‘It’s a can of worms’: understanding primary care practitioners’ behaviours in relation to HPV using the theoretical domains framework

GPs and practice nurses working in Ireland were eligible to participate. GPs were recruited from a group of 145 participants in the 2007 survey [23], who were originally sampled from a national database and had indicated they were potentially willing to assist with further research. The group was diverse in terms of personal and practice characteristics and HPV infection knowledge and vaccination attitudes (in 2007). A purposive sample was recruited from this group, with strata defined in terms of variables that had been found in the 2007 survey to be strongly associated with attitudes towards smear taking and HPV vaccination, namely: gender, years since graduation, area of practice location, and HPV infection knowledge score in 2007 (questions shown in Additional file 1). GPs were approached by post, in batches in a random order, and invited to take part in an interview about their views and experiences of cervical cancer prevention. Those interested returned a reply slip, and the study co-ordinator (LAMcS) contacted them to arrange an interview. Non-respondents received up to two reminder letters. As interviews progressed, particular attention was paid to recruiting GPs in unfilled strata.

Since there is no national database of practice nurses, two approaches were used to identify potential participants. Firstly, all attendees at the annual national Irish Practice Nurses Association conference received a flyer and were invited to return this if they were interested in taking part. Secondly, invitation packs, compiled by the study team, were provided to professional development co-ordinators (PDCs) in six of the seven administrative areas across Ireland. The PDCs were each asked to select, at random, five nurses in their area to whom to send the packs. Nurses who returned the reply slip were contacted for interview. The only sampling stratum was area of practice and, as recruitment progressed, efforts were focussed on ensuring that nurses were recruited from urban and rural areas across the country.

Ethical approval was obtained from the ethics committee of the Irish College of General Practitioners. Participants provided informed consent.

Interviews were conducted by telephone by the study co-ordinator (LAMcS) and guided by a topic guide (Additional file 2). The topic guide was developed with input from a multidisciplinary team including investigators (LAMcS, LS, and SUD), the head of smear taker training at CervicalCheck, a practicing GP and practice nurse. It was informed by: literature review; a group discussion with the CervicalCheck smear taker training unit, who provide training sessions and a telephone helpline for GPs and practice nurses; discussions with the HPV vaccination team at the HSE National Immunisation Office and observation at one of their open meetings for GPs and practice nurses; and informal discussions with primary care practitioners. The guide included open questions and clinical scenarios designed to elicit information about HPV-related clinical behaviours, roles, and responsibilities in primary care, and drivers of clinical behaviours. It was organised in four clinical areas: cervical screening/smear taking, HPV infection, HPV vaccination, and HPV testing. The topic of cervical screening/smear taking was not the main focus of the study, but was included to set the context of the discussions about HPV. Participants were invited to discuss their experiences, views, barriers, and facilitators to practice, and support needs in each clinical area. The clinical scenarios covered HPV-related issues that could arise in practice and, potentially, present difficulties or challenges. Interviewees were asked to describe what they would do in each situation and how easy/difficult they would find it to handle. During the interview the interviewer chose which scenarios to present; these covered topics that had not already been raised by the interviewee, with the aim of eliciting as much information as possible.

Recruitment continued until new issues ceased to emerge for GPs and practice nurses separately. Interviews lasted 16 to 50 minutes and were audio-recorded; three participants declined to be recorded, and so detailed notes were taken instead. Recordings were transcribed verbatim and anonymised.

Content analysis was conducted following the framework analysis approach [32, 33]. GP and practice nurse interviews were analysed together. The four clinical areas were considered separately. Two investigators (LAMcS, LS) read and re-read all transcripts, independently coded these, combined codes into subthemes and allocated these, and direct quotes from participants, to one of the 12 theoretical domains/themes of the TDF [30]. They held regular discussions to resolve disagreements and reach consensus and discussed uncertainties with a third investigator (JJF). To ensure analytical rigour, a second iteration of this process was performed, with re-review of transcripts to identify any important quotes or subthemes missed or misallocated. It was noted whether subthemes arose solely among GPs, practice nurses or both. The final synthesis and interpretation involved considering each theme/domain and subtheme in the context of the whole set of interviews. The strongest/dominant themes/domains were those: mentioned by most practitioners; where the most subthemes were identified; which were discussed at greatest length; and/or which were judged by the investigators to be invested with considerable intensity, passion, or sentiment by practitioners.

Although interviews covered cervical screening, the results reported here concentrate primarily on HPV infection, vaccination, and testing. Illustrative quotes have been provided to supplement narrative descriptions.

