Skip to main content
Download PDF

Impact of personalised risk predictions on breast cancer risk perceptions: insights from the BREATHE study

Journal of Translational Medicine volume 23, Article number: 517 (2025) Cite this article

Abstract

Objective

Biennial mammography screening is well-established for women aged 50 and above, but guidelines for younger women are less clear. Risk-based screening may provide women with key information to make informed decisions about their breast cancer risk and screening. This study examines how predicted breast cancer (BC) risk shapes women’s perception and confidence in risk prediction.

Methods

Women aged 35 to 59 years were recruited for a prospective multi-centre cohort and stratified into above-average, average, or below-average BC risk categories based on genetic and non-genetic risk factors. Perceived risk was assessed at enrolment and after participants were informed of their predicted risk. We used ordinal models to identify predictors of perceived risk and logistic regression to examine the relationship between changes in perceived risk and confidence in the risk prediction.

Results

At enrolment, 43% and 47% of 4112 participants perceived their BC risk pre-result as low or average, respectively. Thirty-five percent adjusted their perceived risk to align more closely with their predicted risk. Predictors of perceived risk post-result: perceived risk pre-result, predicted risk, ethnicity and having regular menstruation. Participants who underestimated their BC risk were nearly eight times more likely to have low confidence in the accuracy of their predicted risk (OR for underestimation vs. accurate perception: 7.94 [95% CI 5.60–11.28]). Predictors of perceived risk post-result: perceived risk pre-result, predicted risk, ethnicity and having regular menstruation. Confidence in risk prediction was lowest when women’s perceived risk pre-result was lower than their predicted risk (OR-2 vs 0 [95%CI] 5.06 [3.67 to 6.97]).

Conclusion

Many women underestimated their BC risk, and their initial perceptions were influenced by the knowledge of their predicted risk. Women who underestimated their risk had less confidence in their predicted risk scores.

Background

Globally, 13 to 21% of years of life lost from preventable cancer mortality is due to breast cancer [1]. A 10-percentage-point increase in uptake of mammography according to current screening guidelines averts 84 breast cancer deaths per 100,000 screened [2]. Mammography screening has the potential to reduce 33% mortality in women who participated [3], provided that the screening uptake rate reaches a minimum of 70%. However, screening also carries risks, such as false-positive results and the overdiagnosis of less aggressive lesions [4,5,6,7]. The evidence of benefits and risks of mammography screening vary with age and ethnicity, leading to differing recommendations across major guidelines regarding the optimal age to begin or cease mammography screening and screening interval for average-risk women [8]. For example, the American College of Obstetricians and Gynecologists (ACOG) recommends that women at average risk of breast cancer begin mammography screening at age 40 or by age 50 if not started earlier, with screening every 1 or 2 years based on a shared decision-making process between doctor and patient, evaluating the benefits and harms of screening [9]. Singapore’s Ministry of Health (MOH) guidelines recommend starting at age 50 and offer specific recommendations for different age groups (MOH Clinical Practice Guidelines 1/2010 [10]).

Many women under 50 years, who may be at higher risk based on individual factors, are not included in routine screening recommendations in Singapore. This age group often lacks tailored advice about their personal breast cancer risk, which can influence their decision-making and screening behaviour. Moreover, the variability in risk perception among different populations, such as by ethnicity and socio-economic status, suggests that a one-size-fits-all approach to screening may not be optimal. Risk-based screening, which considers individual risk profiles, could offer a more personalized and potentially more effective approach, especially for women at elevated risk who might otherwise not participate in regular screening. By addressing both the benefits and risks more precisely, risk-based screening could bridge the gap between current guidelines and the actual needs of diverse populations.

Perceived risk or susceptibility is considered an important determinant of precautionary health behaviours and is thus central to several theoretical models of health behaviour, such as the Health Belief Model and the Precaution Adoption Process Model [11]. Perceived risk refers to an individual’s subjective perception about their likelihood of experiencing personal harm [12]. Perceived risk plays a role in motivating health behaviours, with individuals who perceive their risk as low being less likely to engage in cancer screening [13]. External factors, such as a diagnosis in friends or family, can influence risk perception [14]. Additionally, women who are aware of risk factors associated with a higher likelihood of cancer or have higher perceived risk are more inclined to attend mammography screenings [15, 16]. Inaccurate risk perceptions often lead to inappropriate health behaviours, making it essential to understand these underlying mechanisms to develop effective interventions. In a meta-analysis of 42 studies by Katapodi et al., examining the role of perceived risk in predicting the adoption of health-protective behaviours, specifically breast cancer screening, it was found that women often have inaccurate perceptions of their breast cancer risk, showing an optimistic bias about their personal risk [17]. The study identifies several factors influencing perceived risk, such as family history, race/culture, and worry, with weaker influences from age and education. The analysis found a weak but significant association between perceived risk and mammography screening adherence. Given this context, it is important to investigate whether interventions such as objective risk assessments can correct misguided perceptions of breast cancer risk [18].

Changes in risk perception after an objective risk assessment can provide valuable insights into how effectively risk-based interventions in promoting beneficial health behaviours. The aim of this study is to assess the impact on perceived breast cancer risk when women are informed of their predicted risk and confidence levels related to breast cancer risk prediction.

