Dataset for a randomised factorial experiment to optimise an information leaflet for women with breast cancer

Background Adherence to adjuvant endocrine therapy (AET) is low in women with breast cancer, which increases the risk of recurrence and mortality. A consistently reported barrier to adherence is low perceived necessity of AET and high concerns. Existing interventions to support medication beliefs have mixed effectiveness and rarely target medication beliefs specifically. We developed an information leaflet with five candidate components aiming to increase necessity beliefs about AET and reduce concerns; (1) diagrams explaining how AET works; (2) icon arrays displaying the benefits of AET; (3) information about the prevalence of side-effects; (4) answers to common concerns and (5) quotes and pictures from breast cancer survivors. Guided by the multiphase optimisation strategy (MOST), we aimed to optimise the content of the information leaflet. We planned for the dataset to be open access to provide an exemplar for other investigators to use. Methods The content of the leaflet was optimised in a fully powered online 2 5 factorial experiment. Each candidate component of the leaflet was operationalised as a factor with two levels; on vs off or enhanced vs basic. Healthy women (n=1604) completed the beliefs about medicines questionnaire and were randomised to view one of 32 versions of the information leaflet. The 32 versions comprised unique combinations of the factor levels corresponding to the five candidate intervention components. Time spent on the information leaflet page of the survey was recorded. After viewing the information leaflet, participants completed the beliefs about medicines questionnaire again, a true/false questionnaire assessing their objective knowledge of AET, a subjective rating of their knowledge of AET, and a questionnaire evaluating their satisfaction with the information they received. Importance of this dataset The factorial dataset provides the opportunity for other investigators interested in using the MOST framework to learn about complex factorial designs, using a real dataset.


Methods
The content of the leaflet was optimised in a fully powered online 2 5 factorial experiment.Each candidate component of the leaflet was operationalised as a factor with two levels; on vs off or enhanced vs basic.Healthy women (n=1604) completed the beliefs about medicines questionnaire and were randomised to view one of 32 versions of the information leaflet.The 32 versions comprised unique combinations of the factor levels corresponding to the five candidate intervention components.Time spent on the information leaflet page of the survey was recorded.After viewing the information leaflet, participants completed the beliefs about medicines questionnaire again, a true/false questionnaire assessing their objective knowledge of AET, a subjective rating of their knowledge of AET, and a questionnaire evaluating their satisfaction with the information they received.

Importance of this dataset
The factorial dataset provides the opportunity for other investigators interested in using the MOST framework to learn about complex factorial designs, using a real dataset.

Plain Language Summary
Most women with breast cancer are treated with adjuvant endocrine therapy (AET) to reduce the chance of breast cancer coming back.However, many women do not take the medication as recommended.Women's beliefs about the medication are a common reason for not taking AET.Some women do not think AET will help them, and some women have lots of concerns about AET.At the moment, we do not know the best way to change women's beliefs about AET.Therefore, we ran a study to help us understand what combination of information might help change women's beliefs about AET.
We developed a written information leaflet with five parts; (1) diagrams about how AET works; (2) visual figures of the benefits of AET; (3) information about how likely each side-effect is; (4) answers to common concerns about AET; and (5) pictures and quotes from women who have taken AET.In an online survey, 1,604 healthy women answered questions about their beliefs about the medication.Each woman was shown one version of the information leaflet picked at random.There were 32 possible versions of the information leaflet, which contained unique combinations of the five parts of the leaflet.After women read the leaflet, they were asked to complete the same questionnaire about their beliefs about the medication.They were also asked questions about how satisfied they were with the information they received, true or false questions about AET to assess their knowledge after reading the leaflet, and a rating of how informed they felt about AET.We also recorded how long women spent looking at the leaflet.One of our aims was to make the dataset from this experiment openly available so other scientists could use it to learn how to conduct similar experiments.
Educational interventions are commonly used to address medication beliefs 11 .However, evidence is mixed regarding the effectiveness of such interventions in changing medication beliefs 11,12 .Moreover, medication beliefs are often targeted within a larger complex intervention aiming to support AET adherence.Therefore, the effectiveness of components specifically targeting medication beliefs is unclear 11,12 .
To increase necessity beliefs and reduce concerns in women prescribed AET, we developed a theory-informed multicomponent educational information leaflet intervention 13 .The information leaflet had five candidate components; (1) diagrams explaining how AET works; (2) icon arrays stating the benefits of AET; (3) side-effect prevalence information; (4) answers to common concerns; and (5) quotes and pictures from other breast cancer survivors 13 .
Typically, an information leaflet intervention may be evaluated using a parallel group randomised controlled trial (RCT), comparing whether the leaflet as a whole is more effective than a suitable comparator, such as usual care.However, in a parallel group RCT, effects of the individual components of the leaflet, and interactions between intervention components cannot be estimated 14 .It is possible some intervention components could be redundant, or could have a negative effect on beliefs about AET, meaning the effectiveness and efficiency of the leaflet may be compromised 14 .
The multiphase optimisation strategy (MOST) is an engineeringinspired framework aiming to optimise complex interventions to balance effectiveness with efficiency 14 .MOST proposes an optimisation phase prior to definitive evaluation of complex interventions.In the optimisation phase, highly efficient and fully powered experimental designs, such as factorial designs, can be used to estimate the individual and combined effects of intervention components 14 .Empirical data from an optimisation trial can be used to select an optimised intervention 15 ; for example, selecting an intervention comprising only intervention components estimated to contribute to a positive effect on beliefs about AET.The MOST framework therefore offers the potential to balance intervention effectiveness with efficiency.
As the MOST framework is a novel approach to optimising and evaluating complex interventions, there are relatively few open access examples of data of optimisation trials.Open access datasets of factorial experiments would be useful to provide scientists learning about the framework with some real data to practice analysing and interpreting the data.Our aim was to create a dataset of our factorial experiment that is available for other investigators to use.The original aim of the factorial experiment leading to creation of this dataset was to optimise the content of an information leaflet intervention targeting beliefs about AET 13 .

