In April 2009, H1N1 pandemic influenza rapidly spread throughout the world challenging the response capacity of health care organizations and public health authorities. When the influenza A/H1N1 influenza monovalent vaccine became available in October 2009, public health authorities ranked health care personnel as one of the highest priority groups to receive the vaccine [1]. This recommendation was based on health care providers' risk for acquiring infection while caring for patients with influenza, their risk for infecting vulnerable patients if ill with influenza themselves, as well as the public safety need to ensure the presence of adequate personnel to treat an influx of persons ill with influenza. In the absence of a vaccine, initial control strategies for health care workers focused on infection control measures within health care facilities, use of personal protective equipment, establishing triage and other engineered control measures, non-punitive leave policies, and cough and respiratory etiquette to mitigate the pandemic's impact based on past pandemic planning [2], [3].

In Minnesota, preparation for the arrival of the 2009 H1N1 monovalent vaccine began in August of that year [4]. Health care facilities pre-registered with the Minnesota Department of Health (MDH) designated the number of highest priority groups within their facilities including both patients and health care providers. Significant educational efforts regarding the virus and the vaccine were carried out at the state and local levels by public health agencies and health care organizations through electronic messaging, press coverage, online resources, meetings, and webinars, among many other communication modes. 2009 H1N1 influenza activity in Minnesota peaked the week of October 11–17, 2009, with the number of hospitalizations exceeding the previous influenza season of 2008–2009 by over 500% [4]. Around that same time, the vaccine arrived in Minnesota and was distributed to pre-registered health care facilities. By December 2009, there were adequate supplies to meet all requests from health care facilities and most conducted mass vaccination clinics for staff statewide in November and December. As of April 2010, there were 1,824 hospitalized patients and 63 deaths due to laboratory confirmed 2009 H1N1 cases in Minnesota representing only a small fraction of the burden of disease associated with H1N1 influenza [5].

Despite adequate vaccine supply and high priority designation, U.S. health care worker H1N1 vaccination rates were only 37% as of January 2010 [6]. Leading up to the pandemic, the U.S. national annual seasonal influenza vaccination rate among health care workers was estimated to be 45% (ranging from 36% to 52% depending upon health care worker group) over the years 2004–2008 [7].

Previous studies have found that a number of factors influence whether or not health care workers choose to be vaccinated for both seasonal and pandemic influenza, including previous knowledge about the vaccination, risk perception, attitudes toward vaccination such as anticipated regret, perceptions of professional norms, age, and race [8]–[12]. While there have been numerous studies examining the knowledge, attitudes, and practices of professional groups specifically in the context of pandemics [13]–[17], few have examined differences in these areas in a population of U.S. health care workers and across disciplines. Professional groups with distinct cultures and histories may view their perceived obligations toward vaccination during a pandemic differently in light of their particular professional culture or ethos. For instance, the American Medical Association recently adopted a policy that acknowledges physicians' ethical and professional responsibility to be immunized and supporting the requirement that influenza immunization be a condition of initial and continuing employment [18]. The American Nurses Association endorses voluntary immunization and has rejected requirements for influenza immunization of nurses [19].

Following initiatives encouraging health care workers to be vaccinated against 2009 H1N1 influenza, we conducted a state-wide survey of Minnesota physicians and nurses to assess their vaccination behaviors as well as the attitudes and perceptions that may underlie those behaviors, including perceptions of their professional obligations around vaccination and their acceptance of potential mandatory pandemic vaccination policies.

Of the 1,600 sampled Minnesota health professionals, 49 could not be contacted due to undeliverable addresses. Of the remaining 1,551 participants, we received completed surveys from 1,073 for a cooperation rate of 69% [20]. For physicians, 486 of 772 (63%) with deliverable addresses returned completed surveys, while 587 of 779 (75%) nurses with deliverable addresses returned completed surveys.

Respondent characteristics are shown in Table 1. A majority of respondents were female (67%) and the mean age was nearly 48 years. Compared to physicians, nurses were significantly more likely to be female (p<0.001). The majority of both physicians (85%) and nurses (62%) reported receiving vaccination against pandemic H1N1, but physicians were significantly more likely to be vaccinated than nurses (unadjusted p<0.001). Respondents and non-respondents differed in that females were more likely to respond than males (p = 0.001), nurses were significantly more likely to respond than physicians (p<0.001), and those in the metropolitan area of Minneapolis and St. Paul were less likely to respond than those residing in other Minnesota metropolitan or non-metropolitan regions (p<0.001).

These data from a state-wide survey of health professionals in the wake of the 2009 H1N1 pandemic suggest high (but not universal) rates of vaccination in key health care worker populations and several key attitudes and perceptions that may have influenced whether health care professionals were vaccinated. Chief among these associations include perceptions about risk of serious side effects, professional obligations, and sources of information, all of which varied by professional group. These data raise important questions about inter-professional education and differences in response to pandemics and how states might enlist the optimal support of health care workers in achieving vaccination targets during pandemics.

The self-reported vaccination rates of 85% for physicians and 62% for nurses reported here are among the highest we have seen in population-based assessments of voluntary vaccination programs, such as the 13.5%, 36.2%, 36.5%, and 41.3% rates seen in Hong Kong, Singapore, France, and the United Kingdom, respectively [17], [21]. Compared to early reports of health care worker seasonal vaccination rates as of November 2010, self-reported rates of pandemic vaccination from 2009 may represent a best case scenario of vaccination rates in which pandemic concerns combined with shortages and intense public appetite for the vaccine created a temporary enthusiasm for vaccination in response to an extraordinary threat [6].

