Shared decision‐making for supporting women’s decisions about breast cancer screening

Abstract

Background

In breast cancer screening programmes, women may have discussions with a healthcare provider to help them decide whether or not they wish to join the breast cancer screening programme. This process is called shared decision‐making (SDM) and involves discussions and decisions based on the evidence and the person's values and preferences. SDM is becoming a recommended approach in clinical guidelines, extending beyond decision aids. However, the overall effect of SDM in women deciding to participate in breast cancer screening remains uncertain.

Objectives

To assess the effect of SDM on women's satisfaction, confidence, and knowledge when deciding whether to participate in breast cancer screening.

Search methods

We searched the Cochrane Breast Cancer Group's Specialised Register, CENTRAL, MEDLINE, Embase, CINAHL, PsycINFO, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform on 8 August 2023. We also screened abstracts from two relevant conferences from 2020 to 2023.

Selection criteria

We included parallel randomised controlled trials (RCTs) and cluster‐RCTs assessing interventions targeting various components of SDM. The focus was on supporting women aged 40 to 75 at average or above‐average risk of breast cancer in their decision to participate in breast cancer screening.

Data collection and analysis

Two review authors independently assessed studies for inclusion and conducted data extraction, risk of bias assessment, and GRADE assessment of the certainty of the evidence. Review outcomes included satisfaction with the decision‐making process, confidence in the decision made, knowledge of all options, adherence to the chosen option, women's involvement in SDM, woman‐clinician communication, and mental health.

Main results

We identified 19 studies with 64,215 randomised women, mostly with an average to moderate risk of breast cancer. Two studies covered all aspects of SDM; six examined shortened forms of SDM involving communication on risks and personal values; and 11 focused on enhanced communication of risk without other SDM aspects.

SDM involving all components compared to control

The two eligible studies did not assess satisfaction with the SDM process or confidence in the decision. Based on a single study, SDM showed uncertain effects on participant knowledge regarding the age to start screening (risk ratio (RR) 1.18, 95% confidence interval (CI) 0.61 to 2.28; 133 women; very low certainty evidence) and frequency of testing (RR 0.84, 95% CI 0.68 to 1.04; 133 women; very low certainty evidence). Other review outcomes were not measured.

Abbreviated forms of SDM with clarification of values and preferences compared to control

Of the six included studies, none evaluated satisfaction with the SDM process. These interventions may reduce conflict in the decision made, based on two measures, Decisional Conflict Scale scores (mean difference (MD) −1.60, 95% CI −4.21 to 0.87; conflict scale from 0 to 100; 4 studies; 1714 women; very low certainty evidence) and the proportion of women with residual conflict compared to control at one to three months' follow‐up (rate of women with a conflicted decision, RR 0.75, 95% CI 0.56 to 0.99; 1 study; 1001 women, very low certainty evidence).

Knowledge of all options was assessed through knowledge scores and informed choice. The effect of SDM may enhance knowledge (MDs ranged from 0.47 to 1.44 higher scores on a scale from 0 to 10; 5 studies; 2114 women; low certainty evidence) and may lead to higher rates of informed choice (RR 1.24, 95% CI 0.95 to 1.63; 4 studies; 2449 women; low certainty evidence) compared to control at one to three months' follow‐up. These interventions may result in little to no difference in anxiety (MD 0.54, 95% −0.96 to 2.14; scale from 20 to 80; 2 studies; 749 women; low certainty evidence) and the number of women with worries about cancer compared to control at four to six weeks' follow‐up (RR 0.88, 95% CI 0.73 to 1.06; 1 study, 639 women; low certainty evidence). Other review outcomes were not measured.

Enhanced communication about risks without other SDM aspects compared to control

Of 11 studies, three did not report relevant outcomes for this review, and none assessed satisfaction with the SDM process. Confidence in the decision made was measured by decisional conflict and anticipated regret of participating in screening or not. These interventions, without addressing values and preferences, may result in lower confidence in the decision compared to regular communication strategies at two weeks' follow‐up (MD 2.89, 95% CI −2.35 to 8.14; Decisional Conflict Scale from 0 to 100; 2 studies; 1191 women; low certainty evidence). They may result in higher anticipated regret if participating in screening (MD 0.28, 95% CI 0.15 to 0.41) and lower anticipated regret if not participating in screening (MD −0.28, 95% CI −0.42 to −0.14).

These interventions increase knowledge (MD 1.14, 95% CI 0.61 to 1.62; scale from 0 to 10; 4 studies; 2510 women; high certainty evidence), while it is unclear if there is a higher rate of informed choice compared to regular communication strategies at two to four weeks' follow‐up (RR 1.27, 95% CI 0.83 to 1.92; 2 studies; 1805 women; low certainty evidence). These interventions result in little to no difference in anxiety (MD 0.33, 95% CI −1.55 to 0.99; scale from 20 to 80) and depression (MD 0.02, 95% CI −0.41 to 0.45; scale from 0 to 21; 2 studies; 1193 women; high certainty evidence) and lower cancer worry compared to control (MD −0.17, 95% CI −0.26 to −0.08; scale from 1 to 4; 1 study; 838 women; high certainty evidence). Other review outcomes were not measured.

Authors' conclusions

Studies using abbreviated forms of SDM and other forms of enhanced communications indicated improvements in knowledge and reduced decisional conflict. However, uncertainty remains about the effect of SDM on supporting women's decisions. Most studies did not evaluate outcomes considered important for this review topic, and those that did measured different concepts. High‐quality randomised trials are needed to evaluate SDM in diverse cultural settings with a focus on outcomes such as women's satisfaction with choices aligned to their values.

Plain language summary

Does shared decision‐making help women when making decisions about whether or not to participate in breast cancer screening?

Key messages

Shared decision‐making could help women feel less unsure or regretful and assist with learning during the decision‐making process for breast cancer screening. However, it is important to note that our understanding of how exactly it may affect women's screening decisions is incomplete.

What is shared decision‐making?

Shared decision‐making is when a doctor and a patient work together to choose the best care. They talk about different options, the pros and cons, and what matters to the patient. They use tools like booklets or online guides (decision aids) to provide clear information and decide together.

Why does shared decision‐making matter for breast cancer screening?

Breast cancer screening helps save lives and reduces health issues during treatment. However, it may sometimes give incorrect results or lead to too much treatment. When women and doctors make choices together, they can make informed decisions that align with women's values.

What did we want to find out?

We wanted to know if shared decision‐making could help women feel more satisfied, confident, and knowledgeable when deciding whether to participate in breast cancer screening.

What did we do?

We included studies that looked at how shared decision‐making affects women making choices about breast cancer screening. We chose studies that compared some or all the important aspects of shared decision‐making with routine care. We judged how certain we could be in the findings based on factors like study methods and sizes.

What did we find?

We looked at 19 studies with 64,215 women. Women were given information about the pros and cons of breast cancer screening. Most studies used tools to provide this information. Six studies did not include a discussion with a healthcare professional, and 11 studies did not consider a woman’s values and preferences. The studies followed women for a short time, usually from one to three months, and were conducted in the USA, Europe, Australia, and one in Iran. Most studies were funded by government or schools, and some by private groups.

Shared decision‐making involving all key components

Two studies included discussions with healthcare professionals and considered values and preferences. Based on a single study, shared‐decision making may not affect women’s knowledge about when to start screening and screening frequency, but the results are uncertain. The two studies did not look at outcomes like women’s satisfaction with the shared decision‐making process, confidence in screening choices, adherence to decisions, active participation in decision‐making, effective communication with doctors, or changes in mental health.

Shortened forms of shared decision‐making with clarification of values and preferences

Six studies used decision‐making tools and considered values and preferences but did not include conversations with a healthcare professional. This type of shared decision‐making could make women feel more confident and knowledgeable about their choices, although it may not affect anxiety or cancer worry. These studies did not look at outcomes like women’s satisfaction with the shared decision‐making process, adherence to decisions, active participation in decision‐making, or effective communication with doctors.

Enhanced communication about risks without other components of shared decision‐making

Eleven studies provided women with information about options and the pros and cons but did not include a conversation with a healthcare professional or consider women’s values and preferences. This type of shared decision‐making helps women feel more knowledgeable about their choices, although it is unclear if it increases confidence. It does not affect anxiety or depression, but does reduce cancer worry. These studies did not look at outcomes like women’s satisfaction with the shared decision‐making process, adherence to decisions, active participation in decision‐making, or effective communication with doctors.

What are the limitations of the evidence?

Although there were many studies involving a total of over 60,000 women, the studies used different approaches to look at shared decision‐making, presented data in varied formats, and did not look at outcomes considered important in our review. These differences prevented us from combining information in some cases for clear results. Also, some studies had issues with their methods. As a result, we cannot be certain about some of the conclusions in this review.

How up‐to‐date is this information?

The information is current to August 2023.

Summary of findings

Summary of findings 1. Shared decision‐making (all components) versus control.

Shared decision‐making (all components) versus control for women eligible for breast cancer screening
Patient or population: women aged 40 to 75 years at average to moderate risk of breast cancer Setting: outpatient and primary care in the USA Intervention: audit and feedback to support SDM Comparison: audit (control group)
Outcomes	№ of participants (studies)	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Anticipated absolute effects* (95% CI)
				Risk with control intervention	Risk difference with audit and feedback to support SDM
Satisfaction with the decision‐making process	Not measured
Confidence in the decision made	Not measured
Knowledge of all the options Assessed with: questionnaire of adequate knowledge; more women with adequately responded questionnaires indicates greater knowledge Follow‐up: 12 months	133 (1 RCT)	⊕⊝⊝⊝ VERY LOWa,b	RR 1.18 (0.61 to 2.28)	313 per 1000; age to start screening	56 more per 1000 (122 fewer to 401 more)
			RR 0.84 (0.68 to 1.04)	906 per 1000; frequency of testing	145 fewer per 1000 (290 fewer to 36 more)
Adherence to the chosen option	Not measured
Choice of an option aligned with each woman’s values and preferences	Not measured
Woman's involvement	Not measured
Mental health outcomes	Not measured
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio; SDM: shared decision‐making
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded two levels due to severe imprecision: this estimate is based on a single study, which was adjusted by clustering and was underpowered for the outcome. The confidence interval shows substantial benefits and harms or no effect crossing both lines for minimally important differences (RR reduction or increase of 25%).
^bDowngraded one level due to concerns about bias: we identified some concerns about the randomisation process, deviation from intended interventions, and bias in the selection of the reported result.

Summary of findings 2. Some components of shared decision‐making, including clarification of values and preferences, versus control.

Some components of SDM, including clarification of values and preferences, versus control for women eligible for breast cancer screening
Patient or population: women aged 38 to 71 years at average to moderate risk of breast cancer Setting: outpatient and community setting in Europe and Australia Intervention: using decision aids with clarification of participant values and preferences Comparison: no use of a decision aid (control group)
Outcomes	№ of participants (studies)	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Anticipated absolute effects* (95% CI)
				Risk with control intervention	Risk difference with decision aid (with clarification of values and preferences)
Satisfaction with the decision‐making process	Not measured
Confidence in the decision made Assessed with: Decisional Conflict Scale Scale from 0 to 100 (higher scores indicate more conflict) Follow‐up: 1 to 3 months Assessed with: number of conflicted participants (SURE score ≤ 3, fewer cases indicate fewer participants with conflict) Follow‐up: 7 to 10 days	1714 (4 RCTs)	⊕⊕⊕⊝ VERY LOWa,b,c	MD −1.60 (−4.21 to 0.87)**	The mean confidence score ranged from 16 to 32.	1.60 lower (4.21 lower to 0.87 higher)**
	1001 (1 RCT)		RR 0.75 (0.56 to 0.99)	144 per 1000 were conflicted.	36 fewer per 1000 (63 fewer to 1 fewer)
Knowledge of all the options Assessed with: knowledge score (range 0 to 10, higher scores indicate better knowledge) Assessed with: informed choice (dichotomous composite of knowledge, attitudes, and intentions, higher rates of informed choice indicate greater knowledge) Follow‐up: 1 to 3 months	2114 (5 RCTs)	⊕⊕⊕⊝ LOWa,d	MD ranges from 0.47 to 1.44.***	The mean knowledge score was 6.	MDs in scores ranged from 0.47 to 1.44 higher.***
	2449 (4 RCTs)		RR 1.24 (0.95 to 1.63)	382 per 1000 made an informed choice.	92 more per 1000 (19 fewer to 241 more)
Adherence to the chosen option	Not measured
Choice of an option aligned with each woman’s values and preferences	Not measured
Woman's involvement	Not measured
Mental health outcomes Assessed with: State‐Trait Anxiety Inventory Scale from 20 to 80 (higher scores indicate greater anxiety) Follow‐up: 4 to 6 weeks Assessed with: women with worries about cancer; fewer participants indicate fewer cases of worry Follow‐up: 4 to 6 weeks	749 (2 RCTs)	⊕⊕⊕⊝ LOWa,b	MD 0.54 (−0.96 to 2.14)**	The mean anxiety score was 29.	0.54 higher (0.96 lower to 2.14 higher)**
	639 (1 RCT)		RR 0.88 (0.73 to 1.06)	383 per 1000 had worries about cancer.	46 fewer per 1000 (103 fewer to 23 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). **SMDs were re‐expressed to the scale for illustrative purposes using the baseline SD of the larger study in meta‐analysis (SD 14.5 for Decisional Conflict (Analysis 2.1), SD 1.57 for Knowledge (Analysis 2.4), and SD 10.72 for Mental Health (Analysis 2.7)). ***MDs across different scales (0 to 10, 0 to 7, 0 to 5) could not be pooled into SMDs because they contained changes from baseline and end values. We have presented the range of MDs across categories converted to a 0‐to‐10 scale. CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio; SD: standard deviation; SDM: shared decision‐making; SMD: standardised mean difference; SURE: Sure of myself; Understand information; Risk‐benefit ratio; Encouragement
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level due to risk of bias: most included studies were at an overall high risk of bias.
^bDowngraded one level due to imprecision: few events or participants that led to a wide confidence interval crossing minimally important differences (0.2 for a minimal effect of SDM and 25% reduction in RR).
^cDowngraded one level due to inconsistency: high statistical heterogeneity. 
^dDowngraded one level due to inconsistency: high statistical heterogeneity. We did not downgrade due to imprecision, considering that the wide confidence intervals can be mostly attributed to heterogeneity.

2.1. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 1: Confidence ‐ decisional conflict ‐ continuous

2.4. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 4: Knowledge ‐ continuous

2.7. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 7: Anxiety ‐ continuous

Summary of findings 3. Studies focused only on enhanced communication, without clarification of values and preferences, versus control.

Shared decision‐making (all components) versus control for women eligible for breast cancer screening
Patient or population: women aged 40 to 75 years old at average to moderate risk of breast cancer Setting: outpatient and primary care setting in Europe and Australia Intervention: enhanced communication, without clarification of values and preferences (e.g. enhanced leaflets or decision aids) Comparison: control
Outcomes	№ of participants (studies)	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Anticipated absolute effects* (95% CI)
				Risk with control intervention	Risk difference with enhanced communication (without clarification of values and preferences)
Satisfaction with the decision‐making process	Not measured
Confidence in the decision made Assessed with: confidence (Decisional Conflict Scale 0 to 100; higher scores indicate higher conflict) Assessed with: anticipated regret (1 to 5; higher scores indicate higher regret) Follow‐up: 2 weeks	1191 (2 RCTs)	⊕⊕⊝⊝ LOWa	MD 2.89 (−2.35 to 8.14)**	The mean score was 12.2.	2.89 points higher (2.35 lower to 8.14 higher)**
			Lower anticipated regret if not screening (MD −0.28, 95% CI −0.42 to −0.14) Higher anticipated regret if screening (MD 0.28, 95% CI 0.15 to 0.41)
Knowledge of all the options*** Assessed with: knowledge score (range 0 to 10; higher score indicates better knowledge) Assessed with: informed choice (dichotomous composite of knowledge, attitudes and intentions; higher rates of informed choice indicate greater knowledge) Follow‐up: 1 month	2510 (4 RCTs)	⊕⊕⊕⊕ HIGH	MD 1.14 (0.61 to 1.62)**	The mean knowledge score was 4.	1.14 points higher (0.61 to 1.62 higher)**
	1805 (2 RCTs)	⊕⊕⊝⊝ LOWa,b	RR 1.27 (0.83 to 1.92)	542 per 1000 made an informed choice.	146 more per 1000 (92 fewer to 458 more)
Adherence to the chosen option	Not measured
Choice of an option aligned with each woman’s values and preferences	Not measured
Woman's involvement	Not measured
Mental health outcomes Assessed with: Anxiety: State Trait Anxiety Inventory, 20 to 80; higher scores indicate higher anxiety Depression: Hospital Anxiety and Depression Scale, range 0 to 21 Follow‐up: 2 weeks to 1 month Assessed with: Cancer Worry Scale (1 to 4; higher scores indicate greater worry) Follow‐up: 2 weeks to 1 month	1193 (2 RCTs)	⊕⊕⊕⊕ HIGH	MD 0.33 (−1.55 to 0.99)** MD 0.02 (−0.41 to 0.45)**	The mean score for anxiety was 29.71. The mean score for depression was 0.73.	0.33 points higher (1.55 lower to 0.99 higher)** 0.02 points higher (0.41 lower to 0.45 higher)**
	838 (1 RCT)	⊕⊕⊕⊕ HIGH	MD −0.17 (−0.26 to −0.08)	The mean cancer worry score was 1.67.	0.17 points lower (0.26 lower to 0.08 lower)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). **SMDs were re‐expressed to the scale for illustrative purposes using the baseline SD of the larger study in the meta‐analysis (SD of 18.09 for Confidence (Analysis 3.1), SD of 4.06 for Knowledge (Analysis 3.3), SD 11.08 for Anxiety and SD 1.96 for Depression (Analysis 3.7)). ***Three studies could not be incorporated into meta‐analyses, and their results were summarised in forest plots indicating the direction and magnitude of effects and risk of bias (Analysis 3.6). Their results were mostly consistent with those summarised in this table. CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio; SD: standard deviation; SMD: standardised mean difference
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded two levels due to inconsistency across study results.
^bDowngraded one level due to imprecision (the confidence interval includes substantial benefits and harms).

3.1. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 1: Confidence

3.3. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 3: Knowledge ‐ continuous

3.7. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 7: Anxiety and depression

3.6. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 6: Knowledge ‐ dichotomous (correct answers)

Background

Description of the condition

Breast cancer is the most commonly diagnosed cancer among women (Bray 2018). The incidence rate of breast cancer increases with age, peaking at around 60 years (Winters 2017). In addition to age, other risk factors include race, family history, age at first menstruation, reproductive patterns, breast characteristics, hormone use, and lifestyle factors, such as weight, diet, and alcohol and tobacco use. Some women may inherit a faulty gene, such as a mutation in the breast cancer susceptibility genes (BRCA1 DNA repair associated (BRCA1) and BRCA2 DNA repair associated (BRCA2)), which explains approximately 3% of breast cancers (Winters 2017). By 70 years of age, the risk of breast cancer is about 65% for women with BRCA1 and 45% for women with BRCA2 gene mutations (Winters 2017).

Early detection through breast cancer screening programmes reduces breast cancer mortality and the incidence of advanced breast cancer, which is why it is a widely recommended practice throughout the world (Klarenbach 2018; Siu 2016). However, a mammogram can look normal even though breast cancer is present (a false negative), or look abnormal when the disease is not there (a false positive). A false‐negative result can give a false sense of reassurance, and a false‐positive result requires additional testing to determine that the abnormal finding is not cancer, leading to unnecessary examinations and anxiety (Seely 2018). Most additional tests are images of the breast, but sometimes more invasive procedures, such as biopsies, are needed. Mammographic screening has been shown to be associated with an increased incidence of small breast cancers (less than 2 cm), but not with a decreased incidence of large breast cancers, nor a substantial and proportional decline in mortality rates (Autier 2011; Bleyer 2012; Harding 2015). This population‐level phenomenon is described as overdiagnosis: a diagnosis (in this case, cancer) that would never have caused harm if left undetected (Autier 2017). For breast cancer, this may be partially due to increased detection of ductal carcinoma in situ (DCIS), a non‐invasive earlier form of breast cancer (Harding 2015; van Seijen 2019). Overdiagnosis may lead to harm due to overtreatment (treatment of a diagnosis that would never have caused harm), and the psychological burden of carrying a disease label (Treadwell 2016).

The available evidence regarding breast cancer screening has limitations and contradictions. Trials might not reflect the current advancements in screening technology and treatments for breast cancer, both of which can affect the estimates of the benefits and harms of screening. Furthermore, screening strategies vary according to a woman's breast cancer risk. It is recommended that women at high risk of breast cancer due to a family history of breast, ovarian, tubal, or peritoneal cancer are assessed if they are at risk of hereditary BRCA1‐ or BRCA2‐related breast and ovarian cancer (Nelson 2019). Women at high risk therefore face different screening decisions, for example regular or intensive screening and BRCA mutation testing (if eligible), than women at average risk.

Description of the intervention

Shared decision‐making (SDM) is a process in which a healthcare team works with a person to make decisions based on the best available evidence, and the person's values and preferences (Elwyn 2012; Makoul 2006; Weston 2001). This process helps to reduce the information and power imbalance of the doctor‐patient relationship and to increase the person's autonomy and control when deciding about their health (Charles 1997; Coulter 2011).

SDM becomes especially relevant for making healthcare decisions when more than one reasonable option is available, and people may value the benefits and harms differently. The two core elements of SDM are risk communication and values clarification, that is elucidating what is meaningful to the person and their family (Grad 2017). In relation to risk communication, it is estimated that "for every 10,000 UK women invited to screening from the age of 50 years for 20 years, about 681 cancers will be discovered, of which 129 will represent overdiagnosis, and 43 deaths from breast cancer will be prevented" (p 1784; UK Panel on Breast Cancer Screening 2012). With regard to values clarification, the evidence shows that women's values and preferences about breast cancer screening can vary when presented with complete information about the benefits and harms of screening (Pillay 2018). Some studies show that most women are unaware of the harms of screening, such as overdiagnosis and overtreatment, but they value information about it and can understand it, highlighting the importance of discussing these issues (Hersch 2013; Nagler 2017). In this context, clinical practice guidelines incorporate SDM as a suggested approach to support women's decisions about breast cancer screening in women with average risk (Keating 2018; Oeffinger 2015). The Canadian Task Force on Preventive Health Care included an assessment of women's values and preferences about breast cancer screening to inform their recommendations, suggesting that SDM be incorporated in discussions with all women aged 40 years to 74 years (Klarenbach 2018).

