Supporting Informed Decisions about Breast Cancer Screening in Communities with Known Environmental Contamination: A Pre-Post Study

Abstract

Purpose:

This study tested the impact of a breast cancer screening decision aid (DA) that was adapted to include environmental risk information.

Methods:

This was a pre-post study in communities with known per- and poly-fluoroalkyl substances (PFAS) contamination. Participants were 100 women ages 39–49, with no history of breast cancer and no mammogram in the past year. Participants completed a baseline survey, used the DA, then completed a final survey. Baseline surveys measured demographics, PFAS exposure and PFAS-related risk perceptions. Paired pre-post survey measures included breast cancer risk perceptions, screening knowledge, decisional conflict, and screening intentions.

Results:

Mean age of participants was 43 years; most non-Hispanic white (97%); 43% resided in micropolitan, small town, or rural areas. Potential PFAS exposure included private well use for 35%; 19% had tested their water for PFAS contamination; 54% used a home water filter. While most (62%) did not experience PFAS-related stress, 27% felt stress about family well-being, and 25% often/constantly worried about the health effects of PFAS in the prior month. Pre-DA, 29% felt likely that they would develop breast cancer in the next 10 years, versus 19% post-DA. Median knowledge scores increased from 8 to 10 out of 10 before and after DA use. Intentions to get a mammogram in the next year slightly decreased from 85% to 78%.

Conclusion:

Use of a DA incorporating information about environmental risk information reduced perceptions of breast cancer risk and improved knowledge and decisional conflict about breast cancer screening among women living in PFAS-contaminated communities.

Introduction

Per- and polyfluoroalkyl substances (PFAS) are a class of highly persistent and ubiquitous pollutants found in a variety of oil- and water-resistant consumer products, some formulations of which are known carcinogens [1–4]. PFAS are increasingly common drinking water and food contaminants in the U.S. and are detected in the blood of nearly the entire American general population [5–7]. PFAS are colloquially known as “forever chemicals,” as they are resistant to degradation and persist in the air, water, and soil indefinitely, and in the human body with half-lives of 4–7 years [8–11].

A 2022 report identified PFAS contamination in more than 2,858 communities across all 50 states, but risk and mitigation are unevenly distributed geographically [5, 12]. PFAS contamination is an especially salient concern for northern New England, where much of the landscape is rural. Rural communities are disproportionately exposed to environmental contaminants like PFAS, particularly through drinking water, as rural residing populations often rely on private water systems (e.g., wells, springs). Private water systems are used by about half of the New Hampshire [13] and Maine populations [14], and about 40% of Vermont residents [15]. These systems are not regulated by the U.S. Environmental Protection Agency (EPA), nor are they routinely monitored for potentially toxic substances, including PFAS [12, 13]. PFAS “hot spots” have been identified in rural regions in these states, including licensed sludge and septage sites in rural Maine [16], and industrial water contamination in Bennington, Vermont [5, 17] and Merrimack, NH [5].

Research continues to emerge linking PFAS exposure to numerous adverse health outcomes, including thyroid disease, liver damage, fertility and fetal development, and multiple types of cancer [2, 18]. A growing body of research has identified associations between PFAS and breast cancer risk. Two specific PFAS (perfluorooctanoate (PFOA) and perfluoro octane sulfonate (PFOS), are known endocrine disrupters and classified as Group 1 and 2a carcinogens, respectively [4, 18]. Studies have demonstrated higher rates of breast cancer in women with higher PFAS blood concentrations, particularly estrogen receptor positive (ER+) breast cancers in women under 50 years of age [19, 20]. A 2024 meta-analysis found a heterogeneity of effects across epidemiological studies, although associations by tumor hormonal characteristics and menopausal status were consistent [21].

