METHODS

Section:

ChooseTop of pageAbstractMETHODS <<RESULTSDISCUSSIONReferencesCITING ARTICLES

To determine whether specific ventilatory strategies were associated with an increased risk of SARS development in healthcare workers, we undertook retrospective cohort analyses involving an intensive care unit (ICU) in which nine patients with SARS had been treated with various modes of ventilatory support. Seven patients with SARS were treated during the outbreak period; two patients were treated and discharged greater than 3 weeks before the outbreak and are not included. We considered and evaluated ventilatory exposures for the 2-week period before the outbreak, corresponding to the upper duration latency between exposure and SARS symptom onset in other series (2–4, 7, 8). Because of the concern of dispersal of respiratory droplets and secretions and limited available data, we chose to restrict our analysis to three particular respiratory practices to determine whether they conferred a high risk. We examined (1) the risk of developing probable or suspect SARS (1, 2) for physicians who performed endotracheal intubation for patients with SARS and respiratory failure versus physicians who cared for patients with SARS but did not perform endotracheal intubation, as well for nurses who assisted with endotracheal intubation versus nurses who cared for patients with SARS but were not present for endotracheal intubation; (2) the risk of developing SARS for nurses caring for patients with SARS with high-airflow, noninvasive positive-pressure ventilation (NIPPV); and (3) the risk of developing SARS for nurses caring for patients with SARS treated with high-frequency oscillatory (HFO) ventilation. In the later two groups, the comparison was to nurses caring for patients with SARS treated with conventional ventilation. Physicians formed the “at-risk” and control group for the first analysis, as all intubations were performed by physicians. For analyses of ventilatory techniques, nurses were considered rather than other healthcare workers (such as respiratory therapists or physicians) because of the greater likelihood for a longer duration of exposure and greater specificity of exposure (in the ICU studied, nurses generally care for one patient per shift, whereas respiratory therapists and physicians typically care for many different patients, with shorter durations of direct exposure to individual patients).

For each measure of association, factors previously identified to confer a greater risk of contracting SARS, such as age and presence of diabetes mellitus, were considered as potential confounders (7, 8). During the period of investigation, all patients were treated in negative-pressure isolation rooms, and healthcare workers wore gloves, gowns, N-95/PCM 2,000 masks, and hairnets. Eye or face shields were variably employed. Bronchoscopy and aerosol therapy were not performed. We explored the dose–response relationship between nursing exposure and the risk of developing SARS by examining not only absolute numbers of nurses affected in relationship to all exposed, but also the number of nursing shifts exposed in relationship to the total nursing shifts during the period of evaluation. We similarly compared dose–response relationships for physicians based on duration and degree of difficulty of intubation (greater than two attempts at passing the endotracheal tube, longer than 10 minutes of direct laryngoscopy/bag-mask ventilation, or copious patient secretions). Small sample sizes and limited information precluded determination of independent measures of association through multivariable logistic regression. We compared baseline normally distributed continuous characteristics (e.g., mean age ± SD) using Student's t test. As many comparisons contained small sample sizes, Fisher's exact tests were used to assess risk factors. Two-sided tests were used for all comparison, and 95% confidence intervals (CIs) were calculated for all relative risks (RRs).

RESULTS

Section:

ChooseTop of pageAbstractMETHODSRESULTS <<DISCUSSIONReferencesCITING ARTICLES

The epidemic curve of healthcare workers who contracted SARS after caring for critically ill patients is presented in Figure 1

Figure 1. Epidemic curve of intensive care unit (ICU) healthcare workers who contracted severe acute respiratory syndrome (SARS) through patient care. BiPAP = biphasic positive airway pressure; HFO = high-frequency oscillatory; MD = physician; RN = registered nurse; RT = respiratory therapist.

