Acceptability of risk‐stratified population screening across cancer types: Qualitative interviews with the Australian public

Abstract Background There is mounting evidence of the benefit of risk‐stratified (risk‐tailored) cancer population screening, when compared to standard approaches. However, shifting towards this approach involves changes to practice that may give rise to implementation challenges. Objectives To explore the public's potential acceptance of risk‐stratified screening across different cancer types, including reducing screening frequency if at low risk and the use of personal risk information, to inform implementation strategies. Method Semi‐structured interviews were conducted with 40 public participants; half had received personal genomic risk information and half had not. Participants were prompted to consider different cancers. Data were analysed thematically as one dataset. Results Themes included the following: (a) a sense of security; (b) tailored screening is common sense; (c) risk and the need to take action; (d) not every cancer is the same; and (e) trust and belief in health messages. Both groups expressed similar views. Participants were broadly supportive of risk‐stratified screening across different cancer types, with strong support for increased screening frequency for high‐risk groups. They were less supportive of reduced screening frequency or no screening for low‐risk groups. Findings suggest the public will be amenable to reducing screening when the test is invasive and uncomfortable; be less opposed to forgo screening if offered the opportunity to screen at some stage; and view visible cancers such as melanoma differently. Conclusions Approaching distinct cancer types differently, tailoring messages for different audiences and understanding reasons for participating in screening may assist with designing future implementation strategies for risk‐stratified cancer screening.


| BACKG ROU N D
In comparison with the mostly 'one size fits all' approaches used in existing population screening for cancer, there is mounting evidence that risk-stratified (risk-tailored) screening can both improve the early detection of cancer for those most likely to benefit and reduce the well-recognized harms of screening for those at lower risk. [1][2][3][4] Stratification of individuals into risk groups may be based on assessment of factors such as personal genomic and/or lifestyle risk information, as well as traditional risk factors such as age, family history or ethnicity. Shifting towards a risk-stratified screening approach will involve changes to current practice, including recommending more frequent screening for those at high-risk, different modalities of screening, and modified start and stop times. These changes may present challenges for implementation, for example the use of personal risk information to inform screening advice and a move towards less screening or potentially no screening for those at low risk.
To add to the evidence from modelling studies, several large trials are more definitively assessing the relative benefits and harms of risk-stratified screening. [4][5][6][7][8][9] Successful implementation of tailored screening approaches will depend on it being accepted by end-users, such as the general public. [10][11][12] Acceptability in this setting refers to the perception that the use of additional risk factors and tailored screening advice is agreeable 13 and in the target population's interest 11 ; therefore, increasing the likelihood that individuals will adhere to recommendations and benefit from improved clinical outcomes. 14 Understanding the public's perceptions in the pre-implementation stage of risk stratification will help inform strategies for successful implementation. 4,15,16 Most existing research about the acceptability of risk-stratified screening is in breast cancer, 11,12,[17][18][19][20] but very little is known about people's attitudes towards risk-stratified screening for other cancer types. In Australia, publicly funded national screening programmes exist for breast, cervical and bowel cancer. Opportunistic screening is available for skin cancer 21 and prostate cancer. 22 Multifactorial personal risk information is known to increase cancer risk prediction accuracy, 16 but there is limited research evidence on the acceptability of using such information to guide population screening recommendations. Recent studies suggest that the general public are interested in receiving personal genomic risk information for breast cancer, 11,12,23 prostate cancer 11 and kidney cancer 24 that provision of this information is unlikely to result in high levels of distress and uncertainty 25 or have a negative impact on cancer worry, anxiety and fear. 26 Awareness of personal risk factors has also been associated with willingness to participate in bowel cancer screening. 27 To realize the benefits of risk-stratified screening, low-risk individuals (who could still develop cancer) would undergo fewer screenings. This has the aim to balance the goal of cancer detection with the occurrence of harm from more frequent screening, 1,2 including overdiagnosis 28 (defined here as cancers that are correctly diagnosed but would not have produced symptoms or been identified clinically 29 ). Studies have shown that most people support risk-stratified screening for breast cancer 11,[17][18][19][20]30 and prostate cancer 11 more often if required; however, they are reluctant to reduce screening frequency if at lower risk, even when presented with evidence about the potential risks and harms of screening. Varying the frequency of ovarian cancer screening by risk, however, has been shown to be acceptable by the general public. 30 The screening tests, programmes and the values the public ascribes to screening different types of cancer vary 31 and could impact on acceptability, as could the experience of receiving personal cancer risk estimates including genomic risk information. Exploration of these issues is ideally suited to qualitative research, to gain an understanding of individuals' perspectives and the reasons for their views 32 . Therefore, this study aimed to explore acceptability of riskstratified screening in the Australian population across different cancer types in particular breast, cervical and bowel cancer (publicfunded national screening programmes) and melanoma/skin cancer (a future potential programme), to inform future implementation strategies.