Scholl 2011 proposed that instruments for measuring SDM be classified into three categories. First, instruments assess what one considers before making a decision, by measuring personal preferences, confidence and attitudes, or preparing to be active participating in decision‐making, such as the Control Preference Scale. Second, instruments capture the decision process, such as the Observing Patient Involvement in Decision Making (OPTION) scale, or the 9‐item Shared Decision Making Questionnaire (SDM‐Q‐9). Third, instruments evaluate decision outcomes, such as the Decisional Conflict Scale, the Decision Regret Scale, or the Sure of myself, Understand information, Risk‐benefit ratio and Encouragement (SURE) scale. These tools measure the process from the clinicians' and the patients' perspectives and include observational or self‐rated data. Many tools have been developed and tested in languages other than English (Scholl 2011).

How the intervention might work

SDM offers a framework for providing person‐centred care, which is considered one of the fundamental approaches for improving healthcare quality (Barry 2012; Schrager 2017). Many people report that they wish to be more involved in making decisions about their health (Caress 2005; Sleath 2011). Moreover, evidence suggests that people who become engaged in their healthcare decisions have better results, and experience more satisfaction with the overall care experience than those who take a more passive role (Légaré 2007; Makoul 2006). The potential benefit of SDM is dependent on the willingness and ability of all parties to interact. This ability might depend on factors such as ethnicity, literacy, understanding of health concepts (health literacy), and numeracy (Wigfall 2018). As such, SDM will not necessarily be equally acceptable to all people or caregivers, and its application may vary across healthcare contexts. Decision aids are tools that can complement counselling from a health practitioner and help people become involved in the different stages of SDM. For example, decision aids can help make the decision explicit, provide information about the risks and benefits of each option and how often these occur, and clarify personal values (OHRI).

Why it is important to do this review

There appears to be a gap between women's and clinicians' perspectives on SDM in breast cancer screening. Women expressed concern that they did not receive information about potential false alarms or callbacks that happen with detection, while clinicians considered that they had provided such information (DuBenske 2017). Both parties want women to understand the potential risks, but health professionals acknowledge that they lack the skills to communicate probabilities of the benefits and harms of screening (DuBenske 2017). Statistical illiteracy of clinicians and patients has also been reported as a barrier to making informed choices in cancer screening (Wegwarth 2018).

SDM is increasingly becoming a recommended approach in clinical guidelines (Keating 2018; Klarenbach 2018; Oeffinger 2015). SDM is not limited to using decision aids (Légaré 2014); other SDM interventions, such as coaching or question prompting, may improve knowledge and patient satisfaction (Shepherd 2016; Stacey 2012). However, their effects on decisional conflict in breast cancer screening, and adherence to the chosen option, remain uncertain.

A systematic review focusing on the comprehensive process of SDM in breast cancer screening, with or without the use of decision aids, will allow a better understanding of its effects in this clinical situation. We hope to gather evidence about populations with different social determinants of health that can impact decision‐making. It is recognised that there might be language and cultural barriers to implementing and measuring SDM with reliable and valid tools (Scholl 2011). This systematic review will contribute to the growing body of evidence focusing on SDM interventions for different health conditions, which inspired our approach on this topic (Duncan 2010; Kew 2017).

Objectives

To assess the effect of shared decision‐making on women's satisfaction, confidence, and knowledge when deciding whether to participate in breast cancer screening.

Methods

Criteria for considering studies for this review

Types of studies

We included parallel randomised controlled trials (RCTs) and cluster‐RCTs.

Because SDM in breast cancer screening is a relatively recent research field, we planned to include a wide range of study designs to increase the amount of relevant data. Furthermore, SDM may rely on the organisational structures of the healthcare system, and it is likely that, in some instances, it would not be feasible for trial designs to use random allocation of individual participants within healthcare facilities. Although we planned to include quasi‐randomised controlled trials (q‐RCTs), controlled before‐after studies (CBAs), and interrupted time series (ITS), this was not required given the RCTs identified on this topic. We included studies irrespective of their publication status or the language of publication.

We excluded single‐arm studies.

Types of participants

We included studies of asymptomatic women with an average and above‐average risk of breast cancer, facing the decision to participate in breast cancer screening within the usually recommended ages (e.g. 40 to 75 years old).

We included women with above‐average risk (Warner 2011), defined as:

• women with a lifetime risk of 15% to 20% (according to risk models; Evans 2007);

• women with a five‐year risk above 1.66% (e.g. those with a first‐degree relative who had breast cancer before the age of 65 years or those with a previous breast biopsy specimen showing atypical hyperplasia or lobular carcinoma in situ).

We excluded women at high risk of breast cancer (Saslow 2007), defined as:

• women with a breast cancer susceptibility (BRCA) gene mutation;

• personally untested women with a first‐degree relative who is a BRCA carrier;

• women with a lifetime risk of breast cancer above 20% or 25% (according to risk models; Evans 2007);

• women who received radiation to the chest between the ages of 10 years and 30 years;

• women with Li‐Fraumeni syndrome, or Cowden and Bannayan‐Riley‐Ruvalcaba syndromes, and their first‐degree relatives.

If we found studies that included women at both average risk and high risk of breast cancer, we attempted to gather disaggregated data for the former group from the study reports or by contacting the study authors. If we could not obtain this information, we presented the findings of these studies separately.

Types of interventions

SDM interventions include four key components: (i) two or more participants (one of them being a woman facing a breast cancer screening decision), (ii) information must be shared between participants, (iii) both parties must participate in the decision‐making process (discussing values and preferences), and (iv) a decision must be made or actively deferred (Charles 1997). One of the participants is usually a healthcare professional. The process may or may not include the use of decision aids.

We included interventions that aimed to increase the degree of SDM between a woman and a healthcare professional. These could be aimed at healthcare professionals, women and their families, caregivers, or both.

We included studies focused on enhanced communication strategies or shared information, provided the aim was to help women make decisions about breast cancer screening. However, as this is one component of SDM, we differentiated studies that included all four criteria suggested by Charles 1997 from those that had fewer than four.

We included studies that compared the intervention against usual care or another intervention that did not have SDM components.

Types of outcome measures

The outcomes measured in studies did not form part of the criteria for study inclusion. We chose the primary and secondary outcomes based on previously reported and known effects of SDM and decisions aids in breast cancer screening (Lillie 2014; Martínez‐Alonso 2017), and a core set of domains for measuring the effectiveness of SDM interventions developed by healthcare professionals and participants (Toupin‐April 2019).

Primary outcomes

These included:

• satisfaction with the decision‐making process, measured with a validated scale, such as the Satisfaction With Decision (SWD) scale (Holmes‐Rovner 1996), or another validated or commonly used scale;

• confidence in the decision made, measured with a validated scale, such as the Decisional Conflict Scale (DCS; O'Connor 1995), and SURE (Sure of myself; Understand information; Risk‐benefit ratio; Encouragement) (Ferron 2013), or another validated or commonly used scale;

• knowledge of all options and their potential benefits and risks, measured with scales developed by investigators (Gummersbach 2015; Mathieu 2007; Stager 1993), with the Multidimensional Measure of Informed Choice (MMIC) scale (Michie 2002; van Agt 2012), which combines knowledge, attitudes, and uptake/behaviour/intentions about undergoing the test (Michie 2003), or another validated or commonly used scale.

The primary outcomes encompassed the main possible benefits and harms of SDM, that is dissatisfaction or satisfaction, low or high confidence, and poor or good knowledge regarding the presented options.

Secondary outcomes

• Adherence to the chosen option, measured with the framework proposed by Trenaman 2016, defined as the proportion of women who proceeded with the agreed decision at follow‐up. For instance, the proportion of women who chose to undergo screening and proceeded with it over the following two years; or conversely, the proportion of women who decided not to undergo screening and maintained this decision over the following two years.

• Women's involvement in the SDM process, measured with a validated scale, such as the 9‐item Shared Decision‐Making Questionnaire (SDM‐Q‐9; Kriston 2010), SDM‐Q‐Doc (from the physician's perspective; Scholl 2012), Observing Patient Involvement in Decision Making (OPTION) scale (Elwyn 2003), or another validated or commonly used scale.

• Woman‐clinician communication, measured with scales such as the Combined Outcome Measure for Risk communication And treatment Decision making Effectiveness (COMRADE), which includes other patient‐based outcomes not included in the SWD or DCS scales (Edwards 2003), or other validated or commonly used scales.

• Mental health outcomes, including anxiety, depression, stress and distress, measured with validated or other commonly used scales.

We assessed all outcomes at short‐term (< 12 months) and long‐term (≥ 12 months) follow‐up.

Search methods for identification of studies

Electronic searches

We searched the following sources from database inception to 8 August 2023, with no restrictions on the language of publication.

• The Cochrane Breast Cancer Group's (CBCG's) Specialised Register. Details of the search strategies used by the CBCG to identify studies and the procedure used to code references are outlined in the CBCG's module (breastcancer.cochrane.org/sites/breastcancer.cochrane.org/files/public/uploads/specialised_register_details.pdf). We extracted and considered trials with the keywords 'shared decision making, shared decision, decision support, decision aid, informed decision making, informed decision, informed choice, decision support techniques, patient participation, patient involvement, patient preference, patient choice, patient empowerment, patient values, and directive counselling' for inclusion in the review.

• Cochrane Central Register of Controlled Trials (CENTRAL; Issue 8 of 12, 2023) in the Cochrane Library, see Appendix 1.

• MEDLINE PubMed (1946 to 8 August 2023), see Appendix 2.

• Embase (Embase.com; 1974 to 8 August 2023), see Appendix 3.

• CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature) (1937 to 8 August 2023), see Appendix 4.

• PsycINFO (APA PsycNET; 1967 to 8 August 2023), see Appendix 5.

• The World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (apps.who.int/trialsearch/Default.aspx), see Appendix 6.

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov/), see Appendix 7.

Refer to the Appendices for detailed search strategies.

Searching other resources

We attempted to identify further studies from the reference lists of identified relevant studies or reviews. We searched abstracts from the following events:

• International Shared Decision Making Conference 2022 (www.isdm2022.com/; last accessed in June 2022 and website is no longer active);

• Society for Medical Decision Making: European and North American Conferences in 2020, 2021, and 2022 (smdm.org/meetings; last accessed in February 2023, and there were no Asia Pacific Conferences during this period).

Data collection and analysis

Selection of studies

Two review authors (PR and MVRY) independently screened the titles and abstracts of retrieved records for potential eligibility. We obtained a copy of the full article for each reference reporting a potentially eligible study. Two review authors (PR and MVRY) independently assessed the full texts, classifying them as included studies, excluded studies, studies awaiting classification, or ongoing studies, in accordance with the criteria for each provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022a). We attempted to contact study authors to provide additional information to determine eligibility as needed. Any disagreements were resolved through discussion or by consultation with a third review author (JVAF or KK).

We included studies irrespective of whether the data on outcomes were reported in a 'usable' way. There were no language restrictions during study selection, and we planned to translate studies in languages other than English. When a language was beyond the review team's expertise, we planned to use the Cochrane Engage platform for assistance. We documented reasons for the exclusion of ineligible studies that may have reasonably been expected to have been included in the review in the 'Characteristics of excluded studies' table.

We collated multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review. We documented the study selection process in sufficient detail to complete a PRISMA flow diagram (Page 2021). We used the online platform Covidence for this process (Covidence).

Data extraction and management

Two review authors (from PR, MVRY, NJS, or CAR) independently extracted the characteristics of the included studies using a standard data collection form that had been tested on at least one eligible study. Any disagreements were resolved through discussion or consultation with a third review author (JVAF or KK).

We collected the characteristics of the included studies in sufficient detail to populate a 'Characteristics of included studies' table, including the following information.

• Study design (as assessed by review authors)

• Number of study centres and location

• Study setting

• Date of study

• Participant eligibility criteria

• Number of participants by study and by study arm

• Number of withdrawals and reasons for withdrawal

• Participants' demographic details at baseline, such as age, ethnicity, and assessment of the risk of breast cancer

• Details about the experimental intervention and comparison:

  • details of the key components of SDM, as described by the study authors: a) two or more participants (one of them a woman facing a breast cancer screening decision and another individual, usually a healthcare professional), b) complete and balanced information must be shared between participants, c) both parties must participate in the decision‐making process (discussing values and preferences), and d) a decision must be made or actively deferred;

  • type of intervention: aimed at healthcare professionals, women and their families or caregivers, or both.

• Details about co‐interventions

• Details about the primary and secondary outcomes assessed: definition, method and timing of outcome measurement, and data on relevant subgroups

• Sources of study funding

• Conflicts of interest reported by the study authors

Two review authors (from PR, MVRY, NJS or CAR) independently extracted outcome data from the included studies. We noted in the 'Characteristics of included studies' table if outcome data were reported in an unusable format. Any disagreements were resolved by consensus or by involving a third review author (JVAF or KK). One review author (PR) copied data from the data collection form into RevMan software (RevMan 2024). We double‐checked that the data were entered correctly by comparing the study reports with the presentation of the data in the systematic review.

We extracted the most detailed numerical data that might facilitate similar analyses of included studies. For dichotomous outcomes, we collected the number of events and totals to populate a 2 x 2 table or effect estimates (e.g. odds ratios, risk ratios) with corresponding measures of variance if more detailed data were unavailable. For continuous outcomes, we collected the mean and standard deviation (SD), or the data needed to calculate these if they were unavailable.

We attempted to contact study authors at the end of the data extraction process if any relevant information was missing, including information needed to assess risk of bias.

Assessment of risk of bias in included studies

Randomised controlled trials and quasi‐randomised controlled trials

We assessed the risk of bias in each result using the Cochrane RoB 2 tool (Flemyng 2023; Higgins 2022b; Sterne 2019a). Two review authors (NJS and CAR) independently assessed five domains of bias for each result, considering the effect of being assigned to the intervention. These five domains were bias due to (1) the randomisation process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of the reported results. By collectively assessing the answers to signalling questions and supporting information, we formulated a domain‐level judgement of low risk, some concerns, or high risk of bias. These domain‐level judgements informed an overall risk of bias judgement for each result based on the least favourable assessment across all domains. Any discrepancies between review authors were resolved by discussion to reach a consensus, or by consulting a third review author (JVAF) if necessary. Where inadequate study details were provided, we contacted the study authors to obtain further information. We assessed the results of the following outcomes.

• Satisfaction with the decision‐making process (short‐ and long‐term results)

• Confidence in the decision made (short‐ and long‐term results)

• Knowledge of all the options (short‐ and long‐term results)

• Adherence to the chosen option (short‐ and long‐term results)

• Choice of an option aligned with each woman’s values and preferences (short‐ and long‐term results)

• Woman's involvement (short‐ and long‐term results)

• Mental health outcomes (anxiety; short‐ and long‐term results)

For cluster‐RCTs, we used the RoB 2 tool and added another domain specific to cluster‐RCTs, from the archived version of the tool (Domain 1b. ‘Bias arising from the timing of identification and recruitment of participants’, available at www.riskofbias.info/), with its corresponding signalling questions, following the guidance in Section 23.1.2 and Table 23.1.a in the Cochrane Handbook (Higgins 2022b).

We use a modified version of the RoB 2 Excel tool to manage the data supporting the answers to the signalling questions and risk of bias judgements (available at www.riskofbias.info/). All these data were made publicly available as supplementary material in the Open Science Framework platform.

Controlled before‐after studies and interrupted time series

We did not include controlled before‐after studies and interrupted time series in the current review given the identification of multiple RCTs. If needed, in future review versions, two review authors (PR and MVRY) will independently assess the risk of bias of these studies using the Cochrane tool for bias assessment in non‐randomised controlled studies, the Risk Of Bias In Non‐randomised Studies of Interventions (ROBINS‐I) tool (Sterne 2016; Sterne 2022b). We planned to assess the risk of bias of studies based on the following seven domains in the ROBINS‐I tool: bias due to confounding, bias in the selection of participants into the study, bias in classification of interventions, bias due to deviations from the intended intervention, bias due to missing data, bias in the measurement of outcomes, and bias in the selection of the reported result. We would have assessed the same results as we assessed for RCTs and q‐RCTs.

We planned to label each domain as critical risk, serious risk, moderate risk, low risk, or no information based on our risk of bias judgements for each included study. We would have used these domain‐level judgements to inform an overall risk of bias judgement for each result based on the least favourable assessment across all domains. Any discrepancies between review authors would have been resolved by discussion to reach a consensus, or by consulting a third review author (JVAF) if necessary. Where inadequate study details were provided, we would have contacted the study authors to obtain further information.

Measures of treatment effect

For dichotomous outcomes (e.g. adherence to the chosen option and choice of an option aligned with each woman’s values and preferences), we presented results as risk ratios (RR) with their associated 95% confidence interval (CI). If the studies reported odds ratio, we would have converted them to RR, as indicated in Section 6.4.1.1 of the Cochrane Handbook (Higgins 2022a).

For continuous outcome data (all remaining outcomes), we presented results as mean differences (MD) with their associated 95% CI when the eligible trials used the same instrument to measure a given construct, or standardised mean differences (SMDs) with 95% CIs if different measurement scales were used for the same outcome, and we were confident that they were sufficiently comparable. For SMDs, we re‐expressed the results and presented them as units of a familiar measure, as indicated in Section 15.5.3 of the Cochrane Handbook (Higgins 2022a). We prioritised postintervention (after the SDM conversation) over change from baseline measurements, as defined in Section 7.7.3.1 of the Cochrane Handbook (Higgins 2022a).

Unit of analysis issues

The unit of analysis was the individual woman. We included only the relevant arms where multiple trial arms were reported in a single trial. If two comparisons (e.g. intervention A versus usual care and intervention B versus usual care) were included in the same meta‐analysis, we would halve the control group to avoid double‐counting.

For cluster‐RCTs, we evaluated whether clustering was accounted for in determining the effective sample size; whether assessment for design effect was carried out; and whether the methods used in the analysis were appropriate to the cluster design. If we found an inappropriate analysis, as though individuals, rather than clusters, were randomised, we calculated effective sample sizes and adjusted the number of events and sample size according to the calculated design effect as described in Section 23.1.4 of the Cochrane Handbook (Higgins 2022c). These calculations are described in the footnotes of the analyses, including transformed data, and in Appendix 8.

Dealing with missing data

We performed an available‐case analysis. However, if data were missing, we proceeded as follows:

• contacted the original investigators to request missing data whenever possible;

• made explicit the assumptions of any methods used to deal with missing data, e.g. that the data were assumed to be missing at random, or that missing values were assumed to have a particular value, such as a poor outcome;

• performed sensitivity analyses to assess how sensitive results were to reasonable changes in the assumptions made;

• discussed the potential impact of missing data on the findings of the review in the Discussion section, as recommended in Section 16.1.2 of the Cochrane Handbook (Higgins 2022a).

We did not impute missing values. However, we calculated missing SDs from standard errors, exact P values, or CIs if these were available. Where possible, we conducted intention‐to‐treat analyses for both continuous and dichotomous data. When necessary, we approximated means and measures of dispersion from figures in the reports using Plot Digitalizer (Jelicic 2016). During data extraction, we paid particular attention to missing data due to inappropriate postallocation exclusions, such as due to adverse events, and to cross‐overs, where women did not receive their allocated intervention. We indicated contact with the original authors in the 'Notes' section of the characteristics of each study.

Assessment of heterogeneity

We assessed the presence of clinical heterogeneity across studies by examining the characteristics of the study population, interventions, comparisons, and outcome measurements from the extracted data (see Data extraction and management).

We identified inconsistency across studies through visual inspection of the forest plots to assess the overlap of CIs, and used the I² statistic to measure heterogeneity among the trials in each analysis (Higgins 2003). We interpreted the I² statistic as follows:

• 0% to 40%: might not be important;

• 30% to 60%: may represent moderate heterogeneity;

• 50% to 90%: may represent substantial heterogeneity;

• 75% to 100%: represents considerable heterogeneity.

If we identified substantial heterogeneity, we explored it through prespecified subgroup analysis.

Assessment of reporting biases

We attempted to obtain study protocols to assess the presence of selective outcome reporting.

Had we been able to pool at least 10 studies for a given outcome, we would have used funnel plots to investigate small‐study effects. As asymmetry in funnel plots may have several explanations besides publication bias, we would have interpreted the results carefully.

Data synthesis

Two review authors (PR and JVAF) analysed the data in RevMan (RevMan 2024). We pooled data from comparable studies groups following the recommendations in the Cochrane Handbook (Higgins 2022a). We undertook meta‐analyses only if we judged the participants, interventions, comparisons and outcomes to be sufficiently similar to ensure a clinically meaningful result. Unless good evidence showed homogeneous effects across studies of different methodological quality, we primarily summarised data using a random‐effects model (Deeks 2022).

When pooling of data was considered inappropriate, we presented the range of effects from the analyses on the summary of findings tables for illustrative purposes, following the guidelines for Synthesis Without Meta‐analysis (SWiM) in Chapter 12 of the Cochrane Handbook (Higgins 2022a).

We planned that when results were estimated for individual studies with low numbers of events (< 10 in total), or when the total sample size was less than 30 participants, we would report the proportion of events in each group, together with a P value, obtained from a Fisher’s exact test, instead of estimating and reporting risk ratios.

We included all studies in the primary analysis and explored the effect of bias through a sensitivity analysis.

We did not pool outcome data from different study designs.

Subgroup analysis and investigation of heterogeneity

Insufficient data precluded subgroup analyses. For future versions of this review, we plan to conduct the following subgroup analysis for the primary outcomes to examine possible interactions and possible sources of heterogeneity:

• risk of breast cancer: average or above‐average risk;

• targeted age: 40 to 49 years, 50 to 59 years, etc.;

• the setting of the intervention or mode of SDM delivery: during the consultation, web‐based, etc.;

• target audience: women, healthcare professionals, or both.

Sensitivity analysis

If sufficient studies were available, we determined the robustness of results by excluding studies:

• with an overall assessment of 'some concerns' or 'high risk' of bias;

• that are included with borderline definitions for SDM, e.g. the details of one of the key components of SDM is inadequate (see Data extraction and management).

Summary of findings and assessment of the certainty of the evidence

We created three summary of findings tables based on three categories of SDM:

• shared decision‐making, including all components of SDM, versus control;

• some components of SDM, including clarification of values and preferences, versus control;

• studies focused only on enhanced communication, without clarification of values and preferences, versus control.

We included in each table the following outcomes at short‐term follow‐up (< 12 months):

• satisfaction with the decision‐making process;

• confidence in the decision made;

• knowledge of all the options;

• adherence to the chosen option;

• choice of an option aligned with each woman’s values and preferences;

• woman's involvement;

• mental health outcomes (anxiety).

