Few older women are aware of guidelines on when to stop breast cancer screening

INTRODUCTION

The benefit of breast cancer screening can be delayed for 10 years whereas harms, such as false positive tests and overdiagnosis, may occur in the short term.¹ The likelihood of benefit from screening declines while the likelihood of harm increases with older age and shorter life expectancy.¹

Some guidelines recommend breast cancer screening only in those with 10+ years life expectancy, some recommend screening until age 75 but comment on the importance of comorbidities and life expectancy, others recommend considering comorbidities and life expectancy without a specific stopping threshold (Supplement).^2–5

Despite these recommendations, 38% of US women 75+ and with <10-year life expectancy was screened for breast cancer.⁶ One potential contributor to over-screening is not being aware of the recommendations on screening cessation. A systematic review on women’s screening knowledge found only 3 studies that examined awareness about screening cessation; all were conducted in Australia where screening is age-based and found that only 13–25% of women ages 30–69 were aware of the screening cessation age.⁷ We aimed to assess US older women’s awareness of guideline recommendations around stopping breast cancer screening, identify characteristics associated with guideline awareness, and examine the association between guideline awareness and support for and personal intention of stopping screening.

METHODS

This project was part of a survey experiment with women 65+ on messaging about over-screening using a national survey panel (KnowledgePanel).⁸ We found that older women exposed to a message about rationales for stopping breast cancer screening were more likely to support screening cessation, especially when the message was repeated from multiple sources.⁹ Here we report on responses from the control group (n=751) who did not receive any messaging about breast cancer screening prior to the questions and who responded to an item about awareness of screening guidelines.

The survey first described a hypothetical 75-year-old woman with serious health problems and functional limitations as an exemplar of someone with <10-year life expectancy for whom stopping screening would be appropriate. It then assessed support for stopping screening in older women like the hypothetical patient and personal intentions to stop screening, both measured on 7-point scales with higher scores indicating stronger support and stronger intentions to stop screening. We asked: “Have you ever heard or read …for some women over age 65 who have a lot of health problems, medical guidelines recommend against regular mammograms?” We chose age 65 since 9% of women ages 65–69 and 14% of women ages 70–74 have significant comorbidities and <10-year life expectancy.⁹ We also asked about awareness of screening harms including overdiagnosis and false positives. Covariates included age, race, education, geographic region, prior screening, breast cancer risk factors, cancer worry, health literacy, health/functional status (which were used to estimate 10-year life expectancy), and medical maximizing preference.

We used logistic regression to assess characteristics associated with guideline awareness. Next, we assessed the association between guideline awareness, as a covariate, with support for stopping screening and intention for stopping screening, respectively. Given right-skewed distributions of scores on support and intention for stopping screening, we used generalized linear models (GLM) with log link and gamma distribution. Models included all covariates. Analyses used STATA 17.0.

RESULTS

Participants were mostly non-Hispanic Whites (81.6%) and 33.2% were 75 or older (Table 1). Only 13.3% were aware of guidelines on screening cessation. In contrast, 44.5% had heard of overdiagnosis, 65.0% had heard of false positives.

Table 1.

Participant characteristics (n=751)

Participant characteristics a	Number (%)
Age (n=751)
65 to <75	502 (66.8)
75+	249 (33.2)
Life expectancy b (n=738)
10+ years	589 (79.8)
<10 years	149 (20.2)
Race (n=751)
White, non-Hispanic	613 (81.6)
Black, non-Hispanic	70 (9.3)
Hispanic	35 (4.7)
Other c	33 (4.4)
Geographic region (n=751)
Northeast	127 (16.9)
Midwest	178 (23.7)
South	270 (36.0)
West	176 (23.4)
Self-reported mammogram within last 2 years (n=749)
No	151 (20.2)
Yes	598 (79.8)
Cancer worry (n=751)
Somewhat, a little, not at all worried	666 (88.7)
Moderately or extremely worried	85 (11.3)
Education (n=751)
High school or less	210 (28.0)
Some college or more	541 (72.0)
Health literacy d (n=749)
Normal	685 (91.5)
Low	64 (8.5)
5-year probability (%) of breast cancer, e mean (SD) (n=725)	2.4 (1.3)
Medical maximizing preference, higher score indicates more likely to take action,f mean (SD) (n=749)	3.4 (1.5)
Aware of guidelines on screening cessation (n=751)
No	651 (86.7)
Yes	100 (13.3)

^aSome totals do not add up to 751 due to incomplete or missing data.

^bLife expectancy was estimated using the Schonberg mortality index.¹⁰ Scores for participants ranged from 0 to 19. Scores ≥10 are associated with >50% chance of 10-year mortality. Thus, women who score ≥10 are estimated to have <10- year life expectancy.

^c“Other” included non-Hispanic 2+ races and non-Hispanic other race.

^dHealth literacy was assessed in a single validated question – “How confident are you filling out medical forms?” (Chew et al. Fam Med. 2004;35(8):588–594.) Responses of “not at all”, a little bit”, “somewhat” confident were categorized as low health literacy; responses of “quite a bit” and “extremely” confident were categorized as normal health literacy.

