Study aims and hypotheses
The aims of this study are to evaluate the impact of Navigate for men with low risk prostate cancer and their partners on: uptake of AS as a first-line management option; men’s preparedness for decision-making; men’s decisional conflict, regret and satisfaction; the quality of men’s illness communication; and men’s prostate cancer-specific quality of life. The secondary aims are to: estimate study arm differences in men’s anxiety; explore the impact of Navigate on partner’s decisional conflict, regret and satisfaction, and quality of illness communication; assess the healthcare cost impact and cost-effectiveness of Navigate (economic sub-study); and determine the specific patterns of use of Navigate (web analytic sub-study).
Specific hypotheses are that: a higher proportion of men randomised to Navigate will select AS as a first-line management option when compared with men randomised to the control group; and men randomised to Navigate will feel better prepared for decision-making, experience lower levels of decisional conflict and regret and higher levels of decisional satisfaction, reported better quality of illness communication and better prostate cancer-specific quality of life when compared to men randomised to the control group.
Research questions, rather than hypotheses, were developed for secondary aims, since no specific predictions were made. Research questions relevant to the main study include:
Do men randomised to Navigate report higher or lower levels of anxiety than men in the control group?
Do the partners of men randomised to Navigate experience lower levels of decisional conflict and regret, and higher levels of decisional satisfaction when compared with the partners of men randomised to the control group?
The research question for the economic sub-study is:
What is the cost-effectiveness of the decision-aid intervention compared with usual care for men newly diagnosed with prostate cancer?
The research questions for the web analytic study include:
What are the general patterns of use of the website; are there any identifiable areas for improvement from a user experience perspective?
Do partners and men use the website differently?
What are the patterns of use across individuals relating to the values clarification exercise?
Do patterns of use relate to the primary and secondary outcome variables?
Trial design and randomisation
This study will use a parallel group, prospective randomised controlled trial (RCT) with one baseline and three post-baseline assessments (i.e. follow-ups 1, 2 and 3), along with economic and web analytics substudies. In this study, ‘partner’ refers to the support person nominated by the patient. Both patients and their partners (if one is nominated) will be invited to participate. Following informed consent and completion of baseline study measures, participating patients and their partners will be randomly assigned to Navigate or the control arm with a one-to-one allocation, stratified by recruitment site. Randomisation will be undertaken remotely and independently by the trial coordinator using a purpose-built Microsoft Access randomisation database. Post-baseline assessments will occur approximately one, three and, six months after consent. Participants will be informed of their allocation after completing baseline questionnaires. Researchers involved in data collection will be blinded to group allocation. Statisticians will be unblinded to allocation before preparation of the participant flow diagram and outcome analysis. Ethical approval was received from the Peter MacCallum Cancer Centre Human Research Ethics Committee (HREC16 PMCC114).
Participants and study setting
Australian patients will be eligible to participate regardless of whether or not their partners participate, whereas partners are not eligible without a participating patient.
Men will be eligible to participate in this trial if they: are 18 years or older; have been diagnosed with LRPC in the last three months and have yet to make a treatment decision; have been deemed eligible for AS by their treating clinician; and have internet access and sufficient English to complete study requirements and use the Navigate website. Men will be ineligible if they have a severe psychiatric or cognitive disorder or are too unwell to participate as deemed by their treating clinican, self-report or the research team at the time of approach.
Partners are eligible to participate in this trial if they: are 18 years or older, are the designated partner identified by the consenting patient; and have internet access and sufficient English to complete study requirements and use the Navigate website.
Participating sites include up to five treatment centres within Victoria, Australia, and one in Queensland, Australia. Men with LRPC and their partners from either public or private health care settings are also be able to self-refer online to the trial.
Intervention: Intervention participants are emailed login details to access the Navigate online decision aid. Using Navigate, men and their partners are led through information on each treatment option for LRPC with its potential benefits and side-effects. An interactive values clarification exercise assists the men with weighing up the advantages and disadvantages of each management option to clarify what matters to them and to guide their preferences.