Two review authors (PR and MVRY) used the five GRADE considerations (overall risk of bias, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence as it relates to the studies that contributed data to the meta‐analyses for the prespecified outcomes. In the case of disagreement, they consulted a third review author (JVAF).

We used the methods and recommendations described in the Cochrane Handbook (Higgins 2022a), employing GRADEpro GDT software (GRADEpro GDT). We justified all decisions to upgrade or downgrade the certainty of the evidence using footnotes, and added comments to aid the reader's understanding of the review where necessary. We considered whether there was any additional outcome information that could not be incorporated into the meta‐analyses; indicated if this was the case in the comments; and stated if it supported or contradicted the information from the meta‐analyses.

We used the controlled vocabulary suggested by Santesso 2020 to summarise the findings of the summary of findings table in the Plain language summary.

Results

Description of studies

Results of the search

We identified 12,961 records through database searching and no additional documents through other sources. After removing duplicates, we screened the titles and abstracts of 9554 records and excluded 9427 studies. We assessed the full text of 127 records and excluded 82 studies (see Characteristics of excluded studies) and identified one ongoing study (see Characteristics of ongoing studies) and two studies awaiting classification (see Characteristics of studies awaiting classification). We included 19 studies in the review. Refer to Figure 1.

1.

Study flow diagram.

Included studies

Refer to Characteristics of included studies.

Study design and setting

We included 16 parallel‐group RCTs and three cluster‐RCTs (Elliot 2022; Pérez‐Lacasta 2019; Price‐Haywood 2014). Most were multicentre studies, and women were typically recruited through breast cancer screening programmes. Elliot 2022 recruited women with an appointment in primary care clinics; Hersch 2021 recruited women from electoral registries; and Seitz 2016 and NCT04741503 recruited women online, without specifying the setting. The studies were mainly conducted between 2005 and 2021, except for one study conducted between 1997 and 2000 (Rimer 2002).

One study was from a lower‐middle‐income country (Iran) (Akbari 2020). The remaining studies were from high‐income countries: Australia (Hersch 2021; Mathieu 2007; Mathieu 2010), France (Bourmaud 2016), Germany (Gummersbach 2015; Reder 2017), Italy (Giordano 2012; Roberto 2020), the Netherlands (Kregting 2020), Spain (Baena‐Cañada 2015; Pérez‐Lacasta 2019), and the USA (Elliot 2022; Haakenson 2006; NCT04741503; Price‐Haywood 2014; Rimer 2002; Schapira 2019; Seitz 2016).

We did identify any quasi‐randomised studies in our search.

None of the studies included a patient and public involvement statement or reported if research was actively carried out with patients or members of the public.

Participants

The studies included 64,215 women in total who were between 21 and 75 years old.

The majority of studies excluded high‐risk patients (i.e. with a previous breast cancer diagnosis or more than 20% lifetime risk), except for Gummersbach 2015, Haakenson 2006, and Schapira 2019. However, these high‐risk patients represented less than 4% of the total participants in each of these studies. Most studies selected women with average and moderate risk (e.g. due to first‐degree relatives with breast cancer); of these, most reported participants' risk factors (if present), except for Bourmaud 2016, Giordano 2012, Kregting 2020, and Mathieu 2007. Akbari 2020 and NCT04741503 excluded women with personal and family history of breast cancer (NCT04741503 also excluded women with atypical hyperplasia, BRCA1/BRCA2 mutation, and chest wall radiation therapy). Akbari 2020 did not clarify exclusion based on other risk factors, and did not report any data on women's breast cancer risk (i.e. to confirm if all were average or above‐average risk). Rimer 2002 and Elliot 2022 did not have exclusion criteria based on risk or report data about women's breast cancer risk in their study.

Seven studies reported women's ethnicity, with "white" being the most common category in five studies (Elliot 2022; Haakenson 2006; NCT04741503; Rimer 2002; Seitz 2016), and "black" being the most common category in two studies (Price‐Haywood 2014; Schapira 2019). None of the studies reported women's health literacy levels. All studies except three (Bourmaud 2016; Elliot 2022; NCT04741503) reported women's education level, with the majority having a high school education or higher.

Interventions and comparisons

The composition of SDM differed across studies (see Table 4 for a summary). While all studies included some form of enhanced communication of information on the benefits and harms of screening, we identified three main variations across studies, and categorised these studies into the following three groups.

1. Components of shared decision‐making across studies.

Studies	n	Characteristics of participants*	a) two or more participants...	b) ...share complete and balanced information...	c) ... and discuss values and preferences...	d) ... making or actively deferring a decision
All components of shared decision‐making
Price‐Haywood 2014	168	40 to 75 years; average‐moderate risk	YES	YES	YES	YES
Elliot 2022	27,599	21 to 74; no risk dataa	YES	YES	YES	YES
Some components of shared decision‐making, including clarification of values and preferences
Mathieu 2007	734	70 to 71 years; average‐moderate risk	NO	YES	YES	YES
Mathieu 2010	511	38 to 45 years; average‐moderate risk	NO	YES	YES	YES
Reder 2017	1206	50 years; average‐moderate risk	NO	YES	YES	YES
Roberto 2020	2119	> 45 and > 50 years; average‐moderate risk	NO	YES	YES	YES
Schapira 2019	204	39 to 48 years; average‐moderate‐high riskb	NO	YES	YES	YES
Studies focused only on enhanced communication without clarification of values and preferences
Akbari 2020	202	40 to 69 years; average risk	YES	YES	NO	YES
Baena‐Cañada 2015	434	45 to 67 years; average‐moderate risk	YES	YES	NO	YES
Bourmaud 2016	15844	50 to 74 years; average‐moderate risk	NO	YES	NO	YES
Giordano 2012	5649	40 to 45 years; average‐moderate risk	YES	YES	NO	YES
Gummersbach 2015	792	48 to 49 years; average‐moderate‐high riskc	NO	YES	NO	YES
Haakenson 2006	668	scheduled for screening mammogram; average‐moderate‐high riskd	NO	YES	NO	YES
Hersch 2021	879	48 to 50 years; average‐moderate risk	NO	YES	NO	YES
Kregting 2020	1312	49 to 75 years; average‐moderate risk	NO	YES	NO	YES
Pérez‐Lacasta 2019	524	49 to 50 years; average‐moderate risk	NO	YES	NO	YES
Rimer 2002	1091	40 to 44 and 50 to 54 years; no risk dataa	YES	YES	NO	YES
Seitz 2016	3955	35 to 49 years; average‐moderate risk	NO	YES	NO	YES

*Average risk: women with no additional risk factors for breast cancer. Moderate risk: the study included women with a family history of breast cancer, usually with an estimated 5‐year risk > 1.666% or lifetime risk of 15% to 20%.

^aThis study did not provide exclusion criteria based on estimated risk.
^b3.4% with high risk (> 20% lifetime risk).
^c< 20% with a family history of breast cancer, < 3% with a personal history of breast cancer.
^d52% with a family history, < 2% with a personal history of breast cancer.

• All components of SDM: two studies included all components of SDM, where healthcare professionals received SDM training or were provided with decision aids that included risk communication and elicitation of participants' values and preferences (Elliot 2022; Price‐Haywood 2014). Elliot 2022 measured adherence to breast cancer screening ("being up to date"), while Price‐Haywood 2014 evaluated some of the core SDM outcomes chosen for this review (knowledge).

• Some components of SDM, including clarification of values and preferences: six studies used decision aids targeted at participants prior to an encounter with a healthcare professional (Mathieu 2007; Mathieu 2010; NCT04741503; Reder 2017; Roberto 2020; Schapira 2019). Participants had to read the material alone, which provided balanced information about the benefits and risks of screening, included values clarification exercises, and prepared them to decide on participation in breast cancer screening.

• Studies focused on enhanced communication without clarification of values and preferences: none of these studies included a discussion or an instance for clarification of values and preferences. Seven studies included two components of SDM (i.e. shared complete balanced information and made or deferred a decision) (Bourmaud 2016; Gummersbach 2015; Haakenson 2006; Hersch 2021; Kregting 2020; Pérez‐Lacasta 2019; Seitz 2016). Four studies included three components because two participants shared the information (researcher, counsellor or health advisor with participants) (Akbari 2020; Baena‐Cañada 2015; Giordano 2012; Rimer 2002). Although they all provided balanced information, three studies focused on changing behaviour rather than making a decision (Akbari 2020; Rimer 2002; Seitz 2016).

Overall, three studies assessed the fidelity of the planned clinician‐participant (Elliot 2022; Price‐Haywood 2014) or researcher‐participant interaction (Baena‐Cañada 2015).

Outcomes

Few studies assessed the core outcomes of this review. Four studies did not report on any of the outcomes of this review (Akbari 2020; Bourmaud 2016; Elliot 2022; Giordano 2012).

None of the included studies reported satisfaction with the decision made. Seven studies assessed confidence (Gummersbach 2015; Hersch 2021; Mathieu 2007; Pérez‐Lacasta 2019; Reder 2017; Roberto 2020; Schapira 2019), which could be measured as decisional conflict, regret (Hersch 2021; Reder 2017), or anticipated regret (Hersch 2021; Pérez‐Lacasta 2019; Schapira 2019). Fourteen studies assessed knowledge (Baena‐Cañada 2015; Gummersbach 2015; Haakenson 2006; Hersch 2021; Kregting 2020; Mathieu 2007; Mathieu 2010; Pérez‐Lacasta 2019; Price‐Haywood 2014; Reder 2017; Rimer 2002; Roberto 2020; Schapira 2019; Seitz 2016). Seven studies measured informed choice, a composite outcome including knowledge, attitudes, and intentions (Hersch 2021; Kregting 2020; Mathieu 2007; Mathieu 2010; Pérez‐Lacasta 2019; Reder 2017; Roberto 2020).

Five studies measured mental health outcomes including anxiety, depression, and cancer worry (Baena‐Cañada 2015; Hersch 2021; Mathieu 2007; Pérez‐Lacasta 2019; Schapira 2019). None of the included studies assessed clinician‐participant communication, or adherence to the chosen option aligned with women's values and preferences.

Twelve studies assessed the participation rate in screening; however, this was not a predefined outcome for this review (Bourmaud 2016; Elliot 2022; Giordano 2012; Hersch 2021; Kregting 2020; Mathieu 2007; Pérez‐Lacasta 2019; Price‐Haywood 2014; Reder 2017; Rimer 2002; Roberto 2020; Schapira 2019).

Most studies assessed outcomes either immediately after the intervention or up to three months' follow‐up (Baena‐Cañada 2015; Gummersbach 2015; Haakenson 2006; Hersch 2021; Kregting 2020; Mathieu 2007; Mathieu 2010; NCT04741503; Pérez‐Lacasta 2019; Reder 2017; Roberto 2020; Schapira 2019; Seitz 2016). Three studies provided data at 12‐ and 24‐month follow‐ups (Hersch 2021; Price‐Haywood 2014; Rimer 2002).

Funding and conflicts of interest

Most studies reported funding from government agencies or other academic institutions, and some by private associations or foundations (Bourmaud 2016; Price‐Haywood 2014). Two studies did not report funding sources (Gummersbach 2015; Haakenson 2006).

Excluded studies

We excluded 82 studies for the following reasons:

• incorrect study design (28 studies): these studies focused on developing and uncontrolled testing of decision aids or other decision support systems, qualitative studies on women's views and preferences regarding breast cancer screening, and different observational designs that did not meet our inclusion criteria (Borrayo 2005; Davey 2005; DuBenske 2017; Eden 2015; Elkin 2017; Fechtelpeter 2019; Fredrick 2020; Geller 2007; Gibbons 2018; Goel 2011; Hersch 2014; Krist 2017; Lawrence 2000; Lippey 2022; Lo 2018; Mambourg 2018; Mann 2000; NCT04601272; Percefull 2020; Phillips 2018; Scariati 2015; Schonberg 2020; Stencel 2011; Tolma 2016; Ufomata 2016; Wegwarth 2018; Wong 2015; Wu 2013);

• ineligible patient population (5 studies): these studies included either women who were already regularly participating in screening, and decisions were on management, or women with breast cancer or at high risk of breast cancer (Bouton 2012; Bowles 2021; Lerman 1994; NCT00150917; NCT00247442);

• incorrect interventions (48 studies): these studies primarily focused on increasing the uptake of mammography through reminders or community‐based participatory interventions, including navigators or health promoters, or the use of different types of visual aids to promote screening (Allen 2014; Allgood 2016; Beauchamp 2020; Boling 2005; Bowen 2006; Bowen 2017; Coronado 2016; Curry 1993; del Junco 2008; Fagerlin 2005; Fernández‐Feito 2015; Fiscella 2011; Ghosh 2008; Goldzahl 2018; Haas 2019; Hurdle 2007; Kearins 2009; Kernohan 1996; Larkey 2012; Lewis 2003; Luckmann 2019; Molina 2018; Narasimmaraj 2016; NCT01336257; NCT02964234; NCT02986230; Orlando 2018; Petrova 2015; Reder 2018; Ruffin 2004; Russell 2007; Saver 2017; Saywell 2003; Schoenberg 2013; Segura 2001; Seven 2015; Shieh 2017; Sinicrope 2020; Slater 1998; Smith 2020; Stover 2017; Street 1998; Taylor 1999; Thompson 2002; Tobin 2022; Urban 1995; Wolosin 1990; Yang 2020);

• one terminated study (ISRCTN15366380).

Ongoing studies

One ongoing study assessed the effects of decision aids for breast cancer screening (NCT04948983). The expected completion date was October 2022.

Studies awaiting classification

We identified two studies for which we could not find further details on their status (NCT02914197; NCT03631758).

Risk of bias in included studies

Complete details of the risk of bias assessments per result can be found on the Open Science Framework website (osf.io/an4jp). We have summarised here the main characteristics per domain.

• Randomisation process: most included studies were at low risk of bias, except four studies that did not fully describe the randomisation process (Kregting 2020; Price‐Haywood 2014; Rimer 2002; Seitz 2016).

• Deviation from intended interventions: most results were at low risk of bias given that, although blinding was not thoroughly described across studies, women received their usual care except for the randomised intervention and were analysed according to randomisation. Only the two cluster‐RCTs indicated this might not be the case, but the impact is uncertain (Pérez‐Lacasta 2019; Price‐Haywood 2014).

• Missing outcome data: we deemed results from seven studies with substantial and sometimes unbalanced attrition of participants to be at high risk of bias (Baena‐Cañada 2015; Mathieu 2010; Pérez‐Lacasta 2019; Reder 2017; Roberto 2020; Schapira 2019; Seitz 2016). We deemed results that reported nearly null attrition (i.e. 95% of data were available for each result) to be at a low risk of bias. Although the results from one study were available for 95% of participants at two‐week follow‐up, more significant attrition of almost 20% at longer‐term follow‐up raised concerns about the risk of bias (Hersch 2021). Two other studies reported balanced attritions of between 10% and 20% and were judged as having 'some concerns' (Baena‐Cañada 2015; NCT04741503).

• Measurement of the outcome: we judged most results as at low risk of bias due to the use of well‐known valid tools or tools carefully developed and tailored to the research question, with adequate blinding of participants. In three studies, lack of blinding might have biased reporting by participants and data collection by researchers, therefore we deemed the results of these studies as at high risk of bias (Pérez‐Lacasta 2019; Reder 2017; Roberto 2020)

• Selection of the reported results: the results of seven studies were not adequately reported or prespecified in protocols or registries, and were therefore deemed as having 'some concerns' (Haakenson 2006; Kregting 2020; Mathieu 2007; Price‐Haywood 2014; Rimer 2002; Schapira 2019; Seitz 2016). The results from the remaining studies were prespecified in protocols or registries and were classified as at low risk of bias.

Effects of interventions

See: Table 1; Table 2; Table 3

1. Shared decision‐making, including all components, versus control

Two cluster‐RCTs included all SDM components. One study involving 168 women randomised 18 clinicians to SDM training (including risk communication and elicitation of participants' values and preferences) or usual care (audit of medical records). The study evaluated participants' outcomes (i.e. knowledge) after the clinical encounter (Price‐Haywood 2014). The second study randomised 34 primary care clinics to electronic health records linked with web‐based clinical decision support to promote all the steps of SDM for deciding on various cancer screening options; however, this study did not measure any outcome of interest to this review (Elliot 2022).

See Table 1.

1.1. Satisfaction with the decision‐making process

Not measured.

1.2. Confidence in the decision made

Not measured.

1.3. Knowledge of all options and their potential benefits and risks

Based on a single study, SDM training may result in little to no difference in participant knowledge regarding age at which to start screening (risk ratio (RR) 1.18, 95% confidence interval (CI) 0.61 to 2.28; 133 women) and frequency of being screened (RR 0.84, 95% CI 0.68 to 1.04; 133 women), but the evidence is very uncertain (Analysis 1.1; Figure 2). The certainty of the evidence is very low due to risk of bias and imprecision.

1.1. Analysis.

Comparison 1: Shared decision‐making (all components) versus control, Outcome 1: Knowledge

2.

1.4. Adherence to the chosen option

Not measured.

1.5. Women's involvement in the SDM process

Not measured.

1.6. Woman‐clinician communication

Not measured.

1.7. Mental health outcomes

Not measured.

2. Some components of SDM, including clarification of values and preferences, versus control

Six studies involving 3793 randomised women included some SDM components using decision aids or decision support tools targeted to women prior to an encounter with a healthcare professional (Mathieu 2007; Mathieu 2010; NCT04741503; Reder 2017; Roberto 2020; Schapira 2019). Women read the material alone, which provided balanced information about the benefits and risks of screening, included exercises clarifying values, and prepared them to decide on whether or not to participate in breast cancer screening.

See Table 2.

2.1. Satisfaction with the decision‐making process

Not measured.

2.2. Confidence in the decision made

Confidence

Five studies assessed confidence in women through their Decisional Conflict Scale (DCS) scores (Mathieu 2007; NCT04741503; Reder 2017; Roberto 2020; Schapira 2019).

Using decision aids and clarifying values and preferences may decrease DCS scores and women with residual conflict about the decision compared to women who did not receive such aids at one to three months' follow‐up (standardised mean difference (SMD) −0.11, 95% CI −0.29 to 0.06; 4 studies; 1714 women; Analysis 2.1; RR 0.75, 95% CI 0.56 to 0.99; 1 study; 1001 women; Analysis 2.2). The SD of the largest study was 14.5 (Mathieu 2007), and on a scale of 0 to 100, the SMD result represents a mean difference (MD) of −1.60 (95% CI −4.21 to 0.87; 4 studies; 1714 women). This finding is very uncertain as the evidence is of very low certainty due to risk of bias, inconsistency, and imprecision.

2.2. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 2: Confidence ‐ decisional conflict ‐ dichotomous

Anticipated regret and decisional regret

Two studies assessed regret before the decision (anticipated regret; Schapira 2019) or after the decision (decisional regret; Reder 2017) at six weeks to three months' follow‐up. We did not pool these results as they measured different concepts.

Regarding anticipated regret, there were little to no differences in the assessed scores about having a mammogram (MD 0.20, 95% CI −0.53 to 0.93; 1 study; 113 participants; Analysis 2.3.1; decisional regret scale ranged from 1 to 7) or delaying a mammogram (MD −0.3, 95% CI −0.97 to 0.37; 1 study; 113 participants; Analysis 2.3.1). Similarly, there was little to no difference between groups in regret after the decision had been made (MD −0.51, 95% CI −2.83 to 1.81; 1 study; 707 women; regret scale 0 to 100; Analysis 2.3.2).

2.3. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 3: Confidence ‐ regret/anticipated regret

2.3. Knowledge of all options and their potential benefits and risks

Based on five studies, using decision aids and clarifying values and preferences may increase knowledge of participants compared to control. A pooled effect estimate was not derived due to the use of different scales and change from baseline or end values across studies; however, the results of each study are visually displayed in Figure 3 (Analysis 2.4). The SMDs in scores ranged from 0.30 to 0.92 higher based on studies involving 2114 women. When these scores were converted to a 0‐to‐10 scale (using the SD of 1.57 of Mathieu 2007), the results corresponded to a range of MD 0.47 to 1.44 points. The certainty of the evidence is low due to risk of bias and inconsistency.

3.

Based on four studies, decision aids and clarifying values and preferences may result in higher rates of informed choice compared to control at one to three months' follow‐up (RR 1.24, 95% CI 0.95 to 1.63; 4 studies; 2449 women; Analysis 2.5). The certainty of the evidence is low due to the risk of bias and inconsistency. Inconsistency was primarily driven by a single study using a web‐based decision aid that measured the outcome immediately after the presentation of the information (Mathieu 2010), unlike the other studies, which assessed informed choice with some delay.

2.5. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 5: Knowledge ‐ informed choice (composite of knowledge, attitudes and intentions)

One study involving 1001 women assessed knowledge with 13 questions and found that there was a higher rate of correct answers for three questions and a lower rate for one in those receiving a web‐based standard brochure compared to control (Analysis 2.6) (Roberto 2020).

2.6. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 6: Knowledge ‐ dichotomous (correct answers)

2.4. Adherence to the chosen option

Not measured.

2.5. Women's involvement in the SDM process

Not measured.

2.6. Woman‐clinician communication

Not measured.

2.7. Mental health outcomes

Using decision aids and clarifying values and preferences may result in little to no difference in anxiety levels (SMD 0.05, 95% CI −0.09 to 0.20; 2 studies; 749 women; Analysis 2.7) or cancer worry (RR 0.88, 95% CI 0.73 to 1.06; 1 study; 639 women; Analysis 2.8) compared to the control group at four to six weeks' follow‐up. When the SMD result was converted to a scale from 20 to 80 (using the SD of 10.72 from Mathieu 2007), the effect corresponded to an MD of 0.54 (95% CI −0.96 to 2.14). The certainty of the evidence is low due to imprecision and risk of bias.

2.8. Analysis.

Comparison 2: Some components of shared decision‐making, including clarification of values and preferences, versus control, Outcome 8: Anxiety ‐ dichotomous

3. Studies focused solely on enhanced communication without clarification of values and preferences

Eleven studies did not include a discussion or an instance for women to clarify values and preferences. Seven studies included two components of SDM (i.e. sharing complete and balanced information and making or deferring a decision; Bourmaud 2016; Gummersbach 2015; Haakenson 2006; Hersch 2021; Kregting 2020; Pérez‐Lacasta 2019; Seitz 2016). Four studies covered three components because two participants shared the information (researcher, counsellor or health advisor with patients; Akbari 2020; Baena‐Cañada 2015; Giordano 2012; Rimer 2002). Although they all provided balanced information, three studies focused more on changing behaviour rather than making a decision (Akbari 2020; Rimer 2002; Seitz 2016). Three studies did not report any outcomes of interest to this review (Akbari 2020; Bourmaud 2016; Giordano 2012).

