and frequent interaction with the patient, etc. Four hundred ninety-four of 556 exposed employees (88.8%) were compliant with recommended postexposure evaluation. Three hundred thirty-seven (known prior PPD of less than 10 mm) underwent PPD testing. Of these, 2 were found to be convertors (0.6%) and 2 were found to be reactors (0.6%). One hundred fifty-seven employees (known prior PPD of at least 10 mm) were given TB symptom questionnaires after an exposure. Only 2 (1.3%) answered yes to any of the questions. Six chest radiographs were performed on personnel following an exposure. All were normal.

INH Preventive Treatment Of the 894 convertors and reactors, 259 (91 convertors, 167 reactors) were felt to be eligible for INH preventive treatment by the skin testing technical staff and were offered an appointment with a physician for evaluation (Table 4) . Forty-tvvo individuals (16.2%) did not attend their appointment despite repeated requests. Of those that did attend, only 169 (65.3% of all employees eligible to take INH) agreed to take INH. One hundred fifty-nine employees and volunteers who were started on INH were evaluable to determine completion of therapy. Seventy-seven (48 .4%) completed at least 6 months of therapy. For those who did not finish treatment, the drug was stopped by the employee in the vast majority of cases . CHEST/113/5/MAY, 1998

1187

Table 4-INH Preventive 1'herapy No. (%)

INH dinie evaluation Offered appointment Did not attend Physkian felt INI-1 not indkated Did not agree to take INH Agreed to take INH Started INJ-1 Total evaluable Completed INH Did not complete: stopped by physi<:ian Did not complete: stopped by patient

2.59 42 (16.2%)

lO (3.9%) 38 (14.7%) 169 (6.53%) 159* 77 {48.4%) 12 (7.5 %) 70 (44.0%)

*10 employees were sti ll con tinuing INH therapy at th e end of 1995 and we re excluded from this analysis DISCUSSION

In 1994 the CDC developed new guidelines for prevention of transmission of TB in health-care facilities based on a three-level system of administrative, engineering, and personal respiratory protection controls. Among the administrative controls, it has been recommended that all health-care facilities that care for patients with active TB institute annual tube rculin screening for their employees in order to monitor the risk of nosocomial TB transmission.2.7 vVhen the administrative, engineering, and personal respiratory protection controls have been properly instituted, it is expected that nosocomial TB infection of health-care workers will be minimized, as manifested by low PPD conversion rates.l 0 A survey of health-care institutions done in 1992 (Society for Healthcare Epidemiology of AmericaCenters for Disease Control and Prevention, or SHEA-CDC) revealed that th e annual risk of employee skin test conversion was significantly higher in hospitals with 6 or more TB patients per year (1.2% vs 0.6%). 11 The range of annual conversion rates was fairly large (0.0 to 7.7%) and comparable to that previously reported by individual institutions.I 2-17 Even though UCSD cares for about 25 to 30 sputum culture-positive TB patients per year (81.2% of whom are smear positive), the annual PPD conversion rate was 0.6%, which is lower than the average found in the SHEA-CDC smvey. This provides evidence that the administrative and engineering controls in place at UCSD work well to prevent nosocomial TB infection in employees. Specifically, early diagnosis of TB patients at UCSD may have contributed to the low conversion rate. There was no delay in isolation for the majority of sputum culturepositive patie nts (71 of 85, or 83.5%). In addition to early recognition of disease; subsequent isolation, enginee1ing, and personal respiratOI)' protection control measures were apparently effective in preventing nosocomial transmission. 1188

The PPD reactor rate of 12.8% for new employees was quite high compared with a mean initial reactor rate of 1.91% found in the SHEA-CDC smvey. This high rate is likely attributable to two facts. First, TB is significantly more prevalent in San Diego than in the nation as a whole . Second, individuals at high risk for TB infection make up a significant number of UCSD's employees and volunteers. These include foreign-born persons and members of racial and ethnic minorities. It should be noted that the high rate of reactors among new personnel does not contribute to the relatively low conversion rate among established employees seen at our institution. Once an individual has a PPD of 10 mm or more, he or she is not retested. Therefore, these individuals do not appear in the denominator (ie, total PPD tests for established employees) of the skin test conversion rate for established personnel. In the period from 1993 to 1995, one case of active TB was discovered among the approximately 5,700 employees and volunteers screened at UCSD each year. Careful investigation did not reveal any link to a hospitalized case of TB . It is likely that this employee acquired TB infection in the community. Of all personnel who were skin-tested after an unprotected exposure to an active case, t\vo convertors were detected (conversion rate, 0.6%). The relatively low conversion rate probably results from several factors. The risk of transmission of TB infection from one individual to another is dependent on th e infectiousness of the source case, duration of exposure, proximity of contact, and environmental factors (especially ventilation).l 8 It is likely that the TB patients in whom th e diagnosis was delayed did not have advanced disease and therefore were not as infectious. Additionally, in many cases the employee exposure was probably vel)' brief compared to a household contact. Compliance with INH therapy among health-care workers is poor. Prior studies have revealed compliance rates ranging fi·om 13% to 66%, with the majority shovving rates of 50% or less .12,19 -21 Employees and volunteers at UCSD fared no better despite various suppmtive practices of the TB control program. Improved employee education and possibly even incentives (such as are used to recruit and maintain subjects in clinical trials) might be considered in the future to increase compliance. In summary, we found a low PPD conversion rate at our institution despite the fact that the medical center cared for a moderate number of patients vvith TB . The conversion rate was low even when there was known exposure to an active case. Our baseline PPD reactor rate was fairly high, consistent with the relatively high TB case rate in the community and the demographic composition of our employees and Clinical Investigations

