How does HPV vaccination status relate to risk perceptions and intention to participate in cervical screening? a survey study

Women in several countries will soon be covered by two preventive programmes targeting cervical cancer: HPV vaccination and cervical screening. Screening for cervical cancer has been followed by a reduction in cervical cancer incidence and mortality assumedly caused by screening [1]. The two HPV vaccines currently on the market prevent infection with HPV types 16 and 18, which are responsible for approximately 70 % of cervical cancers [2]. Thus, a significant reduction in cervical cancer incidence is expected among the women vaccinated before sexual debut. The first women vaccinated as adolescents have recently entered the cervical screening programme, and some countries have revised their screening programmes [3, 4].

Women’s knowledge about cervical cancer is generally poor, and they overestimate their risk of cervical cancer and the effect of cervical screening [5‐7]. After implementation of the vaccine, women’s knowledge and attitudes to cervical screening have been investigated [8‐13]. The concern has been raised, that the HPV vaccine gives “a false sense of security” because women believe that they have eliminated their risk of cervical cancer [11, 14]. However, these studies were conducted a considerable time before the study populations were invited for screening. A few studies have investigated actual screening behaviour among the first women vaccinated in catch-up programmes: An Australian study suggests that vaccinated women are being screened at lower rates than unvaccinated women [15], while a study from the UK indicates the opposite [16]. However, we still do not know how screening behaviour will be among women vaccinated as adolescents.

Therefore, our aim was to investigate if there was an association between HPV vaccination status and perceptions of cervical cancer risk; perceptions of vaccine effect; and intention to participate in cervical screening. Furthermore, we investigated associations between risk perceptions and intention to participate in cervical screening.

HPV vaccinated women more often than unvaccinated women intended to participate in screening and they perceived cervical cancer risk to be higher and the vaccine effect to be larger than unvaccinated women did. However, in our analyses, risk perceptions could not explain screening intentions among neither vaccinated nor unvaccinated women.

The response rate of 29.7 % is low, but the age group might be considered a “hard-to-reach” group in the context of survey research. Drop-out analyses reveal socio-demographic participation bias: respondents seem to come from better educated families. Completing the questionnaire would potentially have been more cognitively challenging for the non-respondents, which might have created increased random error due to incomplete understanding of the questionnaire. We, however, have no reason to believe that our results are highly sensitive to this response bias.

Some respondents implicitly reported a negative effect of the HPV vaccine. We assume that these results are due to misinterpretations of the questions. Since our measure of respondents’ perceptions of the effect of HPV vaccination are derived indirectly from the former two questions on cervical cancer risk subject to HPV status, any misinterpretations of these two questions will spill over onto the perceived HPV effect. The bias introduced by such random errors was reduced by dichotomising risk-perception outcomes.

One may question web-based survey methods and our use of smartphones via an SMS-link as administration form. However, several studies have found that the validity of web-based methods of administration are equal to paper questionnaires [25, 26]. Moreover, studies have shown that the use of PDA, which is highly comparable to a modern smartphone, as administration method is not inferior to other more commonly used methods in survey research [27, 28]. Since the target group of this survey was young women who are generally very familiar with the use of smartphones [29], we believe that our method of administration is a strength of this study. Furthermore, the use of personal telephones potentially increased the likelihood that the intended individual answered the survey. The electronic format does, however, introduce a potential source of bias: Respondents were to provide their answer to the risk perception questions on a slider ranging from 0 to 1000 with the slider set at the default position 500. This might have caused some women to place the slider near the default position. Although we thought to minimise this problem by designing the questionnaire such that one could not proceed without moving the slider, we still observe a “hump” around 500 for the question concerning cancer risk conditional on negative HPV vaccination status (Fig. 2). Since the use of a slider is a relatively new method in survey research, there are, to our knowledge, no studies validating an electronic slider. However, the slider is comparable to a VAS scale, and the hump we observe could also be caused by central tendency bias, which is known from scales in general.

A strength of this study is that we were able to control for a number of potential confounders by adjusting for a number of socio-demographic variables obtained from registers and questionnaire data.

Our study confirms previously observed socio-demographic inequalities between HPV vaccinated and unvaccinated women [16, 30‐32]. The proportion of women intending to participate in screening in our study is a little higher than current screening rates in Denmark among vaccinated as well as unvaccinated women [33]. Overall, respondents, HPV vaccinated as well as unvaccinated, had highly inflated perceptions of cervical cancer risk. Such overestimations have been observed previously in other survey studies conducted before and after implementation of the HPV vaccine [5, 9]. However, HPV vaccinated women perceived the risk conditional on negative HPV vaccination status as higher and the risk conditional on positive HPV vaccination status as lower than unvaccinated women did. In other words, HPV vaccinated women perceived the vaccine to be more effective than did unvaccinated women. This could possibly be explained by the classic theory of cognitive dissonance [34]: women who are unvaccinated might unconsciously evaluate the HPV vaccine less favourably to resolve any dissonance between their perception of the vaccine and their own vaccination status.

Risk perceptions should expectedly be a predictor of screening intentions. However, we were not able to demonstrate such a relation in this study. This could be due to lack of power, since there was a tendency towards women (vaccinated as well as unvaccinated) with lower perception of cervical cancer risk to be less inclined to participate in screening.

We found a strong association between HPV vaccination status and intention to participate in screening with more HPV vaccinated women than unvaccinated women intending to participate, even after adjusting for a number of socio-demographic variables. This is in line with the findings from a previous survey of Scottish school girls [35] and a cohort study from the UK [16]. Our results suggest that we should not expect that vaccinated women will not attend screening due to a false assumption that they are completely protected against cervical cancer. This is also supported by a median perceived HPV vaccine effect of 77.1 % among HPV vaccinated women, indicating an overall realistic perception of the vaccine effect. That unvaccinated women might choose not to attend cervical screening can, however, raise concern. Socio-demographic inequality is well-known in screening programmes [36‐41]. A previous study has specifically shown that non-attenders in the Danish cervical screening programme have less favourable socio-demographic profiles than attenders, e.g. shorter education and limited use of primary health care [42]. If non-attenders are also less likely to be HPV vaccinated than attenders are, inequity in health will be aggravated.

Our study suggests that HPV vaccinated as well as unvaccinated women have inflated perceptions of cervical cancer risk. Moreover, vaccinated women are much more inclined to participate in screening than unvaccinated women. This is concerning, since those who would benefit the most from screening might not participate, while healthcare resources are spent on those who benefits least. The implications for practice are mainly pertaining to future information interventions directed at women invited for cervical screening. Both groups of women should be provided with clear and balanced information about their personal cervical cancer risk (i.e. according to HPV vaccination status) and the benefits and harms of screening based on best available evidence, so that they have the best possible prerequisites to make an informed choice.