Informed decision making before initiating screening mammography: does it occur and does it make a difference?

Abstract

Objective  Informed decision making regarding screening mammography is recommended for women under age 50. To what extent it occurs in clinical settings is unclear.

Methods  Using a mailed instrument, we surveyed women aged 40–44 prior to their first screening mammogram. All women were members of a large health maintenance organization and received care at a large medical practice in the Greater Boston area. The survey measured informed decision making, decisional conflict, satisfaction, and screening mammography knowledge and intentions to undergo screening.

Results  Ninety‐six women responded to the survey (response rate 47%). Overall, women reported limited informed decision making regarding screening mammography, both with respect to information exchange and involvement in the decision process. Less than half (47%) reported discussing the benefits of screening; 23% the uncertainties; and only 7% the harms. About 30% reported discussing the nature of the decision or clinical issue; and 29% reported their provider elicited their preferred role in the decision; 38% their preferences; and 24% their understanding of the information. Women who were uninformed had higher decisional conflict (2.37 vs. 1.83, P = 0.005) about screening mammography and were more likely to be dissatisfied with the information and involvement. Women’s screening mammography knowledge was limited in most areas; however being presented with information did not diminish their intentions to undergo screening.

Conclusion  Informed decision making before initiating screening mammography is limited in this setting. There appears to be little indication that information about the benefits and harms decreases women’s intentions to undergo screening. Methods to communicate information to women before initiating screening mammography are needed.

Introduction

While the effectiveness of screening mammography for women in their 40s had been previously debated, ¹ , ² most current U.S. guidelines, including the U.S. Preventive Services Task Force ² and the American Cancer Society, ³ recommend routine screening for women in this age group. In addition, both organizations encourage providers to inform women about the benefits and potential harms of screening mammography and engage them in informed decision making. ² , ³ Women initiating screening mammography have also expressed a desire to be informed about the benefits and potential harms of screening and prefer their primary care providers to be the source of information. ⁴ Further, participation in informed decision making regarding screening mammography may enhance women’s satisfaction ⁵ and may correct their misperceptions about breast cancer risk, the benefits of screening and its potential harms. ⁶ , ⁷ Whether informed decision making reduces screening utilization is not clear. ⁸

We surveyed women aged 40–44 who were scheduled for their first mammogram using a mailed survey to measure their engagement in informed decision making prior to initiating screening mammography. ⁹ Specifically, we were interested in assessing the extent of women’s discussions with their primary care providers about the benefits, harms and uncertainties of screening and their engagement in the decision‐making process. To further understand the effect of informed decision making on patient‐oriented outcomes, we tested the association between informed decision making and two patient‐oriented outcomes, decisional conflict and satisfaction. Finally, to examine whether there may be any effects of sharing‐specific information about screening on women’s intentions to undergo screening, we measured women’s screening mammography knowledge and their intentions to undergo screening based on being presented specific elements of information. Elsewhere we reported results from this survey describing women’s information needs prior to initiating screening and their preferences for decision making. ⁴

Methods

Study setting and participants

The study setting and participants have been previously described. ⁴ Using automated data, we identified women aged 40–44 who were scheduled for their first screening mammogram at a large multi‐specialty medical practice in the greater Boston area between March 2002 and September 2003. All women were insured through a health maintenance organization serving approximately 800 000 members in Massachusetts, Maine and New Hampshire. In this setting, we found that most women undergo their first screening by age 50, and based on our prior qualitative study, women decide whether to undergo screening by their early 40s. ¹⁰ We excluded women with a prior mammogram, breast surgery or breast cancer, and those with no primary care visit in the prior year. We offered a $2 donation to a breast cancer organization of the woman’s choice for completing the survey. The study was approved by our institutional Human Subjects Committee.

Measures

We based the development of our mailed survey measures on published literature on informed decision‐making, ⁶ , ⁹ our earlier qualitative work in this area, ¹⁰ and on prior screening mammography literature. ¹ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ The survey included items assessing informed decision making, decisional conflict, satisfaction with the information and involvement, screening mammography knowledge and intentions to undergo screening. Women’s characteristics were also measured. The survey development included content and format review by behavioural sciences experts at the Dana‐Farber/Harvard Cancer Centre. The questionnaire was pilot tested and revised prior to fielding.