See Table 3.

3.1. Satisfaction with the decision‐making process

Not measured.

3.2. Confidence in the decision made

Three studies reported different measures of confidence, and results were not pooled due to conceptual differences (Gummersbach 2015; Hersch 2021; Pérez‐Lacasta 2019). These measures included confidence, anticipated regret, and decisional regret.

Confidence

Based on two studies, enhanced communication interventions without clarifying values and preferences may result in lower confidence in the decision compared with regular communication strategies at two weeks' follow‐up (SMD 0.16, 95% CI −0.13 to 0.45; I² = 81%; 2 studies; 1191 women; Analysis 3.1.2; when converted to the DCS from 0 to 100 using the SD from the largest study Hersch 2021 of 18.09, the SMD corresponded to an MD of 2.89, 95% CI −2.35 to 8.14). This result was based on a sensitivity analysis excluding a study with high risk of bias contributing to substantial inconsistency (Pérez‐Lacasta 2019). The certainty of the evidence is low due to important inconsistency resulting in imprecision.

Anticipated regret and decisional regret

Two studies reported anticipated regret. One study involving 838 women indicated that enhanced communication may result in a higher anticipated regret if women participated in screening (MD 0.28, 95% CI 0.15 to 0.41) and lower anticipated regret if women did not participate in screening (MD −0.28, 95% CI −0.42 to −0.14) compared to regular communications strategies at two weeks' follow‐up (Analysis 3.1; Hersch 2021). Similar results were found at two years' follow‐up: anticipated regret if women participated in screening (SMD 0.22, 95% CI 0.08 to 0.37) and anticipated regret if women did not participate in screening (SMD −0.16, 95% −0.31 to −0.01). The certainty of the evidence is low due to risk of bias and inconsistency (Analysis 3.1; Hersch 2021).

One study involving 400 women could not be included in the forest plot and indicated there was little to no difference in anticipated regret of participating in screening (P = 0.25) or not participating in screening (P = 0.73) (Pérez‐Lacasta 2019).

One study involving 790 women reported "decisional regret", but did not provide P values or SDs to calculate effect measures and only indicated that there was lower decisional regret at six months ("significant group difference"; only means presented in Analysis 3.1.4) and not at 12 months (Analysis 3.2.2).

3.2. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 2: Confidence ‐ long‐term follow‐up

3.3. Knowledge of all options and their potential benefits and risks

Enhanced communication interventions without clarifying values and preferences may increase participant knowledge compared to regular communication strategies at two to four weeks' follow‐up (SMD 0.28, 95% CI 0.15 to 0.40; I² = 54%; 4 studies; 2510 women; Analysis 3.3; Figure 4). When the result was converted to a scale from 0 to 10 based on the SD of the largest study (Hersch 2021 of 4.06), this corresponded to an MD of 1.14, 95% CI 0.61 to 1.62. One study provided 12‐ and 24‐month data for this outcome with similar results (MD 0.84, 95% CI 0.46 to 1.22 and MD 0.68, 95% CI 0.31 to 1.05, respectively). The certainty of the evidence is high as it is based on a sensitivity analysis excluding one study at high risk of bias that contributed to substantial inconsistency (Pérez‐Lacasta 2019).

4.

It is unclear whether enhanced communication interventions without clarifying values and preferences results in higher rates of informed choice compared to regular communication strategies at two to four weeks' follow‐up (RR 1.27, 95% CI 0.83 to 1.92; I² = 88%; 2 studies; 1805 women; Analysis 3.5.2). The certainty of the evidence is low due to inconsistency and imprecision (the CI includes substantial benefits and harms); we did not downgrade due to risk of bias since the analysis is based on a sensitivity analysis excluding one study at high risk of bias (Pérez‐Lacasta 2019).

3.5. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 5: Knowledge ‐ informed choice (composite of knowledge, attitudes and intentions)

Four studies could not be included in the meta‐analysis (Haakenson 2006; Kregting 2020; Rimer 2002; Seitz 2016). Three of these studies could not be incorporated due to the different types of outcome measures, but their results were summarised in forest plots indicating the direction and magnitude of effects and risk of bias (Analysis 3.6). One study involving 668 women assessed knowledge with 10 questions and found that there was a higher rate of correct answers for some questions in those women receiving enhanced communication versus the control group (Haakenson 2006). The second study, involving 988 women, found little to no difference in women with "sufficient knowledge" (≥ 8 on a 0‐to‐11 scale) (Kregting 2020). The third study, involving 717 women, found an increased rate of correct answers in two key questions at 12 and 24 months' follow‐up (Rimer 2002).

One study with eight arms could not be included in the meta‐analysis due to incomplete reporting; this study compared six forms of a brief intervention compared to no information and basic information (Seitz 2016). The brief intervention involved different models of communication to discuss screening in three different ways (expository, e.g. using facts, untailored examples and tailored examples) and adjusting the length of the communication (i.e. either brief or extension). Five of these six interventions (brief or extended untailored exemplars and extended expository) led to a greater improvement in knowledge of the woman's perceived risk of breast cancer compared to the control groups (P = 0.043; study at a high risk of bias).

3.4. Adherence to the chosen option

Not measured.

3.5. Women's involvement in the SDM process

Not measured.

3.6. Woman‐clinician communication

Not measured.

3.7. Mental health outcomes

Enhanced communication interventions without clarifying values and preferences result in little to no difference in anxiety scores compared to regular communication at two to four weeks' follow‐up (SMD 0.03, 95% CI −0.14 to 0.09; 2 studies; 1193 women; Analysis 3.7; Figure 5). When the result was converted to a scale from 20 to 80 using the SD of 11.08 from Hersch 2021, this corresponded to an MD of 0.33 (95% CI −1.55 to 0.99). The certainty of the evidence is high as it is based on a sensitivity analysis excluding one study at high risk of bias that contributed to substantial inconsistency (Pérez‐Lacasta 2019). One of the studies also reported similar results at 12 and 24 months (SMD −0.04, 95% CI −0.18 to 0.11; 746 women and SMD −0.10, 95% CI −0.25 to 0.04; 712 women, respectively; Analysis 3.8; Hersch 2021).

5.

3.8. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 8: Anxiety and depression ‐ long term

One study involving 355 women reported that enhanced communication results in little to no difference in depression (SMD −0.01 95% CI −0.20 to 0.22; Analysis 3.7). When the result was converted to the Hospital Anxiety and Depression Scale, range 0 to 21, this corresponded to an MD of 0.02, 95% CI −0.41 to 0.45.

Enhanced communication interventions without clarifying values and preferences reduce cancer worry compared to regular communication at two to four weeks' follow‐up (MD −0.17, 95% CI −0.26 to −0.08; Cancer Worry Scale range 1 to 4; 1 study; 838 women; Analysis 3.9; Figure 6) (Hersch 2021). The certainty of the evidence is high as it is based on a sensitivity analysis excluding one study at a high risk of bias that contributed to substantial inconsistency (Baena‐Cañada 2015). The study by Hersch 2021 also reported longer‐term data indicating little to no effect at 12 and 24 months' follow‐up (Cancer Worry Scale MD −0.12, 95% CI −0.20 to −0.04; 746 women and MD −0.05, 95% −0.14 to 0.04; 712 women, respectively; Analysis 3.10).

3.9. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 9: Cancer worry

6.

3.10. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 10: Cancer worry ‐ long term

Subgroup analyses

While there were sources of heterogeneity (case mix of baseline risk, different age ranges, and modes of delivery), too few studies per comparison precluded subgroup analyses. Most estimates had low statistical heterogeneity, except for one outcome in one comparison, which was explored through sensitivity analyses.

Sensitivity analyses

We performed sensitivity analyses based on SDM category, focusing solely on enhanced communication without clarifying values and preferences and outcomes of confidence, knowledge, and mental health. By excluding studies with some concerns or high risk of bias, the results were robust in all cases, and they provided a plausible explanation for the inconsistency found in the outcome knowledge, considering the differences in design and risk of bias in the excluded study (Pérez‐Lacasta 2019). We therefore incorporated the results from the sensitivity analyses in the main analyses and Table 3.

We did not perform the prespecified sensitivity analysis based on borderline definitions of SDM. Our categorisation of SDM components helped identify comparable studies based on their description.

Discussion

Summary of main results

This review included 19 studies with 64,215 women; however, most studies did not evaluate outcomes considered important for this review topic, and those that did measured different concepts. SDM was grouped into the following three main categories.

Shared decision‐making (all components) versus control

Based on a single cluster‐RCT recruiting 18 physicians and 168 women, SDM training may result in little to no difference in participant knowledge in terms of age to start screening and frequency of testing, but the results are very uncertain. Satisfaction with the SDM process, confidence in the decision made, adherence to the chosen option, women's involvement in the SDM process, woman‐clinician communication, and mental health outcomes were not measured.

Abbreviated forms of SDM with clarification of values and preferences versus control

This form of SDM may reduce conflict and risk of residual conflict about the decision compared to control at one to three months' follow‐up. This intervention may increase knowledge and may result in higher rates of informed choice compared to control at one to three months' follow‐up. This intervention may result in little to no difference in anxiety levels or risk of cancer worry compared to control at four to six weeks' follow‐up. Satisfaction with the SDM process, adherence to the chosen option, women's involvement in the SDM process, and woman‐clinician communication were not measured.

Enhanced communication about risks without clarification of values and preferences versus control

Enhanced communication interventions without clarifying values and preferences may result in lower confidence in the decision compared with regular communication strategies at two weeks' follow‐up. Findings from a single study indicated that enhanced communication may result in a higher anticipated regret of participating in screening and lower anticipated regret of not participating in screening compared to regular communications strategies at two weeks' follow‐up. These interventions increase knowledge compared to regular communication strategies at two to four weeks' follow‐up, while it is unclear if it results in higher rates of informed choice compared to regular communication strategies at two to four weeks' follow‐up. These interventions result in little to no difference in anxiety and depression and lower cancer worry compared to regular communication at two to four weeks' follow‐up. Satisfaction with the SDM process, adherence to the chosen option, women's involvement in the SDM process, and woman‐clinician communication were not measured.

Overall completeness and applicability of evidence

All of the included studies were conducted in high‐income countries, except for Akbari 2020, which was conducted in Iran, a lower‐middle‐income country. Nine studies were from English‐speaking countries (three from Australia (Hersch 2021; Mathieu 2007; Mathieu 2010), six from the USA (Elliot 2022; Haakenson 2006; Price‐Haywood 2014; Rimer 2002; Schapira 2019; Seitz 2016), two from Italy (Giordano 2012; Roberto 2020), two from Spain (Baena‐Cañada 2015; Pérez‐Lacasta 2019), two from Germany (Gummersbach 2015; Reder 2017), one from France (Bourmaud 2016), and one from the Netherlands (Kregting 2020)). These findings are consistent with a systematic review that analysed the inclusion and quality of SDM proposals in breast cancer screening clinical practice guidelines (Maes‐Carballo 2021). The guidelines that included SDM (about half of all those reviewed) were mainly from Europe and the USA.

SDM uptake in low‐middle‐income countries has been slow, which could explain the predominance of studies from high‐income countries included in this review. Language barriers affect implementation by limiting the use of decision aids and measurement scales developed in English‐speaking countries. Translation, transcultural adaptation, and validation of these tools are needed, which is a lengthier process than just validation (Ruiz 2019). But even if these tools are available, there are many other organisational and cultural barriers in low‐ and middle‐income countries. These barriers can include that the concept of SDM is still foreign; the paternalistic model is predominant; there are more fragmented healthcare systems; there are fewer to inexistent national policy initiatives; SDM is usually not on the research agenda with fewer funding opportunities; there are scarce opportunities for training of healthcare professionals and a lack of public awareness of patient‐centred care (Abbasgholizadeh‐Rahimi 2022; Alarcón‐Yaquetto 2022; Gogovor 2022; Lee 2022; Riganti 2022). These factors hamper SDM implementation and scalability, that is the possibility of moving beyond isolated implementation strategies fuelled by individuals interested in the topic to a widespread national initiative.

Regarding the components of the SDM intervention, two studies included the four key elements of SDM described in this review (Elliot 2022; Price‐Haywood 2014), and six had two or more participants involved in the decision‐making process, for example a healthcare professional or a researcher sharing complete and balanced information with a participant (Akbari 2020; Baena‐Cañada 2015; Elliot 2022; Giordano 2012; Price‐Haywood 2014; Rimer 2002). Many studies conducted in Australia and European countries (Italy, Germany, Spain, the Netherlands, France) recruited participants through their national breast cancer screening programmes. Having these programmes already in place facilitates the recruitment of women eligible for screening. The invitation letter can be adapted to include SDM components (e.g. share complete and balanced information). In such a scenario, women can read this information and decide to book an appointment independently. This might explain why most of the studies did not include conversations between participants and healthcare providers.

Over half of the included studies did not incorporate the clarification of values and preferences. Earlier studies were more focused on informed choice and providing balanced information, whereas more recent studies started to incorporate the discussion of values and preferences (from the 11 studies that did not include clarification of values and preferences, eight were published before 2016, and from the eight studies that did include it, five were published after 2017). Among the studies focusing on informed choice, three used behavioural change theories to inform the intervention design (Akbari 2020; Rimer 2002; Seitz 2016). Even though the description of their interventions focused on making a decision and fulfilled some of the SDM criteria that warranted their inclusion in this review, they used theoretical frameworks that correspond to motivational interviews, which focus on helping participants adopt a medically indicated healthcare behaviour (in this case, getting a mammogram) rather than making a healthcare decision (Elwyn 2014). With SDM, the preferred choice is the one that aligns with the participant's values and preferences after receiving complete and balanced evidence‐based information (either getting or not getting a mammogram).

Moreover, out of the six studies that involved a healthcare provider or a researcher communicating or presenting the information to the participant, three reported the fidelity of the intervention (Baena‐Cañada 2015; Elliot 2022; Price‐Haywood 2014). Fidelity pertains to the degree to which an intervention was delivered as intended, and is crucial to assessing its effectiveness. Agbadjé 2022 found that most studies that describe SDM interventions were incompletely reported, fidelity being one of the least reported aspects.

In recent years, more efforts have been made to clarify and reach a consensus on the components of SDM interventions. A recent publication defined the core elements of SDM for women considering breast cancer screening in the clinical setting through a Delphi survey (including women, healthcare providers, and healthcare decision scientists) (Croes 2020). These core elements are similar to those included in this review and consist of 22 items relating to information delivery and patient education (the content that needs to be included, e.g. pros and cons in a balanced matter), nine items pertaining to clinician‐patient communication (the way the information is transmitted, e.g. clear, checking patient understanding), and 13 items related to the framework of the decision (which includes clarification of values and preferences). Referring to these core elements, focusing on choice rather than behaviour change, and completely reporting all the components of the intervention could improve the consistency of SDM interventions in future studies.

Regarding SDM outcomes, knowledge was the most commonly reported outcome. There is no unique scale to measure knowledge (often designed and tailored to test the information provided in the interventions), which meant that our results were heterogeneous. Moreover, there was little to no difference in anxiety and cancer worry. However, it is important to consider that these mental health scales were designed and validated in the context of a cancer diagnosis and not for making screening decisions (which might carry different levels or types of anxiety and worry than decisions regarding breast cancer treatment and prognosis).

Furthermore, four studies did not incorporate any of the outcomes of interest to this review. One of these studies was Akbari 2020, who measured the stage of behavioural change (in line with motivational interview theory). The other three studies measured adherence to mammography screening (Bourmaud 2016; Elliot 2022; Giordano 2012), similar to other studies that did report outcomes relevant to this review (Hersch 2021; Kregting 2020; Mathieu 2007; Pérez‐Lacasta 2019; Price‐Haywood 2014; Reder 2017; Rimer 2002; Roberto 2020; Schapira 2019). None of the included studies measured adherence to the chosen option, choice of an option aligned with each woman's values and preferences, or women's involvement or satisfaction with the decision‐making process. These findings show that, other than confidence in the decision made, the included studies did not measure most of the outcomes focused on the participant perspective and what really matters to them (McCormack 2018).

It is also notable that none of the included studies reported information about patient or public involvement in the research process. In recent years, there has been an emphasis on promoting the description of patient and public involvement in research publications; however, its application into practice is still slow (Jones 2021).

Quality of the evidence

One of the main reasons for downgrading the certainty of the evidence was some concerns or high risk of bias in many of the results (see Risk of bias in included studies). We also identified imprecision across results, considering different thresholds for minimally important differences following a minimally contextualised approach (Zeng 2022). We did not downgrade the certainty of the evidence due to inconsistency, as there was a plausible explanation due to the inclusion of a cluster‐RCT at high risk of bias, which we explored through sensitivity analyses. While some studies included women with a higher risk of breast cancer, they were a minority population, therefore we did not downgrade due to indirectness. Finally, we could not formally assess publication bias due to the few studies per comparison.

Potential biases in the review process

The primary focus of this review was SDM, which includes a spectrum of older interventions with some components, and newer studies including all predefined elements. In the protocol, we specified a broad set of study designs, including experimental and quasi‐experimental studies. Nonetheless, we found RCTs and cluster‐RCTs primarily for interventions with some of the components of SDM, and few with all the components, and no quasi‐experimental studies. While observational studies can be a complementary source of information, we believe that high‐quality RCTs are feasible and essential in this research area, given the experimental intervention's characteristics.

We included all studies that incorporated at least some components of SDM, but we excluded those focusing on motivational interventions, reminders, or other approaches to improve mammography uptake. We believe that such studies address a different objective, in which the decision to have a mammography is already made. This narrows the scope and applicability of our findings, but we believe it reduces the heterogeneity across interventions, thereby increasing the validity of our synthesis.

We could not incorporate the results of four studies in our meta‐analyses due to substantial differences in outcome measures, and instead reported the results separately. Moreover, due to the few studies per comparison, we could not explore effect modification in our predefined subgroup analysis or assess publication bias with funnel plots. Four studies met our inclusion criteria but did not report any of our predefined outcome measures. We did not exclude these studies, as this is part of the Methodological Expectations of Cochrane Intervention Reviews (MECIR) (MECIR 2023, Standard C40).

Agreements and disagreements with other studies or reviews

The results of this Cochrane review were similar to other systematic reviews that evaluated the effectiveness of decision aids for treatment and screening decisions for breast cancer (Gao 2021; Martínez‐Alonso 2017; Stacey 2017; Yu 2020; Yu 2021). Stacey 2017 included some breast cancer genetic testing and treatment decisions, among many other health conditions (105 studies in total), with two studies about mammography screening (Mathieu 2007; Mathieu 2010). The review authors found a lack of results regarding adherence to the chosen option at that time, highlighting the need for further research. Four studies included in our review were conducted in 2017 or after (Akbari 2020; Elliot 2022; Kregting 2020; Roberto 2020); nevertheless, this outcome was not included in any of those studies.

The Gao 2021 systematic review included 22 studies about the effectiveness of decision aids in breast cancer decisions, of which five were about screening choices and three about mammography screening in average‐moderate‐risk women that were included in our review (Hersch 2015; Mathieu 2010; Pérez‐Lacasta 2019). Three other reviews that focused specifically on decision aids for breast cancer screening with mammography also found that SDM increased knowledge, reduced decisional conflict, and had little to no effect on other outcomes (Martínez‐Alonso 2017; Yu 2020; Yu 2021). These reviews restricted the intervention to mammography decision aids rather than SDM, hence including fewer studies. They also found significant heterogeneity in all assessment results, as the amount and type of information varied among different decision aids. Some of these studies included uncontrolled before‐and‐after studies that were not included in our review (Martínez‐Alonso 2017; Yu 2020). None of these four systematic reviews that focused on SDM and breast cancer decisions included the secondary outcomes specified in this Cochrane review (i.e. adherence to the chosen option, women's involvement in the SDM process, and woman‐clinician communication), and one systematic review included mental health outcomes (Gao 2021).

Authors' conclusions

Implications for practice.

The majority of shared decision‐making (SDM) interventions in this review involved abbreviated forms of SDM with values‐clarification exercises or enhanced communication. Using abbreviated forms of SDM may decrease decisional conflict, increase knowledge and informed choice, and have little to no effect on anxiety or cancer worry in the short term, but the certainty of the evidence is low. Enhanced communication without clarification of values and preferences may result in lower confidence in the decision, lower anticipated regret if not participating, and higher anticipated regret if participating in screening; it may increase knowledge and reduce cancer worry in the short term. It results in little to no difference in anxiety and depression. The certainty of the evidence varied from high to very low across outcomes and comparisons.

A limited number of studies incorporated all components of SDM (i.e. two or more participants, information must be shared between participants, both parties must participate in the decision‐making process discussing values and preferences, and a decision must be made or actively deferred). This means there is still uncertainty about the overall effect of SDM on supporting women's decisions about breast cancer screening.

Implications for research.

The concept and implementation of SDM has evolved, shifting from focusing on informed choice to incorporating clarification of values and preferences of participants. However, there are misconceptions or different views regarding its components, purpose (make a decision rather than change behaviour), and the outcomes that define it. The studies included in this review highlighted the diversity in defining SDM across studies.

High‐quality randomised controlled trials are required to evaluate the effectiveness of SDM beyond the use of decision aids and that measure outcomes more relevant to participants, such as satisfaction with the SDM process, adherence to the chosen option, women's involvement in the SDM process, and woman‐clinician communication. None of the included studies evaluated these outcomes. Further research is needed in low‐resource countries, low‐health literacy, and culturally diverse populations, and could include participants and members of the public in its design and implementation. Studies need to clearly describe the baseline breast cancer risk of the participants (the options being discussed and the impact on the outcomes might be different, e.g. cancer worry), who is the target of the intervention (professionals, participants, etc.), when it happens (e.g. prior or during consultation), what are the materials (e.g. web‐based, paper‐based leaflet, etc.), and to evaluate fidelity of the intervention (i.e. if SDM delivered as intended).

History

Protocol first published: Issue 12, 2020

Risk of bias

Risk of bias for analysis 1.1 Knowledge.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.1.1 Knowledge ‐ age to start screening

Price‐Haywood 2014

Some concerns

Allocation concealment was not described, although no concerns about recruitment and baseline characteristics were identified.

Some concerns

Participants (patients) were not blinded and no information is provided about deviations from the intervention (fidelity was not checked)

Low risk of bias

Attrition in the communication and audit‐only groups were 3 % and 8 %, respectively.

Low risk of bias

A valid reference was used to assess knowledge. While participants (patients) may have not been blinded, it is unlikely that this could have affected the result.