| Study design and participants
We conducted a qualitative interview study with members of the public. Participants were purposefully selected to invite for an interview from the database of the Melanoma Genomics Managing Your Risk Study 33 a two-arm, parallel-group randomized controlled trial to evaluate the impact of personal genomic risk of melanoma information on sun-related prevention and screening behaviours, psychosocial and economic outcomes. Participants were eligible for this qualitative study if they had completed the 12-month followup in the Managing Your Risk Study and given consent to being contacted about future research. Participants in the Managing Your Risk Conclusions: Approaching distinct cancer types differently, tailoring messages for different audiences and understanding reasons for participating in screening may assist with designing future implementation strategies for risk-stratified cancer screening.

K E Y W O R D S
acceptability, cancer, implementation, risk-stratified, risk-tailored, screening Study were initially recruited from the Australian general population through the Medicare database. In the Managing Your Risk Study, those in the intervention arm provided a saliva sample, received a booklet with personal genomic risk information derived from a polygenic risk score, 33 received a call from a genetic counsellor, and a general educational booklet on skin cancer prevention and early detection. Those in the control arm received only the general educational booklet. For the qualitative study, we aimed to recruit participants from the intervention and control arms, to explore potential differences in views in the two groups.

| Data collection
Participants in the qualitative study were invited to participate in semi-structured face-to-face or telephone interviews approximately 3-12 months after completing participation in the Managing Your Risk Study. Recruitment was purposive, 34 with the aim to include a range of characteristics by age, sex, personal genomic risk of melanoma (low, average and high) and geographic location.
Interview guides were developed for control and intervention participants and piloted with three members of the public. Most interview questions were common to both groups (see Appendix 1: Interview guide questions for public participants) and were informed by Proctor et al's conceptual framework of implementation outcomes. 13 During the interview, participants were asked how they felt about tailored screening for cancer (in the broad sense), in particular to imagine how they would feel about being asked to decrease the frequency of screening or stopping altogether if they were given a low-risk estimate of developing cancer. This was then asked in relation to different cancer types: breast, bowel, cervical and melanoma/skin cancer. The term 'risk-stratified screening' was replaced by 'tailored screening' during interviews to optimize participant understanding of the term.
All interviews were conducted by one member of the research team and lasted between 17 and 49 minutes (mean = 32 minutes).
Participants were reimbursed for their time with a $50 AUD gift voucher.
Interviews were first conducted with participants from the intervention group. Two transcripts (third and fourth interviews) were reviewed by three team members and discussed jointly before subsequent interviews. The interviewer met regularly with the research team after completing further interviews to reflect on the process, to review and revise questions to ensure they were being answered in enough depth before remaining interviews were completed. Data saturation was reached when no new information relating to the acceptability of reducing screening frequency for different cancers was identified. Interviews were then conducted with the control group. In discussion with the research team, it became clear that subthemes were similar to the intervention group but reference to specific cancers was limited. Further in-depth exploration was undertaken through probing questions related to specific cancers in remaining interviews.

| Data analysis
Interviews were audio-recorded and transcribed verbatim. The data were analysed according to thematic analysis as described by Braun and Clarke. 35 This involved familiarization with the data, generating initial codes using a data-driven strategy, developing themes by collating codes, generating a thematic map of potential themes and further analysis of the coded data. All members of the research team initially read 4 transcripts (2 intervention and 2 control) to familiarize themselves with the data. Broad top-level codes were agreed. Analysis was undertaken as one dataset as subthemes were similar. All transcripts across the entire dataset were coded by two researchers and then data were compared across interviews to identify common patterns and initial themes.
In consultation with the research team, codes and themes and subthemes were then reviewed and further analysed to summarize the full range of views and experience. This process ensured rigour of analysis as well as coder reflexivity through guidance and discussion. Themes were defined as an important pattern or idea in the data in relation to the research question (ie the acceptability of risk-stratified cancer screening) and subthemes as a specific idea/s related to the theme.