Recent surveys of Canadian, English, French, and Spanish health care workers also found concern over vaccine side effects to be one of the top reasons workers declined to receive the H1N1 influenza vaccine [10], [16], [17], [22]. The actual risk of serious side effects associated with the influenza vaccine is estimated to be one event per 4–8 million doses based on data from the Vaccine Adverse Events Reporting System [23]. We found that physicians were more likely to most closely estimate the order of magnitude of the risk (i.e. 1/1,000,000) for pandemic vaccine side effects compared to nurses who were more likely to respond “I do not know” or to over-estimate the risk. The absolute risk of serious vaccine side effects for influenza vaccine is similar to the risk associated with hepatitis B vaccine, a vaccine that health care institutions are required to track and offer free of charge for all employees [24]. This illustrates the importance of accurate education to health care providers regarding both level of risk and vaccine efficacy.

Our finding that lower rates of vaccination among nurses was associated with an inaccurate perceived estimate of vaccine side effects also mirrors those by Zhang et al in their survey of nurses in the United Kingdom [12]. In that study, nurses' over-estimation of the seasonal influenza vaccine's side effects and less frequent use of public health authorities for vaccine information suggests that public health messaging either does not sufficiently target nurses or uses communication channels and strategies that may not be optimally effective. For instance, suboptimal trust by nurses of their employers because of labor concerns may translate to suboptimal trust in public health messages about vaccination communicated through employers [25]. Such may be the case for Minnesota nurses, in whom major labor negotiations and strike threats were occurring in the winter of 2009–2010 concurrent with the pandemic [26]. Although this study by Zhang et al. focused on nurses' perceptions of seasonal, as opposed to pandemic, influenza vaccination side effects, the findings nevertheless provide important context for the relevant and related scenario of pandemic vaccination.

That nurses are less likely to view vaccination as a professional obligation raises the possibility that either fewer nurses do not consider such actions as central to their professional identity, or that the way they view their professional role is not fundamentally about “obligations” to perform certain actions per se, but rather about fidelity to particular individual caring relationships. Although not directly addressed by this survey, nurses may endorse strong professional norms related to patient safety primarily in the context of particular relationships.

From these data we speculate on several possible implications for public health and organizational practice. Pandemic influenza messaging may need to target nurses more directly with facts about the significant benefits of vaccination for themselves and their patients in comparison to the associated extremely low risk of negative side effects. These data also raise questions about professional differences in the sources of professional identity generally (i.e. what it means to be a nurse or physician). While aspirational ethical statements may establish a professional obligation to receive vaccinations, as these data suggest, direct appeals to nurses' sense of general professional obligation may not be the most effective framework for bringing about behavior change. Rather, we speculate that focusing on caring relationships as a source of professional behavior change may be more in line with nurses' professional self-identity. For example, perhaps messaging campaigns could appeal directly to nurses' sense of advocacy for the many vulnerable patients they care for daily, as a means of promoting vaccination in a manner more in line with their professional identity. Such efforts deserve further investigation.

Whether health care worker mandates should be legislated is an ongoing policy question that these data address indirectly. While some respondents voiced objections to mandates, recent data suggests that mandates are generally well accepted once in place [27], [28]. Mandates may be more likely to be accepted by doctors and nurses who claimed they did not receive the vaccination because they “don't want to be vaccinated,” simply because they otherwise voice no strong reasons for avoiding it [29]. More nurses may view vaccination as a personal health choice rather than an obligatory patient safety measure [30]. If so, it is easier to see how they may not find the time to fit in a “personal health choice” as a priority in a busy life. Despite disagreement about endorsement for mandates, the percent of health care workers who cite reasons for not being vaccinated due to personal choice was low (16% physicians and 21% nurses).

These data have several limitations. First, the sample was limited not only to the United States context, but specifically one state – a state in which there is generally strong public health infrastructure and coordinated responses to public health emergencies. Our respondent population was also comprised of more women than men, and more nurses than physicians – characteristics which may limit the representativeness of the findings for the health care worker population as a whole. The use of survey data also raises the possibility of response bias if responders systematically differ from non-responders on the characteristics being studied. Further, self-report can be subject to measurement error due to how items were worded. We cannot exclude the possibility of a social desirability bias for such a sensitive topic whereby respondents answer in a way that will cast them in a more positive light. Using a self-administered survey without the presence of an interviewer somewhat minimizes this concern. Moreover, our relatively high response rate and explicitly stated commitment to confidentiality should have minimized this.

Furthermore, our approach (i.e. surveying with closed-ended items) limited our ability to explore the actual concepts such as professional obligation that nurses and physicians may use to form judgments on these matters. Such an approach like that used by Rhudy et al may be warranted in future studies [30]. Finally, cross-sectional measurements cannot establish causation between predictors and outcomes. Despite these limitations, however, these data suggest that a variety of perceived risks, professional norms, information sources, and professional identity may impact how health care workers respond optimally to pandemic action plans at the state level. Crafting state-level responses to pandemics that account for potential misperceptions about vaccine risks and side effects, as well as inter-professional differences, may enhance the ability of the public health workforce to meet the challenges of the next influenza pandemic.