Article Text
Abstract
Objective To assess social patterns in human papillomavirus (HPV) vaccine coverage in a school-based, government-funded vaccination programme located within a single-payer universal healthcare system.
Design We conducted a cross-sectional analysis of HPV vaccine uptake data for the 2013–2014 school year for 131 local authorities in England, and then evaluated the association between vaccine uptake and socioeconomic status at the aggregate level.
Data sources HPV vaccination coverage data from Public Health England’s vaccine uptake guidance and the UK’s March 2011 Census.
Main outcome measures We measured three-dose local authority-level vaccine series initiation to completion.
Results We found that in local authorities where there are more high-income families, the vaccination rate is lower than in local authorities with more low-income families. Local authorities with a higher percentage of whites, compared with non-whites, had higher HPV vaccination rates. Additionally, local authorities with more non-migrants had higher rates of vaccination. Local authorities with more education deprivation had higher rates of vaccination. Local authorities’ higher proportions of high-status occupations had worse vaccination coverage. In bivariate analyses across all the socioeconomic indicators, a 1 SD change in the indicators was associated with about a 2.25 percentage point decrease (for income, education and occupation) or increase (for race and migrant composition) in HPV dose coverage in the local authority. In multivariable analyses, only race remained as a significant predictor of HPV coverage at the local authority level.
Conclusions Across all three doses, there are notable variations by socioeconomic status, with steep reverse gradients in three socioeconomic indicators. More quantitative and qualitative research needs to be conducted to determine the effects of the 2014 transition from a three-dose regimen to two-dose regimen on vaccination coverage, especially in groups that experience lower rates of vaccination.
- HPV
- adolescent
- vaccination
Statistics from Altmetric.com
Key messages
Local authorities with a higher percentage of whites, compared with non-whites, had higher human papillomavirus vaccination rates.
Local authorities that have more families with someone working in higher managerial, administrative or professional occupations had worse vaccination coverage.
Local authorities with low vaccination rates are a heterogeneous group that require continued assessment.
Introduction
Cervical cancer is the second most common cancer in women under the age of 35 worldwide.1 Nearly nine women are diagnosed with cervical cancer each day in the UK.2 The quadrivalent vaccine against human papillomavirus (HPV) is effective in preventing genital warts and precancerous lesions in women, ultimately reducing the burden of cervical cancer in vaccinated populations. In the USA, racial and ethnic minorities have higher rates of HPV vaccine initiation, but lower rates of HPV vaccine completion than their white counterparts. Lower income girls also have higher initiation rates compared with higher income girls.3 One possible explanation for this relationship is that low-income girls are Medicaid eligible, have access to the State Children’s Health Insurance Program, or are Vaccines for Children eligible. Initiation rates are lowest among uninsured, lower income girls, and then increase at the highest income levels.3 These findings are consistent with access to care being an important determinant of vaccination in the USA. Many clinical interventions have been tested to improve initiation and completion rates and decrease disparities across socioeconomic strata, including patient reminders,4 provider reminders,5 and vaccine-only and nurse-only office visits.6
The social patterning of HPV vaccine coverage might be different in countries with a single-payer universal healthcare system, such as the UK, where access to care should not be a barrier to vaccination. Unlike the USA, the UK also provides the HPV vaccine in schools (offered to all girls in year 8, ages approximately 12–13). These two factors may combine to produce a more flattened gradient in HPV vaccine uptake compared with those observed in the USA. The aim of this study is therefore to describe social patterns in HPV vaccine coverage within the context of a school-based, government-funded vaccination programme located within a single-payer universal healthcare system.
Methods
We used data on dose-specific HPV vaccine coverage in England for the 2013–2014 school year by local authority, from Public Health England’s vaccine uptake guidance and latest coverage data.7 Additionally, we obtained indicators of socioeconomic status (weekly income, ethnic group, education deprivation, migration status and National Statistics Socio-economic Classification), by local authority, from the March 2011 census.8 The unit of analysis was local authority, which is a level of subnational division of England used for the purposes of local government. We were able to match HPV vaccination coverage data to census data for 131 out of 152 local authorities in the UK. The primary outcomes were local authority-level dose-specific HPV vaccine series completion (completion rates for first, second and third doses). The primary exposures of interest were local authority-level average weekly income, ethnic group, education status, migration status and socioeconomic classification. For descriptive statistics, weekly income was denominated in pounds; all other socioeconomic variables are local authority-level percentages. Ethnic group was categorised as White (English/Welsh/Scottish/Northern Irish/British, Irish, Gypsy or Irish Traveller, any other White background), mixed/multiple ethnic groups (White and Black Caribbean, White and Black African, White and Asian, any other mixed/multiple ethnic background), Asian/Asian British (Indian, Pakistani, Bangladeshi, Chinese, any other Asian background), Black/African/Caribbean/Black British (African, Caribbean, any other Black/African/Caribbean background) and other (Arab, any other). Education deprivation includes individuals with no qualifications or the lowest level of school qualification. The lowest level of qualification includes level 1 qualifications, which are first certificate, General Certificate of Secondary Education grades 3, 2, 1, or grades D, E, F, G, or level 1 award. Migration status was categorised as white migrant, non-white migrant or non-migrant. Socioeconomic classification was divided into three levels: higher managerial, administrative and professional occupations (level 1); intermediate occupations (level 2); and routine and manual occupations (level 3).
