What are the views of three key stakeholder groups on extending the breast screening interval for low‐risk women? A secondary qualitative analysis

Abstract Introduction There is increasing interest in risk‐stratified breast screening, whereby the prevention and early detection offers vary by a woman's estimated risk of breast cancer. To date, more focus has been directed towards high‐risk screening pathways rather than considering women at lower risk, who may be eligible for extended screening intervals. This secondary data analysis aimed to compare the views of three key stakeholder groups on how extending screening intervals for low‐risk women should be implemented and communicated as part of a national breast screening programme. Methods Secondary data analysis of three qualitative studies exploring the views of distinct stakeholder groups was conducted. Interviews took place with 23 low‐risk women (identified from the BC‐Predict study) and 17 national screening figures, who were involved in policy‐making and implementation. In addition, three focus groups and two interviews were conducted with 26 healthcare professionals. A multiperspective thematic analysis was conducted to identify similarities and differences between stakeholders. Findings Three themes were produced: Questionable assumptions about negative consequences, highlighting how other stakeholders lack trust in how women are likely to understand extended screening intervals; Preserving the integrity of the programme, centring on decision‐making and maintaining a positive reputation of breast screening and Negotiating a communication pathway highlighting communication expectations and public campaign importance. Conclusions A risk‐stratified screening programme should consider how best to engage women assessed as having a low risk of breast cancer to ensure mutual trust, balance the practicality of change whilst ensuring acceptability, and carefully develop multilevel inclusive communication strategies. Patient and Public Contribution The research within this paper involved patient/public contributors throughout including study design and materials input.


| BACKGROUND
The NHS breast screening programme (NHSBSP) in the United Kingdom invites women aged 50-70 years for 3-yearly mammography screening. Since establishing the NHSBSP, a considerable dispute over the benefits and harms of screening has arisen. 1 Harms include overdiagnosis, where a cancer diagnosis is given but if had been left undetected would not have become life-threatening, leading to unnecessary treatment. Moreover, false-positive test results can cause invasive testing and psychological distress, which may be sustained for up to 3 years. 2 However, as approximately 1300 breast cancer deaths are prevented by screening each year, the Independent Panel concluded that the benefits do outweigh these harms. 1 It has been suggested that the benefit-to-harm ratio of breast screening could be improved by utilizing risk-stratified screening, whereby high-risk women receive more frequent screening and prevention options than low-risk women. 3 Based on cohort studies in the United States of America, Canada, Australia and the United Kingdom, the risk of breast cancer can be accurately predicted in screening age women [4][5][6][7] using algorithms such as the Tyrer-Cuzick model. 8 These models produce breast cancer risk estimates that can incorporate individual breast cancer risks based on breast density obtained from mammography, single nucleotide polymorphisms and self-report information, such as family history, hormonal and reproductive factors. These three sources of information have been found to make an equal but distinct contribution to accurate predictions of breast cancer risk in a UK cohort study. 9,10 Although it is unclear whether this is the case for all groups of women as the participants were primarily of White ethnicity across the studies.
Additionally, the models currently predict the risk of breast cancer overall, rather than by breast cancer type so it is yet unclear whether they can determine whether a woman is likely to be diagnosed with a more aggressive, less treatable tumour.
Nevertheless, risk stratification could involve high-risk women being offered more frequent screening to identify breast cancer at an earlier stage and the uptake of risk-reducing medication. 11 Research has predominantly explored the potential for identifying high-risk women, where care pathways have already been developed. 12 The MyPeBS trial in Europe and the US-based WISDOM trial, are both examining the effectiveness of this approach in reducing the development of stage 2+ breast cancer through earlier detection and prevention offers to women at high risk. 13,14 A neglected aspect of risk-stratified screening concerns low-risk women, who could experience greater harms from screening, such as overdiagnosis. Furthermore, cancers detected in women identified as low-risk in a cohort study were more likely to be early-stage and slow-growing, potentially indicating that increased screening intervals would not reduce the chance of treatable breast cancer. 15 When using the Tyrer-Cuzick model with all major predictors included, 13.5% of women received a 10-year breast cancer risk estimate of less than 1.5%. 15 This is equal to the mean risk of women aged 40 years who are not yet invited to the NHSBSP due to similar issues of harms, such as overdiagnosis, outweighing the benefits. This suggests that reduced screening for low-risk women, in addition to improving the benefit-to-harm ratio, could further improve the costeffectiveness suggested for risk-stratified screening. 16 For example, women may be less likely to experience false-positive test results if having fewer mammograms, which would in turn allow for the reallocation of resources to those who would benefit from more frequent screening.
around the complex issue of communicating extended screening intervals to low-risk women. Recommendations of how to communicate are important to ensure that risk-stratified screening is successful in evading undermining trust in health services and to ensure that women understand the rationale behind reducing screening is to maximize their personal benefits and reduce risks. 21 The present study, therefore, had the central aim of understanding: what are the views of key stakeholders on extending the breast screening interval for low-risk women in relation to how this should be communicated to women.

