Ethical approval
Ethical approval was obtained from the Ethics Commission of the Medical Association Westfalen-Lippe and the Medical Faculty of the University of Münster. Data handling will conform to the data protection legislation of the federal state of North Rhine-Westphalia. We have also obtained approval from the data protection officer at Bielefeld University.
This research project is independent of the MSP. Invited women will be informed about the content, purpose, and procedure of the study, and we will obtain their written informed consent before proceeding. Personal data will be stored separately from research data. The first, second, and third assessments (T1, T2, and T3) can be linked to each other by means of a self-generated code consisting of the following elements: day of the month the mother was born, first letter of the first name of the mother, day of the month the father was born, first letter of the first name of the father, and first letter of own place of birth. This way the questionnaires cannot be associated with any personal data. All personal data will be deleted after completion of the study. Participation in the study will be voluntary and the participants may revoke consent at any time.
Decision aid
The DA was developed and piloted as part of this study. The DA and questionnaires for evaluation were programmed in Unipark, an online survey tool (QuestBack GmbH, Cologne, Germany). The structure of the DA was based on that of Mathieu et al. [
19]. The DA was conceptualized as an interactive online tool consisting of a static information part and an interactive part.
We conducted a pilot study in March 2014, which showed that our study methods were feasible and acceptable to women aged 50. This quantitative pilot, to which 300 women were invited of which 53 participated, also showed that women accessed the DA and worked through it and that the questionnaires were accessed and answered. The pilot indicated that about a third of women dropped-out of the study prematurely. In a preceding qualitative pre-test both DA and questionnaires were evaluated with women of the target group through think-aloud protocols. The DA was additionally qualitatively piloted with experts from the National Network Women and Health.
In line with IPDAS criteria [
25], the DA presents the options of choice (participation or non-participation in the MSP) in a decision-relevant context. The advantages and disadvantages of the MSP and their probabilities are described. These include the probabilities of a positive and negative screening result in absolute numbers supported by graphic presentations to allow women to form a realistic expectation regarding the initial outcome of the screening. We explain overdiagnosis, overtreatment, and the procedure following a positive result. Positive and negative information are presented in a balanced way according to current evidence. Step-by-step navigation through the websites of the interactive online DA facilitates a structured progress. Participants receive tailored feedback based on previous entries directly before making the decision. The studies from which the data for the DA originate are cited and provide the participants with easy access to the original sources via links.
A structured decision-making process in several steps is at the core of the DA. Participants have the opportunity to express personal values that are relevant for the decision and to clarify these for themselves. Participants have the option of printing their personal results and the static information section at the end [see Additional file
1] and to discuss matters with others. There is also an input window for remaining questions at the end of the DA.
The information part of the DA contains the standard information that women receive together with the invitation to the MSP [
22]. The DA specifies all event probabilities based on the same population of 200 women over 20 years. In the presentation of outcome probabilities, the information part goes beyond the standard information. All probabilities are additionally represented as crowd-figure pictograms [see Additional file
1], thereby simultaneously indicating the positive and negative frames of outcomes. This type of pictorial information is beneficial [
26]. Additionally, the probability of dying from breast cancer is presented in relation to the likelihood of all-cause mortality. Uncertainty in the evidence is described.
Interactive part
The interactive part consists of three steps. In the first step, the women assign the information items to the following categories: (1) in favour of mammography screening; (2) neither for nor against such screening; and (3) against the screening. In the second step, the women rate each point according to its importance in the decision. In the third step, the women make a decision. Through this approach, the importance of a personal value-based assessment of information is highlighted. An additional PDF file shows this in greater detail [see Additional file
1].
Consent and data collection
The letter of invitation to the study, detailed study information, a consent form, and a return envelope will be sent by post 3 weeks in advance of the estimated date of receipt of the invitation to the MSP. The consent form will also provide for the possibility of specifying that the recipient does not have access to an email address. To increase the response rate, 1 week after the written invitation to participate in the study, a reminder and thank-you postcard will be mailed out.
Three weeks after the invitation is sent out, all women who gave informed consent and provided us with an email address will be emailed the link to the baseline questionnaire (T1). Women whose consent form is received 3-5 weeks after the invitation will receive the link at 5 weeks after the invitation. Women whose consent form is received later than 5 weeks after the invitation will receive the link at 7 weeks.
The link to the second assessment (T2) will be sent to participants 2 weeks after the first email. Women will be randomly assigned to the intervention or control group, and accordingly, receive a link either to the DA and second assessment or only to the second assessment. The link to the third assessment (T3) will be sent to the women 3 months after T2 when the screening appointment is assumed to have passed. A reminder and thank you will be emailed to all women who initially agreed to participate 10 days after sending the survey link at T1, T2, and T3.
Secondary outcome measures
Intention to participate in the MSP will be measured using one item regarding intention to participate in the next 3 months (yes/no/undecided).
Uptake of the MSP will be measured using one item regarding participation in the last 3 months (MSP/opportunistic screening/none).
