In a prospective, multicenter, simulation exercise, we assessed the initial 6 hours of management of a patient (actor) who appeared for treatment at a hospital emergency department with a history consistent with API. Tertiary-level university teaching hospitals across eastern Australia were invited to participate. The inclusion criteria were willingness to join the simulation and possession of a formal local infection control protocol for the management of API that followed Australian (3) or WHO guidelines (7). The study was approved as a quality assurance project by the ethics committee at each participating site.

Nine tertiary-level university teaching hospitals in 3 states of eastern Australia participated in the study (Table 2). The simulations occurred in the winter season, from May through August 2006. All sites conducted targeted staff education sessions 1–2 weeks before their exercise. Seven of the 9 simulations proceeded for the planned 6 hours of observation, and 2 were curtailed because of a critical need for the emergency department bed. Had these latter 2 sites continued, the patient would almost certainly have spent the entire study period isolated in the emergency department, as suitable ward beds were not available. The time spent in each clinical area for each site is summarized in Table 2. All sites performed radiography within the emergency department.

The number of close contacts and total exposures to the potential API patient are summarized in Table 3. The highest number occurred in the first hour of hospital care (triage and emergency department), which correlated with the initial intensive clinical and radiologic assessment and specimen collection. Patient transfer between areas was another peak time for exposures. The average number of close contacts for each API patient during the study period was 12.3 (median 11, range 6–17), with 19.3 exposures (median 20, range 15–26). HCW close contacts constituted 85% of all close contacts; the remainder were patients or visitors who were generally exposed in the triage area.

All 9 sites processed the respiratory specimen, with an average of 2.9 HCWs (median 3, range 2–6) handling or transporting the specimen, predominantly in the pathology department. Two sites used a vacuum transport system to deliver specimens from the emergency department to the laboratory, contrary to WHO recommendations (7).

Environmental decontamination of clinical areas after departure of the suspected API patient was performed haphazardly at all sites. The triage area was appropriately cleaned in none of the 9 sites, whereas the emergency department and ward areas at sites that completed the full simulation were cleaned appropriately in 6 of 7, and 4 of 7 instances, respectively; 1–2 cleaners were required per clinical area to appropriately perform this task.

Large quantities of N95 masks, disposable gowns, gloves, and eye protection were used and indicated during the study period (Table 4). Adherence to appropriate use by HCWs (HCW close contacts and ancillary HCWs) was variable and depended on the particular PPE item, clinical area, and participating institution. Appropriate use of N95 masks by HCWs occurred in 93% of exposures (actual use/total opportunities for PPE use, 18/19.4), although the corresponding figures for disposable gowns, gloves, and eye protection were lower (77%, 83%, and 73%, respectively).

HCW close contacts were stratified into either moderate- or low-risk groups, depending on whether an appropriate PPE set was worn during every exposure. The proportions of HCW close contacts who appropriately wore a PPE set, rather than an N95 mask alone, for every exposure were 59% and 92%, respectively. Thus, depending on how rigorously WHO antiviral medication guidelines (9) were followed, from 8% to 41% of all HCW close contacts would be classified as having experienced a medium-risk exposure and therefore would potentially require postexposure antiviral prophylaxis. This amounts to an average of 0.8 to 4.3 courses of antiviral medication per suspected API patient during the initial 6 hours of management.

To our knowledge, this is the first multicenter study to estimate the quantity of PPE and antiviral therapy that may be required to manage patients with suspected API admitted to hospitals. During the initial 6 hours of hospital assessment, the number of close contacts of a single suspected API patient was high (mean 12.3), with a mean number of exposures of 19.3. Not surprisingly, most (85%) close contacts were HCWs, and PPE use was at its most intense in the first hour of emergency department assessment. Our data suggest that in the initial 6 hours alone, HCWs managing suspected API case-patients would require ≈20–25 PPE sets (mean quantities: 19.4 N95 masks, 22.1 gowns, and 25.1 pairs of gloves). Although a high proportion of HCW close contacts (mean 92%) wore an N95 mask appropriately for all exposures, appropriate concomitant use of other PPE items was less (mean 59% of exposures). Even with the widespread availability of PPE, this observed inadequate utilization rate meant that from 8% to 41% of HCW close contacts were likely to require postexposure antiviral prophylaxis if current WHO recommendations were followed (9). If appropriate PPE, especially N95 masks, were not available, the number of HCWs who would experience moderate-risk API exposure requiring postexposure antiviral prophylaxis would increase substantially.

