Management of a Tuberculosis Exposure in an Oncology Hospital
15 July 2005
The transmission of tuberculosis within health care institutions is a well-known phenomenon(1). Early identification and appropriate treatment of active cases are fundamental tuberculosis control strategies. As identification can sometimes be problematic, an assessment of hospital ventilation also plays an important role in determining the potential extent of exposure(2). Should staff and patient exposure inadvertently occur, contact tracing and tuberculin skin testing (TST) are employed to identify those contacts at high risk of developing active disease. Such screening traditionally follows an "onion skin model", in which the closest contacts are screened first, and more remote contacts are only screened if the inner circle shows evidence of infection(3).
While this model may be appropriate in the household setting, it is less relevant in the hospital setting, as many potential contact episodes are not well defined. The risk of transmission will be different for each exposure and is dependent on the patient, and on procedural and environmental factors. Patient factors include sputum smear status(4), chest radiographic findings, and the presence of cough(5). Procedural factors include cough-generating procedures and the use of appropriate masks, and environmental factors include the duration of exposure, the size of the room, and the number of fresh air exchanges per hour where the exposure occurred(1).
Once an exposure has been determined to be significant, TST of contacts can be problematic for logistic reasons, i.e. follow-up reading, but also because of characteristics of the test itself, such as false-positive and false-negative reactions. False-negative reactions are well known to occur in patients with impaired cellular immunity, such as HIV-positive individuals with low CD4 counts(6), patients with end-stage renal failure(7,8), and those taking immunosuppressive medications(9). Patients with hematologic malignancies and bone marrow transplant recipients are commonly considered to be at risk of false-negative TST reactions as a result of defects in cellular immunity, although this has not been well studied.While a positive TST result is helpful, it is unclear whether a negative result represents a true negative or the inability to mount an immune response. Accordingly, decisions regarding recent infection have to take into account other factors, such as exposure histories and chest radiographs.
In April 2004 a patient being treated for acute myeloid leukemia (AML) was discovered to have active pulmonary tuberculosis. A review of his visits to our facility revealed that there were two inpatient admissions and several outpatient appointments that may have potentially exposed staff and highly immunocompromised patients. This article will outline our investigation and management decisions in a setting where the role of traditional TST-based screening was uncertain and presumed to be of benefit.
Infection Prevention and Control was informed in April 2004 that an 84-year-old patient with AML (patient X) admitted to the allogeneic bone marrow transplant unit was smear-positive for Mycobacterium tuberculosis from a mediastinal lymph node. A chart review revealed that he had previously been admitted in December 2003 for chemotherapy with ARA-C (1-ß-D-arabinofuranosylcytosine) and daunorubicin. In January 2004, he experienced worsening fevers and shortness of breath despite receiving broad spectrum antibiotics and antifungal medications, including ciprofloxacin, cefepime, flagyl, tobramycin, and fluconazole. It is of note that he had no cough. A computed tomography (CT) scan of the thorax revealed mediastinal adenopathy and diffuse parenchymal changes, but it was difficult to comment on the presence of infiltrates because of motion artefact. It was believed that his symptoms were secondary to his underlying hematologic malignancy, and he was treated with high doses of dexamethasone. His symptoms promptly resolved, and he was discharged home in mid-February. He subsequently visited the hospital 14 times at several outpatient clinics.
In April 2004, patient X was readmitted with symptoms of superior vena cava syndrome and worsening shortness of breath. A CT scan of the thorax showed increased nodularity of the lung parenchyma and a dramatic increase in the previously noted mediastinal adenopathy. He was treated with broad spectrum antibiotics but remained afebrile with no cough. Shortly thereafter, a biopsy of a left mediastinal lymph node revealed M. tuberculosis, and a spontaneous sputum sample revealed numerous acid-fast bacilli on smear. Antituberculous therapy with isoniazid, rifampin, pyrazinamide, and ethambutol was started, but the patient died unexpectedly 5 days later.
Because patient X had pulmonary symptoms in January that were unresponsive to antibiotic and antifungal therapy, and an abnormal CT of the thorax that subsequently worsened, it was postulated that his symptoms in January were, in fact, related to undiagnosed tuberculosis. His symptoms likely improved and then worsened because of the use of high doses of corticosteroid therapy. On the basis of this information, it was decided that he was infectious from the beginning of January and that his infectiousness increased over time, to April, when he was found to have numerous acid-fast bacilli on smear. The patient was no longer considered infectious once the diagnosis had been made as he was moved to a negative-pressure room and staff were required to wear submicron filtering masks.