Current clinical guidance advises that individuals with moderate to high PFAS blood concentrations should screen for breast cancer according to age and other risk factors as recommended by the US Preventive Services Task Force (USPSTF) [3]. Meanwhile, USPSTF guidelines have seen repeated changes in age of breast cancer screening initiation and interval in the past 15 years, complicating decision-making for patients and clinicians as there is no single evidence-based, optimal screening schedule for women in their 40s [22, 23]. For these women, guidelines call for the use of risk assessment to inform screening and prevention activities. Available risk tools use personal, familial, genetic, and clinical factors to predict cancer risk, but lack inclusion of environmental risk factors [24–28].

Our team previously designed and tested one such tool to support decisions about screening initiation and interval among women in their 40s, called MyMammogram [29, 30]. Using focus groups and cognitive interviewing with 18 women, we adapted content to address concerns about environmental risks (Figure 1). In this pilot study, we aimed to test the impact of the adapted version of MyMammogram on decisional conflict, breast cancer risk perceptions, screening knowledge, and screening intentions among women living in PFAS-affected communities. We hypothesized that knowledge would increase and decisional conflict, risk perceptions, and screening intentions would decrease after using MyMammogram. Our secondary objective was to establish acceptability of the modified version of MyMammogram as an intervention for our target population.

Figure 1.

MyMammogram Environmental Risk Information

Materials & Methods

Overview/Design

This was a pre-post pilot study conducted from May to December of 2024. Upon enrollment, participants were asked to complete a baseline survey before viewing the decision aid. Immediately after using the decision aid, participants were redirected to a post-intervention survey. All study activities were completed online and participation was estimated to last about 45 minutes.

Setting & Sampling

This study was approved by the Dartmouth College Institutional Review Board (STUDY00032987). We aimed to enroll 100 women ages 39–49, with no personal history of breast cancer, ductal carcinoma in situ, or lobular carcinoma in situ, who were living in counties with known PFAS contamination across New Hampshire, Vermont, Maine and New York. We employed multiple community recruitment methods, including local advertising, online community listservs, and social media posting. We partnered with the University of New Hampshire Survey Center to facilitate outreach via the Green Mountain State Poll in Vermont and the Granite State Poll in New Hampshire. We also partnered with Merrimack Citizens for Clean Water, a nonprofit organization dedicated to improving water quality in response to PFAS contamination in the Merrimack, New Hampshire region, who assisted in advertising our study within their community networks.

Prospective participants were directed to complete an intake survey to assess eligibility. We installed reCAPTCHA bot protection measures in all surveys and manually reviewed each response to reduce the risk of fraudulent participation. Eligible participants were emailed by the research team with a customized link to the study materials and detailed instructions for participation. The beginning of each survey contained a study information sheet which explained the purpose, risks, and benefits of the survey. All participants indicated their agreement to participate prior to completing any study activities.

MyMammogram Intervention

MyMammogram is an online breast cancer screening decision aid designed for use by women in their 40s who are deciding about screening initiation and interval (annual, biennial). It was developed in partnership with women with limited health literacy and primary care clinicians. The decision aid is written at a Grade 6.3 Flesh Kincaid grade level [31] and has a Gunning Fog index of 8.08 [32]. It includes interactive personal risk assessment using the National Cancer Institute’s Breast Cancer Risk Assessment Tool and displays an estimate of an individual’s 5-year risk of developing breast cancer. The decision aid provides information about the process of receiving a mammogram, shares the risks and benefits of screening options including outcome probabilities, and elicits and summarizes mammography preferences. It was designed in accordance with criteria set forth by the International Patient Decision Aid Standards.

Data Collection

Participants completed a baseline survey assessing self-reported PFAS exposure, PFAS-related risk perceptions, and demographic characteristics. Following the baseline survey, participants were provided a link to MyMammogram. Participants were instructed to use the decision aid and upon completion, were redirected back to the post-intervention survey. Paired pre-post survey measures included decisional conflict, breast cancer risk perceptions, screening knowledge, and breast cancer screening intentions, each described below. The post survey also included the secondary outcome of acceptability of MyMammogram. After completing all study activities, participants were given the option to enter a raffle for one of five $200 gift cards. Five participants were randomly selected. The raffle was conducted after the study was closed to recruitment.