Some concerns

The outcome was well described in the methods but not pre‐specified.

Some concerns

There are some concerns related to the randomisation, deviation from interventions and selective outcome reporting.

Subgroup 1.1.2 Knowledge ‐ frequency of testing

Price‐Haywood 2014

Some concerns

Allocation concealment was not described, although no concerns about recruitment and baseline characteristics were identified.

Some concerns

Participants (patients) were not blinded and no information is provided about deviations from the intervention (fidelity was not checked)

Low risk of bias

Attrition in the communication and audit‐only groups were 3 % and 8 %, respectively.

Low risk of bias

A valid reference was used to assess knowledge. While participants (patients) may have not been blinded, it is unlikely that this could have affected the result.

Some concerns

The outcome was well described in the methods but not pre‐specified.

Some concerns

There are some concerns related to the randomisation, deviation from interventions and selective outcome reporting.

Appendices

Appendix 1. CENTRAL

#1 MeSH descriptor: [Breast Neoplasms] explode all trees

#2 (Breast near (cancer* OR tumor* OR tumour* OR neoplasm* OR carcinoma* OR malignan*)):ti,ab,kw:ti,ab,kw

#3 #1 OR #2

#4 MeSH descriptor: [Mass Screening] explode all trees

#5 (Screening*):ti,ab,kw

#6 MeSH descriptor: [Early Detection of Cancer] explode all trees

#7 ((Early AND (detection* OR Diagnos*) AND (tumor OR tumors OR Tumours OR Tumour OR Neoplasm* OR Cancer* OR carcinoma))):ti,ab,kw

#8 #4 OR #5 OR #6 OR #7

#9 MeSH descriptor: [Decision Making] explode all trees

#10 MeSH descriptor: [Decision Making, Shared] explode all trees

#11 MeSH descriptor: [Decision Support Techniques] explode all trees

#12 MeSH descriptor: [Directive Counseling] explode all trees

#13 (Shared Decision Making*):ti,ab,kw

#14 (decision aid* OR shared decision* OR decision support* OR informed decision making* OR informed decision* OR informed choice*):ti,ab,kw

#15 MeSH descriptor: [Patient Participation] explode all trees

#16 MeSH descriptor: [Patient Preference] explode all trees

#17 (SDM):ti,ab,kw

#18 (Patient AND (Preference* OR participation OR Empowerment OR involvement OR Value*)):ti,ab,kw

#19 (values and preferences):ti,ab,kw

#20 #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19

#21 #3 AND #8 AND #20

Appendix 2. MEDLINE PubMed

#1 "Breast Neoplasms"[Mesh]

#2 (Breast) AND (tumor[Tiab] OR tumors[Tiab] OR Tumours[Tiab] OR Tumour[Tiab] OR Neoplasm*[Tiab] OR Cancer*[Tiab] OR carcinoma[Tiab])

#3 #1 OR #2

#4 "Mass Screening"[Mesh]

#5 Screening*[Tiab]

#6 "Early Detection of Cancer"[Mesh]

#7 (Early[Tiab] AND (detection*[Tiab] OR Diagnos*[Tiab]) AND (tumor[Tiab] OR tumors[Tiab] OR Tumours[Tiab] OR Tumour[Tiab] OR Neoplasm*[Tiab] OR Cancer*[Tiab] OR carcinoma[Tiab]))

#8 #4 OR #5 OR #6 OR #7

#9 "Decision Making"[Mesh]

#10 "Decision Making, Shared"[Mesh]

#11 "Decision Support Techniques"[Mesh]

#12 "Directive counseling"[Mesh]

#13 Shared Decision Making*[Tiab]

#14 decision aid*[Tiab] OR shared decision*[Tiab] OR decision support*[Tiab] OR informed decision making*[Tiab] OR informed decision*[Tiab] OR informed choice*[Tiab]

#15 "Patient Participation"[Mesh]

#16 "Patient Preference"[Mesh]

#17 SDM[Tiab]

#18 Patient[Tiab] AND (Preference*[Tiab] OR participation[Tiab] OR Empowerment[Tiab] OR involvement[Tiab] OR Value*[Tiab])

#19 values and preferences[Tiab]

#20 #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19

#21 #3 AND #8 AND #20

Appendix 3. Embase (Embase.com)

#1,"'breast tumor'/exp"

#2,"breast tumor*:ti,ab OR breast tumour*:ti,ab OR breast neoplasm*:ti,ab OR breast cancer*:ti,ab OR breast carcinoma*:ti,ab"

#3,"#1 OR #2"

#4, 'mass screening'/exp"

#5,"screening*:ti,ab"

#6,"'early cancer diagnosis'/exp"

#7,"early AND (detection* OR diagnos*) AND (tumor OR tumors OR tumours OR tumour OR neoplasm* OR cancer* OR carcinoma:ti,ab)"

#8,"#4 OR #5 OR #6 OR #7"

#9 "'shared decision making'/exp"

#10 "'directive counseling'/exp"

#11,"'decision making'/exp"

#12,"'decision support system'/exp"

#13,"'shared decision making*':ti,ab"

#14,"'decision aid'/exp"

#15,"(decision aid*:ti,ab OR shared decision*:ti,ab OR decision support*:ti,ab OR informed decision making*:ti,ab OR informed decision*:ti,ab OR informed choice*:ti,ab)"

#16,"'patient participation'/exp"

#17,"'patient preference'/exp"

#18,"sdm:ti,ab"

#19,"patient AND (preference*:ti,ab OR participation:ti,ab OR empowerment:ti,ab OR involvement:ti,ab OR value*:ti,ab)"

#20,"'values and preferences':ti,ab"

#21 #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20

#22 #3 AND #8 AND #21

Appendix 4. CINAHL EBSCOhost

S22 S3 AND S8 AND S21

S21 (S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15 OR S16 OR S17 OR S18 OR S19 OR S20)

S20 TI (directive counseling OR decision support* OR shared decision* OR informed decision* OR informed choice* OR informed decision making* OR patient participation) OR AB (directive counseling OR decision support* OR shared decision* OR informed decision* OR informed choice* OR informed decision making* OR patient participation)

S19 TI ( values and preferences ) OR AB ( values and preferences )

S18 TI ( Patient AND (Preference* OR participation OR Empowerment OR involvement OR Value*) ) OR AB ( Patient AND (Preference* OR participation OR Empowerment OR involvement OR Value*) )

S17 TI SDM OR AB SDM

S16 (MM “Decision Making”) OR (MM “Decision Making, Clinical”) OR (MM “Decision Making, Patient”)

S15 MH patient preference

S14 MH patient participation

S13 TI decision aid* OR AB decision aid*

S12 TI Shared Decision Making* OR AB Shared Decision Making*

S11 (MH "Decision making, shared")

S10 MH decision support techniques

S9 MH decision making

S8 (S4 OR S5 OR S6 OR S7)

S7 TI ( (Early AND (detection* OR Diagnos*) AND (tumor OR tumors OR Tumours OR Tumour OR Neoplasm* OR Cancer* OR carcinoma)) ) OR AB ( (Early AND (detection* OR Diagnos*) AND (tumor OR tumors OR Tumours OR Tumour OR Neoplasm* OR Cancer* OR carcinoma)) )

S6 MH early detection of cancer

S5 TI Screening* OR AB Screening*

S4 MH Mass Screening

S3 (S1 OR S2)

S2 TI ( Breast tumor*) OR TI ( breast Tumour*) OR TI (breast Neoplasm*) OR TI (breast Cancer*) OR TI (breast carcinoma*) OR AB ( Breast tumor*) OR AB (breast Tumour*) OR AB ( breast Neoplasm*) OR AB (breast Cancer*) OR (AB breast carcinoma*)

S1 MH Breast Neoplasms

Appendix 5. PsycINFO APA PsycNET

((MeSH: (Decision Making)) OR (MeSH: (Decision Support Techniques)) OR (title: (Shared Decision Making*)) OR (abstract: (Shared Decision Making*)) OR (title: (decision aid*)) OR (abstract: (decision aid*)) OR (MeSH: (Patient Participation)) OR (MeSH: (Patient Preference)) OR (title: (SDM)) OR(abstract: (SDM)) OR (title: (Patient) AND (title: (Preference*) OR title: (participation) OR title: (Empowerment) OR title: (involvement) OR title: (Value*)) OR (title: (Decision Support)) OR title: (Directive Counseling)) OR title: (Choice*)) OR title: (shared decision*)) OR title: (decision support*)) OR title: (informed decision making*)) OR title: (informed decision*)) OR title: (informed choice*)) OR (abstract: (Patient) AND (abstract: (Preference*) OR abstract: (participation) OR abstract: (Empowerment) OR abstract: (involvement) OR abstract: (Value*)))OR (title: (values) AND title: (preferences)) OR (abstract: (values) AND abstract: (preferences))) OR (abstract: (Decision Support)) OR (abstract: (Directive Counseling)) OR (abstract: (Choice*)) OR abstract: (shared decision*)) OR abstract: (decision support*)) OR abstract: (informed decision making*)) OR abstract: (informed decision*)) OR abstract: (informed choice*)) AND ((MeSH: (Mass Screening)) OR (title: (Screening*)) OR(abstract: (Screening*)) OR (MeSH: (Early Detection of Cancer)) OR (title: (Early) AND (title: (detection*) OR title: (Diagnos*)) AND (MeSH: (Breast Neoplasms)) OR (title: (breast cancer*) OR title:(breast neoplasm*) OR title:(breast carcinoma*) OR title: (breast tumour*) OR title: (breast tumor*)) OR (abstract: (breast cancer*) OR abstract: (breast neoplasm*) OR abstract: (breast carcinoma*) OR abstract: (breast tumour*) OR abstract: (breast tumor*))

Appendix 6. WHO ICTRP

Basic Searches

breast cancer AND shared decision

Advanced Searches

1. Condition fields: breast cancer

Intervention fields: decision making OR decision‐making OR decision aid OR decision support
Recruitment status: All

2. Condition fields: Breast cancer

Intervention fields: informed decision OR informed choice OR Patient Participation OR Patient Preference

Recruitment status: All

Appendix 7. ClinicalTrials.gov

Basic searches

breast cancer AND shared decision

Advanced searches

1. Condition or disease: Breast cancer

Intervention/treatment: decision making OR decision‐making OR decision aid OR decision support

Study type: All types

2. Condition or disease: Breast cancer

Intervention/treatment: informed decision OR informed choice OR Patient Participation OR Patient Preference

Study type: All types

Appendix 8. Cluster trial adjustment

Design effect for adjustment of cluster trials: 1 + (M − 1)*ICC

M: cluster size
ICC: intracluster correlation coefficient

Pérez‐Lacasta 2019 (information from the trial)

M: 10

ICC: 0.1

Design effect: 1.9

Price‐Haywood 2014 (information from the trial)

M: 6.72

ICC: 0.1558

Design effect: 1.892

The details calculations on each result can be found in the footnotes of each analysis.

Data and analyses

Comparison 1. Shared decision‐making (all components) versus control.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Knowledge	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1.1 Knowledge ‐ age to start screening	1	70	Risk Ratio (M‐H, Random, 95% CI)	1.18 [0.61, 2.28]
1.1.2 Knowledge ‐ frequency of testing	1	70	Risk Ratio (M‐H, Random, 95% CI)	0.84 [0.68, 1.04]

Comparison 2. Some components of shared decision‐making, including clarification of values and preferences, versus control.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Confidence ‐ decisional conflict ‐ continuous	4	1714	Std. Mean Difference (IV, Random, 95% CI)	‐0.11 [‐0.29, 0.06]
2.1.1 Decisional conflict scale (0 to 100) ‐ higher scores/more conflict	3	1007	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.32, 0.17]
2.1.2 SURE scale (0 to 4)	1	707	Std. Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.36, ‐0.06]
2.2 Confidence ‐ decisional conflict ‐ dichotomous	1	1001	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.56, 0.99]
2.3 Confidence ‐ regret/anticipated regret	2		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.3.1 Anticipated regret	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.3.2 Regret	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.4 Knowledge ‐ continuous	5		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
2.4.1 Scale 0 to 10	2	1010	Std. Mean Difference (IV, Random, 95% CI)	0.92 [0.28, 1.56]
2.4.2 Scale 0 to 7	1	707	Std. Mean Difference (IV, Random, 95% CI)	0.30 [0.15, 0.44]
2.4.3 Scale 0 to 5	1	113	Std. Mean Difference (IV, Random, 95% CI)	0.45 [0.08, 0.83]
2.4.4 Scale 0 to 100	1	284	Std. Mean Difference (IV, Random, 95% CI)	0.06 [‐0.18, 0.29]
2.5 Knowledge ‐ informed choice (composite of knowledge, attitudes and intentions)	4	2449	Risk Ratio (M‐H, Random, 95% CI)	1.24 [0.95, 1.63]
2.6 Knowledge ‐ dichotomous (correct answers)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
2.7 Anxiety ‐ continuous	2	749	Std. Mean Difference (IV, Random, 95% CI)	0.05 [‐0.09, 0.20]
2.8 Anxiety ‐ dichotomous	1	639	Risk Ratio (M‐H, Random, 95% CI)	0.88 [0.73, 1.06]

Comparison 3. Studies focused only on enhanced communication, without clarification of values and preferences, versus control.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Confidence	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.1.1 Decisional conflict ‐ post intervention to 2 weeks	3	1329	Std. Mean Difference (IV, Random, 95% CI)	0.00 [‐0.32, 0.32]
3.1.2 Decisional conflict ‐ post intervention to 2 weeks ‐ sensitivity analysis	2	1191	Std. Mean Difference (IV, Random, 95% CI)	0.16 [‐0.13, 0.45]
3.1.3 Anticipated regret ‐ 2 weeks	1	1676	Std. Mean Difference (IV, Random, 95% CI)	0.01 [‐0.54, 0.55]
3.1.4 Decisional regret ‐ 6 months	1	790	Std. Mean Difference (IV, Random, 95% CI)	Not estimable
3.2 Confidence ‐ long‐term follow‐up	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected
3.2.1 24 months ‐ Anticipated regret	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected
3.2.2 12 months ‐ Decisional regret	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected
3.3 Knowledge ‐ continuous	5		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.3.1 Analysis with all included studies	5	2648	Std. Mean Difference (IV, Random, 95% CI)	0.54 [0.20, 0.87]
3.3.2 Sensitivity analysis (excluding studies at high risk of bias)	4	2510	Std. Mean Difference (IV, Random, 95% CI)	0.28 [0.15, 0.40]
3.4 Knowledge ‐ continuous ‐ long term	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.4.1 12 months	1	746	Mean Difference (IV, Random, 95% CI)	0.84 [0.46, 1.22]
3.4.2 24 months	1	712	Mean Difference (IV, Random, 95% CI)	0.68 [0.31, 1.05]
3.5 Knowledge ‐ informed choice (composite of knowledge, attitudes and intentions)	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
3.5.1 All studies	3	2016	Risk Ratio (M‐H, Random, 95% CI)	1.65 [0.87, 3.12]
3.5.2 Sensitivity analysis (excluding studies at high risk of bias)	2	1805	Risk Ratio (M‐H, Random, 95% CI)	1.27 [0.83, 1.92]
3.6 Knowledge ‐ dichotomous (correct answers)	3		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.6.1 Post intervention	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.6.2 2 weeks	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.6.3 12 month follow‐up	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.6.4 24 month follow‐up	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3.7 Anxiety and depression	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.7.1 Anxiety ≤ 1 month	3	1331	Std. Mean Difference (IV, Random, 95% CI)	‐0.02 [‐0.12, 0.09]
3.7.2 Anxiety ≤ 1 month ‐ sensitivity analysis	2	1193	Std. Mean Difference (IV, Random, 95% CI)	‐0.03 [‐0.14, 0.09]
3.7.3 Depression	1	355	Std. Mean Difference (IV, Random, 95% CI)	0.01 [‐0.20, 0.22]
3.8 Anxiety and depression ‐ long term	1		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
3.8.1 Anxiety ‐ 12 months	1	746	Std. Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.18, 0.11]
3.8.2 Anxiety ‐ 24 months	1	712	Std. Mean Difference (IV, Random, 95% CI)	‐0.10 [‐0.25, 0.04]
3.9 Cancer worry	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.9.1 All studies	2	959	Mean Difference (IV, Random, 95% CI)	‐0.46 [‐1.28, 0.37]
3.9.2 Sensitivity analysis	1	838	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.26, ‐0.08]
3.10 Cancer worry ‐ long term	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.10.1 Cancer worry ‐ 12 months	1	746	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.20, ‐0.04]
3.10.2 Cancer worry ‐ 24 months	1	712	Mean Difference (IV, Random, 95% CI)	‐0.05 [‐0.14, 0.04]

3.4. Analysis.

Comparison 3: Studies focused only on enhanced communication, without clarification of values and preferences, versus control, Outcome 4: Knowledge ‐ continuous ‐ long term

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Akbari 2020.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: December 2018 to April 2019 Setting and number of centres: Healthcare centres in Tabriz (not specified how many) Country: Iran
Participants	Inclusion criteria: Being within the age range of 40 to 69 years old Having no history of mammography, ultrasound, and computed tomography in the last 2 years Demonstrating no intention to perform the above‐mentioned actions in the next 30 days Having no history of breast clinical examination by a physician in the last year Representing no intention to perform such an examination in the next 30 days Having no history of breast cancer in the patient or first‐degree relatives Having speech and hearing ability Exclusion criteria: Being pregnant Having galactorrhoea Self‐reporting a mental and motor disability Sample size: 202 included (101 intervention, 101 control) Withdrawals: 5 in intervention arm (lost to follow‐up, reason not mentioned) 4 in control arm (lost to follow‐up, reason not mentioned) Age (years): Intervention: mean 48.5 (6.7 SD) and 49.9 (7.7 SD) Ethnicity: Not measured Data on risk of breast cancer at baseline: Measured (age, age of menarche and pregnancy, age of menopause, breastfeeding) but not reported
Interventions	Group intervention (n = 101): Decision‐aid‐based individual counselling was held by the researcher for the intervention group. The provided content included issues relating to the incidence of breast cancer among Iranian women and the risk of cancer development in terms of age, risk factors, early diagnosis techniques, breast control and mammography, as well as the advantages and disadvantages of screening tests. The intervention group was also asked to read a booklet and contact the researcher if they needed further information on breast cancer or decided to undergo screening. Group control (n = 101): Not reported Co‐interventions: None
Outcomes	None of the measured outcomes were relevant for this review.
Notes	Funding sources: Tabriz University of Medical Sciences Declarations of interest: None Other outcome measures: Stages of behavioural changes (checklist of the stages of behavioural changes), breast control behaviour, breast clinical examination behaviour, and mammography behaviour

Baena‐Cañada 2015.

Study characteristics
Methods	Study design: Randomised controlled trial Study dates: January 2011 and September 2012 (December 2013 in the protocol) Setting and number of centres: Residents of Cadiz‐La Janda Health District (invited to the BCS programme) (in protocol: Oncology Unit, Puerta del Mar University Hospital Cádiz) Country: Spain
Participants	Inclusion criteria (published protocol): Women living in the Bay Health District Cádiz‐La Janda Women aged between 45 and 69 years (manuscript says 67) Women invited to breast screening programme and who attend it Qualified women to give informed consent to participate in the study Exclusion criteria: Women with a history of breast cancer Last participation in mammography examination at 68 to 69 years old (mentioned in the manuscript, not in the protocol) Sample size: 434 randomised participants Withdrawals: 40 control group: non‐response to 2nd survey 1 month later 39 intervention group non‐response to 2nd survey 1 month later The analysis of worries about cancer was performed with only 62 control and 59 intervention group participants, because this scale was introduced in the study sometime after it had been initiated. Age (years): Group 1 intervention (mean 54, SD 6.8); Group 2 control (mean 54, SD 6.5) Ethnicity: Not available Data on risk of breast cancer at baseline: Family history of breast cancer Group 1 (65 total; direct (1st degree?): 23); Group 2 (56 total; direct: 27)
Interventions	Group 1 (n = 218): They received the same information as intervention group but did not receive verbal and written information based on the Cochrane leaflet. Group 2 (n = 216): Each woman assigned to the intervention group received precise verbal and written information on the benefits and risks of the screening programme. The intervention was no mere delivery of documentation; the contents of the Cochrane leaflet were verbally explained to the women, as is standard in informed consent procedures. This information was based on the 1st edition (2008) of the document created by the Nordic Cochrane Centre, Copenhagen (Denmark), a Spanish translation of which can be consulted on the following websites: www.screening.dk and www.cochrane.dk. The leaflet was translated by a native Spanish speaker, revised, and back‐translated verbally with a person to check validity. The document obtained a satisfactory readability score (the Flesch‐Kincaid Reading Ease score was 84.2) and gives quantitative information on the benefits of mammography screening (1 death from breast cancer will be avoided in every 2000 women submitted to screening by mammography in 10 years) and on the risks of the programme (10 women in every 2000 will be diagnosed and treated unnecessarily, and in every 200 a false‐positive result will be produced that will affect the woman psychologically). Additionally, information is offered on other possible harmful effects such as breast pain secondary to the compression of the breasts, exposure to ionising radiations, and the sensation of false security. Co‐interventions: None
Outcomes	Knowledge How measured: Adapted questionnaire (scale 0 to 10) from Health Decision Group of the Sydney School of Public Health, Australia, modified to take into account BCS benefits and risks described in Nordic Cochrane Centre based in Copenhagen, Denmark. 4 MCQ with 3 possible answers (1 point each), 3 free‐responses questions with 1 possible answer (2 points each). A score of 5 or more is considered "good knowledge". Time points measured: prior intervention Time points reported: 1‐month after intervention Anxiety/depression (Mental Health): How measured: Hospital Anxiety and Depression Scale (HADS) Time points measured: before randomisation Time points reported: 1‐month after intervention Cancer worry (Mental Health): How measured: Cancer Worry Scale (CWS) Time points measured: before randomisation Time points reported: 1‐month after intervention
Notes	Funding sources: Grant to Baena from the Andalusian Consejería de Salud y Bienestar Social SAS 111205. Declarations of interest: JNV worked as an epidemiologist in the Screening Program in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work. Other measured outcomes: Attitudes: assessed by 4 questions, each scoring between 0 (most positive) and 6 (most negative), for a maximum score of 24. The questions were as follows: “For me, BCS is (1) a good thing/bad thing, (2) beneficial/harmful, (3) important/not important, and (4) pleasant/unpleasant”. Scores of 13 or more denoted a negative attitude. The internal reliability of the attitude scale was acceptable (alpha coefficient 0.83). Decision made: participants whether they would participate again in the future, given the new information. Measures were as follows: “I have decided to participate”, “I have decided not to participate”, and “I am undecided”.