An environmental assessment was conducted to determine the air changes (ACH) in the various areas of the hospital where patient X had visited. The feet per minute (fpm) supply air was determined by measuring at the face of the diffuser duct using the Davis Instruments Turbo Meter Wind Indicator (Hayward, California, U.S.A.). The flow rate was calculated by multiplying the velocity and the area of the duct. The ACH for negative pressure rooms was calculated using cubic feet per minute (cfm) of exhausted air. For the other areas of the hospital the supply cfm was used to calculate the ACH.
Determination of contacts
An exposure list was created from the hospital patient information system of patients in the inpatient and clinic settings. Potential health care worker contacts were determined from staffing lists. Contacts were defined as roommates of patient X on the inpatient floors. Fortunately, patient X spent most of his time in a private room with the door closed; however, two roommates were identified.
All outpatients who attended the same clinics on the same day as patient X were considered as possible contacts. Because of the open structure of the clinic waiting area and the transfusion centre, and lack of specific patient appointment times, all patients present during patient X’s appointment and 1 hour after his departure were considered exposed. The 1-hour cut off was chosen on the basis of the ventilation parameters for the exposure areas (see Results). Some areas had specific arrival and departure times for patients, which assisted in limiting the exposure list.
Once identified, contacts were stratified into arbitrarily defined high-risk and low-risk categories according to the duration and intensity of their exposure. Low-risk patients were defined as those with only one outpatient contact with patient X. High-risk patients included all inpatient roommates and outpatients having two or more outpatient contacts with patient X. A broader definition of exposure was used for staff contacts. Exposed staff included those on the inpatient units during patient X’s admission and those who worked at the various ambulatory clinics during his visits. Inpatient staff included nurses, physicians, support partners, student nurses, and staff providing occupational and physical therapy, respiratory therapy, and dietary and nutritional care.
Follow-up of contacts
An occupational health nurse trained in planting and interpreting tuberculin skin tests conducted testing on all nonpalliative exposed inpatients and outpatients during follow-up appointments. As this testing occurred > 8 weeks after the exposure, baseline TST results were not available for most patients. Follow-up skin testing to assess for conversion was therefore not required. Some patients were unable to return to the hospital to have their tests read and were given a letter asking the family physician to record the induration and return the letter to Infection Prevention and Control. All patients were recommended to have a chest radiograph regardless of their TST status, as the TST was believed to be insensitive in this population.
The same occupational health nurse conducted baseline tuberculin skin testing, if needed, and follow-up testing of exposed health care workers. Chest radiographs were performed for those health care workers with a TST conversion (reaction of 10 mm induration) or whose induration was > 5 mm on their initial testing.
The treating hematology/oncology physicians were educated regarding the possible limitations of TST in this population and were instructed to consider active tuberculosis in any patients who had compatible symptoms, regardless of their final TST result. It was also recommended that patients in the high-risk category be considered as candidates for preventive therapy, although the final decision to provide this therapy was left up to the treating physician and the patient.
During patient X’s first inpatient visit he spent a total of 69 days on unit A. He was initially in semiprivate accommodation for 32 days, then spent 13 days in a single room and the final 24 days in a negative-pressure room. Patient X’s semiprivate room was calculated to provide approximately 12 ACH and the negative pressure room 8 ACH. The air supply is 100% fresh air.
During his subsequent admission he stayed on unit B for 5 days, after which he was put into an airborne isolation room for the remainder of his admission, on unit C. Unit B is a positive-pressure bone marrow transplant unit that keeps the unit and patient doors closed to maintain appropriate ventilation. Each room is supplied with 275 germ-free ACH per hour by recirculating room air through three separate fans and high-efficiency particulate air filters.
In between both inpatient visits patient X attended outpatient clinic appointments on 9 separate days. He spent approximately 4 hours total in various areas, including the ambulance holding area and clinics. His stay could have increased to 12 hours if he attended the transfusion centre as well. The ambulance holding area was calculated to have 7 ACH and the transfusion centre 5 ACH. Both areas are supplied with 100% fresh air.
A total of 177 inpatient and outpatient contacts were identified. The mean age was 56.6 years (range 19 to 90 years). The number of exposures per patient is shown in Figure 1. The three patients with > 5 exposures included two inpatient roommates as well as an ambulatory patient with the same treating oncologist. No direct face-to-face contact could be confirmed for any patient. The diagnoses of exposed patients included leukemia, lymphoma, myeloma, and solid organ tumours, all at various stages of treatment. Eight patients had received bone marrow transplants between 3 months and 7 years before the exposure, and seven had received their bone marrow transplant during the exposure period (January to April)
Figure 1. Number of exposures per patient
The results of patient TST are shown in Table 1. The occupational health nurse was unable to conduct TST in 40 patients (23%), who were either deceased or were in palliative care at the time of follow-up. Eight patients were positive from previous exposures, two patients refused skin testing, and the results for 25 patients were not returned from their family physicians. Thus, results were available for 102 (57.6%) of the patients who underwent TST. Five of these (5%) were TST positive at the > 5mm induration cut off for close contacts. In fact, all five patients had a TST of > 10 mm induration. Four of the positive patients had only one potential contact with patient X, i.e., were at "low risk".