Methods

Study population

The BREAst screening Tailored for HEr study (BREATHE) is a risk-based mammography screening where women aged 35 to 59 were recruited [19]. Eligible women must not have a histologically confirmed diagnosis of any cancer, no cognitive impairment, and were not pregnant during recruitment. Eligibility criteria was self-reported at recruitment and subsequently verified from medical records. Informed consent was obtained by trained study coordinators in either English, Chinese or Malay. The BREATHE protocol for recruitment and follow-ups is published [19]. Recruitment for the study began in October 2021 and continued until December 2023. Participants were recruited from three hospitals, two polyclinics, and one medical centre in Singapore. Of the 4,592 enrolled individuals, 74 individuals withdrew consent and 17 individuals were diagnosed with breast cancer within six months of enrolment (Supplementary Fig. 1). With an 8.6% (n = 389) loss-to-follow-up, the remaining 4112 individuals completed the first follow-up between February 2022 and June 2024. Individuals lost-to-follow-up were not different from those who completed follow-up in their perceived importance of breast cancer screening, perceived risk at enrolment or their predicted risk (Supplementary Table 1).

Perceived breast cancer risk

Participants’ perceived breast cancer risk was assessed at two separate occasions with a seven-point Likert scale question “What do you think is your chance of getting breast cancer?” (a score of 1 being the lowest and 7 being the highest). A seven-point Likert scale was chosen over a five-point scale was to reduce the potential that responders would choose the midpoint and increase dispersion that may result from Asian responders being less likely to respond with the extreme ends [20]. The first assessment was at enrolment before a breast cancer education questionnaire. The second assessment was at the first follow-up after the participants were informed of their predicted risk (above-average, average, below-average), derived using genetic and non-genetic information. Participants were only told their risk classification (above-average, average, or below-average). They were not made aware of the criteria that resulted in their risk classification. Details of the education questionnaire and risk prediction and classification can be found in the BREATHE protocol [19]. To begin, all participants were assigned as average risk. Those who met any of the following criteria were reassigned as above-average risk: (1) five-year absolute risk prediction by polygenic risk score (PRS) > 3%; (2) five-year absolute risk prediction by the Gail model (GAIL) > 1.3%; (3) high mammographic density (BIRAD 4); or (4) recall for additional mammography tests [19]. Participants aged 35 to 49 were classified as below-average risk if they did not meet the above-average risk criteria and had both PRS and GAIL to be < 1.3%.

Confidence in predicted breast cancer risk and acceptability of risk classification

We were also interested in the participants’ confidence in the risk prediction result as it can potentially influence the adoption of breast cancer screening recommendations and behaviour changes. Confidence was measured by “I am confident that my breast cancer risk classification in my report is reliable” (strongly agree, agree, neither agree nor disagree, disagree or strongly disagree).

To assess the acceptability of disease risk classification, we analysed responses to four questions using a scale from 1 (strongly agree) to 5 (strongly disagree). The questions were as follows: (1) “Learning about my breast cancer risk classification has affected my ability to go on with my day-to-day task”; (2) “Knowing my risk classification including my genetic risk for developing cancer is important”; (3) “Knowing my risk classification including my genetic risk for cancer will motivate me to attend cancer screening according to my risk level”; and (4) “I would like to know my genetic risk classification for other health conditions, if available.”

Demographic information and breast cancer screening behaviour

Socioeconomic status and family history of breast cancer may be associated with mammographic screening [21]. Self-reported demographic information was obtained from the baseline questionnaire at enrolment: attained age at enrolment (years), ethnicity (Chinese, Malay, Indian, other), marital status (married, widowed/separated/divorced, never married), employment status (currently, previously, never employed), highest academic attained (primary and below, secondary, post-secondary, and university and above), housing type (public housing by Singapore’s housing development board (HDB) 1–3 room, HDB 4-room, HDB 5-room, HDB executive, and private/other), and annual income (SGD, < 30,000, 30,001 to 72,000, 72,001 to 120,000, 120,001 to 175,000, > 175,000). Participants were asked if they have ever attended breast cancer screening (yes, no), and if they believe in the importance of breast cancer screening (strongly agree, agree, neither agree nor disagree, disagree, strongly disagree).

Breast cancer risk factors

Other variables were obtained from the structured questionnaire at enrolment, including those used in the Gail model such as age at menarche (categorised as < 12, 12 to 14, or ≥ 14 years), age at first live birth (classified as < 20, 20 to 24, 25 to 29, ≥ 30 years, or nulliparous), number of previous benign breast biopsies, presence of atypical hyperplasia on biopsy (yes or no), and the number of first-degree relatives with breast cancer (mother, sisters, or daughters). Five-year absolute risk based on the Gail model was computed using the methodology described in BREATHE protocol [19]. Additional breast cancer risk factors and lifestyle variables, which may influence their general health seeking behaviour, assessed included menstruation status (regular or not), number of children (1, 2, 3 +, or none), body mass index (BMI, kg/m2), physical activity based on the International Physical Activity Questionnaire (low, moderate/high), ever smoked regularly (ever/current, never), and ever drunk alcohol (yes at least more than once a month, no).

Statistical analysis

Differences in demographic variables, breast cancer risk factors, and perceived and predicted breast cancer risk between participants who completed follow-up and those loss-to-follow-up were assessed with univariate analysis (Chi-squared test for categorical variables and Kruskal Wallis test for continuous variables). The associations with age at enrolment of demographic variables, breast cancer risk factors, and perceived and predicted breast cancer risk were assessed with univariate analysis. Missing values were coded as a separate category during analysis, this category was not included in the univariate analysis. For breast cancer risk factors used in the Gail model, missing values were treated the same as the reference category as indicated by the manual.

We applied the ordinal model, using polr from the MASS library, to predict the participants’ perceived risk after receiving their risk prediction results. Demographic information, risk factor information known to the participants, and predicted risk (i.e. their risk classification) were tested univariately. Stepwise selection, using stepAIC, from the full model with all variables was used to select the best model. The full model includes all variables statistically significantly associated in univariate analysis. To obtain the most parsimonious model, the model with the lowest Bayesian information criterion (BIC) was selected.