| RE SULTS
A total of 40 participants were interviewed (from 80 invited to the interview study); 20 of whom had received personal genomic risk information (10 low risk, 3 average risk, 7 high risk from the intervention group) and 20 who had not (control group). The age range of participants was 21-68 years and 60% were female. Socio-Economic Indexes for Areas (SEIFA) scores [for advantage] were above the national average. Table 1 summarizes the participant demographics.
We found that themes were common and views were similar between intervention and control groups. Key themes and subthemes are summarized in Table 2.

| A sense of security
Overwhelmingly, participants felt that early detection of cancer in the broad sense was important and expressed faith in cancer screening to prevent and detect cancer early. Many described screening as a proactive measure in the 'battle' against cancer, often providing a definitive answer.

| Better health outcomes
When the concept of tailored cancer screening in the broad sense was explained, almost all participants described tailored screening using personal risk information as a positive and logical progression from the current mostly one-size-fits-all approach of population A small number of participants did express concern about the potential for unnecessary anxiety as a result of attending screening too often for more than one cancer. One participant discussed this is in relation to two cancers and the potential emotional and logistical burden.
She described both inconvenience and worry associated with frequent colonoscopies and the relief when a decrease/change in screening (because of her age) was recommended. As a previous breast cancer patient, she expressed the high level of worry when her on-going screening for breast cancer was extended from 5 to seven years. She summarized.
If you're liable to be high-risk of a couple of things, it could be quite concerning to be going for these different tests often, just because you're at high-risk.
(Female, 61 years, control group) The potential for anxiety related to false-positives results was also mentioned. In reference to having considered not going as often for mammography screening and cervical screening, one participant explained as follows: Because sometimes you read about a lot of false

| Scepticism around public health messages
When asked how they felt about a public health message that recommends reducing or stopping cancer screening for those at low risk, many responded that it seemed illogical and unbelievable.
I don't even think I personally believe in that statement, so I think people would be quite sceptical. I don't think people would want to take that on, or if they hear that, they might say, why would you say that kind of thing.
(Female, 25 years, low-risk) A participant who had experienced receiving personal genomic risk information, described how risk information would be better provided by a health professional than as a public message.
So I would probably need personalised information as opposed to more general public health information.
Then I'd be totally comfortable not doing it.

| Responsibility for health
Several participants felt that engaging in screening was part of taking responsibility for one's own health.
…everyone needs to be vigilant to protect themselves and it always comes back on yourself, to how you see yourself and what happens. The buck stops with you, as the old saying goes.
(Female, 68 years, control group) Finally, some were concerned that many individuals do not take enough responsibility for their own health and that reducing screening frequency might reinforce this.
I think that there's an element of risk when you tell someone that they don't need to in that it can make them complacent, or it can make them, I guess feel a lack of sense of responsibility to look after themselves.