For evaluation of gradients, socioeconomic variables were operationalised to measure higher socioeconomic status at the local authority level: income level was operationalised as mean income; ethnic group was operationalised as per cent white; occupation status was operationalised as per cent level 1; education status was operationalised as education non-deprived; and migrant status was operationalised as per cent non-migrant. All exposure variables were then standardised, resulting in a mean of 0 and an SD of 1 across all included local authorities for purposes of the gradient analyses.9
We first graphed bivariate relationships between all the socioeconomic exposure variables and each HPV dose completion rate. Regression analysis was then used to examine significant associations between local authority-level socioeconomic status indicators and HPV vaccine coverage. We first tested each predictor in a bivariate model, then entered all variables into a multivariable regression. Correlations between predictor variables were not large enough to cause concerns in multivariable regression. For ease of interpretation, linear regression was used to model the proportion of eligible girls in the local authority receiving each HPV vaccine dose. For robustness, beta regression models were also estimated to account for the non-linear distribution of the outcomes (results available from the author). Beta regression models yielded substantively similar results to the linear regressions. Analyses were conducted in Stata V.14.10
Results
We were able to match HPV vaccination coverage data to census data for 131 out of 152 local authorities in the UK. The 21 excluded local authorities were unable to be matched with HPV coverage data. Vaccination coverage was high across all doses. For the school year 2013/2014, mean coverage was 90.6% for dose 1 (range: 73.0%–99.4%), 89.5% for dose 2 (range: 71.0%–96.9%) and 86.5% for dose 3 (range: 57.3%–96.6%) (see table 1).
Among all local authorities, the average weekly income was £535.3 (range: 403.66–1066.92). Most local authorities had a predominantly white population (mean: 81.6%, range: 29.0%–98.9%), followed by Asian (mean: 9.8%, range: 0.1%–43.5%) and Black (mean: 4.6%, range: 0.1%–27.2%). Local authorities were on average predominantly non-migrant (mean: 98.7%, range: 93.5%–99.8%), about two-thirds education non-deprived (mean: 64.8%, range: 50.2%–89.0%), about one-third of households having an occupant with a level 1 occupation (mean: 31.0%, range: 18.3%–66.7%). In figures 1–3, we plot the relationship between the socioeconomic status indicators and vaccination coverage. Across all three doses, negative gradients emerge for education, income and occupational status: vaccination coverage is highest in local authorities with lower mean income, fewer households with level 1 occupations, and lower educational attainment. Conversely, positive gradients emerge for race and migration status: local authorities with more white and more UK natives have higher vaccination coverage. The vertical line at 80% coverage indicates the point at which elimination of four HPV strains is possible if coverage in girls and boys is reached and if high vaccine efficacy is maintained over time.11 The yellow points to the left of that line indicate local authorities that fall below 80% vaccination coverage for each vaccine dose. We do not discern a strong socioeconomic patterning in the local authorities that fall below this threshold.
Bivariate regression models confirm these findings (see table 2, panels A–C).
We found that in local authorities where there are more high-income families, the vaccination rate is lower than in local authorities with more low-income families. This is significant across all doses (P<0.001). Local authorities with more education deprivation had higher rates of vaccination for all doses (P<0.001). A 1-point SD increase in weekly income or education non-deprivation at the local authority level results in about a 2.4% decrease in HPV vaccination coverage for dose 1 (P<0.001). Results are of a similar magnitude, direction and significance for doses 2 and 3.
Local authorities that have more families with someone working in higher managerial, administrative or professional occupations had worse vaccination coverage. A 1-point SD increase in level 1 occupation status at the local authority level results in about a 1.8% decrease in HPV vaccination coverage for dose 1 (P<0.001). This was significant across all doses.
Local authorities with a higher percentage of whites, compared with non-whites, had higher HPV vaccination rates. A 1-point SD increase in whites at the local authority level results in a 2.4% increase in HPV vaccination coverage. This was also found to be significant across all doses (P<0.001). Additionally, a 1-point SD increase in non-migrants at the local authority level results in a 2.5% increase in HPV vaccination coverage.
In a multivariable regression model, effect sizes were strongly attenuated and only race remained significant across dose 1 and dose 2 completion (P<0.05). No variables remained significant predictors of dose 3 coverage in the multivariable model.