Informed decision making was measured using a 7‐item informed decision‐making scale modified from a previously validated audiotape‐derived measure ⁹ and included items defined as important for informed decision making based on our prior qualitative study. ¹⁰ Women were asked to state whether their health‐care provider discussed the benefits, harms and uncertainties of screening mammography; and elicited the nature of the decision/clinical issue; their individual role in the decision to have a screening mammogram, their preferences for screening, and their understanding of the information provided. The responses were ‘yes/unsure/no’.

Decisional conflict, or women’s comfort with the decision‐making process, was measured using a previously validated 9‐item scale. ¹⁶ The scale measures the state of uncertainty about a course of action to take, and includes three subscales: uncertainty about choosing among alternatives, factors contributing to uncertainty, and perceived effective decision making. The scale includes items such as ‘This decision is easy for me to make’, ‘I feel that I understand the benefits’, ‘I feel that I have made an informed decision’, and ‘I am satisfied with my decision’. The responses are based on a 5‐point Likert scale (strongly disagree to strongly agree). The decisional conflict scale is distinct from the informed decision making scale, in that the women are asked for their perceived comfort with the decision and understanding of the benefits and harms rather than the actual information exchange and involvement in decision making.

We measured satisfaction by asking participants for their degree of satisfaction with the information provided by their clinicians and with their individual involvement in the decision‐making process. Both items were measured using a 4‐point scale with responses ranging from ‘very satisfied’ to ‘very dissatisfied’. ¹⁷

To assess knowledge and intentions to undergo screening, we provided women with statements describing the numerical estimates for each of the following items: lifetime risk of breast cancer, 5‐year risk of breast cancer, risk of false‐positive mammograms (two items), risk of false‐negative mammograms, cost of mammograms and pain associated with screening. Following each of the statements, we asked the women to report whether they had previously been aware of these statements (yes/no). The numerical estimates were based on pertinent literature available at the time of survey preparation. ¹ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁸ We then asked the women to report whether their awareness of the information changed their intentions to undergo screening (i.e. were they more or less likely to undergo screening, or did the information have no effect on their intentions to undergo screening). To compare our findings to prior studies, we also included multiple choice items that measured women’s knowledge about to the age at which to begin screening, the ages for which medical evidence supports screening, and the relative accuracy of screening for a 40‐year‐old compared with a 60‐year‐old woman (4‐point scale) as well as one item measuring their awareness of ductal carcinoma in situ (yes/no).

Data collection

As previously described, ⁴ we sent surveys to all potentially eligible women 2–4 weeks prior to their scheduled screening mammography appointment, with a reminder 1–2 weeks after the initial mailing. Because our goal was to assess the outcomes among women prior to initiating screening, those who had not responded prior to their mammography appointments were no longer contacted. A research assistant entered the responses into a database and verified the coding for accuracy.

Statistical analysis

First, to determine the ‘completeness’ of informed decision‐making regarding screening mammography, we calculated the distribution of women who replied ‘yes’ to the seven elements of informed decision making. The scores ranged from 0 for women who reported discussing no elements to 7 for women who reported discussing all elements. Women who reported ‘no’ to all seven items were considered ‘uninformed’. We calculated a Cronbach’s alpha for the informed decision‐making scale and found that the scale performed reliably in our sample with an alpha coefficient 0.73. Because the seven items measure two distinct aspects of decision making, information and involvement, we also categorized the informed decision‐making scale items into two subscales, respectively. The informational subscale consisted of three elements (‘benefits’, ‘harms’ and ‘uncertainties’) with a range of scores from 0 (no informational items discussed) to 3 (all informational items discussed) and the involvement subscale consisted of four items (‘nature’, ‘role’, ‘preferences’ and ‘understanding’) with a range of scores from zero (no involvement elements discussed) to 4 (all involvement elements discussed).

Second, we determined the ‘content’ of the discussions by calculating the percentage of women who responded ‘yes’ to each of the seven individual elements. To evaluate the effect of women’s recall, we conducted a sensitivity analysis by recoding the ‘unsure’ responses as ‘yes’, assuming that women may not have recalled a discussion that actually occurred.

Third, we reported the decisional conflict as a summary score derived by adding the Likert scores (1–5) and dividing by the number of items. ¹⁶ Based on prior literature ¹⁶ ; we used a score > 2.0 to indicate clinically significant decisional conflict, implying that women were unsure about, whether to undergo screening or were delaying the decision. The reliability of the decision conflict scale in our population was good (alpha coefficient 0.80) and was similar to the prior findings. ¹⁶

Satisfaction items were collapsed into satisfied and unsatisfied. We used descriptive statistics to report the women’s characteristics, their self‐reported knowledge and their screening intentions.