| Design
A multiperspective design was implemented using qualitative secondary data attained from three primary qualitative studies which aimed to elicit the views of key stakeholders regarding the development of new care pathways for women at low risk of breast cancer. Semistructured interviews were conducted with (a) low-risk women 18 and (b) national healthcare policy decisionmakers, also known as NSFs. 19 Focus groups and interviews were conducted with (c) HCPs who implemented risk-stratified screening research. 20 This multiperspective analysis specifically investigated differences and similarities in how these different groups perceived extended breast screening intervals for women at low risk of breast cancer.

| Participants
Eligibility for (a) consisted of women who took part in BC-Predict and received a low-risk estimate, defined for this study as <2% 10year risk. 17

| Primary data analysis
The original data forming this secondary analysis were analysed using thematic analysis. Details of how each analysis was carried out can be found in the corresponding papers. [18][19][20] 2.4.2 | Secondary data analysis Thematic analysis 22 was conducted using principles of framework analysis to organize the data. Thematic analysis is an analytical method for systematically identifying, categorizing and proposing insight into systems of meaning across a data set. 23 This involved the steps of familiarization, coding, developing a working analytical framework, applying the analytical framework, charting data into the framework matrix and interpreting the data. 24 Framework was applied as it allowed the author to systematically deduce and identify summarized data from the large secondary qualitative data set that was relevant to the research question. 25 Codes were generated inductively at a manifest level. Coding was iterative based on discussions with the analysis team, (G. T., D. P. F., L. M.). Initial codes were generated from all three stakeholder groups' data concurrently and collated into a working analytical framework of potential themes and subthemes. A multiperspective approach was used to recognize patterns in the data and identify links and contrasts between the views of key stakeholders from the three primary studies. Codes were therefore generated across all transcripts concurrently and collated. This was a suitable approach to thematic analysis as it allowed developing themes to be reflected on TAYLOR ET AL. | 3289 when applying the coding to subsequent participants. 25 The analytical framework was applied by indexing subsequent transcripts using the existing categories and codes. The data were then charted into the framework matrix to enable thematic analysis. The framework matrix-assisted interpretation of the data and facilitated multiperspectives by recognizing patterns and identifying links and comparisons between the views of the stakeholders on extending low-risk pathways. 24 Initial themes were produced by a female student studying for a Psychology Master's degree, alongside a supervisory team who discussed initial ideas and checked the themes against the data in the original papers. Also, latent themes were generated, involving a deeper level of analysis regarding how the three stakeholder groups construed and conceptualized similar issues in different ways. 26 The final thematic structure was agreed upon with the two researchers who conducted the primary data collection (V. G. W. and L. M.).

| Sample
Twenty-three women participated in individual interviews, 20 face-toface and 3 over the telephone. 18 Interviews with 17 NSFs were conducted face-to-face (n = 10) or via telephone (n = 7). 19 Three focus groups with HCPs (n = 26) were held face-to-face. Two telephone interviews were arranged for those who could not attend the focus groups. 20 Table 1 shows demographic information about the samples.

| Theme 1: Questionable assumptions about negative consequences
HCPs and NSFs identified that there is a strongly held belief among those who attend the programme that screening safeguards against a diagnosis and that breast screening detect all breast cancers. There was concern that a change to the screening programme would cause apprehension amongst those at low risk regarding the frequency of monitoring, reducing the perceived protective status of the service.
How women would perceive a low-risk result was also debated, with both HCPs and NSFs concerned that an inability to understand personal risk may cause undue concern: Radiographer 5: I think the biggest problem is that patients don't understand what risk means, and to some low risk might mean more than what they had before, because they might presume that they have no risk but they will come for screening just because  including giving them all the information available, notably the harms.
However, women generally felt that the responsibility of decisionmaking was difficult and felt reassured being told their screening interval was extended.