Attitude will be measured using four items adapted from Marteau et al. [
11] and according to the reasoned action approach of Fishbein and Ajzen [
27]. Three semantic differentials (important/unimportant; a good thing/a bad thing; beneficial/harmful) will assess instrumental attitude. One semantic differential will assesses experiential attitude (pleasant/unpleasant). Items will be rated on a five-point scale.
Knowledge will be measured using seven multiple choice items, with two to four answer options. The questions will cover the following: (1) screening for people without symptoms; (2) frequency of positive screening results; (3) meaning of a positive result; (4) potential to detect every cancer; (5) more diagnoses with the MSP; (6) more breast cancer deaths without the MSP; and (7) overdiagnosis and overtreatment. For each item, “do not know” will be offered as an option. Most questions will ask for nun-numeric information. Only Question 2 will ask for numeric information through four ranges of values covering the base population used in the question. This information is considered to be important as all women participating in the MSP will receive a result and a realistic expectation on the probability of both positive and negative results is important.
Perceived behavioural control will be measured using two items developed following the reasoned action approach of Fishbein and Ajzen [
27] and rated on a five-point scale (“Whether I participate in the MSP is up to me”; “If I wanted to, I could participate in the MSP”). Additionally, 15 items (four items at T2 and T3) measuring control beliefs will be rated on a five-point scale with the anchors of “agree” and “disagree”. The items will assess logistic barriers (e.g. not having time for the appointment), trust in the MSP, and fear of pain.
Subjective norms will be measured using two items developed according to the reasoned action approach of Fishbein and Ajzen [
27] and rated on a five-point scale (“Most people who are important to me think that I should/should not participate in the MSP”; “Most people like me would/would not participate in the MSP”). Additionally, five items will measure normative beliefs rated on a five-point scale, ranging from “advise” to “disadvise” with the additional option of “no advice”. These items will assess the direction of advice of doctors, family, and friends.
Decisional conflict will be measured using the SURE (Sure of myself; Understand information; Risk-benefit ratio; Encouragement) test [
28]. This four-item test with the answer options yes and no will assesses knowledge of options, clarity of importance of advantages and disadvantages, sufficient level of support, and being sure about the best choice. Since no German translation of this scale is available, the scale was translated as part of this study, and indices of reliability and validity will be published.
Decision regret will be measured using the Decision Regret Scale [
29]. The five items will be rated on a five-point scale (“strongly agree” to “strongly disagree”). Since no German translation of this scale is available, the scale was translated as part of this study, and indices of reliability and validity will be published.
Result of the MSP will be assessed using two items regarding the result of the MSP and the result of further diagnostics. The answers will allow grouping into four categories: (1) negative result; (2) false-positive result; (3) true-positive result, and (4) positive result with further diagnostics pending.
Result and acceptability of the DA will be measured using one item assessing whether the tailored graphic in the DA was balanced, and four items will assess acceptability of the DA regarding length, amount of information (too much/too little/just right), one-sidedness (pro/balanced/against), and helpfulness in decision making (yes/neither/no).
Invitation status will be measured using two items assessing whether an invitation was received and whether the appointment specified in the invitation had already passed.
Decision stage will be measured using one item following IPDAS criteria with the answer options being “not thought about it”, “contemplating it”, “close to deciding”, and “choice already made”.
Use of/experience with other screenings will be measured using items on (1) the use of other screenings and (2) ever having received a false-positive screening result (including false-positive mammogram).
eHealth literacy will be measured using the eHealth Literacy Scale [
30]. Its German translation has recently been validated [
31].
Internet use will be measured using two items regarding time spent seeking information online and importance of the Internet for health information.
Demographics will be measured using items assessing education level, main language, health insurance, participation in a health insurance bonus programme, and breast cancer in a first-degree female relative.
Data analysis
The data will be exported from Unipark and analysed with SPSS version 21.0 (IBM Corp., Armonk, NY) and Mplus version 7.0 (Muthén & Muthén, Los Angeles, CA). Data cleaning and analyses will be performed using SPSS and Mplus syntax operations. We will conduct descriptive analyses to describe the study population. Possible baseline differences between trial arms will be statistically tested. We will adjust further analyses for imbalances in the baseline scores.
The primary analysis will be by intention to treat. As sensitivity analyses, only completers of all assessments will be analysed. The impact of the DA on the primary outcome will be analysed using
χ 2-tests. The numeric secondary outcomes will be analysed using analyses of variance; categorical secondary outcomes will be analysed using
χ 2-tests. Additionally, latent change model analyses will be performed for continuous longitudinal outcome data. Latent transition analyses will be performed for categorical longitudinal outcome data. To handle missing data, full information maximum likelihood estimation will be conducted in MPlus. To investigate the mediation and moderation effects of the secondary and demographic variables, mediation and moderation analyses will be performed using the PROCESS macro version 2.13 (Andrew F. Hayes,
http://www.processmacro.org). in SPSS. As part of the questionnaire adaptation, factor, reliability, and correlation analyses will be performed.