Logistic regression was used to study the association between participant’s confidence with our predicted risk and participant’s characteristics. Stepwise selection was used to identify the combination of factors associated with participants’ lack of confidence (i.e. those that (strongly) disagree or were neutral).

All analysis was done using R version 4.2.2.

Results

A total of 4112 participants completed the follow-up, of which 78% were of Chinese ethnicity, 10% Malay, 7% Indian, and 5% others (Table 1). Most of our population (44%) were between 40 and 49 years old, where, under the 2024 national guidelines, mammogram screening depended on a doctor’s recommendation. This was followed by the 50 to 59 age group (41%), with 15% aged between 35 and 39 (Table 1). Ninety-five per cent of our participants attained above a primary level of education (around age 12 years). Ninety-six per cent of our participants believe breast cancer screening is important. Half of our participants aged 40 to 49 years had a mammogram in the past year and 78% of participants aged 50 to 59 years self-reported as routine screeners.

Table 1 Demographics and perceived and predicted breast cancer risk of BREATHE's participants by their perceived risk at enrolment (low, normal, high)
Full size table

Perceived risk before receiving risk prediction results

At enrolment, before the education survey, 17% (n = 697) of our participants rated their risk to be 1 (lowest risk on the Likert scale) and 1% (n = 21) rated their risk to be 7 (highest risk) (Supplementary Table 2). Forty-four per cent perceived their risk to be below average (Likert scale 1 to 3) and 47% perceived themselves at average risk (Likert scale of 4).

Supplementary Table 3 presents the lifestyle and breast cancer risk factors by perceived risk at enrolment. Perceived risk pre-result at enrolment was predicted by age and ethnicity (Supplementary Table 4). Younger age and being of Chinese ethnicity were associated with higher perceived risk.

Perceived risk after receiving risk prediction results

Based on the BREATHE’s criteria, 40% (n = 1650) of the participants were predicted to be at below-average risk, 29% (n = 1184) at average risk and 31% (n = 1276) at above-average risk (Table 1). Ninety-six per cent of the participants in the 35 to 39 age group and 59% in the 40 to 49 age group were classified as below-average risk.

After receiving their predicted risk results, 73% of participants who received a below-average risk prediction perceived themselves to be at below-average risk (Supplementary Table 2); 58% of whom received an average risk prediction perceived themselves to be at average risk; 29% of whom received an above-average risk prediction perceived themselves to be at above-average risk. Participants adjusted their perceived risk in the direction of their predicted risk (below average, average, above average) (Supplementary Fig. 2). Twenty-eight per cent were accurate in their risk perception pre- and post-result. Thirty-five per cent of participants adjusted their risk perception to align more closely with their predicted risk, while 28% continued to either overestimate or underestimate their risk. Eight per cent became more extreme in their perceived risk and < 1% overcompensated in their change in perceived risk. Notably, among the participants who received an above-average risk prediction, 20% adjusted their perceived risk to be above-average (Fig. 1).

Fig. 1
figure 1

Changes in perceived risk (enrolment and during follow-up) in women who were identified as above-average risk (facet label: below average, average, above average). Perceived risk was categorised as above-average (Likert scale 5 to 7) in blue and (below) average (Likert scale 1 to 4) in white

Full size image

Post-result’s perceived risk can be estimated by predicted risk, perceived risk at enrolment, ethnicity and menstruation status (Supplementary Table 5). The strongest predictors were predicted risk (odds ratio [ORaverage vs below average = 6.00; ORabove-average vs below average = 23.60) and perceived risk pre-result (ORaverage vs low = 3.29; ORhigh vs low = 8.57) (Supplementary Table 5). Figure 2 presents the prediction of hypothetical scenariosof a pre-menopausal Chinese woman, who perceived herself to be of average risk at enrolment and a predicted above-average risk. She is most likely to perceive herself to be at average risk (Probability = 0.59), quite likely to increase her perceived risk to high (Probability = 0.34), and unlikely to perceive her risk to be low (Probability = 0.07).

Fig. 2
figure 2

Hypothetical scenarios of a Chinese pre-menopausal woman. The ordinal model to predict the woman’s perceived risk after risk assessment (y-axis) included perceived risk at enrolment (panel) predicted risk (x-axis), ethnicity, and menopausal status

Full size image

Receiving an above-average predicted risk or having a higher perceived risk at enrolment increases the likelihood that the woman will view herself to be at a higher risk level (average or high) post-result. A significant interaction was observed between predicted risk and perceived risk at enrolment (Supplementary Table 5), indicating that participants’ perception of high-risk post-results tends to be reinforced by an above-average risk prediction or diminished by an average or below-average risk prediction.

The current age-based screening may have influenced women’s perception of risk within their age groups. We repeated the analysis within the age categories and observed similar associations of pre-results perceived risk and predicted risk with post-results perceived risk within each age category (Supplementary Table 6). However, ethnicity was not associated with post-results perceived risk among the younger participants aged 35 to 39 (Supplementary Table 6). In addition, the most parsimonious model did not include ethnicity for participants aged 40 to 49.

