The presence of an area level, social gradient in health behaviors and health outcomes has been commonly observed in the literature [1–3]. People who live in more socially deprived areas tend to experience worse health outcomes [4–8], and have a greater prevalence of behavioural risk factors [9, 10], than those who live in less socially deprived areas. In the UK, for instance, people living in more deprived areas tend to have less healthy diets, higher rates of smoking, and lower levels of physical activity [11, 12]. All of these behaviors are known risk factors for poor health outcomes. The international picture is somewhat similar [9, 13–15]; although, the exact nature of the relationship between area deprivation and health has been found to vary by context [16].

Differences in the rates of health damaging behaviors, such as smoking, form at least a part of any explanation for variations in health outcomes between areas of different levels of deprivation. There is also an area level social gradient in the impact of health promotion efforts, whereby the impact of health promotion tends to be less in more deprived areas. For instance, people living in more deprived areas tend to have lower rates of smoking cessation [13], lower rates of childhood vaccination [17], and lower uptake rates of screening services [18].

Explaining the relationship between area deprivation and health remains an active field of research. Explaining the relationship between area deprivation and the impact of health promotion activities has been the focus of far less research. Broadly, however, two kinds of explanation present themselves. The first kind of explanation relates to the receptiveness of people living in more deprived areas to health promotion messages. People in more deprived areas may "choose" to ignore the health promotion messages more than those in less deprived areas [19]. A second kind of explanation is that people living in more deprived areas are just as receptive to health promotion messages, but they are less likely to be exposed to health promotion messages than their counterparts in less deprived areas. It is this latter kind of explanation that is explored in this paper.

In June 2007 the UK Department of Health (DH) accepted (in principle) the recommendation of the Joint Committee on Vaccination and Immunization to introduce a national, school-based, human papilloma virus (HPV) vaccination campaign, targeting girls [20]. The long-term goal of the campaign was to reduce the incidence of cervical cancer [21]. In England the vaccination campaign commenced in September 2008 - the beginning of the school year [22]. The vaccination was provided free of charge and on a voluntary basis, initially to 12 and 13 year old girls in schools, but the program was to be expanded to capture older female students (up to 18 years) who would otherwise miss out. In June 2008, prior to the commencement of the vaccination campaign, the Royal Society for Public Health (RSPH) developed and distributed to schools a series of health education resources on the HPV vaccine and cervical cancer [23]. The resources were developed in response to the government initiative, and were designed to support the vaccination campaign. The RSPH anticipated that, by providing schools with HPV related, health education resources which were integrated with the regular school curriculum, students (and through them, parents) would be made aware of the risk of HPV, the benefits of vaccination, and ultimately choose to have the vaccination [23].

The distribution of the educational resources to schools by RSPH created the conditions of a "natural experiment" allowing us to examine the effect of area level deprivation on the exposure to health promotion messages. It was a particularly pertinent campaign on which to focus, because cervical cancer is known to be unequally distributed in England, with the most deprived areas having twice the incidence of cervical cancer as the least deprived areas [24]. By analysing the uptake and distribution of the educational resources (i.e., the reach of health promotion), it was possible to measure the extent to which deprivation affected exposure to health promotion messages. If deprivation was associated with a lack of exposure, then one would anticipate that schools in more deprived areas would be less likely to receive HPV educational resources than would schools in less deprived areas.

In March 2008, a letter was mailed by the RSPH to 5,715 schools across the UK inviting them to receive the HPV health education resources. The schools were drawn from a comprehensive database of UK secondary schools provided to the RSPH by a third party. Of the schools to be sent a letter, the majority (4,750, or 83.6%) was located in England, and it was these schools that formed the sample for the present analysis. The remaining 965 schools in Scotland, Wales and Northern Ireland were not included in the analysis, because of difficulties in matching the schools with equivalent measures of area deprivation. For each school additional limited data were available about the type and size of the school.

The take-up of the educational resources varied across the quintiles of deprivation. Schools located in the most (and second most) deprived areas had take-up rates around 25%, while the schools located in the least deprived quintile had take-up rates around 32%. A test for trend showed a significant increase in the take-up of the educational resources as deprivation decreased (z = -3.99, p = .000). School size was significantly associated with the take-up of the HPV educational resources, with odds of larger schools taking up the educational resources more than twice that of the smallest 20% of schools (OR = 2.11, 95%CI: 1.76-2.53). Significantly more government maintained schools requested materials (32%) than independent schools (27%) (OR = 1.44, 95%CI: 1.22-1.69).

There was substantial regional variation in the take-up of the educational materials. London and the North East had the lowest take-up rates (~20%). The West Midlands and the South East had take-up rates around 27%. The North West and Yorkshire and the Humber had take-up rates around 30%. The East Midlands, the South West, and the East had take-up rates around 33%. All regions, with the exception of the North East had take-up rates significantly higher than that of London, with odds ratios ranging from 1.5 up to 2.1.

