Background
Persistent infection with Human Papillomavirus (HPV) causes 99% of cervical cancers [
1]. In the UK, primary HPV screening was instituted in six ‘sentinel sites’ in 2011 (including Bristol) and the high negative predictive value of this test (>99%) [
2] has led the NHS Cervical screening programme to recommend replacement of the current HPV triage of low-grade cytology with HPV primary screening. The screening sample will initially be tested for one of the 30 high risk HPV subtypes and if positive, a woman’s risk of pre-cancerous change will be triaged into low-grade or high-grade by a cytology test. This result helps to stratify who may require treatment; with a HPV positive, high-grade cytology result there is an 82% chance that the excised tissue will contain high grade pre-cancerous cells and a 2.6% chance of cancer. With a HPV positive, low-grade cytology result, this risk is 15.9% and 0.1% respectively [
3].
However, a positive screening result is not a definitive determinant of outcome. The positive predictive value (PPV) of detecting CIN2+ in women who have high risk HPV and low-grade cytology is only 16% [
2]. 80% of immunocompetent women will clear a HPV infection [
4] and the low PPV indicates that HPV testing currently fails to differentiate between these women and the 20% who will have persistent infection.
Therefore, women with a positive cervical screening test are referred to the colposcopy clinic. The purpose of this assessment is to visualise the area infected by HPV and determine who requires treatment and who can be managed with conservative cytological follow-up. Management difficulties arise when the cells of interest are ‘tucked inside’ the cervix and are not visible for assessment. This is known as unsatisfactory colposcopy or a transformation zone (TZ) type 3, the incidence of which is approximately 20% [
5], potentially accounting for 34,555 of the 172,776 women reviewed in Colposcopy each year [
3]. The identification of a TZ type 3 and the inability to provide histological selection for treatment may deter cytological follow-up and lead to higher rates of excisional treatments in women with low-grade screening results [
6] to prevent missing a ‘hidden’ cancer.
The quandary for Colposcopists is the treatment related morbidity as a deeper excision (15-25 mm) is recommended in this cohort [
7]. This treatment can increase the risk of preterm birth [
8] and cervical stenosis which in turn can lead to infertility and difficulty with future cytological assessments [
9,
10]. Furthermore, there is an 8.6 fold increased risk that the excised tissue will be normal when compared to excisions where the cells of interest are visible [
6]. It seems clear that a reduction in unnecessary treatments is needed to improve patient outcomes.
Currently, guidance for cytological follow-up versus excisional treatment is hospital specific [
5,
7,
11] in this cohort and this may lead to disparities in care. Attendance rates for colposcopy and loss of patients to follow-up has been shown to be affected by service inefficiencies [
12], anxiety [
13,
14] and poor accessibility to targeted information [
15,
16]. With non-attendance rates for colposcopy in the UK documented at 24.4%, of which 46.1% are follow-up appointments [
17], areas of heterogeneity in service provision need to be improved.
Clinical decision-making is a complex process and the inconsistent nature of intuitive management has led to the development of evidenced based practice [
18] which aims to minimise morbidity and maximise optimal outcomes. However, when a paucity of evidence exists, decision-making under conditions of uncertainty can be influenced by patient choice or demographics and health care provider attitudes, experience, age, gender or culture [
19,
20]. Colposcopists play an important role in leading research and policy change in cervical screening programmes and there is currently no literature to suggest how their opinions and experiences shape the management or counselling of these patients. The aim of this study was to identify factors that affect Colposcopist’s decision-making, specifically recommendations for excisional treatments over cytological follow-up, and to interpret these findings in line with decision-making theory.
Method
This was a qualitative study, utilising a series of focus groups. Colposcopists working in NHS trusts within the South West of England Region were purposively sampled to achieve maximum variation [
21]. Participants were included if they were active accredited members of the British Society for Colposcopy and Cervical Pathology (BSCCP); membership is a pre-requisite to practice as a Colposcopist as the BSCCP standardises training and audits quality of service provision. In two waves of recruitment, lead Colposcopists from each trust (who are responsible for quality assurance) were contacted by email. They forwarded the study information to all Colposcopists within the unit to request participation. Prior to taking part in the focus groups, Colposcopists’ provided written consent to participate, to be audio-recorded and for anonymised quotes to be utilised in publications.