Significance was determined using parametric and non‐parametric tests. Specific tests used are presented. All analyses were conducted using sas software, Version 8.2 (SAS Institute, Cary, NC, USA).

Results

Study participants

Our study population has been previously described. ⁴ Of the 203 eligible women invited to participate, we received responses from 96 (response rate 47%). Responders and non‐responders were similar in age (mean 41.0 and 41.3 years, respectively), time since their primary care appointment (mean 135 and 133 days, respectively, and clinical centre, a surrogate for area of residence and likely socioeconomic status). Most participants were white people (67%), had an annual income of at least $40 000 (66%), and at least some college education (72%). Four percentage had family history of breast cancer.

Completeness and content of informed decision making

Overall, informed decision making regarding screening mammography was limited (Fig. 1). On average, women reported discussing a median of one of seven elements of informed decision making with their providers (25th–75th percentiles 0 and 3, respectively). As noted in Fig. 1, 27 (28%) of women stated that their discussions with medical providers included none of the seven elements; 22 (23%) reported one element; 12 (12%), two; 16 (17%), three; 7 (7%), four; 7 (7%), five; and 3 (3%), six. Only two (2%) women reported a complete discussion (including all seven elements). The sensitivity analysis, counting ‘unsure’ responses as ‘yes’ increased slightly in each of the categories (Fig. 1). Of the three informational elements of informed decision‐making, women reported discussing a median of one element (25th–75th percentiles 0 and 2, respectively). Specifically, the benefits of screening were most often discussed (47%), the uncertainties were discussed less often (23%) and the harms were rarely discussed (7%) (Table 1). Of the four involvement elements, women reported that their providers elicited a median of one element (25th–75th percentiles 0 and 2, respectively). Specifically, 30% of the women reported discussing the nature of the decision with their provider, 29% reported being asked about their preferred roles in the decision, 38% reported having their preferences elicited and only 24% reported being asked whether they understood the information. The sensitivity analysis, counting ‘unsure’ responses as ‘yes’ increased the percentages slightly for each of the seven items (Table 1). The median number of elements reported increased to three (one in the information scale and two in the involvement scale).

Figure 1.

 Completeness of informed decision making.
Figure represents the percentage of women who answered ‘yes’ to each of the seven individual informed decision‐making items (black columns), and those who answered ‘yes’ and ‘unsure’ (grey columns).

Table 1.

 Content of informed decision making

Informed decision making element	% of women reporting yes to each of the items	95% Confidence interval	% of women reporting yes and unsure to each of the items	95% Confidence interval
Information domain elements
Benefits of screening discussed	47	37–58	57	47–67
Harms of screening discussed	7	2–13	21	13–29
Uncertainties of screening discussed	23	15–32	32	22–41
Involvement domain elements
Nature of the decision/clinical issue discussed	30	21–40	41	31–51
Patient’s role in the decision elicited	29	20–39	44	34–54
Patient’s preferences in the decision elicited	38	28–48	48	38–59
Understanding of information assessed	24	15–33	40	30–50

Informed decision making and decisional conflict

Overall, the mean decisional conflict score was 2.02 (95% confidence intervals 1.90, 2.14) suggesting some conflict with the decision to undergo screening. Decisional conflict was inversely related to the number of informed decision making elements reported. Women who reported being uninformed (no elements discussed) had higher decisional conflict scores than those who were better informed (2.37 vs. 1.83, P = 0.0005, anova) (Table 2), suggesting that women who were less engaged in informed decision making were more conflicted about their decisions. The informational domain of informed decision making appeared to be more predictive of decisional conflict (Spearman correlation coefficient = −0.34, P = 0.008) than the involvement domain (r = −0.13, P = 0.20).

Table 2.

 Relationship between informed decision‐making and decisional conflict*

	Number of women	Decisional conflict score (mean)	95% Confidence interval
No. elements of informed decision making discussed with primary care provider
0	27	2.37	2.09, 2.63
1	22	1.99	1.76, 2.22
More than 1	47	1.83	1.69, 1.97
Overall score	96	2.01	1.90, 2.14

*P = 0.005, anova.