Bourmaud 2016.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: May to June 2009 Setting and number of centres: Outpatient, multicentre, national Country: France
Participants	Inclusion criteria: Women aged 50 to 74 years Exclusion criteria: Women already diagnosed with cancer Sample size: 15,844 randomised participants Withdrawals: Group 1 (DA): 115 women who proceeded to mammography before randomisation Group 2 (Control): 41 women who proceeded to mammography before randomisation Age (years): Group 1 (percentage) DA: 50 to 59: 3716 (47.1%) 60 to 74: 4169 (52.9%) Group 2 (percentage) Control: 50 to 59: 3830 (48.1%) 60 to 74: 4129 (51.9%) Ethnicity: None Data on the risk of breast cancer at baseline: None
Interventions	Group 1 (n = 7885): Women allocated to the decision aid group received an invitation to participate in the national breast cancer screening programme as well as the specially designed decision aid (a leaflet), by mail. The paper‐based leaflet DECIDEO is a 12‐page pocket leaflet providing scientific information about the advantages and disadvantages of participating in the national breast screening programme, understandable by all, and created to conform with the International Patient Decision Aid Standards. Group 2 (n = 7959): Women in the control group received an invitation and the usual standard information leaflet by mail. This invitation is an administrative letter sent to women scheduled to be invited to participate in the national screening programme every 2 years from the age of 50 onwards. Co‐interventions: Both groups were followed up for 12 months.
Outcomes	None of the measured outcomes were relevant for this review.
Notes	Funding: This study was supported by the French National Association against Cancer (Ligue National Contre le Cancer). This organisation was not involved in the design and conduct of the study, at any time. Conflicts of interest: PSM declares a conflict of interest through her activity, being a practitioner involved in breast cancer screening promotion at a local level. All the other authors declare no financial support for the submitted work; no relationships that might have an interest in the submitted work in the previous 3 years; none of their spouses, partners, or children have financial relationships that may be relevant to the submitted work; and none have non‐financial interests that may be relevant to the submitted work. Other outcomes reported not relevant to this review: Attendance at breast screening and delay in attendance

Elliot 2022.

Study characteristics
Methods	Study design: Parallel‐group cluster randomised trial Study dates: 1 August 2018 to 15 March 2020 Setting and number of centres: Multicentre, 34 primary care clinics Country: USA
Participants	Inclusion criteria: Aged 21 to 74 years Exclusion criteria: Pregnant, cognitively impaired, or in hospice care Up‐to‐date for breast, cervical, colorectal, or lung cancer screening at an index visit at a randomised clinic Sample size: 27,599 (total randomised), 25,198 (total analysed) Withdrawals: (Number per arm and reasons) CDS: 831 (excluded), 132 (opt‐out), 699 (ineligible for analysis*) CDS + SDM: 702 (excluded), 111 (opt‐out), 591 (ineligible for analysis*) UC: 868 (excluded), 182 (opt‐out), 686 (ineligible for analysis*) *Delayed documentation of screening, patients whose screening status at follow‐up could not be determined from electronic health record data, those no longer at risk, and those whose age at follow‐up was higher than ages included in algorithms. Age (years) mean: Group 1 CDS (52.3); Group 2 CDS + SDM (52.4); Group 3 UC (51.9) Ethnicity, N (%): Group 1 CDS: (African American, 107 (1.3); Asian 61 (0.8); Native American 203 (2.5); Pacific Islander 15 (0.2); white 7493 (93.8); Hispanic 87 (1.1); unknown 108 (1.4)) Group 2 CDS + SDM: (African American, 71 (1.0); Asian 51 (0.7); Native American 138 (1.9); Pacific Islander 7 (0.1); white 6752 (95.0); Hispanic 61 (0.9); unknown 86 (1.2)) Group 3 UC: (African American, 125 (1.2); Asian 56 (0.6); Native American 120 (1.2); Pacific Islander 19 (0.2); white 9669 (95.7); Hispanic 123 (1.2); unknown 117 (1.2)) Data on risk of breast cancer at baseline: Not reported
Interventions	Group 1 (n = 7987): CDM The CDS intervention was a web‐based, EHR‐linked CDS system integrated within the primary care clinic workflow. Web‐based CDS cancer prevention algorithms were based on USPSTF guidelines, and the CDS output provided personalised recommendations to both PCPs and patients in high‐literacy (provider) and low‐literacy (layperson) printed and electronic formats. All recommendations were presented as suggestions, and the interface emphasised that CDS suggestions do not take the place of clinical judgement or override a PCP’s detailed knowledge of a patient. Group 2 (n = 7105): CDM + SDM In the CDS + SDM group, the same CDS materials were provided, along with short‐form SDM tools for breast, colorectal, and lung cancer screening for all patients overdue for these screenings. At each CDS + SDM encounter, when one of these cancer screenings was due, patients and PCPs received the indicated printed short‐form SDMT(s). Short‐from SDMTs were implemented early in the study because providers felt overwhelmed by the 3‐ to 4‐page‐long forms. All eligible patients in the CDS + SDM group clinics received the short‐form SDMT. However, only those patients and/or providers who chose to access the long‐form SDMT received the printed form during the clinic visit with their provider. Although all patient‐provider dyads received the short forms, only a subset of the dyads received the long forms. All reported results included dyads that were potentially exposed to both short‐ and long‐form SDMTs. These SDMTs were developed using relevant International Patient Decision Aids Standards (IPDAS) checklist criteria for cancer‐screening tests for primary care SDM. The breast cancer‐screening long‐form SDMT had 2 options: mammography or not, but presented other decisions: age of onset and frequency of testing. Group 3 (n = 10,106): UC UC group, clinics and PCPs had no access to the CDS output or SDMTs
Outcomes	None of the measured outcomes were relevant to this review.
Notes	Funding sources: Financial support for this study was provided entirely by a grant from National Cancer Institute of the National Institutes of Health. Declarations of interest: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Other outcomes: The primary outcome was a patient‐level binary indicator of being up‐to‐date at 12 months following the index date on all cancer screenings (breast, cervical, or colorectal) that were overdue on the index date.

Giordano 2012.

Study characteristics
Methods	Study design: Randomised controlled trial Study dates: Not reported Setting and number of centres: Multiple centres with general practitioners Country: Italy
Participants	Inclusion criteria: Women 40 to 45 years old participating in the intervention arm of the Eurotrial 40 (programmed to get a breast cancer screening with mammography) Exclusion criteria: Women not eligible for breast cancer screening (previous history of breast cancer, recent mammogram (within the last 6 months), pregnancy, breastfeeding, and severe psychiatric symptoms or disease) Sample size: 5649 participants (1615 letter with appointment + informative leaflet, 807 letters with appointment + more detailed booklet, 607 letter with appointment + more detailed booklet + opportunity for face‐to‐face conversation, 2420 letter without appointment) Withdrawals: Not reported Age (years): Not reported Ethnicity: Not reported Data on the risk of breast cancer at baseline: Not reported
Interventions	Group 1 (n = 1615): Letter of invitation with a fixed appointment for mammography + informative leaflet (breast cancer prevention by mammography screening, incidence/mortality of BC, test procedures, test risks and safety, benefits, false positive, false negative, overdiagnosis) Group 2 (n = 807): Letter of invitation with a fixed appointment for mammography + informative booklet with more specified information (breast cancer prevention by mammography screening, incidence/mortality of BC, test procedures, test risks and safety, benefits, false positive, false negative, overdiagnosis) Group 3 (n = 807): Letter of invitation with a fixed appointment for mammography + informative booklet with more specified information (breast cancer prevention by mammography screening, incidence/mortality of BC, test procedures, test risks and safety, benefits, false positive, false negative, overdiagnosis) + opportunity to arrange a face‐to‐face conversation with a counsellor to get more information Group 4 (n = 2420): Letter of invitation without an appointment for mammography but inviting women to contact the centre to receive additional information Co‐interventions: May receive 1 of the following 2: only informed brochure or invited to schedule a personal encounter with a counsellor to get supplemental information
Outcomes	None of the measured outcomes were relevant to this review.
Notes	Funding: Italian Ministry of Health Conflicts of interest: The authors declare that they have no conflicts of interest. Other outcomes reported that were not relevant to this review: Attendance rate to the mammography (participation rates)

Gummersbach 2015.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: July 2012 to September 2013 Setting and number of centres: Multicentre in 33 family practices in the German federal state of North‐Rhine–Westphalia Country: Germany
Participants	Inclusion criteria: Women aged 48 to 49 who have not yet received their first invitation to be screened but are just about to receive it Exclusion criteria: Women from their lists who are not sufficiently fluent in German, and women with obvious cognitive limitations Sample size: 792 women (396 new leaflet, 396 old leaflet) Withdrawals: 218 in the new leaflet arm (211 did not return the questionnaire, 6 age not stated, 1 did not feel willing to participate) and 221 in the old leaflet control arm (211 did not return the questionnaire, 8 age not stated, 2 did not feel willing to participate) Age (years): New leaflet (mean 48.67, SD 0.79); Old leaflet (mean 48.76, SD 0.80) Ethnicity: New leaflet (93.26% native German language); Old leaflet (89.71% native German language) Data on the risk of breast cancer at baseline: Personal experience of breast cancer: New leaflet group (22.60% none, 57.63% in remote acquaintances, 18.08% in relatives, 1.69% myself have had BC); Old leaflet group (20.11% none, 62.07% in remote acquaintances, 15.52% in relatives, 2.30% myself have had BC)
Interventions	New leaflet (n = 178): An edge flyer format leaflet with evidence‐based information about the potential benefits and harms of breast cancer screening. The information included: benefit, mortality reduction, sensitivity, specificity, number needed to screen, overdiagnosis, false‐positive results rate, increase of operation and radiation of women who do not benefit from mammography screening, the rate of a pathological result of screening, interval cancer, and recommended to self‐check the breast for breast cancer screening (however, studies have shown that BSE does not affect breast cancer mortality). Old leaflet (n = 175): An edge flyer format leaflet that promotes breast cancer screening with limited information about the benefit and sensitivity of mammography screening, the rate of a pathological result of screening, interval cancer, the potential side effects of X‐ray Co‐interventions: None
Outcomes	Confidence How measured: Decisional confidence regarding participation in breast cancer screening (mammography) using a 6‐point scale, ranging from “not confident at all” to “very confident” Time points measured: Immediately after reading the leaflet Time points reported: Immediately after reading the leaflet Knowledge How measured: Objective knowledge about benefits and risks of breast cancer screening using 5 items, with scores from 0 to 10 (each item scored with 2, 1, or 0). Subjective knowledge using 2 items (1 about benefits and 1 about risks), with a 4‐point scale, ranging from “Not at all informed” to “Very well informed” Time points measured: Immediately after reading the leaflet Time points reported: Immediately after reading the leaflet
Notes	Funding sources: Not available Declarations of interest: The authors declare that they have no conflicts of interest. Other outcomes reported that were not relevant to this review: Willingness to participate in breast cancer screening Most important factors to determine their decision about screening

Haakenson 2006.

Study characteristics
Methods	Study design: Parallel‐group, randomised controlled trial Study dates: 10 March 2005 to 1 July 2005 Setting and number of centres: Convenience sample of participants in the Mayo Mammography Health Study (MMHS), which is a cohort study of approximately 16,500 patients from Minnesota, Iowa, and Wisconsin who received screening mammograms at the Mayo Clinic Country: USA
Participants	Inclusion criteria: Women who were scheduled for their screening mammograms during a period of 4 months Exclusion criteria: Not available Sample size: 668 (420 intervention, 248 control) Withdrawals: 778 non‐responders Age (years): Intervention (mean 61.9 ± 10.7); Control (mean 61.2 ± 12.0) Ethnicity: Intervention (97.4% white); Control (98.4% white) Data on the risk of breast cancer at baseline: Previous abnormal mammogram results, previous breast biopsy, previous diagnosis of breast cancer, and positive family history of breast cancer
Interventions	Group 1 (n = 420): Educational intervention with 2 educational pamphlets about mammography: a) “10 Tips for Getting a Mammogram” issued by the American Cancer Society: Pamphlet contains 10 brief facts about mammography, including annual screening recommendations, mammography facilities, information about the procedure itself, and the risks and benefits of the procedure (including statistics about the percentage of women who require additional mammograms and biopsy); b) “Breast Health Screening” issued by Mayo Clinic: A 23‐page pamphlet that includes text and diagrams, with content addressing normal breast anatomy, breast cancer risk factors, breast changes, clinical breast examinations, breast self‐examination, and the risks, benefits, and limitations of screening and diagnostic mammography Group 2 (n = 248): Not intervention, only invitation, and a short study survey Co‐interventions: None
Outcomes	Knowledge How measured: 10‐item knowledge assessment based on common knowledge about screening mammography and information available in the educational intervention pamphlets. There were 6 items with multiple‐choice responses and 4 items with true or false responses. An overall knowledge score was calculated for each woman by summing the number of correct responses to the 10 questions. Time points measured: Immediately after the intervention Time points reported: Immediately after the intervention
Notes	Funding sources: Not reported Declarations of interest: Not reported Other outcomes reported that were not relevant to this review: Informed decision‐making about screening mammography; 9 items: I understand why I am having this mammogram. I know what to expect when having a mammogram. I would like more information about the risks and benefits of screening mammography. The more information I know about a screening test, the more comfortable I feel. I like to be involved in my health care decisions. I prefer to have my physicians recommend health care decisions for me. My family is very involved in making my health care decisions. I still have questions regarding this mammogram.

Hersch 2021.

Study characteristics
Methods	Study design: Community‐based parallel‐group, randomised controlled trial Study dates: January to July 2014 Setting and number of centres: New South Wales Electoral Commission's electoral register Country: Australia
Participants	Inclusion criteria: Women aged 48 to 50 years who gave oral consent Exclusion criteria: Mammography in the past 2 years, a personal history of breast cancer; increased risk of breast cancer compared with the population (e.g. strong family history); or insufficient spoken English or reading ability to complete telephone interviews and understand study materials Sample size: 879 (440 intervention, 439 control) Withdrawals: 31 in the intervention arm and 31 in the control arm. Reasons: declined, unsuitable, unavailable, unable to reach in time, or did not answer all questions on attitudes Age (years): Intervention (median 49.7, IQR 49.3 to 50); Control (median 49.7, IQR 49.4 to 50.1) Ethnicity: Intervention (81% born in Australia or New Zealand); Control (82% born in Australia or New Zealand) Data on the risk of breast cancer at baseline: Family history of breast cancer (1 close blood relative diagnosed aged ≥ 50 years): Intervention (4%); Control (5%)
Interventions	Group 1 (n = 419): A booklet that contained information about relevance of decision on breast cancer screening and the things to know before making a decision: a) screening to prevent mortality from breast cancer, b) overdetection, and c) false positive and extra testing. Also includes a question and answer section, and finally presents evidence‐based quantitative estimates of breast cancer mortality benefit, false positives, and overdiagnosis, specific for women aged 50 to 69 years from Australia. All information combined text and visual formats. Group 2 (n = 419): The same booklet, but overdetection was omitted. Co‐interventions: None
Outcomes	Confidence (conflict and regret) How measured: Decisional Conflict Scale (10‐item low‐literacy version) on a scale from 0 (no decisional conflict) to 100 (extreme decisional conflict). Scores less than 25 are associated with implementing decisions. Confidence in decision‐making (Decision Self‐Efficacy Scale), 3 items rated from 1 (not at all confident) to 5 (very confident) Regret: 2 items from a validated scale will measure anticipated regret about screening (action regret) and about not screening (inaction regret), with 5 response categories ranging from strongly agree to disagree strongly Time points measured: 2 weeks, 12 months,* and 24 months after the intervention Time points reported: 2 weeks, 12 months, and 24 months after the intervention *The authors clarify that "Decision regret was not assessed at 2‐year follow‐up (programming error)". Knowledge How measured: Informed choice is assessed by combining measures of knowledge, attitudes, and actual choice (adequate knowledge and her attitudes and intentions were consistent (positive attitudes and intentions, or negative attitudes and intentions)). Knowledge (conceptual and numerical information on breast cancer mortality, false‐positive screening, and overdetection). Women had to score at least 50% of available marks, including at least 1 numerical mark, on all 3 screening outcome subscales to be considered adequate knowledge. Time points measured: 2 weeks, 12 months, and 24 months after the intervention Time points reported: 2 weeks, 12 months, and 24 months after the intervention Mental health How measured: Anxiety using the 6‐item short form of the Spielberger State‐Trait Anxiety Inventory. Scale from 20 to 80, with higher scores indicating greater levels of anxiety Time points measured: 2 weeks, 12 months, and 24 months after the intervention Time points reported: 2 weeks, 12 months, and 24 months after the intervention Breast cancer worry How measured: A single validated item will measure women's level of worry about developing breast cancer, using 4 verbal response categories ranging from "not worried at all" to "very worried". Time points measured: 2 weeks, 12 months, and 24 months after the intervention Time points reported: 2 weeks, 12 months, and 24 months after the intervention
Notes	Funding sources: Australian National Health and Medical Research Council Declarations of interest: Reported as "none" Other outcomes reported: Perceived importance of screening benefits/harms Perceived personal chances of screening benefit/harms Time perspective Perceived personal risk of breast cancer Booklet utilisation and acceptability

Kregting 2020.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: November to December 2018 Setting and number of centres: Women in the South West screening region of the Netherlands invited for breast cancer screening from the local screening organisation “Bevolkingsonderzoek Zuid‐West” and Erasmus MC Country: Netherlands
Participants	Inclusion criteria: Women 49 to 75 years old who were due to be invited for breast cancer screening and provided informed consent Exclusion criteria: Registered at the screening organisation as “Not willing to participate in research”; having a previous diagnosis of breast cancer during the last 5 to 10 years; having no email address or internet access Sample size: 1312 women (703 intervention, 609 control) Withdrawals: 173 intervention, 152 control Age (years): Mean age: 60.1 SD: 6.9 (intervention), 59.9 SD: 6.7 (control) Ethnicity: Not reported Data on the risk of breast cancer at baseline: Not reported
Interventions	Intervention (n = 531): Official breast cancer screening information leaflet from the Dutch National Institute for Public Health and the Environment. The leaflet was developed based on the opinion of experts and contains information about: a) the screening invitation, b) the screening process, c) possible screening outcomes, and d) benefits and harms (overdiagnosis, overtreatment, false‐negatives, and interval cancers) of screening. From the key components of SDM, we identified: Risk communication (sharing balanced information) However, the value clarification exercise, the encounter with a healthcare professional, and the final decision were not addressed in the intervention. Control (n = 609): Not intervention Co‐interventions: None
Outcomes	Knowledge How measured: About the breast cancer screening programme, 11 statements (based on expert consultations); response options were "true", "false", or "I don’t know". "Sufficient" knowledge was operationalised as a minimum of 8 correct answers. Informed choice (a woman was considered to have made an informed choice when she had sufficient knowledge, a positive attitude, and did participate in the programme; or when she had sufficient knowledge, a negative attitude, and did not participate in the programme) Time points measured: Baseline and 2 weeks Time points reported: Baseline (combined both intervention and control groups) and 2 weeks
Notes	Funding sources: Centre for Population Screening (CvB) of the Dutch National Institute for Public Health and the Environment (RIVM) Declarations of interest: None Other outcomes reported that were not relevant to this review: Attitudes towards breast cancer screening (positive or negative) Intention to participate in screening (positive, neutral, negative) Participation (participated, not participated)

Mathieu 2007.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: August 2005 to June 2006 Setting and number of centres: Outpatient, multicentre, national Country: Australia
Participants	Inclusion criteria: Women aged 70 to 71 years Had undergone 2 screening mammograms in the past 5 years Were due for their next mammogram within the next 3 months Had not previously been diagnosed with breast cancer Exclusion criteria: N/A Sample size: 734 women randomised Withdrawals: Group 1 (DA): 31 (Knowledge questions not answered: 3; Values questions not complete: 23; No intention given: 5) Group 2 (control): 44 (Knowledge questions not answered: 1; Values questions not complete: 41; No intention given: 2) Age (years): Group 1 mean (SD): 70.4 (0.7) Group 2 mean (SD): 70.3 (0.6) Ethnicity: N/A Data on the risk of breast cancer at baseline: N/A
Interventions	Group 1 (n = 367): The decision aid was developed using the Ottawa Framework. It was a self‐administered, paper booklet that consisted of 2 sections (information and a worksheet with a values clarification exercise) and an appendix. The information section described the options (to continue or stop screening) and the chances of each of the possible outcomes of each option. The appendix contained an explanation of the possibility of detecting a type of breast cancer that might not affect a woman’s health. Group 2 (n = 367): The standard BreastScreen NSW brochure contained a small amount of information regarding screening at different ages. It was selected because it was the only BreastScreen NSW brochure that contained any information specifically for women aged 70 years. It provided no numeric information about the outcomes of screening. Co‐interventions: None
Outcomes	Knowledge How measured: 9 knowledge questions were designed for this study. Items included 4 concept questions and 5 numeric questions. Answers to questions were scored using a marking scheme developed a priori to give a score of 0 to 10. It was decided a priori that a score of 6 or higher would be considered “adequate” knowledge. Women were classified as making an informed choice if they had adequate knowledge and clear values and expressed an intention to either continue or stop mammography screening. Time points measured: 1 month Time points reported: 1 month Confidence How measured: Decisional Conflict Scale Time points measured: 1 month Time points reported: 1 month Anxiety How measured: State‐Trait Anxiety Inventory and a question asking specifically about breast cancer worry Time points measured: 1 month Time points reported: 1 month Worry/Change in worry about cancer How measured: State‐Trait Anxiety Inventory and a question asking specifically about breast cancer worry Time points measured: 1 month Time points reported: 1 month
Notes	Funding sources: This study was supported by grant 211205 from Australia's National Health and Medical Research Council. Declarations of interest: The funding sources had no role in the design or conduct of the study; the collection, management, analysis, or interpretation of the data; or the preparation, review, or approval of the manuscript. Other outcomes reported that were not relevant to this review: Attitudes, values clarity Trial registration: ACTRN12605000695606

Mathieu 2010.