Sampling aimed to ensure heterogeneity, such that a range of demographics and opinions were included to improve the generalisability of the findings. Data regarding years of experience in colposcopy and job title were collected. Different health care professionals such as nurses, oncologists, family doctors and general gynaecologists can accredit as Colposcopists; their background experience may shape their attitudes and opinions and affect the decisions they make. Protocols and educational experience may vary in different units and therefore demographic location of the Colposcopists’ training centre was also collected. To maintain anonymity, age and gender were not collected. Data saturation [
22], such that no new opinions or attitudes were identified, was achieved with a total of twenty-three Colposcopists from four centres. At this point, recruitment ceased.
A topic guide (Table
1) was designed by qualitative and clinical researchers in collaboration with three experts in the field (who did not participate in the focus groups). An expert was defined as a Colposcopist who was respected and nominated by their peers for their expertise in colposcopy, as these practitioners manage complex as well as routine cases. The topic guide consisted of open questions; these aimed to focus the discussion and allow an exploration of the decisions that are made when reviewing women with unsatisfactory colposcopy and a range of clinical and cytological variables. These questions focused on length, clinical setting and technique of follow-up, the depth of excision undertaken in relation to patient and clinical characteristics and how Colposcopist’s practice is influenced by the current literature and guidance. These semi-structured focus groups enabled the researcher to cover the core set of questions and allow a flexible and dynamic discussion that could be led and expanded upon by the participants. To understand the Colposcopists’ decision-making process, participants were asked to identify the criteria they used in management decisions by asking ‘why’ and ‘how’.
Table 1
Focus Group Topic Guide
Conservative management - Why - Effect of age - Effect of parity - Effect of HR HPV - Length of follow up - Place of follow up - Technique of follow up (what) | Depth of LLETZ - How deep? - Why? - Who? - Alternative methods of treatment |
Who to treat? - How - Where - Why - Effect of age - Effect of parity - Effect of HR HPV | Quality monitoring - Reporting methods - Interpreting the reports - HPV education of colposcopists |
Oestrogen - Who? - How? - Alternative uses - Other methods of everting the TZ? | Issues regarding colposcopy management - National guidance - Patient focused. |
A researcher (KM, who is an accredited Colposcopist and trained in qualitative research methodology) conducted the focus groups. Due to the amount of technical content, the challenge of conducting these discussions was reduced by KM being a Colposcopist. A facilitator (SP or RB) were present to aid transcription by recording the speaker order, noting non-verbal communication and assisting in further exploration of raised points. The focus groups were composed of staff from each individual trust (block sampling), conducted in private rooms at the participants’ hospital and lasted 40–50 mins. Refreshments were provided to facilitate participation during Colposcopists’ lunch times. Field notes were written immediately after the focus group to aid in the interpretation of the transcripts and contextualise the discussion. Interviews were transcribed (by KM) with the aid of the speaker list. All participants were anonymised and the transcripts sent to respondents and the facilitator (SP/RB) for validation. One participants’ statement was refined to expand upon how their anxiety of missing a cancer affected their decision-making, but this did not lead to the researchers disregarding their interpretation of the data. The qualitative software package NVivo 10 was used to aid analysis.
Thematic analysis (TA) was selected as the most appropriate method for this study; it is one approach that can be used to identify, analyse and organise patterns of opinions within a data set. We chose this method as the experiences of participants can be analysed without evaluating how they experience reality (such as with IPA or phenomenology). It also provides the flexibility to allow participants to expand upon their worries and interests without deviating from the decision-making process (which was the aim of this study). TA was chosen in preference to grounded theory as data was collected through the use of focus groups and the focus was not on social processes [
22,
23]. Data was inductively coded, using the six stage TA process described by Braun and Clarke [
24]; as outlined below.
