The results of this large multicenter study in France highlight key issues implicated in vaccination decision-making by HCW during the unprecedented 2009 pandemic alert. Vaccination rates differed according to vaccination type and occupational category: 30% and 58% (SIV) and 21% and 71% (PIV) for paramedical and medical HCW, respectively. A striking finding is that this decision-making involved limited and mainly self-centered sociocognitive dimensions, mainly “self-perception of benefits”, “health motivation” and, only for medical HCW, “self-perception of subjective norm”. The statements best discriminating between vaccinated and nonvaccinated paramedical HCW were similar for both vaccinations and concerned only self-perceived benefits, i.e., to protect “oneself” (SIV) or “the family circle” (PIV), and health motivation. Considering SIV, willing to be a role model was the statement that best discriminated between vaccinated and nonvaccinated medical HCW. Among those 46% willing to be a role model, 80% had received SIV. For the 54% of medical HCW not considering themselves a role model, the statements best discriminating between vaccinated and nonvaccinated were the same as for paramedical HCW, i.e., “protect oneself” and “health motivation”. Concerning PIV of medical HCW, discriminatory domains were perceived benefit effects for oneself or patients, being a role model or perceived barriers (side effects).
The A(H1N1) pandemic refocused attention on HCW as a priority group for vaccination to reduce their occupational risk of infection 
, its related absenteeism that might impact care delivery5
and to limit their role as vectors 
. Hollmeyer et al. identified two major reasons for vaccination nonacceptance: a wide range of misconceptions about flu infection and related risks for patients, including the potential risk of transmission by HCW to their patients, and vaccination effectiveness; and a lack of convenient access to vaccination 
The US Healthcare Infection Control Practices Advisory Committee and the Advisory Committee on Immunization Practices recommend five components to improve HCWs' immunization rates: education and promotion, improved access, legislation and regulation, measurement and feedback, and role models 
. However, the appropriate design and components of such multifaceted campaigns are unknown. In a survey involving 418 American centers, only free vaccination, adequate staff and resources, and education of targeted HCW groups predicted institutional vaccination rates 
. Our healthcare group's 2009–2010 SIV–PIV campaign combined the first 4 (education and promotion, free and convenient vaccination, real-time feedback on the vaccination rates to each department, involvement of all hospitals' leaders and administration support) of those five recommended components, as French legislation and regulation cannot impose flu vaccination. Nevertheless, our vaccination rates remained as low as previously described in Europe 
, with only 11% of paramedical HCW and 48% of medical HCW having received both vaccinations.
A variety of factors have inconsistently been associated with HCWs' influenza-vaccination acceptance in previous studies, including individual and occupational characteristics, previous flu-immunization practices, and individual cognitive determinants 
. Consistent with previous findings 
, 5 factors (older age, prior influenza vaccination, working in an ICU/ED, the hospital they worked in and being a medical HCW) predicted both vaccinations herein. However, our use of a sociocognitive approach and segmentation analysis provides new insights into interpreting decision-making and planning future vaccination campaigns.
Although the perceived benefit/risk-balance difference between vaccinated and nonvaccinated HCW supports the need for campaigns to address the misconceptions, those misunderstandings might be the reason for or the consequences of accepting or refusing flu vaccination. Indeed, reduction of cognitive dissonance, also called “rationalization” (i.e., people use of strategies to align their behavior with unconscious cognitions), is one of the most influential and extensively studied concepts in social psychology 
. According to that theory, a paramedical HCW who refused vaccination assumes this refusal by espousing the most common misconceptions; importantly, this concept suggests that interventions, e.g. vaccination campaign, might reinforce misconceptions and paradoxically be counterproductive (reactance phenomenon) 
Our observations demonstrated that vaccination acceptance among paramedical HCW is mainly a self-centered concept. Therefore, approaches aiming at targeting personal benefits of immunization might be more successful than campaigns focused on preventing absenteeism and transmission to patients. Medical HCWs' decisions were marked by professionalism (being a role model and patient protection). These results plead strongly for campaigns targeting paramedical and medical HCW separately. Professional values develop largely through an informal process of socialization and training, and, thus, cannot be addressed in a vaccination campaign 
. However, evidence from the medical and sociological literature suggests that the role model could play a pivotal part in changing human behavior 
Decision-making by previously vaccinated HCW was more complex but remained mainly self-centered, though it might be hypothesized that decision-making could rely on better understanding of or adherence to the recommendations. However, this analysis combined medical and paramedical populations, and our study was not designed to determine what led those HCW to be vaccinated for the first time.
Among the sociocognitive domains not involved in decision-making, “perceived barriers” and “external influence” deserve particular attention. Two themes were continuously raised by the French media during the 2009–2010 campaign, the safety of the pandemic vaccine and the noninvolvement of general practitioners in the information-and-immunization efforts 
. Nevertheless, “external influences” was not identified as a key to decision-making in any of the models, and the fear of side effects was a consideration only for PIV by a few medical HCW.
This study has several strengths. First, its multicenter design with complementary hospitals in a same healthcare group and a high response rate strengthen its representativeness. Second, the large sample size allowed separate regression-tree analyses for medical and paramedical HCW. Third, 10 cognitive domains potentially involved in this complex decision were considered concurrently and not treated as isolated entities, as was done previously. Finally, the study was performed after the pandemic alert, which allowed dispassionate investigation of both vaccinations.
This study also has limitations. First, the prevalence of A(H1N1) influenza did not reach the anticipated pandemic state, with post-outbreak seropositivity rates ranging from 10 to 25% 
, and the prevalence of seasonal influenza remained lower than expected during the 2009–2010 winter. Even if our questionnaire was designed to assess HCWs' perceptions during the vaccination campaign, and not at the time of the study, we cannot exclude that the actual incidence of both diseases had impacted HCWs' responses to the items included in the CART analysis. Second, the worldwide extrapolation of our results is questionable. However, low vaccination rates and ineffective immunization campaigns have been reported in most parts of the world over three decades. Third, influenza vaccination was self-reported. Moreover, the lower response rate of medical HCW may reflect difficulty reaching them during the study days because of their activities, but we cannot exclude that some of them chose not to respond because of the topic's sensitivity. Then, given the nonlongitudinal study design, recall bias cannot be firmly ruled out. Finally, decision-making may be influenced by many factors not considered in the behavioral models.
Ours is the first study to offer a global comprehensive picture of decision-making for vaccination acceptance, a major challenge worldwide, by a large and diversified care group whose multifaceted campaign combined all the recommended components except specific legislation or regulation. We found that, among numerous well-recognized cognitive factors, only a few were involved in deciding to be vaccinated or not. Despite the important role played by professionalism in the medical community, vaccination acceptance is mainly a self-centered act. A multifaceted campaign without specific legislation or regulation policy might not be able to reach an efficient vaccination rate.