Discussion
The UK has offered school-based, three-dose HPV vaccination to all girls aged 12–13 as part of a government-funded vaccination programme located within a single-payer universal healthcare system since 2008 and then a two-dose regimen starting in 2014. However, there were notable variations by socioeconomic status, with steep reverse gradients in three socioeconomic indicators. Our results show that girls in local authorities with more low-income families, more education deprivation, and fewer individuals working in high-level occupations have the highest vaccination coverage. In contrast, we note higher vaccination coverage in local authorities that are predominantly white and have the most native-born residents. Our results do not suggest a single profile of an undervaccinated local authority. Modelling predicts a significant impact on cervical cancer rates will occur when coverage is at or above 80%.11 Our analyses reveal that while vaccination rates are generally above 80% across all socioeconomic indicators for all doses, local authorities that fall below this threshold are a heterogeneous group.
The aggregate nature of data limits our ability to draw inferences about individual-level predictors of HPV uptake. We also cannot distinguish the role of schools not offering vaccines versus students choosing not to be vaccinated in examining coverage rates. While the indicators chosen for our analysis represent multiple dimensions of socioeconomic status, we may have omitted important predictors of coverage.
By 2014, 32% of girls in high-income countries aged 10–20 received a full course of HPV vaccine; 41% had received at least the first dose.12 In Latin America, 19% of girls aged 10–20 received a full course of HPV vaccine with only 22% having received at least the first dose.12 By contrast, these rates are significantly lower than those in the UK where the average vaccination coverage during the 2013–2014 school year was above 88% across all doses.
High overall rates of HPV coverage are achieved in part through school-based vaccination in which students provide consent. In the UK, young people aged 16–17 are presumed by law to be able to consent to their own medical treatment. Younger children who understand fully what is involved in the proposed procedure (referred to as ‘Gillick competent’) can also give consent, although ideally their parents will be involved. If a person aged 16 or 17 or a Gillick-competent child consents to treatment, a parent cannot over-ride that consent.13 In Public Health England’s school-based HPV vaccination programme, girls in year 8 (aged 12–13) are presumed to be Gillick competent (Yarwood J, 2016, oral communication, 11 November). However, there is no standard assessment of Gillick competency for the HPV vaccination programme (Yarwood J, 2016, oral communication, 11 November). A survey of HPV vaccination knowledge among female students in 13 London schools found that about one-fifth of the girls reported no awareness of HPV; among those who reported some awareness of HPV, knowledge levels were relatively low.14 Although there were no statistically significant race ethnicity differences in HPV knowledge in this study, other socioeconomic status indicators were not assessed.
Another study found that girls attending fee-paying or non-mainstream educational settings, which included pupil referral units, young offender units, hospital education service, specialist schools for students with significant additional needs and young women educated at home, were less likely to initiate vaccination in comparison to non-fee-paying educational settings, both statically significant.15
Administering the HPV vaccine to girls at school is a hallmark of the UK HPV vaccination programme. While the programme is directed nationally, it is managed locally, producing variation in programme implementation across local authorities, which may explain some of our observed variation in coverage levels and disparities. For example, some schools hold assemblies to promote and discuss the programme in advance of vaccination days (Yarwood J, 2016, oral communication, 11 November). Other schools offer science lessons associated with HPV and vaccination around vaccination days (Yarwood J, 2016, oral communication, 11 November). Nurses typically deliver vaccination services, but the nurses may be school nurses known to the students, or visiting nurses from the local authority (Yarwood J, 2016, oral communication, 11 November). Although variation in programme implementation could serve as an opportunity to look at best practices, data on complementary programming or promotion are not currently collected or evaluated in a standardised way (Yarwood J, 2016, oral communication, 11 November).
In early 2014, Public Health England and National Health Service (NHS) England announced that the school-based HPV vaccination programme would transition from a three-dose to a two-dose schedule based on compelling results from non-inferiority efficacy trials.16 17 The two-dose schedule would be easier for schools to implement, would increase series completion rates, and would reduce race ethnic and school-type disparities in completion rates. As a result, we can look at the differences between the dose 3 versus dose 2 completion rates in this analysis to anticipate full series completion under the two-dose schedule. The UK should realise a dividend of about 3 percentage points in series completion for local authorities at the mean for all sociodemographic characteristics, given current coverage rates of 86.4% for dose 3 vs 89.4% for dose 2. There is no strong signal that socioeconomic predictors of coverage differ between dose 2 and dose 3, so we would not expect significant shifts in social gradients in coverage under the two-dose schedule.
The UK incorporates both a two-dose vaccination schedule and a school-based vaccination strategy to improve coverage and reduce social disparities in HPV vaccination for girls. The UK has offered school-based HPV vaccination since 2008 on a three-dose schedule; in 2014, the Public Health England and the NHS England transitioned to a two-dose schedule. Continued assessment of the effects of this schedule change on both levels of and disparities in coverage will be important to inform this effort and others around the world to eliminate the burden of HPV infection.
Acknowledgments
We thank Caitlin McKown (University of Wisconsin-Madison) for graphics for the manuscript.
Footnotes
Handling editor Henry John Christiaan de Vries
Contributors SJ drafted the manuscript. SJ and AB analysed and interpreted the data. AB critically revised the manuscript. ME conducted the literature review and contributed to the manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.