Confidence in predicted risk result

Above 94% of our participants reported that they understood their risk classification (94%, “Q3. I have a clear understanding of my breast cancer risk classification from my report.”) and study’s recommendation (95%, “Q4. I have a clear understanding of BREATHE study recommendations from my report.”) (Supplementary Table 7). The majority (82%) of our participants (strongly) agree with the statement “Q5. I am confident that my breast cancer risk classification in my report is reliable” (Supplementary Table 7). Proportion of participants who were neutral or (strongly) disagreed with the statement (Q5) was highest in oldest age group (21%) and lowest in youngest (12%); highest in Chinese (20%) and lowest in Malay (7%); and highest in those who were predicted “above-average” risk (30%) and lowest in “below-average” (12%); chi-square test p < 0.001. Similar trends were observed for the earlier two statements (Q3 and Q4).

Participants whose predicted risk closely matched their initial perceived risk were more likely to feel confident about risk prediction results (Fig. 3A). In contrast, participants who initially perceived themselves as low-risk but received an above-average risk prediction were the most likely to lack confidence in the risk prediction (OR−2vs 0 (95% confidence interval [CI]): 5.06 [3.67 to 6.97] adjusted for perceived risk at enrolment, perceived risk at first follow-up, and ethnicity (Supplementary Table 8). A larger effect of the difference in perceived risk and predicted risk (OR−2vs 0 [95%CI] 7.94 [5.60 to 11.28] and OR−1vs 0 [95%CI]: 1.87 [1.52 to 2.30]) on confidence was observed when perceived risk post-result was used (Supplementary Table 9, Fig. 3B).

Fig. 3
figure 3

Association of the difference in perceived and predicted risk with participant's confidence in their risk classification

Full size image

Overall, participants’ prior knowledge of breast cancer, as assessed by seven questions from the education survey (Q7 to 13), was not associated with their confidence in the reliability of the risk prediction result (Supplementary Table 10). The exception was among participants, aged 40 to 59, who agreed that a lack of family history does not eliminate the possibility of developing breast cancer, whereby they were more likely to be neutral or disagree with the reliability of the risk prediction.

Discussion

Clinical guidelines advocate for a tailored approach to mammography screening for specific age groups and recommend using decision aids to enhance discussions between patients and healthcare providers [22]. In line with this, we investigated how personalised breast cancer risk prediction results affect women’s perceptions of their breast cancer risk. Before receiving their predicted risk results, 43% of participants perceived their breast cancer risk as below average, with 47% considering themselves to be at average risk. However, many participants tended to underestimate their risk when compared to their predicted risk category. Overall, 28% maintained an over- or underestimation of their risk post-results, and 8% became more extreme in their perceptions.

Perceiving breast cancer risk has been positively associated with adherence to screening in Western countries [23]. Our study demonstrated that the strongest predictors of post-result perceived risk were the predicted risk and the initial perceived risk, therefore highlighting the possibility of encouraging mammogram screening uptake by informing one their predicted risk. It is however uncertain if changes in risk perception will be sufficient to translate into actual alterations in screening behaviour in Asian countries, whereby mammogram screening is often viewed more negatively in terms of efficacy and cost than in Western countries [24,25,26]. There is thus a need for more in-depth exploration of women’s perception towards mammogram to enable more alignment of health communication on mammogram towards their values.

Our study also found confidence in the risk prediction to be generally high. Participants whose predicted risk closely matched their initial perceptions were more likely to trust the results. Participants who were most sceptical of the risk prediction results were less likely to adjust their initial perceived risk to their predicted risk. The implication of inaccurate risk perception on subsequent health behaviour is unclear [27,28,29], although Katapodi et al. concluded with a cross-sectional study that underestimation of breast cancer risk did not predict optimum breast cancer screening practice [28]. Other studies suggest that individuals may react to risk information in ways that do not align with rational decision-making [30]. Some high-risk individuals may experience anxiety or fear that discourages them from engaging in screening, while others may adopt a fatalistic attitude. Further research is needed to explore the relationship between confidence in risk predictions, changes in perceived risk and subsequent health behaviour. Notably prior knowledge of breast cancer risk factors had little impact on participants’ confidence in the risk prediction, except among between 40 and 59 years who agreed that a lack of family history did not rule out the possibility of breast cancer. This warrants further research into factors influencing confidence on risk prediction for risk prediction to be used in health behaviour change.

The complexity of risk communication, which has an aspect of uncertainty that is hard to grasp, also poses a barrier. Decision aids, which are designed to enhance understanding and confidence in decision-making, have shown mixed outcomes in practice. Eden et al. found that while decision aids helped reduce uncertainty and increase confidence in decision-making, they did not significantly alter screening intentions [31]. Additionally, a randomised clinical trial of 204 women aged 39 to 48 showed that decision aids improved knowledge but did not significantly affect risk-based screening uptake or decrease decisional conflict [32]. These findings reflect the ongoing challenges in effectively integrating individual risk assessments into practical screening decisions. Even when risk is communicated effectively, trust in healthcare systems, physicians, and genetic testing itself plays a role in determining whether individuals follow screening recommendations. Cultural beliefs, previous healthcare experiences, and perceived accessibility of screening services will affect uptake rates.