Study characteristics
Methods	Study design: Parallel‐group randomised controlled trial Study dates: September 2005 to June 2007 Setting and number of centres: Outpatient, multicentre, national Country: Australia
Participants	Inclusion criteria: Women aged 38 to 45 years Had not been previously diagnosed with breast cancer Exclusion criteria: > 45 years Sample size: 511 randomised participants (321 completed the trial) Withdrawals: 87 prior to accessing DA or outcome questions 72 (accessed DA: 53 did not view follow‐up questions and 19 did not answer any of them) 19 (delayed access to DA: 4 did not view follow‐up questions and 15 did not answer any of them) 12 excluded age > 45 years (6 DA/6 control) Age (years): Group 1 mean (SD): 41.88 (2.2) Group 2 mean (SD): 41.9 (2.0) Ethnicity: N/A Data on risk of breast cancer at baseline: Active smoking (Do you smoke?) Group 1 total affirmative responses (percentage): 22 (13%) Group 2 total affirmative responses (percentage): 21 (10%) Family history of breast cancer (Know anyone with breast cancer?) Group 1 total affirmative responses (percentage) Mother/sister: 8 (9%) Relative: 23 (25%) Group 2 total affirmative responses (percentage) Mother/sister: 16 (10%) Relative: 43 (26%)
Interventions	Group 1 (n = 172): The internet‐based decision aid was based on a previously developed and evaluated paper‐based decision aid, modified to provide age‐appropriate data. Both decision aids were developed according to standard guidelines and underwent extensive consumer pilot testing. They contained information on possible outcomes of screening and a worksheet to assist women to weigh up the outcomes and clarify their personal values and preferences before making a decision. DA provides: general information and explains why there is a decision to make (the benefits of screening when younger than 50 may not outweigh the downsides); diagrams and text describing the outcomes of screening and the chances of each of these outcomes occurring; and a worksheet to assist women to integrate the outcomes data with their personal values and preferences. The screening outcomes were expressed as event rates per 1000 women screened every 2 years over 10 years, and per 1000 women who are not screened over 10 years. The worksheet contained an exercise to help women clarify their values towards screening before making a decision. There were also 2 examples of how other women have completed the worksheet. These were based on worksheets completed during pilot testing by women in the target age group. Group 2 (n = 212): Women randomised to the control group received no information. They were given access to the decision aid after completing outcome data. Co‐interventions: None
Outcomes	Knowledge How measured: Evaluated using previously developed questions specifically designed to assess the understanding of the outcomes of mammography screening among older women and which were adapted for use in this age group. These questions comprised 4 numerical and 5 concept questions. Answers were scored using a predetermined marking scheme. If a woman scored 6 or higher out of 10, she was classified as having adequate knowledge of the issues surrounding mammography screening. Women were classified as making an informed choice if they had adequate knowledge and clear values and intentions to either continue or stop the screening. Time points measured: In a second time point after randomisation (not specified) Time points reported: In a second time point after randomisation (not specified)
Notes	Funding sources: This study was supported by grant 211205 from the National Health and Medical Research Council of Australia. The funding source had no role in the design or conduct of the study, the collection, analysis, or interpretation of the data or the preparation of the manuscript. Declarations of interest: None Other outcomes reported that were not relevant to this review: Acceptability of the decision aid and women’s values and intentions. Anxiety was only measured in the intervention group.

NCT04741503.

Study characteristics
Methods	Study design: Parallel randomised controlled trial Study dates: 2017 to 2022 Setting and number of centres: Single centre Country: USA
Participants	Participants Inclusion criteria: Latina, Black, or non‐Latina white women Between the ages of 40 and 49 years Can write, read, and understand English Exclusion criteria: Women with greater‐than‐the‐average self‐reported risk of breast cancer will be ineligible for participation, as evidenced by any of the following: Self‐reported personal history of breast cancer (invasive, ductal carcinoma in situ (DCIS), or lobular carcinoma in situ (LCIS)) Self‐reported personal history of atypical hyperplasia Self‐reported first‐degree family member with a history of breast cancer (e.g. mother, sister) Self‐reported known underlying genetic mutation such as a BRCA1/BRCA2 gene Self‐reported prior thoracic or chest wall radiation therapy Sample size: 324 randomised Withdrawals: Group 1: 23 (not meeting protocol criteria for free responses to knowledge questions) Group 2: 17 (not meeting protocol criteria for free responses to knowledge questions) Age: mean 44.2 Ethnicity: Group 1 (160): Hispanic or Latino 46 (28.8%), non‐Hispanic or Latino 114 (71.3%) Race: American Indian or Alaska Native 4 (2.5%), Asian 1 (0.6%), black or African American 51 (31.9%), white 81 (50.6%), more than 1 race 13 (8.1%), unknown or not reported 10 (6.3%) Group 2 (164): Hispanic or Latino 52 (31.7%), non‐Hispanic or Latino 111 (67.7%), unknown/not reported 1 (0.6%) Race: American Indian or Alaska Native 1 (0.6%), Asian 1 (0.6%), black or African American 61 (37.2%), white 72 (43.9%), more than 1 race 13 (8.1%), unknown or not reported 17 (10.4%) Data on the risk of breast cancer at baseline: N/A
Interventions	Intervention (Group 1 = 160): "After randomization, participants will complete pre‐questionnaires, review the Breast Cancer Screening Decision support tool, and then complete the post‐questionnaire." Comparator (Group 2 = 164): "After randomization, participants will complete pre‐questionnaires, review the standard breast cancer screening education from the NCI, and then complete the post‐questionnaire."
Outcomes	Confidence How measured: Decisional Conflict Scale (DCS) Time points measured: After intervention (30 min) Time points reported: After intervention (30 min) Knowledge How measured: Change in total knowledge score of screening mammography guidelines as measured by the percentage of correct responses Time points measured: After intervention (30 min) Time points reported: After intervention (30 min)
Notes	Funding: The National Institutes of Health (NIH), National Institute on Minority Health and Health Disparities (5R00MD011485‐04) Conflicts of interest: N/A Other outcomes reported that were not relevant to this review: Decision self‐efficacy as measured by the 5‐Point Decision Self‐Efficacy Scale Preparation for decision‐making scale as measured by PrepDM Acceptability as measured by the Ottawa Acceptability Measures Intention to screen We contacted the author (ahousten@wustl.edu) on 6 November 2023 to ask for the publication of results and details on risk of bias. The author responded on 9 November 2023 and provided data on baseline characteristics (age) and risk of bias (blinding, measurement of the outcome and missing data) and indicated that the study would be submitted for publication.

Price‐Haywood 2014.

Study characteristics
Methods	Study design: Cluster randomised trial Study dates: 2008 to 2012 Setting and number of centres: 5 clinics in New Orleans (1 federally qualified health centre, 2 academic clinics, and 2 clinics with faith‐based affiliations) Country: USA
Participants	Inclusion criteria: men 50 to 75 years, women 40 to 75 years, enrolled in clinics for ≥ 6 months or had seen their PCP at least 3 times, spoke English; were identified as having limited health literacy via the Rapid Estimate of Adult Literacy in Medicine (REALM ≤ 60 equivalent to ≤ 8th grade); and were due for breast/cervical/colorectal cancer screening (based on American Cancer Society’s (ACS) 2009 guidelines) Sample size: 5 clinics, 18 physicians, 168 patients Withdrawals: 1 patient in the intervention arm (communication + audit), 6 patients in the control arm (audit only) Age (years): Group 1 mean (SD) Physicians: 45.4 (13.8), Patients: 55.8 (7.1); Group 2 mean (SD) Physicians: 42.2 (8.5), Patients: 60.9 (7.6) Ethnicity: Group 1 Ph: Physicians, not Hispanic 6 (45.5%), black not Hispanic 5 (54.5%); Patients: black not Hispanic 82 (87.2%); Group 2: Physicians: white, not Hispanic 3 (42.9%), black not Hispanic 3 (42.9%); Patients: 70 (94.6%) black not Hispanic (there are some errors in the report of proportions and totals) Data on the risk of breast cancer at baseline: Group 1: patient's family history of cancer: any type 50 (50.3%), colon cancer 7 (7.5%), breast cancer 14 (14.9), gynecologic cancer 2 (2.1%) Group 2: patient's family history of cancer: any type 45 (60.8%), colon cancer 5 (6.8%), breast cancer 11 (14.9), gynecologic cancer 5 (6.8%)
Interventions	Group 1 (n = 11 physicians, 94 patients): Audit and feedback Intervention physicians received training in cancer risk communication and SDM. At the end of each visit with intervention physicians, study personnel revealed themselves as actors and gave structured verbal feedback. One week after baseline SP visits, intervention physicians underwent one‐on‐one 30‐minute academic detailing with a study investigator (EPH) to review the most recent ACS guidelines, clinical red flags for identifying patients with limited HL, and strategies for effective counselling. Physicians were taught to present information in small “chunks”; use simple language, pictures, and “teach back” to discuss complex concepts; and discuss and check the understanding of cancer risks, discuss potential benefits/risks of screening, explore preferences, and negotiate plans. Communication intervention physicians were directed to WebSP (web‐based service for SP event management) to review SP ratings of their communication and changes in ratings over time. They received written reports of SP ratings, which included narrative summaries of SP perceptions of the clinic, staff, and physicians. Group 2 (n = 7 physicians, 74 patients): Control ‐ audit only Physicians in the audit‐only group did not receive SP feedback or communication training. Both groups’ patient medical records were audited. Co‐interventions: Audit All study physicians received 2 annual cancer screening status reports and aggregate baseline patient ratings of their communication measured using the Perceived Involvement in Care Scale.
Outcomes	Knowledge How measured: Based on 2009 ACS guidelines, patients’ knowledge was coded as correct if they responded as follows: age to start colorectal cancer (CRC) screening, 50; Tests to screen for CRC ‐ stool test/cards, colonoscopy or “full colon/bowel scope”, sigmoidoscopy or “partial colon/bowel scope”, barium enema; age to start breast cancer screening, 40 to 50; frequency of mammograms, every 1 to 2 years; age to start cervical cancer screening ‐ sexually active or age 21; frequency of pap smears, every 1 to 3 years Time points measured: Baseline and follow‐up (12 months) Time points reported: Baseline and follow‐up (12 months) Subgroups: None
Notes	Funding sources: The study was funded by the Robert Wood Johnson Foundation Harold Amos Faculty Development Program (Grant # 63523). Dr Cooper is supported by grants from the National Heart, Lung, and Blood Institute (K24 HL83113 and P50 HL0105187). Declarations of interest: The authors declare that they do not have a conflict of interest. Other outcomes reported that were not relevant to this review: Standardised patients ratings (baseline, 6 months, 12 months), patient screening status (baseline and 12 months' follow‐up) Notes for analysis: Mammography at follow‐up n = 121, intraclass correlation 0.15580, the average number of patients per provider 6.72, variance inflation factor 1.892

Pérez‐Lacasta 2019.

Study characteristics
Methods	Study design: Parallel‐group 2‐stage (cluster) randomised controlled trial Study dates: July 2016 to June 2018 Setting and number of centres: Outpatient, multicentre, national Country: Spain
Participants	Inclusion criteria: Being a woman Aged 49 to 50 years Exclusion criteria: Had a personal history of breast cancer Difficulty speaking Spanish or Catalan Cognitive impairment Sample size: 40 Basic Health Areas and 524 women were randomised. Withdrawals: Group 1 (SD): 57 participants lost to follow‐up Group 2 (SD): 67 participants lost to follow‐up Age (years): Group 1 (SD): 50.14 (0.45) Group 2 (SD): 50.19 (0.46) Ethnicity: N/A Data on the risk of breast cancer at baseline: Family history of breast cancer: Group 1 (SD): 20 (9.9%) Group 2 (SD): 17 (8.6%)
Interventions	Group 1 (n = 20 Basic Health Areas; 260 women): After completion of Q1, women in the intervention arm received a decision aid that was a leaflet with detailed information on the benefits and harms of screening. The decision aid provided information about options and associated benefits/harms and helped clarify congruence between decisions and personal values. Group 2 (n = 20 Basic Health Areas; 264 women): Women in the control arm received a standard leaflet that did not mention harms and recommended accepting the invitation to participate in the biennial exams of the breast cancer screening programme (usual care). Co‐interventions: None
Outcomes	Confidence in the decision How measured: Assessed using the Decisional Conflict Scale (10‐item low‐literacy version) by O’Connor Time points measured: 2 weeks after the intervention Time points reported: 2 weeks after the intervention Knowledge How measured: Knowledge was measured in the following ways: Assessing each of the conceptual knowledge through 10 multiple‐choice questions Assessing numerical knowledge through 4 multiple‐choice questions that measured absolute and relative values of the screening outcomes Adequate knowledge for informed choice is defined as a score of at least 50% of the available marks, including at least 1 numerical mark, on all 3 screening outcome subscales (mortality reduction, overdiagnosis, and false positive) Informed choice was defined as adequate knowledge and intentions consistent with attitudes (positive attitudes and intentions or negative attitudes and intentions) Time points measured: 2 weeks after the intervention Time points reported: 2 weeks after the intervention Anxiety How measured: Measured with the 6‐item short form of the Spielberger State‐Trait Anxiety Inventory Time points measured: 2 weeks after the intervention Time points reported: 2 weeks after the intervention Subgroups: None
Notes	Funding sources: This study was supported by the research grant “Women participation in decisions and strategies on early detection of breast cancer” (PI14/00113) from the Instituto de Salud Carlos III and co‐funded by Fondo Europeo de Desarrollo Regional (FEDER) “Una manera de hacer Europa”. Anna Pons received a grant for PhD students from the Lleida Biomedical Research Institute (IRBLLEIDA). Declarations of interest: The authors have declared that no competing interests exist. Other outcomes reported that were not relevant to this review: Anticipated regret (did not report the scale, but categories for each item in the Likert scale) Time perspective Perceived importance of benefits/harms of screening Perceived risk of breast cancer Breast screening attitudes Breast screening intention

Reder 2017.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: April to November 2014 Setting and number of centres: Outpatient, multicentre, national Country: Germany
Participants	Inclusion criteria: Women aged 50 (birth months of March to May 1964) Exclusion criteria: Women with a potential Turkish migration background (were excluded from this study but assigned to the InEMa study) Women who ever had breast cancer (n = 29) Women who did not respond to the history of breast cancer question (n = 26) Mammography screening programme appointment prior to T2 (postintervention) (n = 84) Sample size: 1206 randomised participants. Accordingly, the data of 913 women were analysed. Withdrawals: Intervention group: E‐mail failed (n = 4) Not wishing to continue (n = 9) History of breast cancer (n = 25) Mammography screening programme appointment prior to T2 (n = 36) Control group: E‐mail failed (n = 7) Not wishing to continue (n = 5) History of breast cancer (n = 30) Mammography screening programme appointment prior to T2 (n = 48) Age (years): Not available Ethnicity: Not available Data on risk of breast cancer at baseline: Mother or sister with breast cancer: Group 1 (percentage): 63 (14.1%) DA Group 2 (percentage): 70 (15.8%) Control
Interventions	Group 1 (n = 602): The intervention group received an online DA which was designed to comply with IPDAS criteria and the Decision Aid Library Inventory where it was registered. This online DA consisted of a static information part and an interactive part. In the information part, the chance of each outcome was expressed as event rate per 200 women screened every 2 years for 20 years using absolute numbers accompanied by crowd figure. The advantages and disadvantages of the mammography screening programme and their probabilities were described. The interactive part of the DA summarised the main points of the information part and encouraged engagement with the information. It consisted of 3 steps. The women assigned the information items to the categories "in favour of mammography screening", "neither for nor against such screening", or "against the screening". They rated the importance of each information item for the decision. They made a choice. At the end, the participants received a tailored summary based on their responses pictograms. Group 2 (n = 604): The control group received a booklet containing standard information on the MSP, quality of the MSP, breast cancer and its risk factors, screening procedure, interval cancers and symptoms, follow‐up diagnoses, advantages and disadvantages of the MSP. Co‐interventions: None
Outcomes	Knowledge How measured: Assessed using the tool by Berens EM, Reder M, Razum O, Kolip P, Spallek J. Informed choice in the German mammography screening program by education and migrant status: survey among first‐time invitees. PLOS ONE. 2015; 10(11):e0142316. This tool includes: Adequate knowledge, Positive attitude, Positive intention/completed screening, Informed choice. Time points measured: The online assessments were conducted at baseline (T1), postintervention (T2, 2 weeks after T1), and 3 months' follow‐up (T3). Time points reported: The online assessments were conducted at baseline (T1), postintervention (T2, 2 weeks after T1), and 3 months' follow‐up (T3). Confidence (regret and conflict) How measured: Decisional conflict was measured using the 4‐item SURE test. To calculate the total score ‘uncertainty’, all items were summed (score range 0 to 4), with a high value indicating high decisional conflict. Decision regret was measured at T3 using the Decision Regret Scale. Time points measured: The online assessments were conducted at baseline (T1), postintervention (T2, 2 weeks after T1), and 3 months' follow‐up (T3). Time points reported: The online assessments were conducted at baseline (T1), postintervention (T2, 2 weeks after T1), and 3 months' follow‐up (T3).
Notes	Funding sources: This study was funded by BARMER, a provider of statutory health insurance and a federal, self‐governing corporation under public law. The financial support of the German Research Foundation (DFG) and the Open Access Publication Fund of Bielefeld University for the article processing charge. The funders had no role in study design, data collection and analysis, decision to publish, or manuscript preparation. Declarations of interest: The authors have declared that no competing interests exist. Other outcomes reported that were not relevant to this review: Intention/uptake, decision stage, attitudes. Contact with authors: On 16 June 2022, we contacted Dr Reder (maren.reder@uni‐bielefeld.de) to clarify the interpretation of the SURE score (reversed from the Manual).

Rimer 2002.

Study characteristics
Methods	Study design: Randomised controlled trial Study dates: November 1997 to August 2000 Setting and number of centres: Patients enrolled in the Personal Care Plan of Blue Cross Blue Shield of North Carolina (BCBSNC) (outpatient) Country: USA
Participants	Inclusion criteria: Women aged 40 to 44 and 50 to 54, who were enrolled in Blue Cross Blue Shield of North Carolina Exclusion criteria: None Sample size: 1091 Withdrawals: 0 Age (years): Group UC: 42 to 47 years 47%/52 to 57 years 53%; Group TP: 42 to 47 years 52%/52 to 57 years 48%; Group TP + Counselling calls: 42 to 47 years 49%/52 to 57 years 51% Ethnicity: Group UC (%): white 81, black 16, other (Asian, American Indian, Hispanic) 3; Group TP (%): white 81, black 16, other (Asian, American Indian, Hispanic) 3; Group TP + Counselling Calls (%): white 85, black 14, other (Asian, American Indian, Hispanic) 1 Data on risk of breast cancer at baseline: N/A
Interventions	Group 1 Usual Care (n = 378): BCBSNC’s UC includes sending postcard reminders to all women for whom no mammography claims have been filed in the past 12 months and providing primary care physicians with lists of their patients who were overdue for screening. Group 2 Tailored Print Materials (n = 374): In addition to these usual care measures, women in TP were mailed a tailored booklet 2 to 3 weeks following their baseline interviews and a brief tailored newsletter 2 to 3 weeks after their 12‐month follow‐up interview. Group 3 Tailored Print Materials + Counselling Calls (n = 339): Women also received brief counselling calls about 2 weeks after the tailored booklets/newsletters were mailed. Co‐interventions: None
Outcomes	Knowledge (e.g. Satisfaction) How measured: Knowledge and risk variables by intervention groups at 12‐ and 24‐month follow‐up periods Time points measured: 12 and 24 months Time points reported: 24 months
Notes	Funding sources: This research was funded by the National Cancer Institute, grant 5P01‐CA‐72099‐03. Declarations of interest: none

Roberto 2020.

Study characteristics
Methods	Study design: Pragmatic randomised clinical trial Study dates: September 2017 to December 2018 Setting and number of centres: 6 regional organised screening programmes. Each programme may have multiple centres. Country: Italy
Participants	Inclusion criteria: Newly invited women to the screening programmes aged > 50 years (in 4 screening programmes) and aged > 45 years (in 2 screening programmes) Exclusion criteria: Personal history of breast cancer Sample size: 2119 participants (1073 intervention, 1046 control) Withdrawals: 601 in the intervention arm and 517 in the control arm. Reasons for withdrawals not specified. Age (years): Group 1 (mean 49.7, SD 3.1); Group 2 (mean 49.7, SD 3.3) Ethnicity: Group 1 (97.4% Italian); Group 2 (96.2% Italian) Data on the risk of breast cancer at baseline: Acquaintance/family w/BC: Group 1 (331), Group 2 (380) Self‐perception of risk of BC. Group 1 (5.1% much lower, 4.9% a bit lower, 74.0% same average, 13% a bit higher, 3% much higher); Group 2 (3.2% much lower, 8.6% a bit lower, 72.6% same average, 12.8% a bit higher, 2.9% much higher)
Interventions	Group 1 (n = 472): Online DA, non‐static, with 19 screens, each covering 1 topic and answering a question. This included short coloured text, figures, bullet points, and hyperlinks. The DA homepage used a nudging‐like approach to highlight the 4 main sections: What is BC?, What is mammography screening?, What are its benefits and harms?, and What results can be expected from mammography screening?. The DA allowed women to decide which sections to access first, and move to other pages linked from the homepage. The DA provided a list of issues and concerns that possibly affect the screening decision about BC. Each woman was asked to state the importance of each of these items. DA presents quantitative data (based on UK Panel, EUROSCREEN) and controversial information (based on Cochrane reviews). It did not include an SDM process. Group 2 (n = 529): Standard brochure combines the best information from the participating centres’ brochures. The brochure was an online static web page divided into 4 sections. This included black‐and‐white text and no figures. Absolute numbers are reported about breast cancer. The brochure gives no information on the controversy about mammography screening. Co‐interventions: None
Outcomes	Confidence How measured: Decisional Conflict Scale ‐ SURE version Time points measured: 7 to 10 days postallocation, an email was sent with 3‐day reminders until the scheduled mammography date. Time points reported: Not reported Knowledge How measured: Using a questionnaire developed based on literature structured in 13 questions (10 conceptual and 3 numeric) with multiple‐choice answers. A score of 8 out of 13 (about 60%) or higher was considered “adequate knowledge”. This outcome is also mapped as "Informed choice (knowledge, attitude and intention to be screened)". Time points measured: Baseline and 7 to 10 days postallocation, an email was sent with 3‐day reminders until the scheduled mammography date. Time points reported: Baseline, but not at follow‐up.
Notes	Funding sources: Italian Association for Cancer Research (Competitive grant IG2015‐17274) Declarations of interest: AR, Colombo, P Mosconi report grants from the Italian Association for Cancer Research (competitive grant no. IG2015–17274) during the conduct of the study. GC, RS, and EP report grants from Mario Negri IRCCS Institute during the conduct of the study. LG reports grants from Mario Negri IRCCS Institute and Gisma (Italian group that organised mammography screening) during the conduct of the study; P Mantellini and MV report grants from Gisma (Italian group that organised mammography screening) during the conduct of the study. Other outcomes reported that were not relevant to this review: Participation rate in the breast cancer screening Satisfaction with the information (not the decision)

Schapira 2019.