Informed decision making and satisfaction

Almost all participants were either very satisfied or somewhat satisfied with the information provided (n = 82 or 88.2%) and their level of involvement in the decision to get a screening mammogram (n = 90 or 94.7%). However, women who were unsatisfied were more likely to be uninformed. Specifically, among the small group of women who were dissatisfied with the information, most were uninformed (9 of 11, P < 0.0001). Likewise, of the five women dissatisfied with their involvement, four of five, were uninformed (P = 0.02, Fisher’s exact test).

Screening mammography knowledge and intentions for screening

Women’s screening mammography knowledge was limited (Table 3). Whereas 40 (43%) were aware of a woman’s lifetime risk of developing breast cancer, only 12 (13%) were aware of the woman’s 5‐year risk. About one‐quarter of the women (23%) were aware of the numerical benefits of screening while only 13% were aware of the chances of false‐negative and false‐positive mammograms. Most (73%) of the women were aware of the pain and discomfort experienced during mammograms. Only eight (9%) were aware of the costs.

Table 3.

 Breast cancer knowledge and effect of information on screening intentions

Item	Aware	Effect of information on intentions to obtain screening mammogram
		More likely	No effect	Less likely
About 12 of 100 women will get breast cancer during their lifetime	40 (43)	56 (60)	37 (40)	0
< 1 of 100 women will get breast cancer during the next 5 years	12 (13)	35 (38)	54 (59)	3 (3)
About 20 of 100 mammograms will not find an existing breast cancer	12 (13)	26 (28)	60 (64)	7 (8)
About 10 of 100 mammograms will be abnormal and will lead to additional tests (sometimes referred to as call‐back or false alarm)	19 (21)	27 (29)	64 (69)	2 (2)
About 95 of 100 call‐backs end up not being cancer	12 (13)	24 (26)	64 (69)	5 (5)
About 1 of 1000 women will live longer by having regular mammograms in their 40s	21 (23)	52 (56)	37 (40)	4 (4)
Some women experience pain and discomfort during mammograms	73 (79)	13 (14)	67 (73)	12 (13)
A screening mammogram costs about $150 if not covered by insurance	8 (9)	9 (10)	65 (71)	18 (20)

Values are given as n (%).

Most (64%) women reported that screening should begin at age 40, 22 (24%) at age 35, and only 3 (3%) at age 50. When asked to identify all ages for which medical studies have proven benefit from screening, 83 (89%) selected ‘40–49’ and 49 (53%) selected ‘50–69.’ A relatively large percentage also selected ‘18–39’ and ‘70 and older’ Most women (n = 53, or 59%) reported that the accuracy of screening mammography for a 40‐year‐old woman compared with a 60‐year‐old woman was the same, 25 (30%) reported the accuracy to be better and 21 (13%) reported the accuracy to be worse. Finally, only 10 (11%) of women were aware of ductal carcinoma in situ.

Women’s intentions to initiate screening either increased or were not affected by most of the information items presented (Table 3). Women’s knowledge of the chances of having a false‐positive and false‐negative mammogram decreased intentions to initiate screening by only 5% and 8% of the women, respectively. The cost of screening mammography resulted in the greatest percentage (20%) of women reporting being less likely to initiate screening.

Discussion

Our study found that informed decision making among women aged 40–44 prior to initiating screening mammography was limited, both in information exchange and involvement in the decision‐making process. Women reported that the information most often discussed with providers included the benefits of screening, but the harms and uncertainties were less frequently discussed. Involvement in decision making was limited to about one‐third of women. We also found that women who not informed were more likely to be conflicted about their decisions to undergo screening and dissatisfied. The limited informed decision making reported by the women is supported by their suboptimal knowledge about the factual information about screening mammography. After being presented with information about screening, women reported minimal effects on their intentions to initiate screening.

Prior studies have shown that informed decision making in general medical settings is limited. Braddock et al. found that < 20% of the decisions were adequately informed, but did not specifically address screening mammography. ⁹ Our prior qualitative in‐depth interviews ¹⁰ among women aged 38–45 revealed similar results. To our knowledge, this is the first study to use quantitative data to report the extent of informed decision making among women initiating screening mammography.

