A Comparison of Perceived Lifetime Breast Cancer Risk to Calculated Lifetime Risk Using the Gail Risk Assessment Tool

Introduction

Breast cancer remains the leading cause of cancer death in women in the United States. It is important that clinicians proactively continue to discuss breast cancer screening with patients and, simultaneously, individualized discussions surrounding the patients' risk for breast cancer. A systematic review assessing risk perception found that individuals have difficulty in accurately quantifying their personal medical risk and have difficulty understanding how their risk can affect their future care. ¹ For example, few pregnant women with an increased risk for fetal congenital abnormalities used their objective risk in deciding whether or not to have an amniocentesis despite clear communication of the calculated risk. ²

Understanding the factors that influence risk perception, and subsequently affect personal health care decisions based on risk perception, is important. Ultimately, clinicians should consider these factors, in addition to actual calculated risk, to improve risk-based decision-making.

The accuracy of perceived breast cancer risk has been previously assessed and quantified. Prior studies show that many women inaccurately rate their personal breast cancer risk. Some reported overestimation of risk and others reporting underestimation of risk. ^1,3,4 Previously identified factors associated with women inaccurately predicting risk of breast cancer include younger age, characteristics of personal and family health history, variation among ethnicities, education level, and income. ^{3,5 –10}

Self-estimated risk may affect behavior and utilization of screening modalities as women with higher self-estimated risk tend to experience higher anxiety regarding breast cancer screening. ⁵ Heightened anxiety has been shown to lead to both overutilization and underutilization of breast cancer screening and risk reduction options. ^{1,3,6,7,11,12} Seitz et al. found that improving accuracy of self-estimated risk in women of average risk led to screening intentions in line with guideline recommendations. ⁴

In this study, we aimed to compare perceived breast cancer risk with calculated lifetime breast cancer risk based on the Gail breast cancer risk assessment tool. We hypothesized that the current study population may overestimate their breast cancer risk given that average-risk women in the United States tend to perceive their risk to be higher. ³ Understanding how women view their perceived breast cancer risk and how this correlates with actual breast cancer risk may provide insight to facilitate optimal shared decision-making conversations regarding breast cancer screening.

Methods

A descriptive cross-sectional study was conducted using a self-administered anonymous questionnaire. The Mayo Clinic Institutional Review Board (IRB) granted an exemption for the study. The survey was designed after review of the literature revealed no existing, validated survey tools to address our specific aim to assess perceived breast cancer risk. The survey was piloted with female office staff, nurses, and patients for clarity and comprehension and revised based on their suggestions. Once adjustments were made, printed surveys, which could be completed anonymously in approximately 10 minutes, were provided to patients before clinic visits and returned to the clinical staff.

The survey was offered to all women aged 40 years or older presenting for outpatient visits at Mayo Clinic in Scottsdale, Arizona (Community Internal Medicine, Women's Health Internal Medicine and Family Medicine) and Mayo Clinic in Rochester, Minnesota (Community Internal Medicine, Women's Health Clinic and Family Medicine) between July and August 2016. Women who chose not to participate were not documented. Women younger than 40 years of age, those with a personal history of breast cancer, those with a known Breast Cancer Gene (BRCA) mutation, and those who chose not to disclose this information were excluded.

In the questionnaire, self-reported demographics and pertinent medical history were collected from all participants. Participants were asked to rate their perceived personal lifetime risk of breast cancer as low, medium, high, or unsure. To objectively assess risk using the Gail risk calculator (https://bcrisktool.cancer.gov), data collected included age, age at menarche, age at first birth, breast biopsy history, family history of breast cancer, and race/ethnicity (questionnaire available in Supplementary Appendix SA1). Completed surveys were collected and data were entered in a single database. Gail risk scores were calculated for each participant based on survey responses. Low to average lifetime risk is defined as Gail lifetime risk <15%, intermediate lifetime risk is defined as 15%–20%, and high lifetime risk is defined as >20%. ¹³

Results

A total of 550 women were eligible for and completed surveys. Participants were predominantly non-Hispanic white, married, and highly educated (defined as more than or equal to a college degree). Those participating at Mayo Clinic Arizona were slightly older and more educated than those at Mayo Clinic in Minnesota. Most women described their health as good, very good, or excellent, and the majority had health insurance. Table 1 includes demographic information by location. Based on calculated lifetime risk, 86.6% were average risk (<15%), 7.7% were intermediate risk (15%–20%), and 5.6% were high risk (>20%). For the total group, women tended to overestimate their risk 4.3 (130/30) times as often as they underestimated risk (Table 2).

In the elevated risk setting, 18 of 27 women in the high lifetime risk category underestimated risk (66.7%) and 12 of 37 women in the intermediate lifetime risk category underestimated risk (32.4%). A total of 130 participants overestimated their risk; 125 of 414 (30.1%) women in the average lifetime risk category overestimated risk, and 5 of 37 (13.5%) women in the intermediate lifetime risk category overestimated risk.