Individual breast cancer risk assessment has the potential to direct women to screening decisions that are tailored to their specific risk profile [33]. This approach is particularly beneficial for those in age groups where shared decision-making with healthcare providers is recommended or as an alternative to the traditional age-based screening guidelines [34]. This is particularly relevant in Singapore, where a substantial proportion of breast cancers occur in women under 50 years who do not have clear recommendations to attend routine screening[35, 36], and thus may not perceive themselves as being at significant risk to participate in screening. We found that most participants were open to breast cancer risk classification beyond age-based guidelines and showed interest in learning about their personalised risk for other diseases. When participants received their predicted risk, they generally adjusted their initial perceived risk to align with it. To make breast cancer screening available to young women at elevated risk, and not overburden the healthcare system, a single time point assessment of breast cancer risk, with or without mammography, may be suitable for women to determine their optimal starting age for mammography screening [37, 38]. The discriminatory ability of breast cancer risk stratification is validated by multiple observational studies [39, 40]. Large, randomised control trials are ongoing to improve the performance of population-based screening [41]. Nonetheless, we found that women who received a predicted risk higher than their initial perception tended to be less confident in the risk assessment. It should be noted that there are concerns that risk stratification might result in many breast cancers being ‘missed’ if women deemed to be at low risk are not screened [42, 43]. As such, healthcare providers need to develop and be trained on effective risk communication strategies to ensure women understand their risk and are confident in the recommended actions. Policymakers also need to consider other health financing model to ensure equitable access to screening despite one’s personal risk. Exploring key areas for future research, such as the long-term effects of risk prediction on screening behaviours, would broaden the study’s contributions to the field and help inform policies that optimize screening strategies.

The limitations of this study include several key factors. Self-reported data on lifestyle and personal risk perceptions may introduce bias and affect the accuracy of the results. In particular, social desirability bias or recall bias may have influenced the responses provided by participants, leading to an overestimation or underestimation of certain behaviours or risk factors. The study population may not fully represent the general population. BREATHE participants were generally well-informed about breast cancer and recognized the importance of screening, which made them more proactive about attending screenings compared to the general population, as reported in a national survey [36]. This higher awareness is likely attributed to the recruitment settings at established mammography providers and wellness centres, where participants were generally from less-deprived backgrounds [44]. This specific recruitment approach may affect the generalizability of the study's findings to the broader population, as the study cohort may not fully represent individuals from different socio-economic backgrounds. While the Gail model and PRS are commonly used for predicting breast cancer risk, they do not capture all possible risk factors. Other factors not included in these models may contribute to breast cancer risk, potentially influencing the accuracy of predicted risk assessments. Perceived risk was measured on a 1 to 7 Likert scale, which differs from the three-category classification of predicted risk (above-average, average, below-average). The mapping of Likert scale scores to risk categories may not align perfectly with participants'understanding of risk, potentially affecting their perception and confidence. The study also did not evaluate whether participants' lay understanding of risk matched the numerical estimates used by experts, which could have led to mismatches between perceived and predicted risk and affected overall confidence in the risk assessment. Finally, the study did not account for external factors such as physician guidance or social influences on participants’ risk perceptions and confidence. These factors, such as personalized advice or societal norms, could have contributed to discrepancies between perceived and predicted risk, impacting decision-making and confidence levels.

Future research should focus on improving how we communicate breast cancer risk to women, making sure they fully understand their personal risk and feel confident in taking action. It would be helpful to explore how personalized tools, like interactive or visual aids, can help boost women’s confidence in their risk assessments. Research should also look into ways to encourage women at higher risk, particularly those with a family history, to participate in screening. Long-term studies would give us a better idea of how a woman’s risk perception changes over time and how this affects her decision to get screened. It is also important to consider the impact of cultural, social, and economic factors on screening, as this will help to create more accessible and effective programs for all women.

Conclusions

Participants tend to underestimate their breast cancer risk both before knowing their predicted risk result and after. The study revealed that participants' risk perceptions often aligned more closely with their predicted risk after receiving their results, indicating a tendency to adjust their perceived risk based on the predictions provided. Although most participants expressed confidence in the accuracy of their risk assessments, there was notable variability based on initial perceptions and the match between perceived risk post-result and predicted risk.

Data availability

The data generated by this study is owned by the providing institutions (NTFGH, NUH, AH, NUP and JMC). Data may be obtained with a reasonable request to the main Principal Investigator Mikael Hartman (mikael_hartman@nuhs.edu.sg). The data is not publicly available due to privacy and/or ethical restrictions. Legal agreements will need to be drawn up between data requesters and providers for access to the de-identified data. The proposed studies need to comply with Singapore’s laws and regulations regarding human biomedical research and clinical investigation including The Declaration of Helsinki, International Good Clinical Practice Guidelines, Good Clinical Practice guidelines by Singapore’s Health Science Authority and the Ministry of Health.

Abbreviations

BREATHE:

BREAst screening Tailored for HEr study

CI:

Confidence interval

OR:

Odds ratio

References

  1. Frick C, et al. Quantitative estimates of preventable and treatable deaths from 36 cancers worldwide: a population-based study. Lancet Glob Health. 2023;11(11):e1700–12.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Knudsen AB, et al. Estimated US cancer deaths prevented with increased use of lung, colorectal, breast, and cervical cancer screening. JAMA Netw Open. 2023;6(11):e2344698–e2344698.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Dibden A, Offman J, Duffy SW, Gabe R. Worldwide review and meta-analysis of cohort studies measuring the effect of mammography screening programmes on incidence-based breast cancer mortality. Cancers (Basel). 2020;12(4):976.

    Article  PubMed  Google Scholar 

  4. Myers ER, et al. Benefits and harms of breast cancer screening: a systematic review. JAMA. 2015;314(15):1615–34.

    Article  CAS  PubMed  Google Scholar 

  5. Moss SM, et al. Effect of mammographic screening from age 40 years on breast cancer mortality at 10 years’ follow-up: a randomised controlled trial. Lancet. 2006;368(9552):2053–60.

    Article  PubMed  Google Scholar 

  6. Farber R, et al. Systematic review and meta-analysis of prognostic characteristics for breast cancers in populations with digital vs film mammography indicate the transition may have increased both early detection and overdiagnosis. J Clin Epidemiol. 2024;171: 111339.