In an iterative process, analysis was conducted after each interview so that future focus group questions were informed by prior analysis [
22]. After familiarisation with the data and discussion between the researchers, a coding list was developed and the first transcript individually coded by three researchers (KM, RS, RB). RB and RS are postdoctoral qualitative researchers who are affiliated with the University of Bristol and have no clinical involvement in the Colposcopy units that were approached. The data informed the coding, rather than using a rigid pre-designed coding structure or framework of behavioural determinants, to reduce the chance of the researcher’s pre-conceived ideas affecting the themes that were identified in the data set. The coding framework was then applied to future transcripts and revised once more as further transcripts were analysed. To achieve a rigorous analysis, consistency of interpretation was assessed: Two research members (KM and RS) independently coded the last three transcripts, results were compared, divergences discussed and disagreements settled by a third researcher (RB). Field notes were compared with the transcripts to define tone and potential meaning of the words transcribed.
On completion of the coding, all researchers met to discuss and refine key themes that were described in the data. Themes were then defined following in-depth consideration of potential alternative interpretations through the use of mind maps and iterative lists. In a semantic approach, illustrative quotes, descriptive accounts and tables of the themes were then developed from the data. A framework of decision-making was developed after a literature search and brainstorming. Themes that were identified within the data were then mapped to the relevant theoretical constructs within this framework.
Discussion
Excisional treatments have helped reduce the overall mortality rate from cervical cancer by 60% in the UK [
17] but the benefits of treatment have to be tempered with the associated morbidity. Women with unsatisfactory colposcopy and low-grade cytology have higher than desired treatment rates [
6] and it is therefore important to explore factors which may influence decision-making in this cohort. This paper addresses an important issue – the ways in which medical practitioners, in this case Colposcopists, make decisions under conditions of uncertainty. A qualitative approach sheds a useful light on the process of decision-making and to the best of our knowledge this is the first study which addresses this issue. Where rational judgement, cognition and affect could be applied, areas of consensus were identified; A multidisciplinary team decision, patient preference, a high-risk screening result or a low-risk result in combination with patient risk factors such as poor compliance, smoking, high parity or older age resulted in recommendations for excisional treatments. In areas of clinical uncertainty Colposcopist’s experience, knowledge, rational judgement, perception and affect influenced decision-making. When faced with an inability to provide colposcopic assessment or diagnostic histology the psychological stress of missing a cancer, even in women with low-grade screening, deterred prolonged or community based cytological follow-up. A paucity of guidance and patient anxiety further compounded decision-making and led to heterogeneity in care.
Decision-making is a complex process which incorporates knowledge, risk assessment, analytical skills, prior experience and affect [
25]. Decision-making can be challenging in areas of clinical uncertainty where guidance is sparse [
26,
27], when an adverse outcome such as a cancer may occur as a result of the decision [
28] or if a large number of variables need to be contemplated when making a decision [
26].
These themes were illustrated in our study when participants, particularly Gynaecological Oncologists, suggested that the possibility of removing high grade disease outweighed the risk of treatment-related morbidity in women with significant risk factors. In women with low-grade cervical screening, the TOMBOLA study [
29] advocates a policy of surveillance rather than immediate treatment to allow regression of pre-invasive disease. This policy however relies on colposcopic visualisation and histological confirmation of the lesion, which cannot be undertaken in women with unsatisfactory colposcopy. Whilst conscious of the risk of over-treatment, particularly in younger women, participants were more concerned about missing a developing cancer. This finding is supported by studies which have shown that in areas of uncertainty, decisions are made faster and more easily by relying on emotion [
30]. Furthermore, when an emotive thought, such as fear of missing a cancer, induces anxiety, this can lead individuals to place more weight on the negative outcomes than the positive [
31,
32]. Once distracted by a negative stimulus it is then difficult to divert attention from these negative thoughts [
33]. Anxiety has been associated with increased amygdala and reduced pre-frontal activity [
34] which suggests that in areas of uncertainty affective components of decision-making may take precedence over rational cognitive elements [
35,
36].
