Of the 34 cases of multidrug-resistant tuberculosis (resistant to at least isoniazid and rifampin) that were reported to the Chicago Department of Health from January 1994 through April 1995, 6 occurred in patients with AIDS who had been admitted to one floor of one hospital. Because nosocomial transmission of multidrug-resistant tuberculosis was suspected, an investigation was initiated.

Methods

Top Methods Results Discussion Author & Article Info References

Potential outbreak cases-cases of tuberculosis in persons whose isolates were resistant to at least isoniazid and rifampin-were identified by matching the names of patients and health care workers at the hospital to state and local tuberculosis registries and by reviewing laboratory records and death certificates. We performed DNA fingerprinting of M. tuberculosis isolates using IS6110 and pTBN12 probes [14, 15].

Medical records were reviewed for patients who were admitted to the outbreak floor when a patient whose sputum smears were positive for acid-fast bacilli was present. Patients who died within 30 days of exposure were excluded.

Private providers of patients who may have been exposed to M. tuberculosis were contacted by telephone or by mail. All were advised to screen their patients for tuberculous disease and infection by tuberculin skin tests and symptom checks. A positive result on a tuberculin skin test was defined as induration of 5 mm or more, and an increase in induration of at least 5 mm was considered a skin test conversion. Severely immunocompromised patients (CD4⁺ T-lymphocyte count < 50 cells/mL) who did not receive skin testing were assumed to be anergic [16].

To assess exposure of health care workers to M. tuberculosis, we reviewed work schedules, chart signatures, and patient assignments. Results of testing for fit of respirators with high-efficiency particulate air (HEPA) filters and results of annual tuberculin skin tests were obtained from employee records. Health care workers also completed a form about infection control practices and exposure to M. tuberculosis.

Acid-fast bacilli isolation rooms (that is, rooms in which the air pressure is negative in relation to the hallway and that have 6 air changes per hour) were installed in December 1994. The design of the ventilation system on the outbreak floor was reviewed, and smoke tube testing was done to assess the direction of air flow.

Data analysis was conducted using Epi Info, version 6 [17]. Categorical variables were compared using the chi-square test, and relative risks with 95% CIs were derived. Continuous variables were compared using the Wilcoxon rank-sum test.

Results

Top Methods Results Discussion Author & Article Info References

Seven persons had outbreak cases of tuberculosis: six patients already known and one health care worker identified during the investigation. All of these persons had AIDS. All of their M. tuberculosis isolates were resistant to isoniazid and rifampin and had identical DNA fingerprints. Two generations of nosocomial transmission were identified (Figure 1). Case-patient 1 was admitted to the hospital with a diagnosis of tuberculosis; was placed in a private room; and began receiving therapy with isoniazid, rifampin, pyrazinamide, and ethambutol. From the time this patient was admitted, persons entering the patient's room were required to use HEPA-filter respirators. Case-patient 1 refused to remain in the room and was the source of infection for case-patient 2, who was hospitalized at the same time across the hallway.

View larger version (13K): [in this window] [in a new window]   Figure 1. Timeline demonstrating two chains of nosocomial transmission of tuberculous organisms between August 1994 and September 1995 in a hospital in Chicago.

Case-patient 2 was readmitted to the hospital 10 weeks later with fever, sinus pressure, and cough. Because the chest radiograph at admission was normal, tuberculosis was not initially suspected. A sputum smear obtained 3 days after admission was positive (3⁺) for acid-fast bacilli. Treatment with isoniazid, rifampin, and ethambutol was started, and persons entering the patient's room were required to use HEPA-filter respirators. It was reported that the door to the patient's room was malfunctioning and failed to remain closed. Case-patient 2 remained in his room and yet was the source of infection for case-patients 3, 4, 5, and 6, who were hospitalized on the same ward. Transmission to case-patient 6, who was restricted to bed rest, occurred during this patient's short 23-hour hospitalization. Case-patient 7, the health care worker, drew blood from case-patient 2 before case-patient 2 received a diagnosis of tuberculosis and occasionally worked on the outbreak floor while case-patients 1 and 2 were present. For case-patients 2, 3, 4, 5, 6, and 7, no exposure to M. tuberculosis other than that which occurred in the hospital was identified. Case-patients 1 and 2 were never in an acid-fast bacilli isolation room while they were infectious because no such rooms were available until December 1994. Case-patients 4, 5, 6, and 7 were all placed in acid-fast bacilli isolation rooms when tuberculosis was suspected; case-patient 3 did not have sputum samples smear-positive for acid-fast bacilli. Smoke tube testing done in April 1995 revealed that the rooms occupied by case-patients 1 and 2 while these patients were infectious had air pressure that was positive in relation to the hallway. The acid-fast bacilli isolation rooms had air pressure that was negative in relation to the hallway. Ultraviolet germicidal irradiation was not used in the hospital.

The outbreak floor was composed of two separate wings with a central nursing station. Of the 169 patients on the floor, 5 had tuberculosis (case-patients 2, 3, 4, 5, and 6). Of the 164 patients who did not have tuberculosis, 71 (43%) had AIDS. Fifty-three of the 164 patients died before completing screening; none (including 37 patients with AIDS) received a diagnosis of tuberculosis. Of the remaining 111 patients, 5 had positive results on tuberculin skin testing, 34 had negative results, 5 had previously had positive results, 27 were anergic, and 40 had unknown test results. All 5 patients with positive test results had other risk factors for M. tuberculosis infection (for example, foreign birth).