^eBreast cancer risk was estimated using the Gail Breast Cancer Risk Assessment Tool.

^fMedical maximizing-minimizing preference uses a 6-point validated measure (Scherer LD, et al. Med Decis Making. 2020;40(4):545–550), with scores 1–3= ”I strongly lean/ I lean/ I somewhat lean towards waiting and seeing” and scores 4–6 = “I somewhat lean/ I lean/ I strongly lean towards taking action”.

In logistic regression model, screening in the previous two years was associated with being less aware of guidelines on screening cessation (OR 0.49, 95% confidence interval [CI] 0.29–0.84, Table 2). In GLM models, awareness of guidelines, but not awareness of screening harms, was associated with higher support (26% higher) and higher intention (23% higher) for screening cessation (ratio of expected scores: 1.26 [95% CI 1.07–1.49] and 1.23 [95% CI 1.01–1.51], respectively).

Table 2.

Associations between participant characteristics, awareness of guidelines, and support and intention of screening cessation.

Participant characteristics	Logistic regression; Dependent variable = awareness of guideline a	GLM model; Dependent variable = support of screening cessation b	GLM model; Dependent variable = intention of screening cessation b
	Odds Ratio (95% CI)	Ratio of expected scores (95% CI) b
Age
65 to <75	Ref	Ref	Ref
75+	0.71 (0.42,1.21)	1.00 (0.88, 1.15)	1.30 (1.10, 1.52)
Life expectancy
10+ years	Ref	Ref	Ref
<10 years	1.38 (0.75, 2.53)	1.02 (0.87, 1.19)	0.93 (0.70, 1.34)
Race
White, non-Hispanic	Ref	Ref	Ref
Black, non-Hispanic	1.14 (0.53, 2.48)	0.78 (0.65, 0.95)	1.04 (0.81, 1.32)
Hispanic	1.51 (0.61, 3.73)	0.98 (0.75, 1.28)	1.24 (0.90, 1.71)
Other	0.61 (0.18, 2.08)	0.88 (0.67, 1.16)	0.97 (0.70, 1.34)
Geographic region
Northeast	Ref	Ref	Ref
Midwest	0.75 (0.35, 1.59)	1.06 (0.88, 1.26)	1.05 (0.85, 1.30)
South	1.13 (0.58, 2.21)	1.04 (0.88, 1.22)	1.17 (0.95, 1.43)
West	1.60 (0.80, 3.20)	1.01 (0.65, 1.21)	1.06 (0.85, 1.32)
Mammogram within last 2 years
No	Ref	Ref	Ref
Yes	0.49 (0.29, 0.84)	0.67 (0.58, 0.78)	0.40 (0.33, 0.48)
Cancer worry
Somewhat, a little, not at all worried	Ref	Ref	Ref
Moderately or extremely worried	1.47 (0.76, 2.87)	0.77 (0.64, 0.93)	0.74 (0.59, 0.92)
Education
High school or less	Ref	Ref	Ref
Some college or more	0.64 (0.40, 1.04)	0.98 (0.86, 1.11)	1.14 (0.97, 1.33)
Health literacy
Normal	Ref	Ref	Ref
Low	0.58 (0.22, 1.54)	1.21 (0.97, 1.50)	1.36 (1.05, 1.76)
5-year probability of breast cancer, per one % increase in probability	1.06 (0.89,1.25)	1.02 (0.98, 1.07)	1.03 (0.97, 1.09)
Medical maximizing preference, per one point increase on 6-point scale	1.00 (0.86, 1.17)	0.92 (0.88, 0.95)	(0.92 (0.88, 0.97)
Aware of guidelines on screening cessation	/
No		Ref	Ref
Yes		1.26 (1.07, 1.49)	1.23 (1.01, 1.51)

^aLogistic regression model included all covariates in the table as independent variables and awareness of guidelines on screening cessation as dependent variable.

^bGeneral linear models included all covariates in the table PLUS awareness of guidelines on screening cessation as independent variables. Dependent variables were support for screening cessation in older women like the hypothetical patient, and intention to stop screening for oneself, respectively. The models used log link and gamma distribution to account for the right-skewed distribution in both of the dependent variables. The coefficients from the models were exponentiated to generate ratios of expected scores where a ratio of 1.20 indicates a 20% higher outcome score compared to the reference group and a ratio of 0.80 indicates a 20% lower outcome score compared to the reference group.

DISCUSSION

In a national survey of US older women, we found that 87% were not aware that guidelines recommend against regular breast cancer screening in older women with significant comorbidities, a prevalence similar to prior Australian studies.⁷ Study limitations include sampling and non-response bias. Compared to national census data, study participants were younger, more likely to be White, and had higher education; these differences may affect the results’ generalizability. We used a single, unvalidated question to assess guideline awareness, although we did pilot test the survey instrument with older women to enhance face validity.

We found that, even after controlling for demographic, health, and attitudinal factors, awareness of guidelines on screening cessation was associated with higher support and higher personal intentions of screening cessation. Our results suggest that raising awareness among older women about timing of screening cessation (i.e., through messaging and patient decision-aids) is critically needed and is a potentially promising strategy for reduce over-screening.