NIOSH investigators and School of Medicine faculty developed a survey that was in part adapted from previous research applying the theory of planned behavior when examining PPE use for influenza control among critical care clinicians.^8,9 The survey included questions regarding personal characteristics, work history, history of exposure to pH1N1, and ILI symptoms. Knowledge, attitudes, and practices regarding influenza infection control measures, including previous seasonal influenza vaccination, hand hygiene, respirator training and fit testing, use of PPE, and adherence to recommended work restrictions, were examined. Knowledge and attitudes were examined by degree of agreement based on a 5-point Likert scale. Respondents were given a list from which to choose PPE included in precautions when caring for pH1N1-infected patients at each hospital and were asked to recall the frequency with which they used PPE when in close contact with patients. Internal medicine residents, cardiology fellows, and pulmonary and critical care fellows in the program at any time between May 1 and June 30, 2009, were invited to participate in the electronic survey administered during August and September 2009.

NIOSH investigators visited the School of Medicine and all 4 affiliated hospitals in September 2009 and convened in-person focus group sessions with residents and fellows to discuss their experiences and concerns regarding pH1N1. Meetings were held with representatives from graduate medical education, hospital and nursing administration, infection control and hospital epidemiology, employee health, safety and environmental health, and emergency management from all 4 hospitals. Medical records of hospitalized patients with confirmed or probable pH1N1 and of ill physicians in training were reviewed during the site visit. Exposure of physicians in training to infected patients was determined retrospectively by authorship of admission, progress, and discharge notes and signature of physician orders. Resident rotation schedule information and program social events were extracted from residency program records. Documentation of employee health program–administered antiviral prophylaxis to exposed physicians in training was reviewed. As a public health response, in accordance with the guidelines of US Title 45 Code of Federal Regulations Part 46, this evaluation was deemed to not require review by an Institutional Review Board.

Although the potential for aerosol generation during specific procedures is unknown, we conservatively defined intubation, suctioning, administration of nebulizing medications, bronchoscopy, acquisition of a nasopharyngeal sample, and ventilation with bilevel positive airway pressure or continuous positive airway pressure to be aerosol-generating procedures. We categorized respondents who rotated at a hospital during the study period or who reported close contact (<6 feet) with a patient with confirmed or probable pH1N1 or ILI as either knowing or not knowing the minimum recommended PPE when caring for such a patient at that hospital. Respondents who indicated a higher level of respiratory protection than the minimum recommendation specified in Table 1 were also classified as knowing the respiratory recommendations. We deemed that a respondent's selection of a full-facepiece powered air-purifying respirator (PAPR) did not necessarily demonstrate knowledge of an eye protection recommendation.

We analyzed survey responses to 4-point frequency scale questions on use of PPE components for those respondents reporting close contact with a patient with pH1N1 or ILI. Based on these responses and minimum hospital PPE recommendations, we categorized the respondents into high and low PPE adherence groups. Respondents indicating “never” or “some of the time” for use of any of the required PPE components composed the low adherence group, and those indicating “most of the time” or “always” for use of all required PPE components composed the high adherence group.

Respondents who reported using a higher level of protection than recommended (eg, use of an N95 respirator or PAPR when only a surgical mask was required or gowning when not required) were classified as high PPE adherence as long as the other required PPE components were used as well. Respondents who reported wearing a PAPR were categorized as wearing eye protection, because full-facepiece PAPRs were provided. Factors associated with high and low PPE adherence were determined.

Likert scale responses were categorized as “expressed agreement” if the respondent answered “agree” or “tend to agree,” and as “expressed disagreement” if the respondent answered “disagree” or “tend to disagree.” Bivariate analyses were conducted with the Student t test, Pearson's χ² test, and Fisher's exact test using SAS version 9.2 (SAS Institute, Cary, NC). Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Statistical significance was set at P < .05.

Initially, all 4 hospitals followed the CDC's interim guidance for PPE use by HCP when caring for patients with pH1N1.⁷ However, beginning in June 2009, 3 hospitals deviated from this guidance because of an insufficient supply of N95 respirators, limited HCP respirator fit testing in a region with a low prevalence of Mycobacterium tuberculosis, and differing guidance at the state level. At these 3 hospitals, a surgical mask was considered appropriate protection. Table 1 presents minimum PPE recommendations when caring for patients with confirmed or suspected pH1N1 or ILI by hospital, as of September 2009. All hospitals required gloves, gown, either a fit-tested N95 respirator or PAPR, and eye protection for aerosol-generating procedures.

Of 210 current and recently graduated residents and fellows, 88 (42%) completed the electronic survey. The median age of the respondents was 30 years (range, 25-53 years), and 57 (65%) were male. The 88 respondents included 75 internal medicine residents, 7 cardiology fellows, and 6 pulmonary and critical care fellows.