    Article  PubMed  Google Scholar 

  7. Welch HG, Prorok PC, O’Malley AJ, Kramer BS. Breast-cancer tumor size, overdiagnosis, and mammography screening effectiveness. N Engl J Med. 2016;375(15):1438–47.

    Article  PubMed  Google Scholar 

  8. Ren W, Chen M, Qiao Y, Zhao F. Global guidelines for breast cancer screening: a systematic review. Breast. 2022;64:85–99.

    Article  PubMed  PubMed Central  Google Scholar 

  9. American College of Obstetricians and Gynecologists. Practice bulletin number 179: breast cancer risk assessment and screening in average-risk women. Obstet Gynecol. 2017;130(1):e1–16.

    Article  Google Scholar 

  10. Ministry of Health S. Cancer screening. MOH clinical practice guidelines. 2010.

  11. Janz NK, Becker MH. The Health Belief Model: a decade later. Health Educ Q. 1984;11(1):1–47.

    Article  CAS  PubMed  Google Scholar 

  12. Weinstein ND, Klein WM. Resistance of personal risk perceptions to debiasing interventions. Health Psychol. 1995;14(2):132–40.

    Article  CAS  PubMed  Google Scholar 

  13. Lin YA, et al. Screening behaviors of high-risk individuals for lung cancer: a cross-sectional study. Asia Pac J Oncol Nurs. 2024;11(4): 100402.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Akiyama M, et al. Screening practices of cancer survivors and individuals whose family or friends had a cancer diagnoses—a nationally representative cross-sectional survey in Japan (INFORM Study 2020). J Cancer Surviv. 2023;17(3):663–76.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Ebrahimi F, Roohparvarzadeh N, Yadegarfar G. Predictors of breast cancer screening behavior in women aged 20–69 referred to public health centers: a cross-sectional study in Isfahan province. Int J Prev Med. 2022;13:115.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Petrocchi S, Ludolph R, Labrie NHM, Schulz P. Application of the theory of regulatory fit to promote adherence to evidence-based breast cancer screening recommendations: experimental versus longitudinal evidence. BMJ Open. 2020;10(11): e037748.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Katapodi MC, Lee KA, Facione NC, Dodd MJ. Predictors of perceived breast cancer risk and the relation between perceived risk and breast cancer screening: a meta-analytic review. Prev Med. 2004;38(4):388–402.

    Article  PubMed  Google Scholar 

  18. de Jonge ET, Vlasselaer J, Van de Putte G, Schobbens JC. The construct of breast cancer risk perception: need for a better risk communication? Facts Views Vis Obgyn. 2009;1(2):122–9.

    PubMed  PubMed Central  Google Scholar 

  19. Liu J, et al. BREAst screening tailored for HEr (BREATHE)—a study protocol on personalised risk-based breast cancer screening programme. PLoS ONE. 2022;17(3): e0265965.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Lee JW, Jones PS, Mineyama Y, Zhang XE. Cultural differences in responses to a Likert scale. Res Nurs Health. 2002;25(4):295–306.

    Article  PubMed  Google Scholar 

  21. Semprini J, Saulsberry L, Olopade OI. Socioeconomic and geographic differences in mammography trends following the 2009 USPSTF policy update. JAMA Netw Open. 2025;8(2): e2458141.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Medicines optimisation: the safe and effective use of medicines to enable the best possible outcomes. Manchester: National Institute for Health and Care Excellence (NICE). NICE Medicines and Prescribing Centre (UK). 2015. (NICE Guideline, No. 5.) 10, Patient decision aids used in consultations involving medicines]. https://www.ncbi.nlm.nih.gov/books/NBK355917/. Accessed 29 Oct 2024.

  23. Walker MJ, et al. Perceived risk and adherence to breast cancer screening guidelines among women with a familial history of breast cancer: a review of the literature. Breast. 2013;22(4):395–404.

    Article  PubMed  Google Scholar 

  24. Wegwarth O, et al. What do European women know about their female cancer risks and cancer screening? A cross-sectional online intervention survey in five European countries. BMJ Open. 2018;8(12): e023789.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Waller J, Osborne K, Wardle J. Enthusiasm for cancer screening in Great Britain: a general population survey. Br J Cancer. 2015;112(3):562–6.

    Article  CAS  PubMed  Google Scholar 

  26. Malhotra C, Bilger M, Liu J, Finkelstein E. Barriers to breast and cervical cancer screening in singapore: a mixed methods analysis. Asian Pac J Cancer Prev. 2016;17(8):3887–95.

    PubMed  Google Scholar 

  27. Persoskie A, Ferrer RA, Klein WM. Association of cancer worry and perceived risk with doctor avoidance: an analysis of information avoidance in a nationally representative US sample. J Behav Med. 2014;37(5):977–87.

    Article  PubMed  Google Scholar 

  28. Katapodi MC, Dodd MJ, Lee KA, Facione NC. Underestimation of breast cancer risk: influence on screening behavior. Oncol Nurs Forum. 2009;36(3):306–14.

    Article  PubMed  Google Scholar 

  29. Ferrer RA, et al. Unrealistic optimism is associated with subclinical atherosclerosis. Health Psychol. 2012;31(6):815–20.

    Article  PubMed  Google Scholar 

  30. Ferrer R, Klein WM. Risk perceptions and health behavior. Curr Opin Psychol. 2015;5:85–9.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Eden KB, et al. Mammography decision aid reduces decisional conflict for women in their forties considering screening. J Womens Health (Larchmt). 2015;24(12):1013–20.