volunteers. One case of TB was detected among employees and volunteers at our institution in 3 years. Compliance with INH prophylaxis was poor. These findings suggest that our administrative and engineering controls for prevention of nosocomial transmission of TB are, in general, effective. However, our INH preventive therapy program is ineffective because of poor compliance, and efforts should be directed at improvement in this area. ACKNOWLEDGMENTS: The authors wou ld like to thank Ellen Logue, Terri Festa, Kathy Miskovish, and Ed Barber for their assistance in collecting data for this study. REFERENCES 1 Sepkowitz KA. Tuberculosis and the health care worker: a historical perspective. Ann Intern Med 1994; 120:71-79 2 Menzies D, Fanning A, Yuan L, et al. Tuberculosis among health care workers. N Engl J M ed 1995; 332:92-98 3 Cantwell MF, Snider DE , Cauthen GE , et la. Epidemiology of tuberculosis in th e United States, 191)5 through 1992. JAMA 1994; 272:535-39 4 Ce nters for Disease Control and Prevention. osocomial transmission of multidrug-resistant tuberculosis among HIV infected persons-Florida and New York, 1988-1991. MMWR Morb Mortal Wkly Rep 1991; 40:585-91 5 Pearson ML, Jereb JA, Frieden TR, et a!. Nosocomial transmission of multidrug resistant Mycobacterium tuberculosis: a risk to patients and health care workers. Ann Intern Med 1992; 117:191-96 6 Beck-Sague C, Dooley SW, Hutton MD, et a.l. Hospital outbreak of multidrug-resistant Mycobacterium t11berculosis infections: factors in transmission to staff and HIV-infected patients. JAMA 1992; 268:1280-86 7 Centers for Disease Control and Prevention. Guidelines for p reventing the transmission of Mycobacterium tuberculosis in health-care facilities. MMWR Morb Mortal Wkly Rep 1994; 43(HR13): 1-132 8 International Union Against Tuberculosis Committee on Prophylaxis. EfHcacy of various durations of isoniazid preventive therapy for tuberculosis: flve years of follow-up in the

IUAT trial. Bull World Health Org 1982; 60:555-64 9 The American Thoracic Society. Treatment of tuberculosis and tuberculosis infection in adults and children. Am J Respir Crit Care Med 1994; 149:1359-74 10 Louther J, Rivera P, Feldman J, et a!. Risk of tuberculin conversion according to occupation among health care workers at a New York City hospital. Am J Respir Crit Care Med 1997; 156:201-05 11 F ridkin SK, Manangan L, Bolyard E, et al. SHEA-CDC TB survey, patt II: efllcacy of TB infection control programs at member hospitals, 1992. Infect Control Hosp Epidemiol 1995; 16:135-40 12 Raad I, Cusick J, Sheretz RJ, et a.!. Annual tuberculin skin testing of employees at a university hospital: a cost benefit analysis. Infect Control Hosp Epidemiol 1989; 10:465-69 13 Fridkin SK, Manangan L, Bolyard E, etal. SHEA-CDC TB survey, patt I: status of TB infection control programs at member hospitals, 1989-1992. Infect Control Hosp Epidemiol1995; 16:129-34 14 Atuk NO, Hunt EA. Serial tube rculin testing and isoniazid therapy in general hospital employees. JAM A 1971; 218: 1795-98 15 Berman J, Levin ML, Orr ST, et al. Tuberculosis risk for hospital employees: analysis of a five-year tuberculin skin testing program. Am J Public Health 1981; 71 :1217-22 16 Ball R, VanWey M. Tuberculosis skin test conversion among health care workers at a militaty medical center. Mil Med 1997; 162:338-43 17 Schwartzman K, Loo V, Pasztor J, et al. Tuberculosis infection among health care workers in Montreal. Am J Respir Crit Care Med 1996; 154:1006-12 18 Hopewell PC, Bloom BR. Tuberculosis and other mycobacte rial diseases. In: Murray JF, Nadel J, eds . Textbook of respiratmy medicine. Philadelphia: W.B. Saunders, 1994: 1094-160 19 Gamins BC, Bock N, Watkins DL, et al. Acceptance of isoniazid preventive therapy by health care workers after tuberculin skin test conversion. JAMA 1996; 275:1013-15 20 Frase r VJ, Kilo CM , Bailey TC, et al. Screening of physicians for tuberculosis. Infect Control Hosp Epidemiol1994; 15:95100 21 Barrett-Connor E. The epidemiology of tuberculosis in physicians. JAMA 1979; 241:33-38

CHEST/113/5/MAY, 1998

1189