To our knowledge, this study is the first to focus on the behaviors of GPs in private practice – rather than on their expressed willingness to be vaccinated against the pandemic A(H1N1) influenza and to recommend this vaccination to patients. French GPs’ pvaccination uptake rate was lower than the vaccine uptake found among Dutch GPs (85%) in a mail survey 
. Those authors attribute this high rate in part to a campaign by the Dutch government and the Dutch College of General Practitioners that strongly and repeatedly urged HCWs to be vaccinated against A/H1N1. We are not aware of other studies of pvaccination uptake in GPs 
. However, our result was high compared to the rates observed in most studies of hospital physicians in Western countries (<15.0% to ≤50.0%) and elsewhere 
, even those based, as ours was, on questionnaires, a method that might overestimate pvaccination uptake 
. The relatively high rate among French GPs may be explained by their first-line role in the care of seasonal and pandemic influenza and their consequently high risk of exposure to the corresponding viruses 
. Under these circumstances, most French GPs wanted to avoid being unable to work during an epidemic when patient demand is highest 
French GPs’ pvaccination uptake was close to the percentage of French GPs who stated they were willing to accept the vaccination (61.7%) in a previous survey during the prepandemic period 
. These positive words about the pvaccination appeared to be transformed into deeds – vaccination – despite the controversy surrounding it in France 
, especially the pandemic severity and the safety of the vaccine (choice of a pvaccine including an adjuvant, risk of Guillain-Barre syndrome 
). Most GPs (67.0%) said that they had agreed in July 2009 with the statement that HCWs should be vaccinated in priority against A/H1N1, and this opinion was confirmed by an even greater majority (>75%) after the pandemic. Actual uptake of the pvaccine among hospital HCWs in other countries was lower than the stated intentions during the prepandemic period 
Our results about the determinants of pvaccination uptake are consistent with previous results for HCWs and with population studies that suggest the role of past vaccination against influenza, degree of perceived threat during the 2009 pandemic, perceptions of the benefits and risks associated with pvaccination, and sources of information about the pandemic (Bish et al. 2011). To our knowledge however, our study is the first to attempt to quantify the relative contribution of these factors, including a factor concerning the pvaccination campaign organization 
. Such information could be useful to help prioritize and design the components of programs aiming at improving pandemic preparedness. GPs' history of regular influenza vaccination was by far the most important factor independently associated with their pvaccination 
). Well-run seasonal influenza vaccination campaigns thus appear essential to prepare the ground for the next pandemic, especially in the face of inaccurate perceptions about its potential risks 
Such campaigns, however, while necessary, will apparently not be sufficient, for a considerable proportion of the GPs regularly vaccinated against seasonal influenza reported doubts about the efficacy of the pvaccine and fears of its side effects. These doubts may have been due to concerns about using new vaccines during a pandemic, probably different from concerns about established products in non-crisis situations 
. Our results suggest that GPs’ decisions to be vaccinated against A(H1N1) were based in part on their assessment and comparison of the perceived risk and the perceived benefits associated with pvaccine uptake (see , the variable for the perception of pandemic vaccine risks and efficacy), as also found among the general population 
. During this process GPs appeared to give more weight to the vaccine's safety than to its efficacy in making their decisions. Our results also suggest that they applied the same kind of reasoning for their patients, taking level of vulnerability to A(H1N1) influenza into account in advising young adults about the pvaccination.
Some dissonance between GPs’ behavior and their perceptions of the pvaccine nonetheless appeared (). A quarter of the unvaccinated GPs (A/H1N1) were quite favorable to the pvaccine: as observed elsewhere 
, they probably did not perceive themselves as at risk of contracting this flu or developing serious consequences related to it. At the same time, nearly one third of those vaccinated expressed doubts about its efficacy.
Although the ranking of explanatory factors might have been different had the pandemic been more severe 
, our results add further evidence to the need to design effective strategies to inform GPs (and all HCWs) about the risks and efficacy of a new pvaccine and to avoid or correct false opinions about it. These strategies should consider multiple channels of information, given that GPs use various sources to obtain their information. The inverse relation between GPs’ pvaccination and their use of mass media for pandemic information () suggests that the controversy about the pvaccine safety in the mass media induced or reinforced GPs’ inaccurate opinions and their doubts about the new vaccine 
GPs’ personal acceptance of pvaccination appeared strongly predictive of their recommendation of it to the young adults in their practice – at risk or not. This correlation is not surprising, as it exists for seasonal vaccinations 
. Such a correlation suggests that ensuring high pvaccination uptake in GPs might be one lever for improving pvaccination uptake in the general population, especially as GPs’ positive advice significantly increased acceptability of the pvaccination among adults 
. However, French health authorities’ decision to conduct the pvaccination campaign in dedicated centers implied that GPs were not involved in its implementation, although they usually play a central role in vaccination education and follow-up of their patients 
. This probably produced a supplementary barrier to acceptance of pvaccination: uptake in the French general adult population was between 3% and 8% depending on age 
and 22.7% in pregnant women. Indeed our results suggest that GPs’ disagreement with this decision was the third most important obstacle to their personal uptake of pvaccination ().
Strengths and Limitations of the Study
The participation rate in this national French private GP panel was 36.8%, a relatively high rate for panels of physicians requiring participation in repeated surveys (see for example 
: response rate
19%). Participants in the panel differed from nonparticipants for gender, age, and 2008 workload, but weighting our data by these variables did not affect our results; this is reassuring regarding the magnitude of a potential selection bias. The survey on pvaccination was cross-sectional and retrospective. Therefore, the observed links should be interpreted with caution, as recall bias and a posteriori rationalization may have affected GPs’ responses about their attitudes and actions 
. However, these would probably have had a much stronger effect had the study been carried out in the midst of the controversy, rather than several months later when the media had left the controversy behind. Using the multi-model averaging approach allowed us to draw inferences from a set of plausible models rather than from a single model 
. Given the set of potential explanatory variables, this technique accounts for all the possible configurations and summarizes all this information in a final composite model. In doing so, one part of the uncertainty linked to the process of selecting the final model is controlled. A weakness of this approach, however, is that uncertainty is controlled only partially: the multi-model inference is made under the assumption that the observed variables are sufficient to explain a given phenomenon and do not account for unobserved heterogeneity or omitted variables 
, a limitation it shares with other modeling approaches.
Analysis and quantification of the relative contribution of the factors associated with private GPs’ pvaccination allowed us to pinpoint priority components of preparedness that can be improved for future influenza pandemics. In particular, the results suggest that efforts should be devoted to encouraging regular uptake of seasonal influenza vaccination among GPs, as well as providing them with clear information on the risks and efficacy of a new pvaccine and to putting in place an organizational framework for future mass vaccination campaigns that would allow their direct involvement in the vaccination process at the population level.