Measures

Demographics included age, race, ethnicity, marital status, state of residency, rurality, mammogram history, and health literacy. Health literacy was assessed using the three-item Brief Health Literacy Screener (Chew items) [33] where items are scored along a five-point Likert scale, and scores greater than or equal to 9 indicate adequate health literacy (range: 3–15).

PFAS-related risk perceptions were assessed using two items adapted from the Environmental Health Risk Perception Scale [34]: Responses to the first item, “How often in the past month have you had thoughts or fears about the health effects of PFAS?”, were dichotomized into low risk perception (Never, Rarely) and high risk perception (Often, Constantly). Responses to the second item, “How worried are you that you or your family will experience health problems as a result of PFAS exposure?”, were similarly dichotomized into low risk perception (Not at all, Somewhat) and high risk perception (Moderately, Enough to worry a great deal). These items were collapsed into a single measure of PFAS risk perception. If a participant were classified as high risk perception for either or both questions, they were grouped into high PFAS risk perception overall. PFAS stress was assessed using the following question: “If you feel that you are struggling now because of living in a PFAS-affected community, please specify what has caused you the most stress in the last month.” Response options included finances, physical health, mental health, wellbeing of family, housing security, job security, or no stress.

The primary outcome of interest was decisional conflict, which is a measure of informed decision-making. Decisional conflict was measured using the four-item SURE scale[35] where individuals rate their confidence about the best choice for them, knowledge of benefits and risks of each option, clarity about which benefits and risk matter most to them and having sufficient support and advice to choose. Each item was scored (1) for “Yes” or (0) for “No.” The final outcome is reported dichotomously as the top box score (all items scored as Yes vs. not).

Secondary decisional outcomes included breast cancer risk perceptions, screening knowledge, and screening intentions. Breast cancer risk perceptions were assessed in relative terms using the following question: ‘How do you think your chances of developing breast cancer in the next ten years compares to the average person of your age and sex?,’ with responses scored along a seven-point Likert scale ranging from Much lower (1) to Much higher (7). For interpretation, we collapsed responses into those who perceived their risk as above average (scores 5–7), to those that considered their risk average or below average (scores 1–4). Breast cancer screening knowledge was measured using ten true/false items [36]. Each item was scored (1) if correct or (0) if incorrect and summed, for a total knowledge score ranging from 0–10. Breast cancer screening intentions were assessed using one item, ‘How likely are you to get a mammogram in the next year?,’ with responses scored along a seven-point Likert scale ranging from Not at all likely (1) to Extremely likely (7). We considered those who were somewhat to extremely likely (scores 5–7) to have intended to screen in the next year, versus those who were neutral or unlikely (scores 1–4).

Acceptability of MyMammogram was assessed using the Acceptability E-Scale [37] and the Acceptability of Intervention Measure (AIM) [38]. The Acceptability E-Scale consists of six items each rated on a five-point Likert scale. We added an additional item to more discretely assess different learning objectives within MyMammogram, for a total of seven items. The AIM consists of four items on a five-point Likert Scale, for a total summary score ranging from 4–20. For both measures, higher scores indicate greater acceptability.

Data Analysis

We compiled descriptive statistics for demographic measures (Table 1) and environmental exposures (Table 2). We assessed pre-post differences in proportions for the categorical outcomes of decisional conflict, breast cancer risk perceptions and screening intentions using McNemar’s tests (Table 3). We report median knowledge scores using the Wilcoxon signed rank test to assess pre-post differences. We report the mean score for each item on the Acceptability E-Scale and the mean summary score for the AIM.

Table 1.

Demographics

Characteristic (N=100)	Percent
Age (Mean, Range)	43 (39–49)
Race
White	97%
Two or more races	2%
Asian	1%
Ethnicity
Not of Hispanic, Latino or Spanish origin	92%
Hispanic/Latino/Spanish origin	2%
Missing	6%
Marital status
Married or Living together	81%
Never married	10%
Divorced or Separated	8%
Prefer not to answer	1%
State of Residency
New Hampshire	66%
Vermont	19%
Maine	12%
New York	3%
Rurality (RUCA)
Rural area	11%
Small town	7%
Micropolitan (Large town or city)	25%
Metropolitan (Urban)	57%
Prior Mammogram	76%
Health Literacy
Adequate	95%
Inadequate	5%

Table 2.

Environmental Exposures

Characteristic (N=100)	Percent
PFAS Exposure
Home water source
Town/city public water	62%
Private well	35%
Unsure	3%
Do you currently use a filter for your drinking water at home?
Yes	54%
No	46%
Has your home water been tested for PFAS?
Yes	19%
No	43%
Unsure	38%
Have you ever had your blood tested for PFAS?
Yes	7%
No	87%
Unsure	7%
PFAS Risk Perception
How often in the past month have you had thoughts or fears about the health effects of PFAS?
Constantly	2%
Often	23%
Rarely	44%
Never	30%
Missing	1%
How worried are you that you or your family will experience health problems as a result of PFAS exposure?
Enough to worry a great deal	4%
Moderately	25%
Somewhat	53%
Not at all	17%
Missing	1%
PFAS Stress
If you feel that you are struggling now because of living in a PFAS-impacted community, what has caused you the most stress in the last month?
I have no felt stress related to PFAS	62%
Stress over wellbeing of my family	27%
Stress over physical health	6%
Stress over finances	5%

Table 3.

Pre-DA and post-DA results for categorical outcomes

Outcome (N=100)	Pre-DA Percent	Post-DA Percent	p-value*
Decisional Conflict
Sure	52%	85%	p <0.001
Unsure	46%	13%
Missing	2%	2%
Breast Cancer Risk Perception
Higher risk	29%	19%	p = 0.044
Lower to same risk	69%	79%
Missing	2%	2%
Breast Cancer Screening Intentions
Likely	85%	78%	p = 0.046
Neutral/unlikely	14%	21%
Missing	1%	1%

^*McNemar’s Chi-squared test with continuity correction

For dichotomous outcomes, we ran logistic regressions including predictors of the pre-DA measures of the outcome, and covariates known to be predictive of screening behaviors and/or PFAS exposures and associated risk (age, prior mammogram (y/n), state, rurality, and PFAS-related risk perceptions). Because rurality, state, and age did not demonstrate significant associations with any outcome and the inclusion of the state variable created some exclusions due to small cell sizes, they were removed from the final models. For screening knowledge, we modeled the change in knowledge scores as the dependent variable with covariates listed above as predictors using multivariable linear regression.

Results

Participant demographics are shown in Table 1. We enrolled 100 women ages 39–49, with a mean age of 43 years. Participants were majority non-Hispanic White (97%) and married or living together (81%), and 41% had an advanced degree beyond a 4-year Bachelor’s. The majority (95%) were found to have adequate health literacy. Most (66%) lived in New Hampshire, and 43% resided in a micropolitan, small town or rural area. The majority (76%) had at least one prior mammogram in their lifetime. Among those 70 participants who were enrolled through community sources, the median time to complete study activities was 20.7 minutes (range: 9.2 minutes to 6.9 days).

Environmental exposure and risk perceptions are reported in Table 2. In terms of potential PFAS exposure, 35% reported using private well water in their homes and 54% used a home water filter. Almost one-fifth (19%) had tested their water for PFAS contamination, and 7% had tested their blood for PFAS. While most (62%) did not experience PFAS-related stress, 27% reported feeling stress about the well-being of their family. Similarly, 25% constantly or often worried about the health effects of PFAS in the prior month. Nearly one-third (29%) of participants reported feeling moderately or a great deal worried about experiencing health problems as a result of PFAS exposure.

The primary outcome of decisional conflict demonstrated significant improvements, with the proportion of respondents who were “sure” increasing from 52% pre-DA to 85% post-DA (p<0.001) (Table 3). Breast cancer risk perceptions were reduced: the proportion of participants who perceived themselves to be at above average risk of developing breast cancer decreased from 29% pre-DA to 19% post-DA (p<0.05). Median knowledge scores about breast cancer screening also improved, from 8 out of 10 pre-DA to 10 out of 10 post-DA (p<0.001) (Figure 2). Screening intentions declined slightly: 85% reported being somewhat/very/extremely likely to get a mammogram in the next year pre-DA, vs. 78% post-DA (p<0.05).

Figure 2.

Distribution of Knowledge Scores

The results of the linear and logistic regression models are presented in Table 4. Age, rurality, and state of residence did not appear to be associated with decisional conflict, breast cancer risk perception, screening intentions or differences in knowledge score. Respondents were more likely to be “sure” of their screening decision if they had a prior mammogram. Those with average or below average breast cancer risk perception pre-DA were significantly more likely to report the same result post-DA. Respondents were more likely to perceive they were at above average risk post-DA had not had a prior mammogram, but this predictor was not statistically significant. Respondents that were likely to screen for breast cancer in the next year pre-DA and had a prior mammogram were significantly likely to also report intent to screen post-DA. Those with low risk perceptions of PFAS were more likely to be “sure” of their screening decision and report average or below average perceptions of breast cancer risk.

Table 4.

Multivariable Logistic and Linear Regression Model Results for Post-DA Outcomes*

	Decisional Conflict^	Breast Cancer Risk Perception	Screening Intentions	Knowledge Score Difference
Characteristic	OR (95% CI)	OR (95% CI)	OR (95% CI)	Beta (95% CI)
Prior mammogram
No	Ref.	Ref.	Ref.	Ref.
Yes	1.75 (0.45, 6.21)	0.70 (0.15, 2.82)	14.6 (2.91, 89.4)	−0.05 (−0.56, 0.46)
Pre-DA PFAS Risk Perception +
High	Ref.	Ref.	Ref.	Ref.
Low	3.40 (1.00, 12.5)	2.89 (0.88, 9.95)	0.32 (0.03, 1.98)	0.04 (−0.40, 0.48)
Pre-DA Breast Cancer Risk
Higher risk		Ref.
Lower to same risk		10.1 (3.24, 36.4)
Pre-DA Screening Intentions
Neutral/unlikely			Ref.
Likely			99.0 (14.1, 2,082)

Abbreviations: CI = Confidence Interval, OR = Odds Ratio, DA = Decision Aid

Statistically significant differences with p< 0.05 are denoted in bold type.

Decisional Conflict, Breast Cancer Risk Perceptions, and Screening Modeled as binary outcomes. Knowledge outcome represents continuous score difference (post-pre).

⁺Pre-DA PFAS risk perception was classified high if respondent indicated worry on at least one of two PFAS risk questions

Acceptability of MyMammogram was high across all measures. Participants found MyMammogram easy to use (4.7), understandable (4.7), helpful for learning about both breast cancer risk factors (4.0) and breast cancer screening options (4.0), acceptable in terms of time needed for use (4.5), and were overall satisfied with the decision aid (4.3). Participants were more neutral about whether they enjoyed using MyMammogram (3.7). The mean AIM score was also highly acceptable at 16.6 out of 20.