Patient and Public Involvement
A panel of four women prescribed AET for breast cancer were involved in this project.The panel were recruited in 2021 via a local charity supporting people affected by cancer.Through regular (approximately quarterly) meetings with the investigator(s), the panel had input into the design of the study (i.e., targeting medication beliefs), the targets of the intervention components and the content and design of the intervention components making up the information leaflet.They additionally provided quotes of their motivations for taking AET and advised on wording of the scenario presented at the beginning of the survey to provide context around being prescribed AET 13,16 .The panel did not contribute to the conduct or recruitment of the study.After further evaluation of the information leaflet in a larger trial 17 , the panel will be consulted on methods of dissemination of study results.

Ethical approval
Ethical approval was granted from the University of Leeds School of Medicine ethical review board (MREC 21-033, 21/03/2022).Written informed consent was obtained electronically from all participants.All procedures, including the informed consent process, were conducted in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000.
Full methods and a detailed description of the development of the candidate intervention components are described elsewhere 13,18 .

Candidate intervention components
The information leaflet was made up of five candidate intervention components, each operationalised as factors with two levels: (1) diagrams explaining how AET works (levels: on/off); (2) icon arrays visually displaying the benefits of taking AET in terms of reduced recurrence and mortality (levels: enhanced/basic); (3) information about potential side-effects from AET and their prevalence (levels: enhanced/basic); (4) answers to common concerns women have about taking AET (levels: on/off); (5) quotes and pictures from breast cancer survivors, stating their motivations for taking AET (levels: on/off).

Participants
Participants were required to be female, over 18 and able to read English.A market research company recruited participants via sending the survey link to potential respondents in the UK.A total of 1,604 women completed the survey.One participant was excluded due to being under 18.

Design
We used a 2 5 (2x2x2x2x2) factorial design (Table 1).After completing basic demographic information, participants were Note.Each component had two levels: on vs off, or enhanced vs basic.
This table was taken directly from Green et al., 13 .
shown a scenario asking them to imagine they had been diagnosed with breast cancer and prescribed AET.This scenario is available elsewhere 13 .Participants could not proceed to the next page of the survey until 30 seconds had passed.
Participants completed the beliefs about medicines questionnaire.They were then randomised to one of 32 experimental conditions.Each condition corresponded to a unique version of the information leaflet made up of different combinations of the factor levels corresponding to the five intervention components.All possible combinations of factor levels made up the 32 experimental conditions.Participants could not proceed with the survey until three minutes had passed.After viewing the leaflet, participants were asked the same questions about their beliefs about AET, in addition to true or false questions about their knowledge of AET, their subjective knowledge of AET and their satisfaction with information they had received about AET.The questionnaires are available online (DOI 10.17605/OSF.IO/ 3CZ9Q) 19 .

Outcome measures
Participant demographics.Age, marital status, education level, ethnicity, menopausal status, previous breast cancer diagnoses and any close relations diagnosed with breast cancer were collected at the beginning of the survey.For women diagnosed with breast cancer, further questions about the stage and whether they had been prescribed AET were asked.