A limited number of HPV-related clinical behaviours were identified. As regards HPV infection, the key behaviour was initiating a discussion on this topic with female patients. This was more often done if the practitioner had a relevant ‘opening,’ such as a patient with HPV reported on her smear result, or a patient presenting with genital warts. A few participants, mainly practice nurses, reported routinely mentioning HPV infection to women attending for smears. For HPV vaccination, three behaviours were identified: offering/recommending vaccination to appropriate patients; discussing vaccination when raised by a patient; and administering the vaccine. The first of these was the most common behaviour. HPV testing was very rare among practitioners and answering patients’ questions was the most important behaviour in this regard.

Taking smears was considered a predominantly female role with responsibility falling on female GPs and practice nurses, who frequently stated that patients should always have the option of a female smear taker. Male GPs were less likely to perform smears and made comments like ‘I do an occasional one when a patient requests it.’

In contrast, HPV infection was discussed with patients by both male and female GPs and practice nurses. Similarly, responsibility for behaviours relating to HPV vaccination and HPV testing fell within the remit of both GPs and practice nurses. Sometimes, a GP described discussing HPV vaccination with patients and administering the first dose, then referring the patient to the practice nurse for the two subsequent doses.

All 12 theoretical domains played a role in relation to discussing HPV infection with patients (Table 2). Memory attention and decision processes was least frequently raised, being mentioned by a single participant. The strongest domains were knowledge, emotion, social influences, beliefs about consequences and beliefs about capabilities.

The knowledge domain had several layers. First, practitioners described a general lack of knowledge and need for more information. Second, the evidence-base was perceived as rapidly evolving, and practitioners reported difficulty keeping up-to-date. Third, practitioners questioned the credibility of some information sources. For example, they noted that much of the information on HPV came from pharmaceutical companies, and they considered it to be biased. In contrast, participants expressed confidence in CervicalCheck as an information source not just as regards smears, but also in relation to HPV. As regards emotion, there was a general belief that the whole area of HPV is ‘sensitive’ and ‘awkward’ as it relates to sexual behaviour. Practitioners were concerned about patient embarrassment and, for some, their own embarrassment. The strength of the emotions involved led practitioners to adopt coping strategies such as being ‘careful’ in what they said, ‘tiptoeing’ around the topic, or avoiding it altogether. Patient lack of interest or knowledge was given as another reason for not discussing HPV infection (social influences). In addition, some practitioners were reluctant to raise the topic because they felt it might discourage women from having smears (beliefs about consequences). In relation to beliefs about capabilities, both GPs and practice nurses indicated that they found it difficult to initiate a discussion on HPV infection without some kind of ‘opening,’ such as a direct question from the patient.

As regards the skill domain, GPs observed that they found it difficult to explain HPV in a way that patients could understand. Practice nurses observed that discussing HPV infection could be seen as passing judgement on an individual’s sexual behaviour, which would be professionally inappropriate (social/professional role and identity). In terms of motivation and goals, some GPs questioned the need to discuss HPV infection with patients at all because the infection can resolve on its own and/or there is no treatment. A lack of time in consultations and lack of aids for discussion were noted in regard to environmental context and resources. Some practitioners recognised opportunities to raise the topic with patients (e.g., when a patient presented with genital warts or asked for a prescription for the contraceptive pill) (behavioural regulation). Overall, however HPV infection was not widely discussed with patients (nature of the behaviour).

The only theoretical domain that did not emerge in relation to HPV vaccination was memory, attention and decision processes (Table 3). The dominant domains were beliefs about consequences, social influences, knowledge and environmental context and resources.

Some practitioners believed that HPV vaccination was effective and safe, but others had concerns about side effects (beliefs about consequences). For example, some raised the case of a schoolgirl in the UK who died shortly after receiving the vaccine in 2009. In general, practitioners who had daughters were more likely to have a positive attitude towards HPV vaccination (social influences). Others perceived that parents would think that HPV vaccination would encourage promiscuity in their children, and negative media publicity was cited by some as a reason for not discussing it with patients. The same knowledge issues arose for HPV vaccination as for HPV infection (i.e., lack of knowledge, need for more information, and credibility of information sources). The evolving evidence base emerged as especially challenging in relation to practice. For example, practitioners expressed considerable uncertainty over whether to vaccinate sexually active girls. Few were clear or certain of what to do when faced with this, and other related situations, in the clinic. Others were certain about what they would do and why, but their reported practice conflicted with current evidence. In relation to environmental context and resources, the major barrier to practitioners recommending HPV vaccination to patients was cost; for most practitioners this was their primary concern about vaccination.