    Article  PubMed  Google Scholar 

  32. Schapira MM, et al. The impact of a risk-based breast cancer screening decision aid on initiation of mammography among younger women: report of a randomized trial. MDM Policy Pract. 2019;4(1):2381468318812889.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Kerlikowske K, Bibbins-Domingo K. Toward risk-based breast cancer screening. Ann Intern Med. 2021;174(5):710–1.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Clift AK, et al. The current status of risk-stratified breast screening. Br J Cancer. 2022;126(4):533–50.

    Article  PubMed  Google Scholar 

  35. Jara-Lazaro AR, Thilagaratnam S, Tan PH. Breast cancer in Singapore: some perspectives. Breast Cancer. 2010;17(1):23–8.

    Article  PubMed  Google Scholar 

  36. Epidemiology and Disease Control Division and Policy, R.S.G. Ministry of health and health promotion board, Singapore. Singapore: National Population Health Survey; 2022.

    Google Scholar 

  37. Bolze A, et al. The potential of genetics in identifying women at lower risk of breast cancer. JAMA Oncol. 2024;10(2):236–9.

    Article  PubMed  Google Scholar 

  38. Bhatt R, et al. Estimation of age of onset and progression of breast cancer by absolute risk dependent on polygenic risk score and other risk factors. Cancer. 2024;130(9):1590–9.

    Article  PubMed  Google Scholar 

  39. Mabey B, et al. Validation of a clinical breast cancer risk assessment tool combining a polygenic score for all ancestries with traditional risk factors. Genet Med. 2024;26(7): 101128.

    Article  CAS  PubMed  Google Scholar 

  40. Darabi H, et al. Breast cancer risk prediction and individualised screening based on common genetic variation and breast density measurement. Breast Cancer Res. 2012;14(1):R25.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Allweis TM, Hermann N, Berenstein-Molho R, Guindy M. Personalized screening for breast cancer: rationale, present practices, and future directions. Ann Surg Oncol. 2021;28(8):4306–17.

    Article  PubMed  Google Scholar 

  42. Burnside ES, et al. Age-based versus risk-based mammography screening in women 40–49 years old: a cross-sectional study. Radiology. 2019;292(2):321–8.

    Article  PubMed  Google Scholar 

  43. Carbillon L, Bricou A, Sellier N. Challenges, benefits, and harms of risk-based screening mammography in women 40–49 years old. AJR Am J Roentgenol. 2016;206(2):W50.

    Article  PubMed  Google Scholar 

  44. Scobie H, et al. Optimising recruitment to a lung cancer screening trial: a comparison of general practitioner and community-based recruitment. J Med Screen. 2024;31(1):46–52.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We want to thank our dedicated research and administrative staff—Hui Ling Tan,Yi Ying Lim, Pooi Yee Wong, Ganga Devi D/O Chandrasegran, Nabilah Binte Supiee, Siti Zulyqha Binte Yazid, Alleza Joeay Balbanero Aquino, Pei Xuan Lim, Jolene Lu Yee Poh, Brenna Jing Jie Quah, Qian Ning Peh, Chun Mei Wang, Cara Wee Ying Wong, Kimiie Wei Lin Chia, Yi Lin Chen, Jinan May Loewen, Hui Min Lau, Varshaa D/O Saravanan, Vannevia Jedidiah Shi Tong Foo, Nurfilya Binte Hamdil, Hian Ching Ng, Yen Shing Yeoh, Amanda Tse Woon Ong, Jing Jing Hong and Siew Li Tan, for their contributions in the planning, preparation and execution of BREATHE. We would also want to thank Dr Chuan Chien Tan for assisting in the initial setup of the project, the doctors from Department of Obstetrics & Gynaecology from National University Hospital – Dr Judith Shan Lin Ong and Dr Susan Jane Sinclair Logan for allowing our team to conduct recruitment at Jade Clinic. We wish to acknowledge the contribution of Singapore Consortium of Cohort Studies-Multi-ethnic cohort (MEC) in providing information on women without breast cancer which is representative of the general population.

Funding

This study is funded by the JurongHealth Fund (reference number JHF-20-RE-003) and the National Research Foundation, Singapore, Precision Health Research Singapore under its Clinical Implementation Pilot (PRECISE CIP) Fund. M.H. is supported by the JurongHealth Fund, PRECISE CIP Fund, the Breast Cancer Prevention Programme under Saw Swee Hock School of Public Health Programme of Research Seed Funding (SSHSPH-Res-Prog-BCPP), Breast Cancer Screening Prevention Programme under Yong Loo Lin School of Medicine (NUHSRO/2020/121/BCSPP/LOA), the National University Cancer Institute Singapore (NCIS) Centre Grant Programme (CGAug16M005), and Asian Breast Cancer Research Fund. J.Li is supported by the Agency of Science, Technology and Research (A*STAR). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author information

Author notes

  1. Jingmei Li, Philip Tsau Choong Iau and Mikael Hartman have contributed equally to this work.

Authors and Affiliations

  1. Saw Swee Hock School of Public Health, National University of Singapore and National University Health System, Singapore, 117549, Singapore

    Peh Joo Ho, Serene Si Ning Goh, Jenny Liu, Nur Khaliesah Mohamed Riza & Mikael Hartman

  2. Agency for Science, Technology and Research (A*STAR), Genome Institute of Singapore (GIS), 60 Biopolis Street, Singapore, 138672, Singapore

    Peh Joo Ho & Jingmei Li

  3. Department of Surgery, Yong Loo Lin School of Medicine, National University of Singapore and National University Health System, Singapore, 119228, Singapore

    Peh Joo Ho, Su-Ann Goh, Serene Si Ning Goh, Jingmei Li & Mikael Hartman

  4. Department of Surgery, National University Hospital and National University Health System, Singapore, 119228, Singapore

