There were some notable differences between groups targeted for the 2009-10 pandemic and seasonal influenza vaccines, and substantial differences in target groups between countries. The greatest inconsistency in target groups was seen in comparisons between S0910 and P0910. Some groups targeted for the pandemic vaccine more commonly than for the 2009-10 seasonal vaccine, including pregnant women and obese persons, and health care workers were subsequently more commonly targeted for the 2010-11 post-pandemic seasonal vaccine. Subsidies to increase vaccine uptake were available to target groups in many countries by either providing the vaccine for free, at a subsidized cost or through national health insurance (Figures and ).
Before further discussing the patterns in target groups, two important issues should be noted: (A) the social and political backgrounds of policy decisions to prioritize certain groups for P0910 are likely to differ from those for seasonal vaccines and (B) the 'best' or 'optimal' strategy to achieve certain public health objectives has yet to be fully clarified in the context of including or excluding certain groups for prioritization. As for (A), compared to seasonal vaccines, public health decision of prioritized groups for P0910 may be more associated with maintaining social security during emergency vaccination and avoiding any confusion that could arise from 'first come, first served' basis among the public [8
]. On the other hand, policymaking of seasonal vaccines is unlikely to face needs to give rapid decisions for implementing emergency vaccination programmes and to maintain social security, and rather, target groups of seasonal vaccines may sometimes reflect the results of surveys of public demands among the potential target groups and may also take into account aspects of cost-benefit and cost-effectiveness. With respect to (B), various studies including modeling exercises considered optimal vaccination strategies [16
] but those exercises have not allowed full clarification of all target groups. It should be emphasized that the limitations of modeling studies and inconsistency in target groups (as seen in the present study) may be attributable to non-uniform public health objectives of vaccination, such as reducing peak hospital burden versus reducing overall mortality, as well as uncertainties with respect to the transmission dynamics, severity and age-specific immunity against influenza viruses before and after vaccination. The objective of this study has been not to criticize certain choices of target groups, but to explore to what extent target groups are similar between countries, or even within countries from year to year and between the seasonal and pandemic vaccines.
Some inconsistent patterns were seen between pandemic and seasonal vaccines in targeting certain age groups, those with underlying medical conditions and those with occupations associated with higher risk. First, mathematical modeling studies have shown that achieving high vaccination coverage among school-age children could substantially elevate herd immunity and protect other risk groups (including elderly and those with underlying diseases) by means of a 'transmission-limiting' strategy [17
]. Nevertheless, many countries experienced substantial attack rates in children before vaccines became available [21
], and in such an instance (e.g. after observing epidemic peak), effectiveness of transmission-limiting strategy tends to be minimal [19
]. While school-age children were targeted for the P0910 vaccine in some countries, most countries included in our review adopted a morbidity-limiting strategy and targeted groups at high risk of severe disease (Figures and ). Vaccine effectiveness in the elderly remains uncertain [6
] but the elderly are commonly targeted to receive interpandemic influenza vaccine because of their high risk of severe disease [4
]. However, early seroepidemiological studies in 2009 suggested that the elderly may be protected against infection [23
], and few countries included elderly in the target groups for the P0910 vaccine (Figure and Table ).
Second, with regard to underlying medical conditions, pregnant and postpartum women appeared to be at higher risk of severe disease if infected from the early stages of the pandemic [24
] and many countries prioritized this group to receive the vaccine when it became available. Whereas pregnant women were less commonly targeted for seasonal influenza vaccination before the pandemic, evidence of an increased risk of severe influenza had already been observed in pregnant women before 2009 [28
]. Influenced by the 2009 pandemic, 22/34 countries targeted this group for S1011 vaccine (Figures and , Table ). Other than pregnant women, a number of studies during the pandemic found obesity to be a risk factor for hospitalization and death [26
], and some countries incorporated obese individuals into the target groups for the P0910 vaccine and again retained them in the target groups for the post-pandemic S1011 vaccine (Figures and , Table ).
Third, regarding roles and occupations, health care workers were targeted for both pandemic and seasonal influenza vaccines by many countries and they have been increasingly targeted during and after the course of the pandemic (Figures , and , Table ), although they may not necessarily face higher risk of infection due to occupational exposures [34
] and the issue of mandatory vaccination in this group remains controversial [36
]. Workers in long term residential care homes were less commonly targeted for the pandemic vaccine compared to the seasonal vaccines, perhaps because of the apparent low risk of pH1N1 in the elderly [7
]. Animal contacts might have been less commonly targeted merely because of displacement by other risk groups as P0910 vaccine was initially expected to be scarce. Later they were more commonly targeted for the S1011 influenza vaccine which included the pandemic strain.
A few limitations should be discussed. We were not able to find information on priority groups in many countries, and thus, the data included in our review may not sufficiently represent the worldwide patterns in target groups. In addition, information available online could differ from what was actually implemented, even though most of the information included in our study was from official government websites. Potentially lower quality sources were excluded by limiting our search to official webpage of the government, news or academic journal articles. Future studies of influenza vaccine policies could consider a survey-based approach interviewing a representative health official from each country. Actual vaccine uptake or influenza-associated morbidity should also be investigated in relation to the policy decisions of target groups when relevant data are available as part of policy evaluations. There were also within country variations; e.g. recommendations on target groups in Canada and the United States differed between states. In these situations, the federal recommendations were used to reflect the central policy, but more detailed investigation of the heterogeneity in target groups between states or regions could be a topic for further research especially for a within-country policy evaluation.