All 5 patients were on the same ward as case-patients 1 or 2 while these patients were infectious. No evidence of transmission among patients who did not have AIDS or patients exposed only to case-patients 3, 4, 5, or 6 was identified. Twenty-six patients with AIDS (all of whom were known or presumed to be anergic) were exposed to case-patients 1 or 2. At the time of exposure, case-patients had lower CD4⁺ T-lymphocyte counts than non-case-patients and were more likely to be ambulatory (Table 1).

Of the 104 exposed health care workers, 17 who had previously had positive results on skin tests and 13 who had resigned from the hospital were excluded from analysis. Of the remaining 74, 11 (15%) had tuberculin skin test conversions (6 nurses, 4 housestaff physicians, and 1 ward secretary) and 4 had positive skin test results, but no baseline test results. None had any other identified exposure to tuberculosis. The remaining 59 (80%) had negative results on skin testing at least 12 weeks after exposure. Except for case-patient 7, no health care worker developed tuberculosis.

Case-patient 2 was the source of most of the episodes of transmission to health care workers. While case-patient 2 was hospitalized with infectious tuberculosis, health care workers with skin test conversions worked more days on the floor than did health care workers with negative test results. Health care workers who had conversion were no more likely to have provided direct care to a case-patient, including case-patient 2, than were workers with negative test results (Table 1).

Forty-two (57%) of 74 exposed health care workers responded to questions about infection control practices. Those with negative test results were no more likely than those with skin test conversion to report always wearing a HEPA-filter respirator when entering the room of a patient with tuberculosis (25 of 29 workers compared with 4 of 5 workers; 8 workers did not recall whether they used HEPA-filter respirators). The results of respirator-fit testing before exposure were available for 68 health care workers; 62, including 80% of those with skin test conversions, passed this testing.

Discussion

Top Methods Results Discussion Author & Article Info References

This investigation provides evidence of nosocomial transmission of multidrug-resistant M. tuberculosis, including 1) exposure of case-patients who had secondary cases of tuberculosis to a source case-patient on the same floor of the hospital; 2) consistency of incubation periods with those of previous nosocomial outbreaks among patients with AIDS [1-3]; 3) isolates with matching patterns of drug susceptibility and DNA fingerprints; 4) lack of any other identified exposure to tuberculosis; and 5) exposure of all health care workers with skin test conversion to a case-patient who was not in an acid-fast bacilli isolation room.

Several factors may have contributed to the transmission of M. tuberculosis, but the evidence for the role of insufficient environmental controls is most compelling. Case-patients 1 and 2 were both infectious, and the flow of contaminated air from their rooms (which had air pressure that was positive in relation to the hallway) facilitated the spread of infectious droplet nuclei throughout the outbreak floor. Organisms were transmitted to health care workers who did not directly care for a patient with infectious tuberculosis and to one secretary who did not have patient care responsibilities. In this situation, a respiratory protection program did not prevent transmission to health care workers; these findings are consistent with previous evidence that inadequate environmental controls were an important factor in nosocomial transmission [8]. Case-patients 1 and 2 were both placed on antituberculous medication either at admission or within 3 days of admission, but the delayed recognition of and inadequate therapy for multidrug-resistant tuberculosis could have contributed to transmission. In many other reports [2-7], delayed recognition was a common reason for transmission of tuberculous organisms.

Case-patient 2 shows that a patient with AIDS can have infectious tuberculosis despite a normal chest radiograph. Active pulmonary tuberculosis should be suspected in patients who have AIDS, fever, and cough, regardless of radiographic findings [10]. The evidence of transmission to a patient exposed for only 23 hours underscores the urgency with which patients who have signs and symptoms compatible with tuberculosis should be immediately isolated and evaluated.

Limitations of this investigation include the possible underestimation of transmission to exposed patients as a result of a high patient mortality rate. In addition, immunosuppression among patients may have made tuberculin skin testing unreliable. Skin test results were not available for 40 of the patients who may have been exposed; however, most of these patients were not known to have HIV infection, and their private physicians did not consider them to be at high risk for tuberculous infection or disease. The number of exposed health care workers who became infected without directly caring for a patient with tuberculosis may have been overestimated because chart signatures and assignment sheets may have been incomplete. Information is lacking for health care workers who resigned from the hospital or did not complete the questionnaire. Use of respirators may have been overestimated as a result of self-reporting.

As of 1993, 38% of hospitals in the United States reported not having acid-fast bacilli isolation rooms consistent with guidelines from the Centers for Disease Control and Prevention [18]. Inadequate isolation facilities and delayed diagnosis of tuberculosis may contribute to the airborne spread and nosocomial transmission of M. tuberculosis. Germicidal ultraviolet irradiation may be used as a supplemental environmental control measure to reduce the concentration of infectious droplet nuclei in the environment [10]. The excellent skin testing program in this hospital facilitated the detection of new infections. A respiratory protection program alone cannot protect all health care workers from infection with M. tuberculosis, especially those who work on the wards but do not enter patient rooms. Hospital infection control programs should emphasize the prompt identification of patients with tuberculosis and isolation consistent with the recommendations of the Centers for Disease Control and Prevention. Although all cases of active tuberculosis in this outbreak were seen in persons with AIDS, inadequate administrative and environmental controls put all patients and hospital staff at risk for tuberculous infection and disease.

Drs. Ridzon, Valway, Castro, and Onorato: Division of Tuberculosis Elimination, Centers for Disease Control and Prevention, 1600 Clifton Road, MS E-10, Atlanta, GA 30333.

Dr. Luskin-Hawk and Ms. Schultz: Lakeshore Infectious Disease Associates, 2900 North Lake Shore Drive, Chicago, IL 60657.

Dr. Paul: Westside Communicable Disease Division, Chicago Department of Health, 2016 West Ogden Avenue, Chicago, IL 60612.