Psychological impact of risk-stratified screening as part of the NHS Breast Screening Programme: multi-site non-randomised comparison of BC-Predict versus usual screening (NCT04359420)

The NHS Breast Screening Programme (NHSBSP) aims to reduce breast cancer mortality through early detection. However, screening causes harms like false positives and overdiagnosis. Risk-stratified screening, offering different options based on risk, could improve the benefit-to-harm ratio. The National Institute for Health and Care Excellence (NICE) recommends increased screening and risk-reducing medication for high-risk women, but this is limited by unknown risk status in most women. Assessing risk at screening would enable this, and trials are underway. Before implementation, it's crucial to determine if risk-stratified screening induces harms, such as increased cancer worry or anxiety. Previous research, particularly the PROCAS study, showed limited harms, but limitations existed: risk estimates were given years after questionnaire data collection, highest risk women were excluded, the comparison group might have had elevated anxiety, and baseline anxiety/worry weren't measured. Potential benefits include personalized risk information leading to behavior change and increased knowledge for informed screening decisions. This study aimed to assess whether BC-Predict increased harms (anxiety, cancer worry) or benefits (knowledge, intentions to attend future screenings) at 3 and 6 months post-screening. Specific objectives included comparing self-reported harms and benefits between BC-Predict and NHSBSP groups, comparing responses of women accepting vs. declining BC-Predict, and examining differential changes based on risk estimates (high, moderate, average, below average).

Existing literature on breast cancer screening highlights the balance between benefits (reduced mortality) and harms (false positives, overdiagnosis). Risk-stratification is proposed as an innovation to optimize this balance by tailoring interventions based on individual risk levels. NICE guidelines advocate for enhanced screening and chemoprevention for high-risk individuals but their implementation is hampered by the lack of widespread risk assessment. The Predicting-Risk-Of-Cancer-At-Screening (PROCAS) study provided preliminary evidence regarding the psychological impact of risk estimation but suffered from several methodological limitations, including delayed communication of risk estimates, exclusion of the highest-risk women, and the absence of baseline anxiety/worry measurements. These factors necessitated the current study to address the critical knowledge gaps concerning the psychological implications of timely risk communication within a robust methodological framework.

This multi-site nested study was embedded within the larger PROCAS2 trial, a non-randomized controlled trial comparing BC-Predict (risk-stratified screening) and standard NHSBSP. Ethical approval was obtained. Women scheduled for mammograms at seven sites in North-West England were eligible. Two groups were invited: those invited for first-time screening and those nearing their 60th birthday. Inclusion criteria included biological female sex and ability to consent. Exclusion criteria included prior breast cancer or bilateral mastectomy. Women offered BC-Predict received an additional invitation letter along with a participant information sheet and instructions for online risk assessment. Risk assessment used self-reported information (family history, parity, BMI etc.), mammographic breast density, and genetic information (SNPs from saliva samples). Women with clear mammograms received a letter with their 10-year risk estimate (high ≥8%, above average/moderate 5-7.99%, average 2-4.99%, below average/<2%). High and above-average risk women were referred to a Family History and Risk Prevention Clinic. Questionnaires were administered at baseline, 3 months, and 6 months post-mammogram. Measures included perceived relative risk, state anxiety (six-item STAI short form), cancer worry (Lerman scale), knowledge about breast cancer screening, attitudes towards screening, intentions to attend future screenings, and (for the BC-Predict group) satisfaction with information. Demographic data included age, IMD decile, and screening type (first-time vs. repeat). Data analysis used t-tests and chi-squared tests for comparing characteristics between groups, ANCOVA to compare changes in outcome variables between groups, controlling for baseline values, age, and IMD. A last-occasion-carried-forward approach was used for sensitivity analysis on missing data. All tests were two-sided with a 5% alpha level.

Of 5901 invited women, 662 (11.2%) participated. There was no significant difference in participation rates between BC-Predict and NHSBSP groups. Women in the BC-Predict group had slightly higher IMD deciles (less deprived) than the NHSBSP group. Both groups were similar in age and mammography attendance. Follow-up rates were 77.2% at 3 months and 71.5% at 6 months. Analyses comparing BC-Predict and NHSBSP groups showed no significant differences in any outcome variable (comparative risk perception, state anxiety, cancer worry, screening knowledge, attitudes, or intentions) at 3 or 6 months. A sensitivity analysis confirmed these findings. Baseline scores strongly predicted follow-up scores. Women accepting BC-Predict had slightly higher IMD deciles, higher screening knowledge, lower risk perceptions, and less cancer worry than those declining. Analysis of the BC-Predict group according to risk estimates revealed that risk perceptions changed in line with the provided risk estimates. Cancer worry also aligned with risk estimates at 6 months. State anxiety increased significantly at 6 months in the below-average risk group compared to other groups. There were no significant changes in screening knowledge or intentions to attend future screenings related to risk group.

This study provides strong evidence that offering BC-Predict did not increase psychological harms (anxiety, cancer worry) compared to standard NHSBSP. The findings are consistent with previous research but overcome limitations such as delayed risk communication and inadequate control groups. The strong correlation between baseline and follow-up psychological scores indicates that pre-existing psychological distress, rather than the intervention, was the primary driver of changes in anxiety and worry. The study's limitations include non-randomization leading to potential group differences in deprivation levels (although analyses controlled for this) and a low response rate (11.2%), potentially introducing response bias, although the response rate was similar across groups. The unexpected increase in state anxiety among women in the low-risk group at 6 months warrants further investigation. The lack of impact on screening knowledge suggests that risk-stratified screening alone may not improve understanding of broader screening-related issues. Overall, this study reinforces the conclusion from the previous PROCAS study and indicates that risk-stratified screening is not associated with major psychological harms.

This study demonstrates no evidence that offering BC-Predict as part of the NHSBSP leads to increased psychological harms. This finding, supported by previous studies but strengthened by methodological improvements, suggests that risk-stratified screening can be implemented without significant psychological concerns. Further research could explore the unexpected increase in anxiety among low-risk individuals and investigate the potential for risk-stratified screening to improve broader understanding of screening complexities.

The primary limitations are the non-randomized design, leading to differences in deprivation levels between groups (though controlled for in analyses), and the low response rate (11.2%), which could introduce response bias. While the response rate was similar between groups, selection bias remains a possibility. The smaller than anticipated sample size (662 compared to the planned 1054) reduced power to detect small effects, however no significant differences between groups were observed. Further research should investigate the unexpected anxiety increase among women in the low-risk group.