After adjusting for the covariates, there remained a significant association between area deprivation and the take-up of the HPV educational resources. Schools in the least deprived areas had odds 1.32 times greater than schools in the most deprived areas of requesting the teaching resources (p = .01). After adjustment, there remained a significant trend, with decreasing levels of deprivation associated with increasing take-up rates of the educational resources (p = .001) [28]. The interpretation of the covariates, post adjustment, was generally the same as the interpretation prior to adjustment. Larger schools were significantly more likely to take-up the teaching resources than smaller schools (OR = 2.16, 95%CI: 1.74-2.68). Geographical region also remained significantly associated with the take-up of the educational resources - indeed there were only minor variations in the odds ratios pre- and post adjustment. The major variation in the results was that the type of school (government maintained or independent) was no longer significantly associated with take-up post adjustment.

In 1986 the Ottawa Charter for Health Promotion detailed the basic goals and objectives for health promotion [29]. One of the goals was to achieve equity in health, and one of the strategies for doing this was to ensure equality of access to information. The motivation behind the strategy was the view that people can make healthier choices (in this case choose to be vaccinated), if they have appropriate information [30]. It is this issue of equality of access to information that lies at the heart of the present study. In an environment in which it is known that the most deprived areas have twice the incidence of cervical cancer as the least deprived areas [24], the most deprived and the least deprived areas should have, at a minimum, the same level of exposure to health relevant information. Indeed, as a matter of policy, it may be preferred for those areas with the highest incidence to have an even greater exposure to relevant information than those less deprived area with a lower incidence.

As anticipated, schools in the least deprived areas were significantly more likely to request the educational resources than schools in the most deprived areas; furthermore, this association held even after adjusting for school type, size and geographical region. Although the relationship between area deprivation and request for teaching resources was not particularly strong (OR = 1.32), given the association between deprivation and incidence of cervical cancer, any significant trend in exposure to health promotion material in the wrong direction is cause for concern and further investigation.

One explanation for the finding that less deprived areas had a higher take up rate of the educational resources may relate to the "inverse equity hypothesis" [31, 32]. According to this hypothesis, higher socioeconomic status groups pick up interventions quicker than lower socioeconomic status groups. This increases the health differences between the groups in the short term. However, with the passage of time, the lower socioeconomic groups begin to pick up the intervention, which then reduces the health differences between the groups. It may be that schools in less deprived areas are better placed to take advantage of the freely available educational resources, explaining their quicker take up. Assuming the hypothesis is correct, and given continued availability of the materials [33], the difference in the take up rate may reduce over time.

In addition to the modest deprivation effect, there was a regional effect, with schools from some regions (the East and South West) having almost twice the odds of requesting the education resources as schools in London or the North East (with the highest percentage of small deprivation areas in the most deprived area quintile [25]). There was also a school size effect, with the smallest 20% of schools half as likely to request the materials as the larger schools.

It is tempting to speculate why factors such as area, school size, or indeed school type might be significantly associated with the take-up of the health promotion resources. One might speculate for example that religion could underpin school size or school type effects; perhaps with more religious (and possibly conservative) schools less supportive of the HPV vaccination program. Alternatively, it might be that smaller schools, with fewer staff, simply lack the capacity to take advantage of support offered through external initiatives. Unfortunately, the data are such that none of these questions can be adequately disentangled, and any response remains purely speculative.

There are two important limitations to the findings. The first limitation relates to the operationalisation of exposure to health promotions materials. The relationship between an individual's exposure to the health promotion materials and a school's request for materials is essentially unknown in this study. It may be that schools that did not request the materials from the RSPH but obtained the materials through a secondary source. Nor is it sufficient simply for school to request the materials, they have to be integrated into the curriculum which could be constrained by competition with other topics [34], school policy or local culture [35]. The extent to which materials are integrated into the curriculum will also affect individual's exposure. Nonetheless it is reasonable to assume that, on average, students in schools requesting the materials had a higher exposure to those materials than students in schools that did not request the materials. The second limitation relates to the design. As a "natural experiment" there was no control, and a range of unmeasured (and unknown) possible confounders, making any inference about a causal relationship impossible.

There is, prima facie, a third limitation, which on reflection is unfounded. The "third limitation" is that the area deprivation of a school does not reflect the area deprivation of the students within the school. This makes less sense when it is actually drawn out. The argument would be that, on average, students attending schools located in the most deprived areas are no more likely to live in most deprived areas than students attending schools in least deprived areas. It is true that students attending independent (privately funded) schools may travel considerable distances to attend school. Schools in receipt of government aid, however, tend to draw their student body from their local area. So the area level of deprivation of a government aid school is going to be similar (the data were in quintiles) to the area in which the students live. The adjustment in the analysis for school type was particularly pertinent to this argument; and even after adjustment, students attending schools in more deprived areas were significantly less likely to be exposed to the health promotion material.

Notwithstanding the limitations, the findings support the central idea that more deprived areas are likely to have a lower exposure to health promotion messages than less deprived areas. The effect is not strong, but the accumulation of weak effects over time, can have important ramifications for population health [36], and for the disparity in the health of more and less socially deprived groups. The interaction between the level of exposure, the level of area deprivation, and the individual response to health promotion messages would be a fruitful line of future inquiry.

There was a social gradient associated with schools' response to the opportunity to receive educational resources supporting the HPV vaccination campaign. Schools in the most deprived areas were less likely to request materials than schools in less deprived areas. This was independent of other significant associations, such as school size and geographical region. This has important implication for the level of exposure to health promotion messages that people from more and less deprived areas are likely to experience.