Linked to cost, practitioners perceived that there was an ethical difficulty associated with recommending to patients something that the practitioner knew patients could not afford (social/professional role and identity). Some practitioners mentioned that they were attempting to provide the HPV vaccine at a cheaper price (e.g., by buying in bulk, or forming a buying consortium with other practices) (behavioural regulation). Some practitioners had experienced a consultation about HPV vaccination where both mother and daughter were present and this was described as particularly ‘tricky’ to manage (skill) and as a ‘minefield’ (emotion). In addition, some noted that it could be difficult to determine whether a patient had been sexually active (skill). In terms of beliefs about capabilities, concerns were expressed about the ‘newness’ of the vaccines. Finally, some practitioners did not see patients in the relevant age group (nature of the behaviour) while others did not feel that HPV vaccination was necessary (motivation and goals).

Nine of the 12 theoretical domains emerged in relation to HPV testing. The dominant ones were knowledge and beliefs about capabilities (Table 4). Fewer subthemes surfaced for HPV testing than for the other clinical areas.

In terms of knowledge, there was very limited awareness of HPV testing and both GPs and practice nurses were unclear about what testing involved, how it was done and whether it was available in Ireland. Practitioners observed that there was a need for HPV testing guidelines or management algorithms, and these would make them more likely to consider offering testing (beliefs about capabilities).

All theoretical domains emerged as influences on clinical practice. Only one—memory attention and decision processes—did not play a significant role but was mentioned by one participant in relation to one clinical behaviour. Even for HPV testing, which was very uncommon, nine of the domains emerged as potential influences on practice. This perhaps reflects how complex practitioners consider the topic of HPV to be. Some domains surfaced more strongly and these are discussed below.

Various studies show that women consider HPV a sensitive topic because it is associated with sexual behaviour [34, 35]. This was echoed in our study: practitioners observed that HPV infection is a sensitive topic for patients and gave this as a reason for not raising it in consultations. While it is likely that this theme would also emerge elsewhere, it is possible that it may be particularly strong in Ireland because of cultural and social norms around discussing or admitting sexual behaviour. For example, 2008 research among women in Ireland found that there was a considerable stigma attached to having smears because it was seen to be an admission of sexual activity [17, 18]. Interestingly, the issue of sensitivity around HPV did not seem to be limited to patients; practitioners’ comments about HPV infection also revealed emotional influences. For example, they frequently talked about the difficulty of raising the subject with patients and the underlying tone of some interviews was tentative and awkward. One might expect that healthcare professionals would have ample experience of dealing with sensitive topics. Therefore, it is unclear whether there is something particularly challenging about HPV itself or whether practitioners’ emotions are influenced by their lack of knowledge, concerns about their ability to deal with patients’ reactions (skills), or other issues. Further research would be valuable to investigate the relationship between relevant theoretical constructs.

Practitioners often stated that they did not discuss HPV with patients because they believed that patients either did not know anything, or did not want to know anything, about it. Studies in various counties have shown that there is limited knowledge about HPV amongst women [36‐40]. However, a lack of knowledge is not universal, and a population survey of women in Ireland conducted in 2010 (O’Connor et al. 2010; personal communication) found that 44% had heard of HPV infection and 55% had heard of HPV vaccination. In addition, qualitative research among women in Ireland in 2008 found that, on learning about the link between HPV and cervical cancer, women were shocked, angry, and felt that this was a ‘secret’ that the medical community had kept from them [18, 19]. Women also wanted to know more about HPV. Thus, it may be counterproductive for practitioners to assume that women have not heard of HPV or are not motivated to discuss it. Our findings also suggest that influencing practitioners’ perceptions about patient influences (e.g., a lack of desire to know about HPV) might increase the behaviour. We also found that practitioners’ reported that HPV infection is difficult to explain in a way that patients understand (skill), and there is insufficient time in consultations (environmental context and resources) suggesting that other factors interact with perceived social influences in relation to discussing HPV infection. It is also possible that reasons cited by practitioners for not discussing HPV infection may mask a reluctance of practitioners themselves to discuss the topic.

Beliefs about consequences was important in relation to both HPV infection and vaccination. For HPV infection, only negative consequences were raised, while for HPV vaccination, both positive and negative consequences emerged, with practitioners tending to discuss one aspect or the other. Of note, practitioners who raised concerns regarding vaccination safety were generally unaware that the death of a schoolgirl in the UK following vaccination was subsequently found to be unrelated to the vaccination [41]. The polarized views about the consequences of vaccination may reflect the fact that the HPV evidence-base is still evolving, particularly in relation to long-term efficacy and safety [12]. However, the fact that this domain was important suggests that influencing practitioners’ beliefs about consequences might influence their behaviours in relation to HPV.

Previous research found gaps in doctors’ HPV infection and vaccination knowledge [22, 23]. This study confirms these and shows that primary care nurses also have limitations in their knowledge. Knowledge gaps were reported directly by practitioners, in their own words, and also became apparent through their responses to the clinical scenarios, meaning that practitioners varied in the extent to which they perceived that their knowledge was limited. Practitioners also described difficulties in keeping abreast of the clinical evidence; this has been identified as a barrier to primary care practice and patient care in other clinical areas [42‐44].