| What this study adds to the existing literature
The previous primary research found how key stakeholders viewed the feasibility of risk-stratified screening, in terms of whether it is acceptable. The current paper developed these findings further by Findings revealed doubt about the role GPs would have in this aspect of risk-stratified screening, from both NSFs and women. The NSFs highlighted the importance of GPs in supporting women, yet questioned their capability of doing so based on a lack of knowledge and evidence of acceptability. Subsequently, a need for training in primary care was emphasized. However, women appeared uneasy in receiving advice about breast cancer risk and extended screening intervals from GPs and placed more trust in HCPs with expertise. This is similar to the evidence in samples of women who desire advice from specialist HCPs to discuss being at higher risk of breast cancer. 30 Supporting this, Spanish HCPs 31 identified that a strong facilitator when implementing risk-stratified screening was when information is communicated by professionals that women trust. This study also identified that implementation of risk-stratified screening requires HCP acceptability and highlighted the importance of training and education of professionals for effective communication. The need for training and education in primary care has also been highlighted in a recent systematic review although there appears to be a lack of specific research focus on women at low risk of breast cancer. 32 In addition, NSFs and HCPs lacked trust in women to understand the changing programme, and the current study findings showed that many women did not particularly consider the harms of screening when deciding to attend mammograms. These findings are consistent with research that demonstrates that although most women adhere to screening guidelines, they generally do not know how they were developed and that there are screening harms. 33 Furthermore, research has shown that women require prompting to consider screening harms in low-risk pathway scenarios, and so, addressing awareness and understanding of the harms is necessary for communicating low-risk pathways to women. 28 Despite heightened concerns from the professional stakeholders that women would not receive the change to low-risk pathways well, women in fact had concerns that were less severe than both the HCPs and NSFs. Women showed that with good communication from professional experts, most worries would be alleviated, and given it is a Government programme, they would be confident in what was advised. This incongruence of views is consistent with previous research which highlighted HCP concerns about women being resistant to low-risk pathways. 31 Furthermore, recent qualitative research found that women overall demonstrated high acceptability towards risk-stratified breast screening, however, were less accepting of a low-risk pathway. 28 An Australian study found that women were reluctant to reduce screening due to uncertainty and insecurity that low risk still means there is some risk. 34 Additionally, participant characteristics such as lower educational level was shown to exhibit less favourable views of low-risk pathways. 35 Given that over 80% of participants in the present study included a sample of White British women, it is therefore uncertain what a wider group of women may think about implementing low-risk pathways. It may be less acceptable to those not willing to take part in the study and women from ethnic minority backgrounds.
Protecting the integrity of the programme was found to be essential to effectively communicate any screening changes. The NSFs identified with this in terms of ethically supporting women and wanting to ensure that women are able to make informed choices to extend their screening interval whilst recognizing this may not be practically feasible. However, most women generally felt they would be more reassured if they were told their screening interval was extended, as the responsibility to make health-related decisions for themselves would be too difficult. There is limited research of views on providing women agency in making their own informed healthrelated decisions. However, in line with the current findings, a study of HCPs showed that 70% of participants believed that women making health-related decisions in relation to risk-stratified screening was important. 31  Similarly, tailored messages will likely be required for different groups of women. 28 The practicalities of extending risk screening intervals need to be considered. Regardless of whether risk evidence is improved, the professional stakeholders lacked confidence in their ability to provide support to women due to their understanding that the limited time available at work would make adequate training unfeasible. Decision-makers could consider identifying the minimum knowledge required for screening professionals and utilize those with greater risk expertise given that women are also identified as desiring this type of support. This is especially important given that the screening HCPs appeared more resistant to the change than the low-risk women in the present study.

| Implications for practice
Additionally, low-risk pathways need to ensure that they are acceptable, yet practical, in terms of the availability of low-risk women have to make decisions. The present research suggests that low-risk women found decisions about their health being TAYLOR ET AL. | 3293 made for them by HCPs reassuring. However, it is important to note that it is uncertain what a wider group of women, who were not willing to take part in this study, would think about this, in addition to the NSFs in this study being concerned that limited choice for women would result in scrutiny of the programme.

| Implications for research
Before extending the screening interval for low-risk women, and to mitigate scrutiny, it will be necessary to provide the key stakeholders with better evidence this it is safe. Despite the Tyrer-Cuzick model being accurate at predicting low risk of cancer, future research should produce evidence regarding how many interval cancers are diagnosed in low-risk women and at what stage they are. Such evidence could advance this model to discriminate between women who will or will not develop cancer. 36 This would benefit risk-stratified screening because if low-risk results can guarantee no interval cancers between screens, it will minimize concerns. 3 Further research with key stakeholders is also needed whereby a decisive low-risk pathway is developed as opposed to being asked about hypothetical concepts. For example, a low-risk pathway could be proposed that decides the number of years between screens and takes into consideration the current study's findings, such as information and reminders women would receive and who would provide support. This, in addition to including a more diverse sample of low-risk women like recent research with British Pakistani women 37 would further confirm the acceptability of extending the screening pathway for low-risk women. This body of evidence will be useful for decision-makers when considering whether to implement extended screening intervals.

| Strengths and weaknesses
Whilst familiarizing with the original data transcripts and analyses, the first author (G. T.) identified that the original analyses (see McWilliams and colleagues [18][19][20] did not consider in any depth the data containing extensive material on alternative perspectives regarding issues relevant to how low-risk pathways should be communicated to women. Thus, the present analysis was worthwhile with these data. A particular strength of this study includes the variety of NSFs and HCPs involved. This included members of the UK National Screening Committee and HCPs currently implementing risk-stratified screening within a feasibility study, including GPs. 19,20 This variation of roles enabled a thorough understanding of the issues based on multiple experiences.
However, as women were willing to take part in the BC-Predict study, they may have been particularly interested in breast cancer risk and breast screening. Also, over 80% of the low-risk women sample were of White British background. 18 It is, therefore, uncertain what a wider group of women may think about implementing low-risk pathways.

| Conclusions
informed consent before taking part in an interview. Participants consented to the use of anonymized quotes in publications.