    Serene Si Ning Goh, Ying Jia Chew, Sujith Wijerathne, Philip Tsau Choong Iau & Mikael Hartman

  5. Department of General Surgery, Ng Teng Fong General Hospital, 1 Jurong East st 21, Singapore, 609606, Singapore

    Jenny Liu, Ying Jia Chew, Han Boon Oh & Philip Tsau Choong Iau

  6. Department of General Surgery, Jurong Medical Centre, Singapore, 648346, Singapore

    Han Boon Oh

  7. Jurong Polyclinic, National University Polyclinics and National University Health System, Singapore, 609788, Singapore

    Chi Hui Chin & Desmond Luan Seng Ong

  8. Bukit Batok Polyclinic, National University Polyclinics and National University Health System, Singapore, 659164, Singapore

    Sing Cheer Kwek

  9. Choa Chu Kang Polyclinic, National University Polyclinics and National University Health System, Singapore, 688846, Singapore

    Zhi Peng Zhang

  10. Department of Diagnostic Imaging, National University Hospital and National University Health System, Singapore, 119074, Singapore

    Swee Tian Quek

  11. Department of Surgery, Alexandra Hospital and National University Health System, Singapore, 159964, Singapore

    Sujith Wijerathne

  12. National Cancer Centre Singapore, SingHealth, 30 Hospital Boulevard, Singapore, 168583, Singapore

    Jingmei Li

Authors
  1. Peh Joo Ho
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Su-Ann Goh
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Serene Si Ning Goh
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Jenny Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Ying Jia Chew
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Nur Khaliesah Mohamed Riza
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Han Boon Oh
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Chi Hui Chin
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Sing Cheer Kwek
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Zhi Peng Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  11. Desmond Luan Seng Ong
    View author publications

    You can also search for this author inPubMed Google Scholar

  12. Swee Tian Quek
    View author publications

    You can also search for this author inPubMed Google Scholar

  13. Sujith Wijerathne
    View author publications

    You can also search for this author inPubMed Google Scholar

  14. Jingmei Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  15. Philip Tsau Choong Iau
    View author publications

    You can also search for this author inPubMed Google Scholar

  16. Mikael Hartman
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Authors Mikael Hartman, Philip Tsau Choong Iau, Jingmei Li, and Peh Joo Ho contributed to the study conception and design. Material preparation and data collection were performed by Mikael Hartman, Philip Tsau Choong Iau, Jenny Liu, Nur Khaliesah Mohamed Riza, Ying Jia Chew, Su-Ann Goh, Han Boon Oh, Christopher Hang Liang Keh, Chi Hui Chin, Sing Cheer Kwek, Zhi Peng Zhang, Desmond Luan Seng Ong, Swee Tian Quek, and Sujith Wijerathne. Analysis was done by Peh Joo Ho. The first draft of the manuscript was written by Peh Joo Ho, Serene Si Ning Goh and Jingmei; all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Jingmei Li.

Ethics declarations

Ethics approval and consent to participate

The BREATHE protocol for recruitment and follow-ups is published. Informed consent was obtained by trained study coordinators in the participant’s preferred language (English, Chinese or Malay). Ethics approval was obtained from the National Healthcare Group Domain-Specific Review Board (ref no. ​​2020/01327, on 7 June 2021). Individuals who withdrew consent before 1 August 2024 (n = 74) were excluded from all analyses presented in this study.

Informed consent

Informed consent was obtained from all individual participants included in the study.

Competing interests

The authors have declared that no competing interests exist.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

12967_2025_6515_MOESM1_ESM.docx

Additional file 1. Supplementary Figure 1. Flowchart of individuals enrolled in BREATHE. Figure 2. Changes in perceived riskby predicted riskand age group.

12967_2025_6515_MOESM2_ESM.xlsx

Additional file 2. Supplementary Table 1. Comparing participants lost-to-follow-up with participants who completed follow-up. Table 2.  Proportion of participants at each level of perceived risk (Likert scale 1 to 7) at enrolment and during follow-up) by predicted risk (below average, average, above average) and age group. Table 3. Lifestyle and breast cancer risk factors of BREATHE's participants by their perceived risk at enrolment. Table 4. Predicting women's perceived breast cancer risk at enrolment, using ordinal models. Variables were selected based on the lowest Bayesian information criterion. Table 5. Predicting women's perceived breast cancer risk after receiving risk reports, using ordinal models. Variables were selected based on the lowest Bayesian information criterion (BIC). Table 6. Predicting women's perceived breast cancer risk after receiving risk reports by age categories, using ordinal models. Variables were selected based on the lowest Bayesian information criterion. Table 7. Perceived and predicted risk variables associated with 1) understanding of the risk classification, 2) understanding of the risk recommendation, and 3) Confidence in the risk classification. Table 8. Association of confidence in risk classification with the difference in perceived risk at enrolment and predicted risk, using logistic models. Variables were selected based on the lowest Bayesian information criterion. Table 9. Association of confidence in risk classification with the difference in perceived risk at follow-up and predicted risk, using logistic models. Variables were selected based on the lowest Bayesian information criterion. Table 10. Association of participants prior knowledge with their perception of the reliability of their predicted risk, by age.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ho, P.J., Goh, SA., Goh, S.S.N. et al. Impact of personalised risk predictions on breast cancer risk perceptions: insights from the BREATHE study. J Transl Med 23, 517 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12967-025-06515-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12967-025-06515-1

Keywords

Journal of Translational Medicine

ISSN: 1479-5876

Contact us