Beliefs about Medication Questionnaire-AET (BMQ-AET).
The BMQ-AET is a 10-item scale used to assess specific medication beliefs 20 .It is made up of two subscales; necessity and concerns, each with five items.Concern scores were subtracted from necessity scores to create a BMQ differential score.A BMQ differential score was calculated for the BMQ completed before and after viewing the information leaflet.
Satisfaction with information about Medicines (SIMS).A modified version of the SIMS was used 21 .Seven of the original SIMS items were removed due to not being relevant.One item was added asking about the benefits of taking AET.Participants were asked to rate their satisfaction with 11 different aspects of information about AET on a 5-point scale; too much, about right, too little, none received or none needed.Responses of "about right" and "none needed" indicated satisfaction with information and scored 1, while all other responses indicated dissatisfaction and scored 0. A total score was calculated (0-11).
Objective knowledge.Eight true or false items assessed objective knowledge of AET, relating to the mechanisms, benefits and side-effects of AET.Items answered correctly were scored as 1, items answered incorrectly were scored as 0. A total knowledge score was calculated.
Subjective knowledge.One item assessed participants subjective knowledge of AET; "How informed do you feel about hormone therapy for women with breast cancer".Participants answered on a 0 (not very well informed at all) to 10 (very well informed) subscale.
Engagement.To observe engagement with the information leaflet, the length of time participants spent on the information leaflet page of the survey was recorded.

Statistical considerations
Missing data.There was no missing data as all survey items were mandatory.Any non-completed responses were not recorded.
Sample size.A sample size of 1,524 was required to detect an effect size of 0.15, with power of 0.9 and alpha set to 0.1.Alpha was set to 0.1 as a decision-priority approach was taken in which Type I and Type II errors are equally detrimental to selecting an optimised information leaflet 15 .Assuming 5% of participants would be speed responders (not completing the survey correctly), we increased the sample size to 1,604.The 'MOST' package in R Studio version 4.2.0 was used to calculate sample size 22,23 .
• Clean dataset-the dataset after data cleaning (csv file).
• R script for data cleaning-the code used to clean the raw dataset (R script).
• Data notes-explanation of the variables in the cleaned dataset (PDF).

Extended data
Open Science Framework: Extended data for 'Dataset for a randomised factorial experiment to optimise an information leaflet for women with breast cancer', https://doi.org/10.17605/OSF.IO/3CZ9Q 19 This project contains the following files: • Information Leaflet Survey V1.1.pdf

Caroline Charlton
Northumbria University, Newcastle upon Tyne, England, UK Thank you for the opportunity to review this exciting article.This article is a great example of extending open science practices and knowledge sharing, outlining an interesting and novel methodology for intervention optimisation, and allowing for other researchers interested within intervention optimisation to utilise the shared dataset and learn more about the MOST framework.
The rationale for this article is clearly outlined, and the reported work is clear, concise, and technically sound.The methods provide a good overview of how the dataset was created, with links provided to previously published resources to further aid understanding on the intervention design and the decisions made during this process.Outcomes measures are clearly stated, although I would be interested to know whether there were any controls/measures in places regarding the engagement outcome, as the length of time spent on the leaflet survey page may not be indicative of actual engagement (e.g., the participant may have done something else whilst waiting to be allowed to move forward in the survey).The dataset is presented with clear instructions on how to access it and what files are contained within the dataset.
Overall, a great article which is important for sharing the novel methodology utilised.I would recommend this article to be indexed without any need for revision.

Is the rationale for creating the dataset(s) clearly described? Yes
Are the protocols appropriate and is the work technically sound?Yes

Are sufficient details of methods and materials provided to allow replication by others? Yes
Are the datasets clearly presented in a useable and accessible format?

Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Health literacy, cancer screening, intervention design and evaluation.I recently published a paper evaluating the NHS cervical screening leaflet.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Emily A. Walsh
University of Miami, Coral Gables, Florida, USA This article, Dataset for a randomised factorial experiment to optimise an information leaflet for women with breast cancer, summarized a study of 1604 health women who completed a single, brief leaflet-based intervention to promote knowledge of adjuvant endocrine therapy.Authors concisely described the methods and procedures of the study, which was based at the University of Leeds, and consisted of recruiting women via a paid market research company.This public dataset would offer open access to data for researchers to practice and examine questions related to a Multiphase Optimization Strategy, or MOST approach, which is a novel and compelling analytic strategy used to identify which components of an intervention offer the most effect.Authors note the lack of current public datasets available to analyze via a MOST approach, and thus, this dataset will provide meaningful and impactful opportunities to researchers practicing and utilizing this approach.Notably, authors recruited healthy women to partake in this study on knowledge of a cancer treatment, which may slightly impact generalizability of related findings of this dataset versus a patient sample of women with breast cancer.However, the benefits of this dataset outweigh this minor consideration, which should be stated in subsequent, related publications from this dataset.

Is the rationale for creating the dataset(s) clearly described? Yes
Are the protocols appropriate and is the work technically sound?Yes

Are sufficient details of methods and materials provided to allow replication by others? Yes
Are the datasets clearly presented in a useable and accessible format?

Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Psycho-oncology, clinical psychology, breast cancer, adjuvant endocrine therapy adherence and medication management I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Open Peer Review Current Peer Review Status: Version 1
This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.