Study characteristics
Methods	Study design: Randomised controlled trial Study dates: 2014 to 2015 Setting and number of centres: University of Pennsylvania Health Care System Country: USA
Participants	Inclusion criteria: Women 39 to 48 years of age with no prior mammogram and enrolled in 1 of 4 primary care practices affiliated with an academic medical centre in the Northern United States Exclusion criteria: Cognitive impairment based on clinical history and inability to speak English Sample size: 204 participants (104 intervention group, 103 control group) Withdrawals: 2 intervention group, 1 control group after enrolment before starting Age (years): Control group median 40.0 (39.0, 42.0); intervention group median 40.0 (39.0, 42.0) Ethnicity: Control group (%): black 72 (70.6), white 32 (31.4), Asian 4 (3.9), Hispanic 2 (2.0), other/unknown 4 (3.9) Intervention group (%): black 60 (58.8), white 23 (22.5), Asian 2 (2.0), Hispanic 1 (1.0), other/unknown 4 (3.9) Data on risk of breast cancer at baseline: Lifetime risk Control group n (%): ≤ 12: 86 (85.1); > 12 to ≤ 20: 13 (12.9); ≥ 20: 2 (2.0) Missing 1 Intervention group n (%): ≤ 12: 76 (74.5); > 12 to ≤ 20: 21 (20.6); ≥ 20: 5 (4.9) Missing 0
Interventions	Intervention group (n = 104): Breast cancer screening patient decision‐aid (BCS‐PtDA) Women were offered assistance with log in to the BCS‐PtDA using a laptop computer in the clinic and had the option of completing the programme in the waiting room. Women also received an email with a link to the BCS‐PtDA so it could be completed at a later time. Summary sheets from the BCS‐PtDA were printed by study staff, attached to the clinical electronic medical record, and available for clinician review. Control group (n = 103): Usual care Women were asked to fill out a breast cancer risk assessment after randomisation and then proceeded with usual care. Co‐interventions: None
Outcomes	Confidence (Decisional conflict) How measured: Decisional conflict was assessed with the 16‐item Decisional Conflict Scale (DCS). This scale includes 5 subscale scores in the domains of Uncertainty, Feeling Informed, Clear Values, Support for Decision, and Effective Decision Making. The DCS is scored from 0 to 100, with scores < 25 associated with increased adherence to decisions made. Time points measured: 6‐week follow‐up survey Time points reported: 6‐week follow‐up survey Confidence (Anticipated regret) Anticipated regret for decisions to delay or initiate mammography was assessed on a 7‐point Likert‐type scale with response options ranging from 1 (strongly disagree) to 7 (strongly agree), with increasing values indicating more significant anticipated regret. Time points measured: 6‐week follow‐up survey Time points reported: 6‐week follow‐up survey Knowledge How measured: Knowledge was assessed with the following 5 multiple‐choice questions (correct answers): 1) Do all women who have an abnormal mammogram have breast cancer? (No), 2) Do mammograms detect every cancer? (No), 3) Which of the following age groups will have the most extra tests, biopsies, or procedures as a result of having mammograms? (40 to 49), 4) Can mammograms detect breast cancer when it is at an early stage? (Yes), and 5) For what age group are more deaths prevented by mammogram. The knowledge scale was scored from 0 to 5, reflecting the number of correct responses. Time points measured: 6‐week follow‐up survey Time points reported: 6‐week follow‐up survey Breast cancer worry (mental health) Breast cancer worry was assessed with the 3‐item Lerman Breast Cancer Worry Scale. Items were summed for a score of 1 to 13. Time points measured: 6‐week follow‐up survey Time points reported: 6‐week follow‐up survey
Notes	Funding: Financial support for this study was provided by the National Cancer Institute‐funded consortium, Population‐based Research Optimizing Screening through Personalized Regimens (PROSPR) U54CA 163313. Declarations of interest: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing of the report. Other outcomes reported that were not relevant to this review: The accuracy of risk perceptions or intentions regarding age of initiation of mammography

Seitz 2016.

Study characteristics
Methods	Study design: Parallel‐group randomised trial Study dates: September 2013 Setting and number of centres: Unspecified (online) Country: Not reported
Participants	Inclusion criteria: Women 35 to 49 years old Exclusion criteria: Prior history of BC or genetic mutation in BRCA1 or BRCA2 Sample size: 3955 randomised (4549 consented to participate) Withdrawals (n = 1037): Left the survey (not reported how many) Completion time was either > 90 minutes or less than half of the mean time for participants who completed their condition (not reported how many). Age (years): Group 1 (mean 38.9) Group 2 (mean 44) Ethnicity: Group 1 (white 69%, African American 11.5%, other 19.5%) Group 2 (white 81.8%, African American 11%, other 7.2%) Data on risk of breast cancer at baseline: Group 1 Low risk (< 1.5% 10‐year risk of BC) n = 1227 Group 2 Moderate risk (> 1.5% 10‐year risk of BC) n = 1691
Interventions	Group 1 (Low risk 194, Moderate risk 259): No information; participants moved directly from the first set of measures to the next. Group 2 (Low risk 150, Moderate risk 196): Basic information: condition included the definition of mammography, a statement that women between the ages of 40 and 50 years old have a choice to make about when to begin mammography, screening options, and a table summarising recommendations from the ACS and the USPSTF. Group 3 (Low risk 141, Moderate risk 210): Brief intervention + Expository Group 4 (Low risk 169, Moderate risk 195): Brief intervention + Untailored exemplar Group 5 (Low risk 121, Moderate risk 200): Brief intervention + Tailored exemplar Group 6 (Low risk 160, Moderate risk 212): Extended intervention + Expository Group 7 (Low risk 154, Moderate risk 205): Extended intervention + Untailored exemplar Group 6 (Low risk 138, Moderate risk 214): Extended intervention + Tailored exemplar Description of intervention conditions: "Basic information" +: "Brief intervention": received individualised 10‐year and lifetime estimates of their objective risk for developing BC and the risk of an average‐risk age‐matched woman, all presented as both numeric frequencies and icon arrays "Extended intervention": "brief intervention" + statistics about outcomes of mammography (e.g. rates of false positives and mammogram‐detected cancers), and information about effects of mammography on reducing BC mortality "Expository": received arguments for having mammograms in one's 40s or waiting until age 50; arguments were didactic, with general statements about women making choices and no example case "Untailored exemplar": received arguments for having mammograms in one's 40s or waiting until age 50, arguments were presented using exemplars (e.g. women making decisions about mammography), and all participants in these conditions read the same text "Tailored exemplars": received arguments for having mammograms in one's 40s or waiting until age 50, exemplars’ ages, BC risks, family history of BC, and parity were similar to the participant and exemplar names were chosen to reflect common names for women of the participant's race and age
Outcomes	Knowledge (Accuracy of the perceived risk of breast cancer) Measured 2 ways: Asking participants, “What is your chance of developing breast cancer in the next 10 years?” as a percentage (0% = no chance of breast cancer and 100% = I definitely will get breast cancer) and as a frequency out of 1000 (“X out of 1000”) (similar measures have been used previously to measure perceived BC risk) A measure of the accuracy of the perceived risk of BC was created by subtracting each woman's estimated objective risk for BC, estimated using the NCI BCRAT, from her perceived risk measurement. Time points measured: After intervention Time points reported: Not reported Subgroups: By the risk of breast cancer: average or above‐average risk
Notes	Funding sources: Penn Center for Innovation in Personalized Breast Cancer Screening (PCIPS), Population‐based Research Optimizing Screening through Personalized Regimens (PROSPR), Grant 1U54CA163313 from the National Cancer Institute Declarations of interest: None Other outcomes reported that were not relevant to this review: Mammography intentions were measured by asking, “Which statement best describes how you feel?” Options included “I will start or continue to have mammograms in my 40s,” “I will have my first or next mammogram at age 50,” “I will never have a mammogram,” and “I am undecided”. The intention measure was recoded into a dichotomous measure. For women with an estimated 10‐year BC risk < 1.5%, the measure was coded “1” for women who intended to have their first or next mammogram at age 50 and “0” for other intentions. For women with an estimated 10‐year BC risk ≥ 1.5%, it was coded “1” for women who intended to start or continue to have mammograms in their 40s and “0” for other intentions.

ACS: American Cancer Society
BC: breast cancer
BCS: breast cancer screening
BSE: breast self‐examination
CDS: computerised clinical decision support
DA: decision aid
IQR: interquartile range
IPDAS: International Patient Decision Aids Standards
MCQ: multiple choice questionnaire
MSP: mammography screening programme
NCI BCRAT: National Cancer Institute Breast Cancer Risk Assessment Tool
PCP: primary care physicians
N/A: not available
SD: standard deviation
SDM: shared decision‐making
SURE scale: Sure of myself; Understand information; Risk‐benefit ratio; Encouragement
UC: usual care
USPSTF: United States Preventive Services Task Force

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Allen 2014	Wrong intervention: church‐based promotion of screening behaviour
Allgood 2016	Wrong intervention: reminders for breast cancer screening
Beauchamp 2020	Wrong intervention: reminders for breast cancer screening
Boling 2005	Wrong intervention: brochure to increase participation in breast cancer screening
Borrayo 2005	Wrong study design: qualitative study on breast cancer screening
Bouton 2012	Wrong patient population: participants with breast cancer
Bowen 2006	Wrong intervention: counselling to increase breast cancer screening
Bowen 2017	Wrong intervention: web‐based intervention without balanced information about benefits and harms
Bowles 2021	Wrong study population: women who had a mammography and were provided a decision aid on further management
Coronado 2016	Wrong intervention: counselling programme to promote screening
Curry 1993	Wrong intervention: invitation to screening with or without counselling about risk factors or risk‐adjusted invitations
Davey 2005	Wrong study design: survey eliciting information about informed choice and mammography
del Junco 2008	Wrong intervention: Project HOME (Healthy Outlook on the Mammography Experience), to promote the uptake of breast cancer screening
DuBenske 2017	Wrong study design: qualitative study on the experiences of provider and patient perspective on screening mammography
Eden 2015	Wrong study design: development of a decision aid (Mammopad)
Elkin 2017	Wrong study design: development of a decision aid and pilot trial (no control group)
Fagerlin 2005	Wrong intervention: women were randomised to make a risk estimate prior to receiving risk information
Fechtelpeter 2019	Wrong study design: development of a decision aid with testing in focus groups and surveys
Fernández‐Feito 2015	Wrong intervention: nursing counselling to reduce stress before mammography
Fiscella 2011	Wrong intervention: multimodal intervention to promote breast and colon cancer screening
Fredrick 2020	Wrong study design: development of an education programme on cancer knowledge, prevention and screening (no randomisation)
Geller 2007	Wrong study design: survey on the methods to enhance communication with women about mammography
Ghosh 2008	Wrong intervention: different types of formatting the same numerical data
Gibbons 2018	Wrong study design: quality improvement project (no randomisation)
Goel 2011	Wrong study design: pre‐ and post‐testing
Goldzahl 2018	Wrong interventions: 1) a new logo, 2) patient‐approved letter content, 3) combination of the previous 2, 4) information on the number of women receiving mammograms in the area of residence
Haas 2019	Wrong intervention: different modes of delivery of mammography results
Hersch 2014	Wrong study design: pilot study of the decision aid that was later included in a randomised trial (Hersch 2021)
Hurdle 2007	Wrong intervention: promotion of general health in women, including breast cancer prevention (not focused on screening)
ISRCTN15366380	"Updated 06/08/2015: The trial did not start because the objectives would have had limited practical application."
Kearins 2009	Wrong intervention: phone calls to invite non‐attenders to mammography screening
Kernohan 1996	Wrong intervention: community intervention to promote screening
Krist 2017	Wrong study design: prospective observational cohort study
Larkey 2012	Wrong intervention: assignment of a "promotora" to increase the uptake of various screenings (including mammography)
Lawrence 2000	Wrong study design: development of a decision aid and preliminary testing (no randomisation)
Lerman 1994	Wrong patient population: all women were at a high risk of breast cancer
Lewis 2003	Wrong intervention: randomisation into 3 videos with positive, neutral, and negative framing of mammography screening
Lippey 2022	Wrong study design: development of a decision aid with consumer testing (no formal evaluation of effects)
Lo 2018	Wrong study design: uncontrolled evaluation of the usability and acceptability of a breask cancer risk assessment and risk managment tool
Luckmann 2019	Wrong intervention: 3 strategies to improve uptake of screening
Mambourg 2018	Wrong study design: development of a decision aid (no testing)
Mann 2000	Wrong study design: uncontrolled study of an educational intervention to improve screening
Molina 2018	Wrong intervention: study focused on motivational interviewing to increase the uptake of screening
Narasimmaraj 2016	Wrong intervention: different risk‐based cancer screening strategy (all women would be invited to be screened)
NCT00150917	Wrong patient population: all women with a higher risk of breast cancer (family history)
NCT00247442	Wrong patient population: women 70 years or older (decision to continue or discontinue screening)
NCT01336257	Wrong intervention: decision support tool (reminders) to improve the uptake of screening
NCT02964234	Wrong intervention: empowering intervention (to increase social outreach and increase uptake of screening)
NCT02986230	Wrong intervention: decision support tool that promotes screening (overdue cancer prevention screening services)
NCT04601272	Wrong study design: retrospective cohort study
Orlando 2018	Wrong intervention: implementation of a risk assessment tool
Percefull 2020	Wrong study design: quality improvement study using reminders to improve uptake of breast cancer screening
Petrova 2015	Wrong intervention: the same information was provided as text, a fact box, or a visual aid (wrong comparison)
Phillips 2018	Wrong study design: acceptability and usability of a decision support tool (uncontrolled study)
Reder 2018	Wrong intervention: numeric vs non‐numeric information for mammography
Ruffin 2004	Wrong intervention: reminder based on screening history to improve the uptake of screening
Russell 2007	Wrong intervention: pamphlet vs interactive instruction to increase the uptake of mammography
Saver 2017	Wrong intervention: both groups received different formats for the same data (comparators were decision aids)
Saywell 2003	Wrong intervention: counselling strategy to increase uptake of mammography
Scariati 2015	Wrong study design: development and testing of a decision aid (no control group)
Schoenberg 2013	Wrong intervention: multicomponent intervention to increase uptake of mammography
Schonberg 2020	Wrong study design: pre‐post trial without a control group
Segura 2001	Wrong intervention: different strategies to recruit women into screening
Seven 2015	Wrong intervention: different educational interventions to improve uptake of breast cancer screening
Shieh 2017	Wrong interventions: annual vs personalised risk‐based screening
Sinicrope 2020	Wrong intervention: literacy intervention to promote the uptake of mammography
Slater 1998	Wrong intervention: Friend to Friend intervention to increase the uptake of mammography
Smith 2020	Wrong intervention: different communication strategies to discontinue breast cancer screening
Stencel 2011	Wrong study design: experimental psychology study on the influence of information on anxiety
Stover 2017	Wrong intervention: 2 types of strategies for breast cancer screening
Street 1998	Wrong intervention: 2 forms of presenting the same information before screening
Taylor 1999	Wrong intervention: enhanced support to facilitate screening (transportation assistance, etc.)
Thompson 2002	Wrong intervention: motivational intervention to increase the uptake of mammography
Tobin 2022	Wrong intervention: intervention to promote screening and reduce anxiety and depression via case managers
Tolma 2016	Wrong study design: participatory intervention to increase uptake of mammography
Ufomata 2016	Wrong study design: uncontrolled study of the implementation of a new handout
Urban 1995	Wrong intervention: phone calls to improve the uptake of mammography (reminders)
Wegwarth 2018	Wrong study design: cross‐sectional study of women's perceptions of mammography
Wolosin 1990	Wrong intervention: reminder system to improve the uptake of mammography
Wong 2015	Wrong study design: developing and testing (uncontrolled) of a decision aid for mammography
Wu 2013	Wrong study design: non‐randomised study of a decision support tool (MeTree) for various types of screening
Yang 2020	Wrong intervention: observational study on different types of communication related to mammography

Characteristics of studies awaiting classification [ordered by study ID]

NCT02914197.

Methods	Study type: Parallel randomised clinical trial Sample size: 608 participants Masking: Triple (participant, investigator, outcomes assessor)
Participants	Inclusion criteria: Patient on the active roster of physicians of MUSIC (McMaster University Sentinel and Information Collaboration; McMaster Family Health Team) Due for a mammogram (have not had a mammogram ≥ 18 months) according to Canadian screening interval recommendations for routine screening Exclusion criteria: Previous history of invasive or non‐invasive breast cancer At higher risk for breast cancer compared to the general population, i.e. identified as having the BRCA1 or BRCA2 gene, immediate family member has had breast cancer Have had a mammogram within the last 18 months (< 18 months)
Interventions	Intervention: Full information on the risks and benefits of mammography through: decision aid; YouTube video; group information session. Control: Standard information leaflet for breast screening from Cancer Care Ontario
Outcomes	Primary outcome measures: Self‐efficacy [1 month] Decision Self‐Efficacy Scale Secondary outcome measures: Informed Choice (Knowledge) [1 month] Informed Choice (Attitude) [1 month] Informed Choice (Intention) [1 month] Decisional Conflict [1 month] SURE (Decisional Conflict Scale ‐ SURE Test for clinical practice) Anxiety [1 month] Trust in Medical System [1 month] MMI (Medical Mistrust Index) Perception of Health Provider Recommendation [1 month] Information Relevant to the Decision Making Process [1 month] Decision Regret [12 months] Decision Regret Scale Screening Participation [6 months and 12 months] Acceptance of a Decision Aid in Primary Care [12 months] Knowledge of the Benefits and Risks of Mammography [1 month]
Notes

NCT03631758.

Methods	Study type: Parallel randomised clinical trial Sample size: 209 participants Masking: Triple (participant, investigator, outcomes assessor)
Participants	Inclusion criteria: English‐speaking women, 40 to 49 years old Residents of British Columbia, Nova Scotia, and Prince Edward Island Exclusion criteria: History of breast cancer or mammography in the last 12 months Classified as above‐average risk for breast cancer (i.e. known genetic mutation linked to BC; no genetic testing, but a parent, sibling, or child with a genetic mutation linked to BC; or received chest wall radiation before age 30)
Interventions	Experimental (Evidence + PDA): Evidence‐based information on mammography such as blog posts, plain language evidence summaries and web resource ratings (quality‐appraised online resources), plus a blog post on PDAs and the relevant decision aid from the Portal database Experimental (Evidence only): The same evidence‐based information as Group 1 (Evidence + PDA), but without the PDA to quantify the effect of accessing evidence through the Portal alone Sham comparator (Attention control): Information on how to distinguish high‐ from low‐quality health information, not specific to cancer screening or PDAs
Outcomes	Primary outcome measures: Decisional Conflict Scale [6 months] Secondary outcome measures: Decision quality [6 months] Preparation for Decision Making Scale [6 months] Screening behaviour [6 months]
Notes

Characteristics of ongoing studies [ordered by study ID]

NCT04948983.

Study name	The Effect of a Patient Decision Aids for Breast Cancer Screening
Methods	Study type: Parallel randomised clinical trial Sample size: 3269 participants Masking: None
Participants	Inclusion criteria: Women 50 to 69 years old Attend the primary care centres Fluent in Spanish Exclusion criteria: Women with a personal history of breast cancer Have a current mammogram (as recommended by the Chilean Ministry of Health) Not having the capacity to consent
Interventions	The intervention group will access a web page, answer a set of questionnaires at baseline, and then the DA (developed according to the IPDAS recommendations). The control group will access a web page, answer a set of questionnaires at baseline, and then receive standardised information given by the healthcare system. Both groups will complete the questionnaires 2 weeks later.
Outcomes	Primary outcome measures: Informed choice for breast cancer screening [2 weeks] Secondary outcome measures: Decisional conflict [2 weeks] Decisional Conflict Scale Depression, anxiety and stress [2 weeks] Satisfaction with the decision [2 weeks] Number of participants who undertake a mammography [2 weeks]
Starting date	1 July 2021
Contact information	Paulina Bravo, PhD (56‐2) 2354 5838 pbbravo@uc.cl Contact: Alejandra Martinez, MSc +56940742491 alejandra.martinez@uc.cl
Notes

Differences between protocol and review

Sensitivity analysis based on the description of components of shared decision‐making (SDM) was precluded by insufficient data to be categorised by SDM elements. Similarly, we did not conduct prespecified subgroup analyses and assessment of publication bias due to few studies per comparison.

We grouped studies into categories considering our predefined elements of shared decision‐making, creating three comparisons and therefore three summary of findings tables instead of the single one defined at the protocol stage.

Contributions of authors

• Draft the protocol: PR, MVRY, CMEL, JVAF, KSK

• Study selection: MVRY, PR, JVAF

• Data extraction: MVRY, NJS, CAR, PR

• Enter data into RevMan: MVRY, NJS, CAR, PR

• Carry out the analysis: JVAF, PR

• Interpret the analysis: all authors

• Draft the final review: all authors

• Disagreement resolution: KSK

• Update the review: PR (guarantor)

Sources of support

Internal sources

• Hospital Italiano de Buenos Aires ‐ Family and Community Medicine Division, Argentina

  Provides in‐kind support for PR, MVRY, and KK for this review.

• Institute of General Practice ‐ Heinrich Heine University Düsseldorf, Germany

  Provides in‐kind support for JVAF

• Cochrane Associate Centre ‐ Instituto Universitario Hospital Italiano, Argentina

  CMEL receives a salary as an Information Specialist for Cochrane reviews.

External sources

• No sources of support provided

Declarations of interest

PR, MVRY, KSK, and JVAF have published commentaries, taught, and conducted research related to shared‐decision making. With a grant from the Hospital Italiano de Buenos Aires, we developed a decision aid for breast cancer screening: decidirmamografia.com.ar/. The development paper and alpha testing of this tool (acceptability and usability, not an implementation study) was submitted for publication and is under review. We have no commercial interest or revenue from this decision aid.

CMEL: none known.

NJS: none known.

CAR: EviSalud ‐ Evidencias en Salud (other business ownership). This online platform provides courses related to evidence‐based medicine and research methods. There are no direct or indirect relationships with the review topic.

JVAF is Managing Editor for the Cochrane Metabolic and Endocrine Disorders Group, Clinical Editor for the Cochrane Urology Group, and Cochrane's Governing Board member. He was not involved in the editorial process of this review. The views of this manuscript do not represent the views of the Cochrane Governing Board, as JVAF has an authoring role in this manuscript.

New