Women who underestimated risk, among those who were calculated to have a high or intermediate lifetime risk, tended to be younger (<60 years), married, had at least one relative with a history of breast cancer or a history of an abnormal mammogram. Subanalysis of those who underestimated risk with a history of abnormal mammograms found that it was more common in those from 51 to 60 years or with a college degree. Those who underestimated risk and had a family history were also more commonly between age 51 and 60 years and with a college degree (Table 3). Overestimation of risk was more common in those with a college degree, those who were married, without a family history of breast cancer, and no history of an abnormal mammogram (Table 4).

Discussion

In a group of predominantly white, married and educated women, women overestimated risk 4.3 times more frequently than underestimating their risk. In our study, women who overestimated risk were more likely to be between age 40 and 60 years, married, did not have a family history of breast cancer, and did not have a history of an abnormal mammogram. These data are similar to a previous study, which showed that 44.1% of women older than age 50 with access to health care overestimated their risk of breast cancer, with 68.1% overestimating their risk of dying from breast cancer. ¹⁴ Existing literature supports that 10%–67% of average-risk women overestimate their breast cancer risk. ⁵

Another study showed that in a group of 11,000 women, only 14.3% of women accurately predicted their 5 years risk of breast cancer, with 44.2% overestimating risk, 13.6% underestimating risk, and 27.9% not answering. ¹⁵ Fehniger et al. found that women with average breast cancer risk correctly predicted their risk, whereas participants at higher risk underestimated their risk and only 18% accurately predicted their increased risk. ³ In our study, a majority of women who underestimated or overestimated risk had a college degree or higher. This is consistent with prior studies demonstrating the lack of a linear relationship between level of education and health literacy. ^5,16

There was a substantial number of participants in the high- and intermediate-risk categories who underestimated their risk. A majority of high- and intermediate-risk women who underestimated risk were between 40 and 60 years, married, had at least one family member with a history of breast cancer, or a history of an abnormal mammogram. This is an important group to target when discussing risk and the importance of enhanced screening in the clinic. Of interest, most of the women who underestimated risk had personal health characteristics that would presumably alert an individual of higher risk (i.e., family history of breast cancer and history of an abnormal mammogram).

These findings may guide clinicians to discuss risk with these patients specifically highlighting how their personal health characteristics may impact their calculated breast cancer risk. Utilizing risk calculators like the Breast Cancer Risk Assessment (aka Gail risk model) or Tyrer Cuzick IBIS model can help facilitate this discussion. ^17,18 Recent studies show that underestimation and overestimation are common in breast cancer risk perception; often the difference in perceived versus actual risk stems from personal health characteristics and/or various experiences with breast cancer. ^{19 –21} Abittan et al. found no association between the presence or absence of a prior personal breast cancer risk discussion with a physician and the accuracy of women's own breast cancer risk estimation. ¹⁵

Thus, more than a simple discussion of risk may be needed. Individual counseling incorporating patients' personal risk factors may be important, and other strategies to improve breast cancer risk assessment accuracy are needed.

Various motivators have been identified as reasons women may inaccurately estimate their risk of cancer. For example, whereas risk calculators help identify and quantify personal risk, variables included in the risk tools may not capture all risk factors perceived as adding to risk, ²² so some women may incorporate these ‘invisible’ factors (such as uncertainty of breast cancer risk and distrust of the health care system) in estimating their personal risk or they may weigh certain factors differently compared with the calculators. In addition, some women distrust clinical guidelines with concern that there are hidden motives for decreased screening, such as lack of funding for women's health issues. ²²

These conclusions may influence women's risk perception or screening behavior such that they desire more or less testing. Factors such as heightened anxiety or access to care may also negatively impact screening frequency. ²³ Discussing differences between a woman's perceived risk and her calculated Gail risk may uncover the reasoning behind discrepancies and can lead to enhanced shared decision making.

Because our study was limited in demographic heterogeneity, the results are most applicable to white women. Previous studies have revealed a discrepancy between perceived and predicted risk of breast cancer based on racial ethnicities. ^3,4,14,15 Haas et al. found that only 43% of women with increased risk accurately self-estimated risk, with black women underestimating their risk more commonly than white women. ⁵ In another study, a majority of higher risk women underestimated their risk in an ethnically diverse cohort of Latina, African American, non-Latina white, and Asian/Pacific Islander participants. ³ Similarly, underestimation was noted to be more common among black women compared with white women, whereas Asian Pacific Islander women were less likely to overestimate their risk compared with white women. ⁵

Conclusion

Among a population of predominantly white, educated, married women, participants overestimated risk at a much higher rate than underestimated risk. Despite this finding in the total group of surveyed women, a majority with high calculated lifetime risk of breast cancer tended to underestimate lifetime breast cancer risk. Most of these women had personal health history characteristics including abnormal mammography history, and family history of one or more first-degree relatives. Although educating women regarding personal breast cancer risk is salient, it is important to take the factors into account that may impact decision-making about breast cancer screening. Identifying and discussing personal health experiences that may impact accurate risk perception could potentially lead to better shared health care decision-making and ultimately, better health outcomes.