In our study, uncertainty of decision-making in women with low-grade cytology was reduced by the perceived increase in risk that a persistent HPV result conferred. However, recent evidence has shown that the proportion of women with a low-grade screening result and subsequent grades of pre-cancerous disease (CIN 1, 2 or 3) is no different following the introduction of HPV testing [
3]. What has fallen is the number of women referred to colposcopy with inadequate or borderline results and those with normal colposcopy [
3]. This could be falsely viewed as an increase in individual risk, leading to a more aggressive management approach when histological selection for treatment is not possible. Most people are naturally risk adverse and look to avoid poor outcomes by selecting the least risky option [
37]. Uncertainty of outcome (inability to visualise the transformation zone) heightens anxiety and compounds this risk aversion. When it is not clear whether the alternative decision may result in further risk or benefit, willingness to take a risk, in this case prolonged cytological follow-up, is avoided [
38]. National guidance on the risk conferred by a HPV result in women with low-grade screening may reduce the dominant role of affect and strengthen the cognitive component of decision-making. Moreover, studies which assess the benefit of HPV genotyping in this cohort may also assist in risk stratification as the 10-year incidence of CIN3 is 17% with HPV 16, 14% with HPV 18 and 3% with other high-risk subtypes [
39].
The study group identified discrepancies in the recommended technique and setting of cytological follow-up. Colposcopy nurses preferred community follow-up and this may be a reflection of the higher volume of patients they see. Whereas the majority of Doctors favoured colposcopy follow-up and this attitude may be influenced by the higher proportion of women with cervical cancer they manage. Although current evidence suggests an increased cytological yield when using a cytobrush in combination with a Cervex-Brush (which samples cells from inside and outside of the cervix) [
40], there is a paucity of evidence correlating this increased yield of cells when used in conjunction with unsatisfactory colposcopy [
41]. This lack of knowledge and the inability of community services to offer a cytobrush compounded decision-making, particularly for doctors. Studies which improve knowledge in this area may aid rational judgement.
Patient choice was cited as a major influence affecting decision-making. Eighty one percent of referrals to colposcopy are for low-grade screening results [
3] but patients report the same level of anxiety irrespective of the cytological grade [
14]. This anxiety is driven by fear of cancer, worries that subsequent cytology will be abnormal and future fertility concerns [
14,
42]. There is a plethora of literature assessing women’s preferences for the management of low-grade cytology when colposcopy is satisfactory, with the majority of studies showing a preference for colposcopic review over cytological (cervical smear) surveillance [
43,
44]. Furthermore, if cytological follow-up is chosen, women have cited a preference for ‘regular’ screening [
45]. Until such time as one outcome is shown to be superior to another it could be argued that Colposcopists should advocate the more cost-effective approach of cytological follow-up. However, in a shared care model, determining patient preferences will improve patient satisfaction and outcomes [
46] – even if this involves, as shown in our study, young women with low-grade screening and low risk factors choosing excisional treatments over cytological follow-up.
To reduce the emotional burden of decision-making, health care providers will defer decision-making [
47] and the majority of our participants felt a reduction in emotional burden following an MDT decision to offer excision, particularly in young and/or nulliparous women. This finding is supported by studies which have shown that the use of the MDT reduces overtreatment in the colposcopy setting [
48] and potential heterogeneity in care. Although it should be noted, that with a paucity of evidence to guide this expert body’s management, homogeneity of care may be achieved within departments or regions but may not occur at a national level.
In areas that lack evidence it is clear that prior expertise forms the basis of decision-making [
49]. This was evident when participants recommended a depth of excision which was incongruent with recent national guidance [
7]. Furthermore, experienced Colposcopists were more likely to recommend longer total length of conservative follow-up and at 12 month intervals in women with low-grade screening. Evidence has suggested that experts are ‘wise risk takers’ [
50], their knowledge reduces anxiety and indecision allowing them to make decisions which deviate from set guidance to individualise care [
51].