| D ISCUSS I ON
These qualitative findings demonstrate that Australian public participants are mostly supportive of risk-stratified screening across different cancer types, at least insofar as it would lead to additional screening for those deemed to be at high risk. Tailoring screening according to individual risk was described as a logical move to improve screening precision and there was optimism around using personal genomic risk information for estimating risk for this purpose.
Consistent with other studies, 11,17,18 there was strong support for increased screening for those at high risk but not all participants felt that less frequent screening for those at low risk would be acceptable. There were some differences by cancer type and screening test in the level of acceptability of stopping screening altogether for people at low risk.
Our study has some important implications for the implementation of risk-stratified screening for different cancers. First, some participants were more amenable to reducing screening frequency when the screening test was invasive and uncomfortable. When prompted, participants identified the potential harms of screening tests including anxiety and inconvenience associated with false positives, and the high level of discomfort in some tests such as colonoscopies and pap smears, as reported elsewhere. 20 However, a significant proportion of the public already opt for unnecessary colonoscopies for bowel cancer screening, a screening intervention that is recommended only for people at increased risk. 36 A trial is currently exploring the use of a risk-stratified screening tool to increase risk-appropriate bowel cancer screening. 37 Second, offering a single screening episode which could include education about prevention and early detection, as opposed to no screen may be more acceptable to those at low risk, providing a connection to a screening programme. Reluctance to reduce or forgo screening for low-risk groups was linked to uncertainty and insecurity, given that residual risk exists even for those with a low-risk estimate; this has also been observed by others. 18 Most previous research on this topic has been conducted in breast cancer. 12 Concern was also reported about the health of those who opt out of screening altogether. 20 Some participants in our study expressed concern for and were critical of individuals who do not act to maintain their own health and saw less screening as likely to have negative consequences for this group. This view that holds individuals responsible for their own health is debated 41 and may reflect the particular characteristics of participants in this research study who primarily live in areas of socio-economic advantage.
Other studies have reported a reluctance to forgo screening linked to scepticism that reducing screening frequency was costcutting rather than evidence-based and in the public's interest. 38 This was observed in Australia when the National Cervical Screening Program changed from Pap smear tests to human papillomavirus (HPV) testing in 2017 and shifted to a later starting age and larger screening interval. 42 Audience segmentation, which tailors messages for different groups, is recommended for social marketing campaigns in cancer prevention in addition to whole-of-population approaches. 43 Our findings help to identify these distinct audiences and concerns.
Health professionals are well placed to provide clarity around public messages particularly for low-risk groups, as participants generally expressed a high level of trust in health professionals.
However, trust in health professionals' knowledge about tailored screening is not universal 23 and a need to educate health-care professionals on all aspects of tailored screening has been reported. 44 Individual expertise and personal experiences impact on risk perception and understanding, and presentation of risks in multiple ways is recommended for effective risk communication. 25,45 In the Managing Your Risk study, participants were given their remaining lifetime (absolute) risk as both a percentage and a frequency (x out of 100), comparing their personal risk with a person of the same age and gender. It was displayed as icon arrays (100 person diagram) and as a risk category of low, average or high genomic risk. We previously showed that people who received a high genomic risk result for melanoma reported higher distress and uncertainty compared with average and low-risk groups although overall these levels were low. 25 Few participants with high-risk results expressed concerns about life insurance or cancer worry, but participants across all groups frequently mentioned the importance of confidentiality and privacy.
The feasibility of providing personal risk information as part of tailored screening through health-care professionals, such as a triaged approach of general practitioner and genetic counsellor contact according to the level of risk, 46 also requires further research.
A strength of this study was that it explored people's perceptions about risk-stratified screening for different cancers, whereas most previous studies have focussed on one cancer type. Our study also provided a rare opportunity to include the views of both those who have experienced receiving personal genomic cancer risk information and those who have not, although they had only received this information for melanoma and not the other cancers. Other than melanoma, we also did not specifically record participants' lived experiences of the cancers being discussed. Another limitation influencing the generalizability of our findings is the potential for bias due to the 50% response rate in this interview study, as individuals supportive of screening may have been more inclined to participate.
Indeed, the overall higher socio-economic status of participants compared with the general population may result in a stronger interest in, and awareness of, cancer prevention and screening than the broader community. Those contacted to take part had previously participated in the parent Managing Your Risk study and nominated their interest in further research, so the response rate may also be related to practical barriers such as responding to a mailed invitation, time or research fatigue. When comparing characteristics among interview respondents and non-respondents, there were no large or significant differences between them in relation to key demographic or behavioural characteristics.

ACK N OWLED G EM ENTS
We thank the people who participated in the qualitative interviews and the Melanoma Genomics Managing Your Risk Study. The authors also thank Ashleigh Sharman for her assistance with recruitment and data management.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R ' S CO NTR I B UTI O N S
KD, AEC, NMR, LK, AKS and AJN were involved in the concept and design of the study, development of methodology, participated in data analysis and interpretation and revised the manuscript critically.
In addition, KD (PhD candidate; experienced health educator) conducted the interviews (40) and wrote the manuscript. ZS was closely involved in data analysis and manuscript revision. AEC, NMR, LK and AJN provided study supervision. All authors read and approved the final manuscript.

E TH I C A L A PPROVA L
Ethics approval was obtained by University of Sydney Human Research Ethics Committee (HREC) 2018/941.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on reasonable request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.