None of the low-risk contacts was considered for preventive therapy. The provision of preventive therapy for high-risk contacts was significantly complicated by extensive comorbid disease, causing signs and symptoms such as fever, weight loss, chills, pulmonary infiltrates, and lymphadenopathy, which could be compatible with active tuberculosis, and by concerns of hepatoxicity or drug interactions. In general, patients were treated conservatively and were followed for the development of active disease. No cases of active tuberculosis have been identified 1 year after the exposures.
A total of 112 staff members were identified as contacts. Baseline test results showed that 36 were previously positive, and 76 were negative. The previously TST-positive staff were assessed by the occupational health nurse for signs and symptoms of TB. All TST-negative staff were recalled at 3 months for follow-up testing when appropriate. One staff member was found to have TST conversion and completed preventive therapy. This staff member worked on Unit B and cared for patient X just before his diagnosis of tuberculosis. No other cases of infectious tuberculosis were known to have been present on Unit B since the health care worker’s previous negative skin test.
Table 1. Tuberculin skin testing results for high- and low-risk patient contacts
|Contacts (%)||TST -ve (%)||TST +ve (%)*||No TST result†||TST not done‡|
|High risk||55 (31.1)||31 (56.4)||1 (1.8)||3 (5.5)||20 (36.3)|
|Low risk||122(68.9)||66 (54.1)||4 (33)||22 (18.0)||30 (25.0)|
|Total||177||97 (54.8)||5 (2.8)||25 (14.1)||50 (28.2)|
|*Positive => 5mminduration
†Result not received from family physician.
‡Patients either had a previous positive result (8 patients), were deceased or in palliative care (40 patients), or refused testing (2 patients).
Exposure to patients with pulmonary tuberculosis in hospitals can be a frequent occurrence in settings where patients at high risk of infection and reactivation are treated. Various guidelines state that administrative controls are central to preventing patient and health care worker exposures(1,10). Possible controls include tuberculosis management programs, strategies to aid in the early identification of TB patients, and airborne isolation policies. In general, it is understood that the follow-up of contacts is far more complicated and less efficient and effective than preventing the exposure from occurring. This is particularly true in the hospital setting, where the definition of what constitutes a significant exposure may be difficult to determine. While TST is a central component to contact tracing, its role in identifying contacts who have underlying immunodeficiency may be less important because of the limitations of the test.
We attempted to measure the ventilation in the various parts of the hospital where patient X was cared for. In general, the ventilation was found meet or exceed recommended parameters(1). Given that adequate ventilation has been shown to be a key factor in preventing tuberculosis transmission in hospitals(2), this information helped us to limit our contact list to only patients who were in the same clinic at the same time as patient X and up to 1 hour after he left. At 5 ACH, over 99% of infectious particles would have been removed within 1 hour(11). Were the ventilation below standard, we would have been obliged to expand our contact list, as infectious M. tuberculosis particles are known to be capable of remaining airborne indefinitely(12). Expanding the contact list despite excellent ventilation would have resulted in more work for little benefit and have caused needless stress for patients who likely were not infected. Inappropriate hospital ventilation has been shown to be at the root of one large hospital outbreak of multidrug-resistant tuberculosis, which resulted in a contact list of 1,400 patients and staff(13).
We were fortunate to have had reliable recent TST data for the majority of our health care workers, as this enabled us to determine that patient X was likely infectious and did transmit tuberculosis to one health care worker. We might have expected that patient infections would have occurred as well, especially for patients in the "high risk" category. Five percent of our patients had a positive TST, and three of the five were from TB endemic countries, namely China, India, and Sri Lanka. It is unknown whether their positive results represent exposure in their country of origin rather than recent infection.
In order to determine whether the TST was a good measure of patient exposure, we attempted to determine the theoretical proportion of patients who should have had a positive TST result at baseline, according to their country of origin (Table 2). Country of origin was available for 100 of the 102 patients with TST results. Using the prevalence rates of latent infection reported by Dye et al. in the various World Health Organization regions(14), we would have expected 10 patients to be positive for TB infection at baseline at >10 mm, although the difference between predicted and observed populations did not reach statistical significance (point estimate = 0.5, 95% confidence interval 0.162 to 1.172, Chi squared test). Given that many of these foreignborn patients would have received BCG vaccination as well, we might have predicted an even higher proportion of positive tests. This crude comparison suggests that using the TST to identify recently infected contacts in this immunocompromised population may have been of limited benefit. It is unknown whether newer interferon-gamma-based assays for tuberculosis infection would aid in identifying high-risk contacts in this situation, as these tests have not been well studied in immunocompromised populations(15).