Notably, a substantial minority of close contacts (15%; ≈2 per API patient) were non-HCWs (e.g., hospital patients or visitors), generated primarily in the triage area. Although the duration of unprotected exposure was often short (<5 minutes) for these persons, they represent a potential risk for subsequent community and hospital spread of API. This highlights the importance, in triage and reception areas particularly, of using appropriate infection control measures and signage to assist in cohorting of potential API patients and minimizing exposure of unprotected bystanders.

The critical importance of effective PPE in hospital infection control was demonstrated during the outbreak of SARS in 2003 (10–14). Nosocomial transmission of SARS was a prominent feature of the epidemic (15) and played a large role in the initiation and maintenance of outbreaks. As reported in a case-control study by Seto et al. (13), staff who used masks (in particular), gowns, and performed hand hygiene were less likely to become SARS infected than those who did not. Similarly, Lau et al. (14) noted that inconsistent use of PPE by HCWs working on wards with SARS patients in Hong Kong was associated with a significantly higher risk for nosocomial disease transmission. Provision of adequate PPE stock is therefore likely to be important in controlling the spread of API.

Many countries are compiling extensive stockpiles of PPE and antiviral medications for use if a new pandemic occurs. Planning for sufficient numbers of resource items is complex and dependent on estimations of pandemic-related additional emergency presentations, hospitalizations, general practice, and outpatient visits. In Australia, official estimates of additional hospitalizations range from 57,900 to 148,000 (4). Our data suggest that management of this number of hospitalizations without regard for suspected influenza patients who are assessed but who are not sufficiently ill to require admission, would require from 1,123,260 to 3,714,800 PPE sets (depending on whether they were N95 masks, gowns, or gloves, or all 3 items). Although ascertaining (from these data) the number of courses of postexposure antiviral prophylaxis required is difficult, if stocks of readily available PPE were inadequate, the number of courses of antiviral medication required would likely increase dramatically, up to 12–13 courses per suspected API case during the initial 6-hour assessment. Thus, adequate stocks of PPE provide a means of protecting valuable antiviral drug stockpiles for use in ill or heavily exposed persons.

An important consideration when extrapolating our data to other healthcare systems is that recommendations regarding the optimal form of respiratory protection vary between countries. The WHO interim guidelines for management of human cases of avian influenza (AI) state, “HCWs working with AI-infected patients should select the highest level of respiratory protection available, preferably a particulate respirator… designed to protect the wearer from respiratory aerosols expelled by others” (7). This recommendation is reflected in the Australian pandemic influenza guidelines (3) and explains the high use of N95 masks in our study. However, pandemic influenza plans in the United Kingdom (5), United States (6), and Canada (16) currently recommend the use of surgical masks for close patient care, unless the HCW is engaged in procedures in which aerosolization occurs. Thus the proportion of N95 masks to surgical masks required will vary between countries with different guidelines, which affects assessment of stockpile adequacy. Our study did not assess the relative efficacy of N95 masks compared with surgical masks for protection against API transmission.

This study has several limitations. First, the duration of the study was short (6 hours), much shorter than the likely in-hospital stay of days for a patient with severe influenza. Thus, total PPE and antiviral agent usage per admission is likely to be substantially higher. Second, the study was conducted at a less busy time of day for emergency departments and therefore may not reflect the greater number of persons who would likely be exposed in the triage and emergency department areas during busier periods. Third, the patient was not clinically unwell or hypoxic; thus, relatively few HCWs were required to assess, manage, or review the API patient’s condition. Fourth, we observed the management of the index API case-patient alone, although we acknowledge that actual patients are likely to come to the hospital with other household members (high-risk close contacts). However, extending observation to include management of asymptomatic but potentially infectious accompanying persons in a standardized manner would have substantially increased the complexity of the exercise. Our findings, therefore, likely underestimate the true resources required and contacts exposed for the management of a genuine API patient. Finally, the presence of observers and the preceding education sessions may have artificially increased compliance with PPE use, although in the event of a true pandemic one might assume that HCW compliance rates would be high as they aim to minimize their personal risk. Also, this study was designed to quantify the use of PPE in an environment with raised awareness of infection control practice, mimicking that which might occur during a pandemic, and thus provide relevant data for health resource planners.

This study suggests that managing a single API patient is resource intensive and exposes a high number of persons to a potentially severe infection. These data represent the likely minimum clinical resources required during an API patient’s initial hospital assessment using current WHO-derived infection control guidelines. Given our findings, if a global influenza pandemic occurs with attack rates even on the lower end of projected estimates, demand for PPE and antiviral medication in healthcare facilities will likely outstrip current supply in industrialized countries, let alone the supply in resource-poor settings. Further studies are needed to assess resource usage in other healthcare settings such as intensive care units, fever clinics, general practice, and the community.