Download Figure | Download Powerpoint

[More] [Minimize]

Healthcare workers and the nature of their exposures are described in Table 1

As the numbers of nurses who assisted and did not assist endotracheal intubation were much greater than physicians, we were able to define their risk more fully. Nurses who assisted in endotracheal intubation were much more likely to develop SARS than were nurses who cared for patients with SARS in the ICU at other times (RR, 21.38; 95% CI, 4.89 to 93.37; p = 0.001) (Table 2). Three of the five nurses who developed SARS assisted during intubation of patients with SARS. The risk was similarly high when examined as a function of nursing shifts of exposure compared with shifts worked when endotracheal intubation was not performed (RR, 33.0; 95% CI, 7.50 to 145.30; p < 0.0001).

Nurses caring for patients receiving NIPPV may have been more likely to develop SARS than nurses caring for patients with SARS treated with conventional ventilation (RR, 2.33; 95% CI, 0.25 to 21.76; p = 0.5), but this was not statistically significant (Table 3)

Nurses caring for patients with SARS receiving HFO did not appear to have an increased risk of developing SARS than did nurses who cared for patients with SARS who received conventional mechanical ventilation (RR, 0.74; 95% CI, 0.11 to 4.92; p = 0.6) (Table 3). The association was similar when examined by nursing shift at risk (RR, 0.55; 95% CI, 0.1 to 3.12; p = 0.8), and neither reached statistical significance. In this group, similar to that observed with other interventions, SARS occurred in nurses caring for patients within 48 hours of their patient's admission to the ICU.

DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

We have shown that physicians and nurses involved in the early critical care period and endotracheal intubation of patients with SARS are at a substantially increased risk of themselves contracting the illness. This association is particularly strong among nurses present during the period of endotracheal intubation. Physicians who actually place the endotracheal tube also appear to be at increased risk, although this association did not reach statistical significance, possibly because of the smaller numbers of physicians caring for patients with SARS during the outbreak. It is also interesting that the RR may be higher for nurses than physicians. This might be because of a longer duration of exposure that nurses likely had in the peri-intubation period, whereas physician exposure is often limited to the procedure itself. The relationship among other ventilation practices that may also lead to increased exposure to viral-laden droplets and subsequent transmission of SARS is less clear—no statistically significant association between NIPPV or HFO could be found by examining an outbreak within one ICU.

Our study also offers insights into the importance of timing of SARS transmission from critically ill patients to healthcare workers. Although symptoms in workers arose over a 7-day period, we feel it was important to examine potential exposures in the 14 days before the date of onset, corresponding to upper limits of latency from exposure (7, 8). Transmission often appeared to occur within 48 hours of patients' admission to the ICU and often involved acute patient respiratory stabilization through endotracheal intubation and mechanical ventilation. There may be a number of potential explanations for this. First, the clinical deterioration that often precipitates transfer to an ICU may be associated with increased tachypnea, dyspnea, or coughing and thus increased dispersal of infectious respiratory droplets. As well, deterioration may involve the changes in respiratory support that also increase the likelihood of secretion dispersal. Although patients were treated in negative-pressure isolation rooms and full droplet precautions were in place before the onset of the outbreak, a compromise in such protective barriers may have been more likely with both patient deterioration and transfer (5). With increased awareness, education, and vigilance to infection control procedures; the addition of face shields, personal respirators for high-risk activities such as endotracheal intubation, the addition of bacterial–viral filters (Clear-Guard 3; Intersurgical Ltd., Berkshire, UK) on the exhalation valve of all mechanical ventilators; diminished exposure to acutely ill patients with SARS requiring stabilization and intubation; and possibly because of fewer other ventilation practices associated with a high risk of droplet dispersal (such as NIPPV), further nosocomial transmission was stopped within this ICU.