Table 2. Expected number of TST positive contacts based on World Health Organization(14) region of origin
|Region||Prevalence rate, %||Number of contacts||Expected number of
|Actual number of positive
|South East Asia||44.0||2||0.88||1|
|*The Canadian-born population was assumed to have a prevalence rate of 0.5%.
Our assigning patients to arbitrarily defined high- and low-risk categories did not significantly assist in our management of the outbreak in that it did not influence who received preventive therapy. The nature of this patient population meant that a substantial proportion of patients were receiving palliative care and, hence, not ideal candidates for preventive therapies. Furthermore, their concurrent malignancies mimicked many of the symptoms of tuberculosis, making it quite difficult to rule out active tuberculosis. There was great reluctance to start such patients on preventive therapy. Finally, a significant proportion of patients were considered at risk of drug-induced hepatitis and interactions.
We report the difficulties in attempting to employ traditional contact tracing methods in a cancer hospital. While we have evidence that our index patient was infectious, it is unknown whether transmission occurred to other patients despite our intensive investigation, although it is reassuring that no additional cases of tuberculosis have been detected in this cohort 1 year after the exposure.We believe that our assessment of the hospital’s ventilation system aided us in containing our contact list. Given our patient TST findings and our experience in providing treatment for latent infection, expanding our contact list would have resulted in substantially more work and anxiety, for little benefit. Our experience supports the current emphasis on administrative and environmental controls to prevent tuberculosis transmission in hospitals.
Menzies R, Conly JM. Tuberculosis control in Canadian health care institutions. In: Long R, ed. Canadian tuberculosis standards. 5th ed. Ottawa: Health Canada, 2000.
Menzies D, Fanning A, Yuan L et al. Hospital ventilation and risk for tuberculous infection in Canadian health care workers. Canadian Collaborative Group in Nosocomial Transmission of TB. Ann Intern Med 2000;133(10):779-89.
Tannenbaum TN, Fanning A. Contact follow-up and outbreak management in tuberculosis control. In: Long R, ed. Canadian tuberculosis standards. 5th ed. Ottawa: Health Canada, 2000.
Rose C, Zerbe G, Lantz S et al. Establishing priority during investigation of tuberculosis contacts. Am Rev Resp Dis 1979;119:603-09.
Loudon RG, Romans WE. Cough frequency and infectivity in patients with pulmonary tuberculosis. Am Rev Resp Dis 1969;99:109-11.
Makowitz N, Hansen NI,Wilcosky TC et al. Tuberculin and anergy testing in HIV-seropositive and HIV-seronegative persons. Ann Intern Med. 1993;119:185-93.
Poduval RD, Hammes MD. Tuberculosis screening in dialysis patients - is the tuberculin test effective? Clin Nephrol 2003;59(6):436-40.
Fang HC, Chou KJ, Chen CL et al. Tuberculin skin test and anergy in dialysis patients of a tuberculosis-endemic area. Nephron 1998;91(4):682-7.
Menzies D, Pourier L. Diagnosis of tuberculosis infection and disease. In: Long R, ed. Canadian tuberculosis standards. 5th ed. Ottawa: Health Canada, 2000.
Centers for Disease Control and Prevention. Guidelines for preventing the transmission of Mycobacterium tuberculosis in health-care facilities, 1994. MMWR 1994;43(RR-13):1-132.
Centers for Disease Control and Prevention. Guidelines for environmental infection control in health-care facilities. MMWR 2003;52(RR-10):1-44.
Riley RL, O?Grady F. Airborne infection: transmission and control. New York: Macmillan, 1961.
. Breathnach AS, de Ruiter A, Holdsworth GM et al. An outbreak of multi-drug resistant tuberculosis in a London teaching hospital. J Hosp Infect 1998;39(2):111-7.
Dye C, Scheele S, Dolin P et al. Global burden of tuberculosis: Estimated incidence, prevalence, and mortality by country. JAMA 1999;282(7):677-86.
Pai M, LeeWR, Colford JL. Interferon gamma assays in the immunodiagnosis of tuberculosis: a systematic review. Lancet Infect Dis 2004;4:761-76.
Source: Z Hirji, MHSc, CIC, S Boodoosingh, MLT, CIC, Infection Prevention and Control Unit, University Health Network, Toronto; R Santos, RN, A Yang, RN, J Weeks, RN, Employee Communicable Disease Surveillance Unit, University Health Network; K Iverson, MHSc, S Lim, MD, M Gardam, MSc, MD, CM, MSc, Infection Prevention and Control Unit, University Health Network, Toronto.
Report a problem or mistake on this page
- Date modified: