The perceived ethical appropriateness of messaging on breast cancer screening cessation among older women

Abstract

Objective:

Messaging about breast cancer screening cessation may reduce over-screening by raising awareness of the harms of screening, but in a background of strongly positive beliefs about screening among the public, such messaging may be perceived negatively. We aimed to assess whether older women perceived it to be ethically appropriate for clinicians to share a message that encourages breast cancer screening cessation.

Methods:

As part of a large national online survey experiment with women 65+ years, we presented a message (hereafter referred to as primary message) describing the rationales for stopping breast cancer screening (e.g., guideline recommendation, harms of screening) and assessed how ethical women thought it would be for doctors to share this information with patients. We assessed open-ended reactions. We also tested two variations of the primary message with different wordings of the recommendation to consider stopping screening.

Results:

Of 683 participants, 75.9% agreed that the primary message is ethically appropriate for doctors to share with patients., 13.2% neither agree or disagree, and 10.9% disagreed. Themes in open-ended responses suggested that the difference in participant response was partly attributed to whether participants perceived the message as informative or persuasive. Comparing across message variations, messages with stronger recommendations to stop screening were perceived to be less ethically appropriate than the primary message.

Conclusions:

Most older women perceived that it was ethically appropriate for doctors to share messages aimed at reducing breast cancer over-screening with patients.

Practice Implications:

Interventions should be developed to deliver messages to reduce over-screening among older women in practice settings, with evaluations to monitor their response.

1. Introduction

Breast cancer screening recommendations are variable in older women(1). The US Preventive Services Task Force recommends screening up to age 74, with insufficient evidence for continued screening among those 75+(2). The American Cancer Society (ACS) and American College of Radiology recommend screening in older women regardless of age unless they have limited life expectancy, with the ACS using 10-year life expectancy as the threshold(3–6). Breast cancer screening’s benefits can be delayed for up to 10 years whereas harms may occur in the short term(7,8). These harms include overdiagnosis and overtreatment of screening-detected cancers that, in the absence of screening, would not have been clinically important in a patient’s lifetime; false positive results that can cause psychological distress and lead to invasive procedures such as biopsies; and diverted attention away from existing health conditions (7–11). However, US national data showed that 50.6% women older than 75 years old and 42.8% with life expectancy < 10 years had a mammogram in the last 2 years, suggesting over-screening (12,13).

One way to try to reduce over-screening is for clinicians to talk with patients about the harms of over-screening and the decision to stop screening. Older adults consider clinicians as the most trusted source of information about cancer screening (13,14). Studies have shown that clinicians support stopping routine cancer screening using age and/or life expectancy criteria in older patients; as such, clinicians are well-positioned to deliver messages aimed at reducing over-screening to older women (15–17).

In our prior work, we developed messaging aimed at older women to support their understanding of the rationale for stopping breast cancer screening(18). We developed candidate messages about the various reasons for stopping screening in older women; we then conducted cognitive testing with older women to iteratively revise the candidate messages and evaluated them in a national survey(18). During cognitive testing, we found that older women preferred less definitive or direct recommendations about stopping mammograms. We found the most highly rated messages mentioned guideline recommendations, anecdotes about women who experienced false positive results, and evidence about overdiagnosis(18). We combined these into a single message and randomized 4570 women ages 65+ without history of breast cancer to 0, 1, or 2 message exposures over two time points, where the message was delivered via clinician, family member, or news story(19). We found that multiple exposures from different information sources increased older women’s support for screening cessation from 18% to 47% in a randomized controlled survey experiment (19), though actual behavior change was not assessed.

Although these prior results suggest messaging is a promising strategy to reduce over-screening, concerns may exist regarding the implementation of such a message into clinical discussions. In the context of high public enthusiasm for cancer screening (20), sharing information about the harms of screening and stopping screening may be perceived as persuasion (e.g., trying to change another person’s beliefs or actions). Controversy exists around the ethics of persuasion in health care, where the ethical principles of promoting benefits and minimizing harms are weighted against the concern that persuasion may threaten patient autonomy (21–24). Some argue that certain forms of persuasion, such as using rational arguments and sharing information based on scientific evidence, may be acceptable in clinical discussions (21–26).

There is literature describing the use of persuasive messaging to shift breast cancer screening behavior but the ethical appropriateness of such approaches was not discussed nor assessed from patient perspectives. Literature in younger women suggests that sharing information around the harms of over-screening, such as over-diagnosis, may be received negatively and be perceived by some as manipulative, which is generally considered an unethical form of influence(27,28). In a qualitative study with 20 older women, we found that older women often did not consider it appropriate for clinicians to persuade a patient to stop mammograms, partly because they did not believe that there was any downside to mammograms so there was no justification for clinicians to persuade a patient to stop (29). Whether sharing information about the downsides of mammograms and rationales for stopping screening would be considered ethically appropriate by older women is not known. Better understanding older women’s views about the ethical appropriateness of such messaging can help mitigate negative reactions and proactively address barriers during implementation and dissemination of the message.

We sought to assess if older women considered it ethically appropriate for clinicians to share a message that encouraged stopping breast cancer screening. Given the feedback during message development about older women’s preference for less directly worded recommendations about screening cessation, we also sought to test two additional message variations with different recommendation wordings.

2. Methods

2.1. Study design and survey instrument

This project was part of a survey experiment with 4570 older women on messaging about over-screening that used a nationally representative survey panel (Knowledge Panel). The study protocol has been previously described (19). Briefly, panel members were invited to participate if they were 65 years or older, women, English-speaking, and had no history of breast cancer. The results reported in this paper were from a survey module given only to the control group (n=683) at the end of the survey experiment. They had not received any messaging about breast cancer screening prior to this module.

We described a hypothetical 75-year-old woman, Ms. Johnson, with serious health problems and functional limitations to depict an older woman with <10-year life expectancy for whom stopping screening would be appropriate. We stated that Ms. Johnson was at a regular visit with her primary care doctor who shared some information about mammograms. We showed participants the message that we previously developed, which mentioned several rationales for stopping screening, which we hereafter refer to as the “primary message” (see Appendix for full message detail).

Next, we showed participants two additional variations of the message with slightly stronger recommendations to stop mammograms. In each variation, the entire message was shown, with the only changes in the last sentence of the message where, instead of saying: “It may be worth considering stopping mammograms” in the primary message, the two message variations said: “I think you should consider stopping mammograms” or “I think it is in your best interest to stop mammograms”.

2.2. Outcome and covariates

The survey instrument was piloted with 6 older women and iteratively revised based on feedback. Our primary outcome was the perceived ethical appropriateness of the primary message, as measured by the extent of agreement with the statement: “I consider this information ethically appropriate for a doctor to share with their patient” on a 5-point Likert scale (1=strongly disagree, 5=strongly agree). Then, in an open-ended question, we asked about why the participant considered the primary message ethically appropriate or inappropriate. We also asked the question about perceived ethical appropriateness of the two message variations.

KnowledgePanel provided information on participant age, race, education. We additionally assessed health literacy(30), mammogram status in past 2 years, breast cancer risk as measured by the Gail model (31), breast cancer worry, intention to get mammogram in the next 2 years, predicted life expectancy(32), and trust in the physician (33).

Statistical analyses

For the quantitative responses, we first summarized the results descriptively. We dichotomized the primary outcome, i.e., the perceived appropriateness of sharing the primary message, as strongly agree/agree vs. neutral/disagree/strongly disagree, and explored its association between participant characteristics. The selection of participant characteristics was based on prior literature. We found no specific prior literature on predictors of perceived ethical appropriateness of messages about stopping breast cancer screening; therefore, we included characteristics that have been shown to be important in screening decisions and/or in understanding screening benefits/harms. In addition to screening intention and cancer risk, which have been shown to be strongly predictive of screening behavior (34,35), our prior work showed that women older in age and those with poorer health were more likely to stop screening; those who were more worried about cancer risk and those of Black race were less likely to stop screening (18,36). Further, trust in clinician and prior screening behavior have been shown to influence screening decision-making (37,38). We sought to explore if these factors known to influence screening decision-making would also influence perception of messages about stopping screening. Lastly, we included education and health literacy, which have been shown to be important in understanding health information, and specifically screening benefits and harms (39,40). We explored univariate associations using chi-squares test.

To compare the primary message with the two additional message variations, we tested the proportion of participants who agreed or strongly agreed that each message variation was ethically appropriate, using the McNemar test for pair-wise comparisons.

2.3. Qualitative analyses

For the open-ended responses, we conducted thematic framework analysis using both inductive and deductive coding. The coding scheme was developed and iteratively revised with input from 3 team members (N.S., S.H., J.M). Deductive coding, informed by the study focus, included three main codes: participants’ views about the message, views about the benefits and harms of breast cancer screening, and other views related to screening decision or discussion. Inductive coding generated sub-codes within each of these categories. The code on views about the message included sub-codes on whether the message was perceived as persuasive or informative. The code on views about the benefits and harms of screening included sub-codes on whether the woman supported continued screening, supported stopping screening, or was unsure. The code on other views included sub-codes on patient choice, doctor’s role, and importance of information.

Two team members (N.S. and J.M.) used the finalized coding scheme to code the responses. First, we coded a random subset of 50 responses together to ensure reliability, reconciling any differences by consensus. Then, we coded the rest of the responses separately. Approximately 10% of the separately coded responses were then reviewed together to assess reliability; we felt confident in the consistency of our coding patterns. Next, the qualitative responses were reviewed, together with the participants’ quantitative ratings about the message’s ethical appropriateness, to identify potential explanations for their quantitative choices. The qualitative responses were summarized using themes, sub-themes and illustrative quotes.

3. Results

A total of 683 older women completed this survey module. Their mean age was 72.7 years (SD 5.9), 82.9% were non-Hispanic Whites, 71.6% had at least some college degree, 79.3% had a mammogram within the past two years and 75.4% intend to have another mammogram within the next 2 years (Table 1). Of the 683 women, 487 (71.3%) responded to the open-ended question about message appropriateness. Those who responded to the open-ended questions, compared to those who did not, had higher educational attainment and health literacy but were not otherwise different (Appendix table 1).

Table 1:

Demographic Information

Participant’s characteristics	All participants (n=683), N (%)	Considers the message appropriate, N(%) a	P-value b
Age
<75	461 (67.5)	350/460 (76.1)
75+	222 (32.5)	167/221 (75.6)	0.88
Race
White, non-Hispanic	566 (82.9)	427/565 (75.6)
Black, non-Hispanic	63 (9.2)	48/62 (77.4)
Others	54 (1.9)	42/54 (77.8)	0.90
Education
≤High School	194 (28.4)	140/193 (72.5)
≥Some College	489 (71.6)	377/488 (77.3)	0.20
Low health literacy c
Yes	56 (8.2)	37/56(66.1)
No	625 (91.8)	479/623 (76.9)	0.07
Mammogram in past 2 years
Yes	540 (79.3)	408/512 (75.7)
No	141 (20.7)	107/140 (76.4)	0.86
Intend to get mammogram in next 2 years
Yes	513 (75.4)	383/512 (74.8)
No	167 (24.6)	133/167 (79.6)	0.20
High breast cancer risk d
Yes	143 (21.7)	107/143 (74.8)
No	516 (78.3)	391/514 (76.1)	0.76
Cancer worry e
Moderate or extremely	73 (10.7)	52/73 (71.2)
Somewhat, slightly, not at all	609 (89.3)	465/608 (76.5)	0.32
Predicted life expectancy f
10+ years	537 (80.2)	406/535 (75.9)
<10 years	133 (19.9)	102/133 (76.7)	0.85
Trust in physician g
Lower trust	314 (46.7)	233/314 (74.2)
Higher trust	359 (53.3)	280/358 (78.2)	0.22

^aPercentages here indicate the percentages within each row that chose “strongly agree” or “agree” that the message was ethically appropriate to share. Denominator may differ from the row total due to two missing responses.

^bChi-square test was used to compare the proportions that perceived the message as ethically appropriate across each covariate’s categories.

^cHealth literacy was measured using a validated question: “how confident are you in filling out medical forms by yourself?” Responses of “somewhat”, “a little bit” or “not at all” were categorized as low health literacy(30).

^d5-year probability of breast cancer risk was assessed using the Gail Breast Cancer Risk Assessment Tool (31). We classified those with 5-year risk ≥3% as high risk, which is the threshold used in guidelines (49,50) for considering cancer prevention medications.

^eCancer worry was measured using a single question: “How worried are you about getting breast cancer?”

^fLife expectancy was estimated using the Schonberg mortality index(32): 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.

^gTrust in physician was measured using a validated 10-question scale, scores range from 10–50 with higher score indicates higher trust(33). We categorized the responses into two groups using the median score of 40 where those with scores <40 were categorized as “lower trust” and those with scores 40 or higher were categorized as “higher trust”.

Most participants (75.9%) perceived the primary message (i.e., It may be worth considering stopping mammograms) as ethically appropriate for doctors to share with their patients and an additional 13.2% were neutral. This perception did not differ significantly by any participant characteristics in the univariate analyses, including age, race, education, health literacy, mammogram status in past 2 years, breast cancer risk, breast cancer worry, predicted life expectancy, and trust in the physician (Table 1). Most participants (383/512 or 74.8%) who intended to continue screening nonetheless considered the message about stopping screening as ethically appropriate.

In the open-ended responses, two major themes emerged as reasons for participants perceiving the message as ethically appropriate (Table 2). First, some participants perceived the message as ethically appropriate because they supported stopping screening in older women and shared concern over the harms of screening. One participant said: “There are so many unnecessary tests done that this information was refreshingly honest. The patient may still choose to have a certain test done but it is still good to know that it’s ok not to have the test.” Second, rooted in a trusting doctor-patient relationship, many believed that it was the doctor’s role to share information. One participant commented: “Scientifically based information should always be shared... by doctors with their patients. Isn’t that why one goes to see a doctor?”

Table 2:

Themes, subthemes, and illustrative quotes from open-ended responses on why participants thought the primary message was or was not ethically appropriate for a doctor to share with their patients.

Perceives message to be ethically appropriate
Supports stopping screening	“Enduring more tests and treatments while already dealing with health issues does not seem to improve the quality of life for the patient.”
Has concern about screening harms	“The patient should know that it is not always in their best interest to get all screening tests. We are not told very often that there is a downside to these tests.” (participant ID 4404)
Trust in doctor	“I am one that follows my doctor’s advice, but then I trust him and he has been my doctor for 25 years... It’s all about trust in your doctor.”
Perceives message as informative	“The doctor is sharing information the patient can use to make an informed decision. So, I consider it ethically appropriate to disclose it.”
Perceives message to not be ethically appropriate
Persistent enthusiasm about screening	“I still feel it’s important to get a mammogram no matter what.”
Has concern about risk of cancer without screening	“I don’t believe it is accurate information... it’s better to have a false positive mammogram rather than to have no mammogram and find out later you have breast cancer..”
Perceives message as persuasive	“To me this is an attempt to persuade someone to avoid a test that would provide a patient with potentially lifesaving information.”
Across both groups
Screening is patient’s choice	“A doctor can advise me whatever they feel is appropriate to tell me, but I am still going to make my own decision whether to have the mammogram or not.”

Participants who did not perceive the message as ethically appropriate (10.9%) more often voiced persistent enthusiasm about screening and/or concerns over the risk of cancer without screening. Several also perceived the message as persuasive rather than factual and considered this inappropriate. One participant said: “The statement sounded more like a scare tactic. It was being persuasive rather than information. Getting a mammogram saves lives.” Across all participants, there were frequent comments that the ultimate decision about screening should be the patient’s choice.

A majority of participants also perceived the stronger-worded message variations as ethically appropriate (although the proportions were both significantly less than for the primary message, p<0.001). Compared to the primary message,. 66.1% of participants perceived the message ending in “I think you should consider stopping mammograms” and 64.7% of participants perceived the message ending in ”I think it is in your best interest to stop mammograms” as ethically appropriate for doctors to share with their patients. No significant difference was found when comparing the two variations with each other (Figure 1).

Figure 1: Perceived ethical appropriateness of message variations on stopping breast cancer screening.

4. Discussion and Conclusion

4.1. Discussion

Messaging is a promising but under-studied approach to reduce over-screening for breast cancer in older women. This is the first national study to explore the perspective of older women regarding the ethical appropriateness of messaging on stopping breast cancer screening. We found that over three-quarters of the participants considered the message—which included reference to guidelines and information on over-diagnosis and false positives, as well as advice to consider stopping mammograms—appropriate for doctors to share with patients during clinical discussion. Approximately two-thirds of participants felt that it was ethically appropriate for clinicians to share more strongly worded recommendations against screening; whereas 20% of participants were neutral and 15% disagreed.

Prior work, including our own, examined older adults’ responses to discussing stopping routine cancer screening but did not specifically evaluate the perceived ethical appropriateness of such a discussion (37,41–43). In this study, we found strongly pro-screening views among our participants but, interestingly, many participants who intend to continue screening themselves still considered the message to consider stopping screening as ethically appropriate. The open-ended responses suggested that one factor underlying the difference in perceived ethical appropriateness of the message was whether the message was perceived as informative versus persuasive. This is a novel finding. Message development for public health campaigns often focuses on identifying messages that are perceived to be both informative and persuasive (44). Our results suggest that messages for clinical discussions may be more acceptable to patients if they are perceived as informative but not necessarily as persuasive. We found that trust in the doctor was another factor underlying the perception that the message on screening cessation was ethically appropriate. This finding is consistent with prior work that found older adults were amenable to stopping screening in the context of a trusting relationship with the clinician (37,45). What remains unknown is whether exposure to messages on screening cessation, in a background of strongly pro-screening views, may negatively impact trust. One prior study found that changing screening guidelines may result in mistrust in African American women(46). Future studies should evaluate the impact of screening cessation messaging on patient trust in clinicians and in guidelines.

We found that more strongly worded recommendations were still accepted by the majority of participants. This finding challenges existing assumptions in the literature and provides clinicians with different options in how they communicate with patients. Our results showed that acceptance of the more strongly worded message variations were lower. This finding may be explained by ours and others’ finding that ultimate decisions about screening should be left to the patient, so that a less prescriptive recommendation was preferred (47,48).

The limitations of our study included that the message was tested in a hypothetical vignette and perceptions and reactions may be different if delivered in a clinical encounter. Further, the study design did not allow participants to ask questions about the message, which they may likely do in a clinical discussion. Although recruitment was from a nationally representative online panel, there was evidence of nonresponse bias; participants in the sample had high health literacy and were mostly White, and thus their responses may not be generalizable to the entire population of older U.S. women. The study design did not allow for follow-up questions or clarifications of the open-ended responses. Although two team members who coded the open-ended responses reviewed a subset of the responses to ensure coding reliability, this was not formally assessed using Cohen’s kappa.

4.2. Conclusion

In summary, we found that most older women perceived a message aimed at reducing breast cancer over-screening as information that is ethically appropriate for doctors to share with patients.

4.3. Practice Implications

The message has already been shown to effectively increase support for screening cessation in our previously reported work, and the current study alleviates concerns for negative reactions, suggesting its utility and acceptability in potentially reducing over-screening among older women(19). Implementing the message in clinical discussions through clinician coaching and/or prompts can be tested as an intervention to reduce over-screening as the next step.

Highlights.

• Messaging to reduce breast cancer over-screening may be perceived as manipulative.

• We tested the perceived ethical appropriateness of a message on screening cessation.

• Among 683 older women, 75.9% agreed that the message was ethically appropriate.

• Messaging can reduce over-screening without significant negative reactions.