On March 23, 2003, a 74-year-old immunocompromised man was transferred to our ICU from a hospital where the original cluster of Toronto’s SARS cases occurred (2). The patient originally had signs and symptoms consistent with a presumptive diagnosis of community-acquired pneumonia. Before transfer, SARS was excluded from the differential diagnosis because the patient had not traveled, had never left the emergency department of the referring hospital, and had only had a single recent outpatient visit to an area of the original hospital in which SARS had not been identified. Upon arrival in our ICU, the patient was placed in precautions for methicillin-resistant Staphylococcus aureus (MRSA) pending admission screening results (6). Therapy with broad-spectrum antimicrobial drugs was initiated. Humidified high-flow oxygen was administered for the first 5 h, noninvasive positive pressure ventilation by oronasal mask for the next 18.25 h, and invasive mechanical ventilation for the subsequent time (7.5 h). Endotracheal intubation required fiber-optic placement. That the extent of the outbreak at the referring institution was larger than originally appreciated became apparent at this time; therefore, the patient was transferred to another facility for placement in negative pressure isolation for possible exposure to SARS. Subsequently, his family members became ill, and the SARS-associated coronavirus was identified in the patient’s respiratory secretions (polymerase chain reaction testing of bronchoalveolar lavage confirmed the diagnosis of SARS).

Once the risk for SARS was identified, all patients in the ICU were considered to have been potentially exposed. To prevent spread of SARS, we closed the ICU to admissions and discharges and implemented strict respiratory and contact precautions for all remaining patients. We quarantined 69 healthcare workers who were considered to be at high risk for developing SARS. On the basis of our understanding of disease transmission, we arbitrarily decided that persons at high risk included anyone who had entered the index patient’s room or who had been in the ICU for >4 hours during the patient’s 30.75-h stay.

After research ethics board approval and informed consent, two researchers used a structured questionnaire to interview quarantined healthcare workers. The questionnaire elicited demographic information, details about health, and information about exposure to the index patient. Time of exposure was categorized as follows: <1 min, 1–10 min, 11–30 min, 31–60 min, 1–4 h, or >4 h. Exposure proximity, procedures performed, and infection-control precautions were documented. Each healthcare worker was asked about symptoms suggestive of SARS that developed during or after the quarantine period.

For healthcare workers in whom suspected or probable SARS developed, additional data were collected about the nature and course of their illness. Suspected and probable SARS were defined according to the definitions issued by the World Health Organization (WHO) (7). Symptoms of suspected SARS were a fever >38°C, respiratory symptoms, and an epidemiologic link with a SARS patient; all quarantined healthcare workers were considered to have an epidemiologic link on the basis of contact with the index patient. Probable SARS was defined as suspected SARS with radiographic lung infiltrates.

Of the 69 quarantined patients, 63 were interviewed. Five declined, and one could not be contacted. SARS did not develop in healthcare workers who were not quarantined and patients who had been in the unit at the time of the exposure.

Our results suggest that proximity and duration of contact to a patient with SARS are associated with risk for viral transmission, an observation suggested by others (2–4). In addition, certain procedures, such as endotracheal intubation, pose increased risk. These findings may be predictable given that SARS is thought to spread primarily by large droplets (9).

Three of the six persons in whom SARS developed after entering the index patient’s room may not have adhered to standard MRSA precautions in that they performed procedures which involved contact with mucous membranes without wearing gloves. Furthermore, we were unable to determine if hand washing impacted SARS transmission, as this information was not collected.