Respondents reported the following exposures: 56 (65%) had close contact with a patient with pH1N1 or ILI (defined as temperature ≥100°F and cough and/or sore throat) while working, 45 (52%) had close contact with a coworker with pH1N1 or ILI (henceforth referred to as ill) at work, 28 (32%) had close contact with an ill coworker outside of work, and 12 (14%) had close with an ill person in their household or in the community. Thirteen respondents (15%) reported no close contact with any of the foregoing possible sources of exposure.

Between May 1 and June 30, 2009, 13 of the 88 respondents (15%) reported symptoms of ILI (defined as fever and sore throat and/or cough). Seven of these 13 sought medical care, 5 of whom had laboratory confirmation of influenza A by viral culture or direct immunofluorescence assay; none were hospitalized. The first onset of ILI in a responding physician was on May 10, and the second onset was on May 30. The next onset of ILI was not until June 13. A cluster of 8 ILI cases occurred within a 96-hour period after an internal medicine residency program dinner held on June 19; 6 of these individuals reported attending the dinner. Prophylactic antiviral agents were prescribed to 68 residents on June 23-24, 2009, and no further cases of ILI were reported. Figure 1 shows an epidemic curve illustrating the number of cases of ILI and influenza A infection among respondents in June 2009.

Among the 13 survey respondents with ILI, the following close contacts in the 7 days before symptom onset were reported: only an ill patient (n = 2), only an ill coworker while at work (n = 2), only an ill coworker outside of work (n = 1), only an ill household or community case (n = 1), and multiple ill sources (n = 5). Two respondents with ILI reported no known contact with a potential source.

Ten of the 13 respondents (77%) who reported ILI worked while ill, with duration ranging from 1 day to 4 days. The following reasons for working while ill (respondents could cite more than one reason) were reported: having a professional obligation to coworkers (n = 4) or to patients (n = 4), not thinking that they were putting their patients (n = 3) or fellow coworkers at risk (n = 2), not thinking that they were contagious (n = 2), and not wanting to admit they were sick (n = 2).

Twenty-five of 59 respondents (42%) at hospital A, 7 of 55 (13%) at hospital B, 28 of 32 (88%) at hospital C, and 3 of 5 (60%) at hospital D were categorized as knowing the minimum PPE recommendations when caring for patients with influenza at the time. Eye protection was the component most often misunderstood by respondents; for example, 41 of 55 respondents (75%) at hospital B believed that eye protection was not recommended or did not know whether it was recommended. Table 2 summarizes the survey results regarding infection control knowledge and attitudes. Most respondents expressed agreement that proper hand hygiene and use of PPE protected both themselves and patients against acquiring influenza.

Seventy-nine respondents (90%) reported receiving the 2008-2009 seasonal influenza vaccination. Sixty-eight respondents (77%) reported always washing their hands before patient encounters, and 71 (81%) reported do so after patient encounters. Thirty-eight respondents (43%) reported receiving training on the use of N95 respirators, 29 (33%) reported receiving training on the use of PAPRs since the start of their program, and 19 (22%) reported undergoing respirator fit testing between the start of their program and June 2009.

Fifty-three of the 88 respondents (60%) reported close contact with a patient with pH1N1 or ILI and provided sufficient information on PPE adherence. Of these 53 respondents, we classified 19 (36%) as having high adherence to the PPE recommendations and 34 (64%) as having low adherence. Reported adherence to recommendations for the use of individual PPE components (Table 1) was 82% (46 of 56) for gloves, 58% (11 of 19) for gowns, 36% (16 of 45) for eye protection, 83% (15 of 18) for N95 respirator or PAPR, and 88% (42 of 48) for surgical mask or higher levels of protection. Fifty-two respondents reported using all recommended PPE when in close contact with a patient with pH1N1 or ILI; 40 (77%) reported not always wearing recommended PPE. The most common reasons given for not using recommended PPE are listed in Table 3.

We found no statistically significant differences in demographic characteristics or rotation hospital between respondents in the high adherence group and those in the low adherence group. Expressing either agreement or disagreement with other infection control knowledge and attitudes statements (listed in Table 2) was not significantly associated with either high or low PPE adherence. In addition, knowledge of the correct, recommended PPE at 3 of the 4 hospitals was not significantly associated with classification into either high or low PPE adherence. At hospital A, knowledge of the correct recommended PPE was associated with high PPE adherence (OR, 3.80; 95% CI, 0.98-14.67; P = .047).

Those recorded as being on an intensive care unit (ICU) rotation during the study period (OR, 3.43; 95% CI, 1.02-11.56) and those present at an aerosol-generating procedure on a patient with pH1N1 (OR, 5.20; 95% CI, 1.54-17.55) were more likely to be in the high adherence group. Those recorded as being on an inpatient ward rotation were less likely to be classified as high adherence (OR, 0.22; 95% CI, 0.06-0.80).