Colposcopists are independent practitioners and it could be argued that guidance may not be necessary in scenarios which lack consensus of opinion. Furthermore, it is clear that not all clinical scenarios can conform to guidance and removing all uncertainty from the medical profession may hinder adaptability, critical analysis, maturity of thought and patient choice. That being said, part of a clinicians’ duty is to reduce patient anxiety and optimise clinical outcomes, but how can this be achieved if the clinician themselves is plagued by anxiety. In situations where there is a lack of clear evidence, affect may compromise rational judgements. Homogeneity of care improves service provision and clinical outcomes through consistent use of evidenced based interventions [
52] and the majority of decisions in colposcopy contain a small number of variables and are fairly unambiguous. Guidelines improve decision-making in areas of ambiguity, recognize shortfalls in the literature, provide assurance that clinicians are advocating appropriate treatments and promote under-recognised and neglected patient cohorts.
This study had some limitations. Assessing practice in one geographical UK region may increase the institutional bias but the inclusion of four centres with varying patient populations and participants who trained in different regions improved the generalizability of the data. Moreover, there was no difference in opinion based on training location. To triangulate these findings we propose a nationwide survey, based on the themes identified, to explore the frequency of these opinions and identify areas of consensus where guidance may be clarified. It is also important to consider why participants agreed to take part; it could be argued that attendees did so to express a particular viewpoint and therefore the data may not resonate with national opinions. However, only four of twenty eight Colposcopists did not participate and this was due to conflicting clinical commitments. Furthermore, two of these gave written statements for clinical scenarios that they wished to be discussed. Although these statements were not used in the analysis, they stimulated animated discussions on the optimal cytology collection device and the risks and benefits of repeating the referral cytology at the first Colposcopy appointment.
Age and gender were not collected for confidentiality reasons. However, gender has been shown to influence clinical decision-making. Female clinicians can have longer consultation times with more time devoted to counselling [
53] whereas male colleagues may spend more time discussing technical aspects [
54]. As a key outcome of this study was factors affecting choice of conservative or surgical management, assessing this association would be useful for guideline implementation and should be explored in future studies such as a national survey. Correlation of experience with management decisions may also be valuable for guideline development and it could be argued that more weight should be applied to recommendations from Colposcopists who have more experience in this area. However, measuring competence and experience can be problematic; most decision-making in Colposcopy contains limited variables and is formulaic, all Colposcopists have to attain the same basic competencies and re-accredit three yearly. Therefore a consensus opinion may provide best recommendations for guidance and in areas of dissonance, further research will be required to aid analytical thinking and reduce uncertainty.
Focus groups, rather than interviews or questionnaires, were chosen as the method of study as numerous viewpoints on a specific issue can be studied in an interactive setting and comments made by individual participants stimulated group discussions. The use of focus groups reduced the interaction of the facilitator and the input or potential bias of the researchers. Moreover, they provided richer data than a questionnaire by expanding upon the decision-making process and enabling targeted suggestions for guidance which was the key component of interest in this study.
Block sampling was chosen as it can be more representative; a heterogenous group ensured differing opinions were shared leading to lively debates in some of the units. Sensitization, with the possibility of pre-set answers which may reduce analytical thinking during the focus groups, was reduced by the provision of a general theme in the participant information sheet rather than set questions. Further strengths included the use of open ended questions, an extensive coding process and an iterative analysis which helped ensure saturation and depth of information was attained. Transcription of the data by KM facilitated reading and interpretation of the data as suggested by Braun and Clarke. None of the participants withdrew their data and the respondents verified the validity of the transcripts. The interpretivist method of data collection, the heterogenous group, double coding of the transcripts and a self-awareness of the researchers own preconceptions by including both Colposcopists and qualitative researchers in the study group will have reduced reflexitivity [
55].