During our study, we made two important observations. First, SARS developed in one healthcare worker despite the fact that the worker wore an N-95 mask, gown, and gloves. Second, SARS developed in another healthcare worker who had no identified contact with the index patient or with any other persons known to have SARS. In the case of the first healthcare worker, the absence of eye protection may have contributed to disease transmission. In addition, although this person wore an N-95 mask while in the patient’s room, he had not been fit-tested for this mask; however, fit-testing should not be necessary if the SARS-associated coronavirus is spread by large droplets (6). As a result of this and similar episodes of SARS transmission in the Toronto area (10), the province of Ontario has now made specific recommendations for healthcare workers performing intubation that involve increased protection (available from: URL: www.sars.medtau.org) (11), and protective eye wear is currently mandated for patient encounters. In the second case, transmission could have occurred in a number of possible routes. The nurse may have come within sufficient range of the SARS patient to be exposed to large droplets. Recent reports indicate that the virus may survive for several hours on fomites or in body secretions (12) and raise the possibility of transmission by indirect contact with contaminated objects or of inadvertent carriage and spread by another healthcare worker. Fecal transmission is unlikely as the patient did not have a bowel movement during his stay. True airborne spread may also have occurred. Although evidence does not support this route of transmission for the SARS-associated coronavirus, existing literature suggests that other coronaviruses may be spread by an airborne route in certain circumstances (13).

Given our lack of knowledge about the transmissibility of SARS at the time this exposure occurred, we made a conservative decision to quarantine for 10 days all persons who were in the unit for at least 4 h or who had a history of entry into the affected patient’s room. In addition, we closed the ICU to admissions and discharges for a 10-day period, markedly affecting our institution’s ability to deliver health care. In fact, during the Toronto outbreak, several of the city’s ICUs were closed as a result of quarantine and illness in staff with similar consequences (14); by infecting healthcare workers, SARS has an impact on the health of an entire community. A less aggressive quarantine approach may have been as effective in controlling transmission and allowed more staff to be available for work. For instance, only persons who have had direct contact with the patient (i.e., entered the patient’s room) could have been quarantined. If we had taken this approach, the quarantine would have excluded six persons with SARS from the workplace but only removed 25 of the 62 persons who remained well. However, this approach would have missed one healthcare worker in whom SARS developed. Another approach might be to monitor staff closely for SARS-related symptoms while they continue their usual activities and quarantine only those in whom symptoms occur. This approach would require evidence that SARS cannot be transmitted before symptom onset, confidence in the facility’s ability to identify symptomatic staff, and reliability of healthcare workers in reporting symptoms. We think that our quarantine approach prevented secondary spread of illness to other persons who may have come in contact with the workers in whom SARS developed.

Our study involved a small number of cases, and definitive conclusions cannot be drawn from a report of this size. For example, although SARS developed in our staff within the 10-day quarantine period, others have demonstrated that the time period from infection to onset of symptoms may be >10 days (15). One of the strengths of our study is that the exposure occurred during a defined period in a contained unit, and as such, there is less potential for confounding caused by the exposure of healthcare workers to multiple SARS patients.

Our observations emphasize the consequences of missing the diagnosis of SARS for even a relatively brief period. In our experience, we would make the following recommendations. First, the possibility of unexpected exposure of healthcare workers to patients with SARS should be anticipated, and once such exposure is recognized, those deemed to be at risk for SARS transmission should be promptly quarantined. Second, vigilant surveillance for symptoms of SARS must be maintained by all healthcare workers who work in institutions with SARS patients; SARS may develop in healthcare workers even when they do not have direct exposure to patients with SARS. In addition, protocols for managing patients with SARS should include not only contact and respiratory precautions but also procedures that minimize patient contact since duration and proximity of contact increase the risk for transmission of SARS. Finally, additional precautions should be taken when performing high-risk procedures, such as endotracheal intubation (11).

Though many of the healthcare workers in our ICU were exposed to the patient with SARS, our experience suggests that the greatest risk for SARS transmission occurs in those healthcare workers with prolonged exposure or direct physical contact with the patient. Use of gowns, gloves, and masks as barriers appears to reduce the risk for SARS transmission in most but not all situations. Additional information will be needed to determine if modes of transmission beyond droplet spread are important. We think this information will be helpful to institutions dealing with similar exposures to patients with SARS and developing quarantine protocols.