There are several limitations of our analysis. The small sample size and limited available data preclude extensive stratification or meaningful multiple logistic regression that might identify other and independent predictors. Certain associations may be altered by important confounding or effect-modifying variables we were unable to consider. Several other critical care interventions may be associated with a greater risk for transmission of SARS to healthcare workers. For example, patients receiving HFO typically have severe respiratory disease and may therefore produce a larger quantity or concentration of virus in sputum or expiratory gas than less ill patients, producing an increased risk for transmission of SARS to healthcare workers on this basis alone, irrespective of mode of ventilation. If this were the case, however, this would strengthen our observation that no conclusive positive association exists between HFO use and SARS transmission. We did consider the effects of known potential common confounders such as age and diabetes mellitus on the associations we report (7, 8). We were able to stratify risk of endotracheal intubation according to physician or nursing exposure. Among physicians, we did not find a clear trend toward a greater or lesser risk based on the level of training. Additionally, and again somewhat surprisingly, we did not detect a clear trend toward an increased risk for performing multiple intubations or among difficult intubations. This quite possibly relates to the small sample size of our associations and may be further clarified in an ongoing multicenter epidemiologic investigation.

Proximity and duration of contact to a patient with SARS have previously been found to be associated with a higher risk of viral transmission (9). We attempted to explore the effects of duration of exposure by examining healthcare worker risk in terms of both patient exposures and number of shifts exposed. Among nurses caring for patients with SARS, we evaluated dose–response exposure among nurses who cared for more than one patient with SARS and also among those who worked more than one shift with individual patients, without demonstrating obvious important trends. Because respiratory therapists were found to have exposures to nearly all patients and ventilation strategies in the ICU over any given shift, we were unable to evaluate their risk in light of particular independent exposures. During the period of investigation, there was extensive bi-daily screening of all healthcare workers for symptoms and signs of SARS. If there was any suspicion of healthcare worker illness, they were assessed and placed under home quarantine observation, minimizing the chance that exposure and illness could have resulted from staff-to-staff contacts as opposed to direct patient care.

Certain high-risk components of SARS nursing care, such as tracheal suctioning, may also be an important confounder in this type of analysis. Tracheal suctioning was generally not performed among patients with SARS ventilated with either NIPPV or HFO. In-line, closed system endotracheal suctioning was used for patients with SARS receiving endotracheal intubation. If indeed endotracheal suction might increase the risk of respiratory droplet dispersion, this would likely act to dilute any estimate of positive association discovered between NIPPV, HFO, and nursing acquisition of SARS.

The ICU SARS Working Group of the Provincial Operations Committee of Emergency Management Ontario (an agency of the Solicitor General of Ontario) and others have developed guidelines for procedures such as endotracheal intubation, cardiopulmonary resuscitation, and mechanical ventilation in critical care areas during a SARS outbreak (10, 11). The guidelines include specific recommendations on the use of personal protection devices specify that the most qualified individual available should perform endotracheal intubation and that therapies such as prolonged NIPPV and aerosolized bronchodilator or humidification therapies generally should not be initiated where safe alternatives are available. It has been recommended by many clinicians involved in the Asian SARS outbreak that noninvasive ventilation may be preferred over early endotracheal intubation and mechanical ventilation because of the risk to healthcare workers involved with endotracheal intubation (12). Although noninvasive ventilation might be considered for patients with the expectation of very near-term improvement, much the same as for patients with hypoxemic cardiogenic pulmonary edema or hypercarbic respiratory failure because of a chronic obstructive pulmonary disease exacerbation (13, 14), observational studies of critically ill patients with SARS report that this form of acute lung injury is unfortunately generally not rapidly reversible.

These preliminary observations and others within the greater Toronto healthcare system have precipitated an ongoing epidemiologic investigation into the specific nature of risk of transmission between patients with SARS and healthcare workers. This investigation, with the assistance of infection control officers of hospitals of the greater Toronto area, local departments of public health, and the Centers for Diseases Control and Prevention, is likely to offer a detailed understanding of the mode of transmission and best means of prevention. In the interim, we advise caution in initiating procedures that may be associated with increased dispersal of respiratory droplets in patients with SARS. Our brief report highlights that there may be risks with many forms of support and that decisions should likely be made on an individual patient basis with due attention to the hazards for both patients and healthcare workers.