How much information and involvement in decision making is needed for the decision to undergo screening mammography? Data suggest that patients almost universally want information shared with them by their medical providers, but the degree of involvement patients want in the decision making process varies. ¹⁹ , ²⁰ Our prior findings showed that the women in this study sample wanted information about the benefits and possible harms of screening and preferred to play an active or collaborative role in the decision making process. ⁴ The results of this study suggest that both the information and involvement preferences were not met. Despite this, most women appeared to be satisfied with both, perhaps because of their satisfaction with their care from their medical providers. As noted in prior literature, patients tend to express satisfaction with their own care, though may be dissatisfied with health care in general. ²¹ However, women who were dissatisfied were less likely to have engaged in informed decision making.

The lack of informed decision making reported by the women is supported by their lack of knowledge of screening mammography. Women’s suboptimal knowledge in this area has been previously reported. ⁶ , ⁷ , ²² , ²³ However, prior studies used items that asked for women’s numerical approximation of risk and placed a strong emphasis on numeracy. In our study, we presented women with the numerical information and asked them to state whether they had previously been aware of it. This method relies on recognition rather than recall; that so many women reported lack of knowledge using this method is even more convincing than previously reported. Consistent with prior findings, the women in this sample reported conflicting information about the age at which to begin screening, ²⁴ and were not aware of ductal carcinoma in situ. ⁶ , ²⁵ , ²⁶ Interestingly, we found no consistent correlation between women’s self‐reported discussion of the benefits, harms and uncertainties of screening with their knowledge of the related items. The fact that women’s intentions to undergo screening were not affected by numerical information is not surprising but important to acknowledge. Our prior qualitative work suggested that although women wanted to have more in‐depth discussions about screening with their providers, they were not making decisions per se. ¹⁰ In a randomized controlled trial, Lewis et al. found that while women’s overall knowledge about the benefits and harms of screening improved following an educational intervention, their positive perceptions of screening did not change. ²⁷ Although prior literature in prostate cancer screening suggested that informed decision making decreased the rates of screening ²⁸ and even resulted in a malpractice suit, ²⁹ it is likely that women look more favourably at breast cancer screening, and screening in general, despite potential limitations and possible harms. ³⁰ , ³¹

Our study adds to the existing literature by assessing women’s intentions to initiate screening based on individual pieces of information. This may help clinicians better understand how to present certain types of information and may reassure them that presenting the possibility of harms does not lead to lesser screening. It is interesting that among a group of insured women, the item that led to the greatest decline in self‐reported intentions to undergo screening was that of mammography cost.

Our study has several strengths. First, it is the only study that we know to assess the completeness of informed decision making among women initiating screening mammography in their 40s. The study adds to the growing literature linking the informed decision making and patient‐oriented outcomes of informed decision making, specifically satisfaction and decisional conflict, ⁵ which had not previously been reported among women initiating screening mammography. Second, the modified informed decision‐making scale that we developed is less expensive and time consuming than those requiring the audiotaping of clinical encounters ⁹ , ³² and is shorter than COMRADE, a recently developed scale. ³³ As recently suggested by Ford et al., patients report greater engagement in informed decision making than measured by observation. ³⁴ Therefore, even though we found limited decision making, our results may overestimate patient engagement in informed decision making. Finally, the generalizability of our findings to women in other clinical settings is supported by an enrolment of a large percentage of minority groups and the distribution of incomes that was similar to that of the state of Massachusetts. ³⁵

Our study is limited by a relatively small sample and a response rate of 47%. We believe that the response rate was mainly affected by the relatively short time window before the women’s mammography appointments. Because we were interested in women’s opinions prior to their mammograms, we did not contact the non‐respondents beyond a 2‐week period. As previously noted, the respondents and non‐respondents were similar, ⁴ suggesting no systematic bias. Another possible limitation is recall bias, with women who reported dissatisfaction and decisional conflict being less likely to recall informed decision making they received; however, we found no association between the time since the clinical encounter with informed decision‐making and decisional conflict, and found no significant differences when we assumed that those who were unsure actually had the discussions. Our results should not be generalized to those women who did not undergo screening.

In summary, we found limited engagement of women in informed decision making before their first screening mammogram, both with respect to information exchange and involvement. When information exchange occurred, it mostly focused on the benefits of screening, but rarely included a discussion of the potential harms and uncertainties. While most women were satisfied and reported minimal decisional conflict, those who reported being uninformed were more frequently dissatisfied with the information they received and their involvement in the decision‐making process, and were more likely to report decisional conflict. Our study also showed that women in their 40s had limited knowledge about screening mammography and that being provided with information did not adversely affect their stated intentions to initiate screening. Given the emphasis on informed decision making and its likely benefits, methods to communicate information to women are needed.