Discussion

Our study demonstrated that a significant minority of women living in PFAS-affected communities were concerned about PFAS exposure and related health effects. Though many participants expressed some degree of worry or stress related to environmental exposures, far fewer reported testing of their drinking water or blood for the presence of PFAS. These findings imply that participants were aware of environmental risk factors, but potentially less knowledgeable or able to access tools to identify exposure and mitigate risk. There is a potential need for improved communication regarding the availability of clinical and community resources for residents of PFAS-affected communities, as well as further guidance about risk management for adverse health outcomes. Public health resources designed to educate the public about PFAS exposure and related health effects have been developed by experts in the field. Future work should examine how these resources can be used to better support these communities [39, 40].

Pre-post survey results proved that MyMammogram was able to improve decisional outcomes among early screening-age women living in areas with environmental exposures. All decisional factors improved, including decisional conflict, breast cancer risk perceptions, and knowledge about mammograms. These results are analogous to those found in previous tests of MyMammogram: a prior study among U.S. women aged 40–54 with no breast cancer history found significant improvements in mean decisional conflict and knowledge scores [29]. Improvements in decisional conflict and knowledge scores were comparable to those observed in similar studies of other web-based breast cancer screening decision aids as well [41, 42]. We also observed a reduction in relative breast cancer risk perception. The proportion of participants who considered their 10-year risk of developing breast cancer to be above average declined from 29% to 19%. This is still high compared to the average population 10-year risk, which is 1.55% for women at age 30 and 2.40% for women at age 40 [43], but significantly reduced after DA use.

Intentions to screen for breast cancer in the next year decreased slightly after using the DA. MyMammogram is designed to improve choice clarity and enable users to decide about breast cancer screening unique to their personal risk factors and preferences. Small changes in screening intentions at the overall group level are likely reflective of individual differences and preferences related to decision-making. These results align with existing literature on breast cancer screening decision aids, which has reported variable results, including improvements, reductions and no changes in intentions to screen, at various levels of statistical significance [41, 42, 44, 45]. Taken in tandem with reductions in breast cancer risk perceptions, it is possible that reductions in screening intentions are a result of lower risk perceptions, aligning intentions and risks more closely. However, this study is not powered to test the mechanisms supporting changes in screening intentions, which will be the focus of future work.

MyMammogram was highly acceptable to this study population. Though we did not measure acceptability of the adapted environmental risk content directly, most participants found MyMammogram helpful for understanding breast cancer risk factors, which include PFAS exposure. These findings support future testing and use of the decision aid for screening-age women in communities with environmental exposures.

Limitations

This study was focused on a narrow geographical region with known PFAS contamination. The sample demographics thus reflect the region, with a predominantly non-Hispanic White sample. Findings are therefore not widely generalizable to other regions with PFAS contamination, or the United States as a whole. The sample represented a group with relatively high mammogram experience: 76% reported having had at least one prior mammogram. By comparison, CDC data from 2022 reported that 60% of women ages 40–49 had gotten a mammogram in the past two years [46]. Mammogram knowledge scores were also high at baseline, introducing the potential for ceiling effects. Mammogram knowledge was assessed using a repeated measure within a short timeframe. We did not test for long-term knowledge gains and are unable to assess the durability of knowledge gains found here. Future studies will include follow up to determine medium to long-term impacts of decision aid use. Finally, we are limited to self-reported data regarding mammography history, breast cancer risk perceptions, and screening intentions. Future studies should further study screening behaviors and associated cancer detection as verified by medical records.

Conclusion

This study revealed moderate levels of stress and uncertainty about PFAS and other environmental exposures among women living in PFAS-affected communities. Use of MyMammogram addressed gaps in several important factors used in decision-making about screening mammography by significantly reducing perceptions of breast cancer risk, increasing knowledge about mammograms and reducing decisional conflict regarding breast cancer screening. In future work, we plan to partner with primary care providers across a wider geographic region to continue to test MyMammogram alongside a PFAS education intervention for clinicians to encourage informed shared decision-making for screening-age women.

Data Availability:

The data that support the findings of this study are available from the corresponding author upon reasonable request.