Knowledge limitations, and the uncertainty expressed about current evidence, are perhaps not surprising given that the HPV vaccination programme in Ireland started only recently, data continues to emerge from the HPV vaccination trials [12], HPV testing is both relatively new and rare in primary care, and no clinical guidelines are in place. However, high-risk HPV infection was established as a necessary cause of cervical cancer a decade ago [2], and the key features of HPV infection have been clear for several years. For example, the same answers would be correct nowadays for all of the questions contained in the instrument developed by Jain et al. in 2004 to assess practitioners’ HPV infection knowledge [22]. Hence, there are probably fewer uncertainties in the evidence-base around HPV infection, and it is not as rapidly evolving, as practitioners perceive.

For many practitioners, uncertainty around evidence flowed into an uncertainty about what to do in practice. Our analysis further suggested that this might have influenced how practitioners felt about the clinical areas and this, in turn, impacted on their practice (e.g., leading to avoidance). We also found a contrast between knowledge and beliefs about capabilities. Some GPs especially were high in self-efficacy although it was clear from the interviews that they lacked knowledge. This suggests knowledge to be a necessary intervention target, but shows that it is unlikely to influence clinical practice if other relevant key variables are not taken into account. Further research thus needs to establish how knowledge and other potentially relevant factors suggested in this study act together to influence behaviour.

This study was the first stage in the development of an intervention, a process being guided by the MRC Framework for the Development of Complex Interventions [45, 46]. Some investigators have observed that developing an intervention is time-consuming [47] and all studies, including ATHENS, have resource limitations. We applied the TDF, within a framework analysis approach [32], to analyse qualitative data collected using a topic guide not specifically based on the TDF domains. One of the main advantages of this was that the analysis was focused and efficient. The structured approach afforded by using the TDF to pre-define themes facilitated cross-checking the allocation of subthemes to themes and reaching analytical consensus. No influences on practitioners’ behaviours arose in the interviews that could not be classified according to the TDF, underlining the comprehensiveness of the framework. The possibility cannot be excluded that if a less structured analytical approach had been taken other aspects of clinical practice and experience might have emerged, but this risk is minimised by the wide range of psychological constructs included in the TDF. Equally, it is possible that a different analytical approach would have resulted in the themes being labelled or interpreted somewhat differently; this, however, could apply to almost all qualitative studies. A further strength of using the TDF was that our analysis was not prematurely confined to a particular psychological theory. This was, of course, a major rationale for the development of the TDF [30]. In our context, we had little a priori information on which to base the selection of appropriate theories. In view of the number of theoretical construct domains that emerged in relation to each clinical area, it was clear that using the TDF had strengthened the study: had we focussed from the outset on one or two theories (e.g., Theory of Planned Behaviour) [29], it is likely that we would have missed important influences on clinical behaviour and this would have impacted on the likely effectiveness of any intervention subsequently developed.

The evidence base for HPV testing in cervical screening and efficacy of HPV vaccination continues to develop and to date there are no guidelines on HPV in primary care in Ireland. One potential limitation of the TDF is that it was designed for use in situations in which clear evidence-based clinical practice guidelines are in place. It has previously been suggested that the theoretical domains might be of less utility in situations without clear-cut guidelines or in which the evidence base is somewhat uncertain, because the effect of potential predictors might be overwhelmed by variations in attitudes [48]. This did not seem to be an issue in our context. Although there were variations in practice and attitudes, between nine and 12 of the domains were relevant to each clinical behaviour.

A further potential limitation is that the TDF does not specify relationships between the construct domains. Our synthesis and interpretation suggested that there may be links between psychological constructs in influencing behaviour (e.g., between beliefs about capabilities and knowledge), but the TDF does not enable formal investigation of these links. Nor were we able to determine, other than in a qualitative way, which of the domains were likely to be the most important drivers of clinical behaviour. However, these limitations were in part a function of the qualitative study design, rather than the TDF per se. This study was the first step in an intervention development process and was intended to be hypothesis generating rather than hypothesis testing. Hence, we chose qualitative methods to provide a detailed picture of roles and responsibilities in clinical practice around cervical cancer prevention and to identify which domains may play a role. The next stage in the development process is a quantitative study in which we will determine the frequency of the behaviours of interest, identify the most important predictors of behaviours and investigate inter-relationships between domains and constructs predicting these behaviours. We have used the results reported here to design a questionnaire for this quantitative study. This proved to be a very efficient method of questionnaire development. Questions were included covering the domains and subthemes that were reported by more than one practitioner; this helped in focusing and limiting the length of the questionnaire, and should reduce the possibility of redundant questions. The language used by interview participants was used to form question stems, which should enhance face validity. The questionnaire study is currently under way.