Usual care: Usual care participants are provided with minimal information to fulfil ethical obligations and emailed a link to the Prostate Cancer Foundation of Australia website page for LRPC treatments. This website provides brief information on LRPC and treatment options, but differs from the Navigate website in that it does not include comparisons of the pros and cons of each management option; a values clarification exercise nor the presentation of material in graphical or multimedia formats.
Recruitment and consent
Potentially eligible men will: i) are identified by their treating clinician/nurse or the site investigator using screening clinic lists; or ii) self-refer to the website via digital marketing strategies. The treating clinician can also refer interested patients directly to the study team. Once men and their partners provide consent and complete baseline questionnaires, they are provided with the appropriate website link as per their group allocation. Non-consenters are asked for de-identified demographic and clinical details. Attrition will be monitored and reasons for withdrawal recorded.
Hospital recruitment and consent
Eligible men are approached after their diagnosis by research nurse/assistant, consistent with the agreement of those patients’ treating doctors. The research nurse/assistant reviews the Patient Information and Consent Form (PICF) with interested patients and patients provide online consent via the Qualtrics survey platform. Using the same consent procedure, partners are invited to participate. Consenting patients are also asked to provide written consent for the study team to obtain data from the Medicare Benefits Scheme and Pharmaceutical Benefits Scheme databases.
For patients or partners who are unsure about participating or do not have time to complete the PICF in clinic, a research team member is able to contact the patient/partner by telephone within a week to determine participation status. If the patient/partner agrees to participate, they are emailed a personalised PICF link.
Self-referral, recruitment and consent
The study is being promoted using digital marketing and community engagement strategies (Google Ads, Facebook, social media, blogs, commentary articles). Men who self-refer to the study are asked to complete an online Expression of Interest form including the details of their treating clinician who will be contacted by the study team to confirm eligibility. If eligible, a research nurse/assistant will undertake the consent process with the patient over the telephone.
Clinician referral, recruitment and consent
The treating clinician can also refer their patients directly the study confirming they meet the eligibility criteria and they have discussed the trial with his/her patient who is happy to be contacted. A research nurse/assistant is then available to undertake the consent process with the patient over the telephone.
Procedure and Assessments
After consent, the research team emails a link to access this questionnaire to be completed at home. A reminder telephone call is made if the study measures have not been completed within 48 hours.
Follow-ups 1, 2, and 3 are emailed to participants respectively at 1, 3 and 6 months post-consent. If follow-up questionnaires have not been completed within two weeks, a reminder telephone call is made. After completion of follow-up 3, participants are contacted via email and telephone to confirm trial completion, and to thank the participant for their time. The trial flow chart is presented in Figure 2.
<Figure 2 about here>
Main study measures
Demographic and clinical information
Demographic information collected from patients and partners includes: age (in years), marital status, postcode, highest education level, occupation and ethnic origin.
Clinical information (Gleason score and PSA) is obtained from the referring site; the referring clinician, or collected via a medical record audit.
The primary outcome is self-reported uptake of AS as the first-line management option for LRPC assessed as a dichotomous outcome (AS or curative treatment option) at follow-up 1.
Preparedness for decision-making is assessed at follow-up 1 with the preparedness for decision-making scale (PrepDM). The PrepDM total scale measures the perceived usefulness of the decision aid in preparing the patient to communicate with his doctor(s) to make a health decision and has shown acceptable internal consistency (Cronbach’s α=0.92 to 0.96), and Item Response Theory analyses demonstrated that all ten scale items function well (38).
Decisional conflict is assessed at follow-up 1 with the 16-item Decisional Conflict Scale (DCS). The DCS measures patients’ perceptions of uncertainty, being uninformed, unsupported and having unclear values in decision making. The DCS is suitable for use with cancer patients and has shown acceptable internal consistency (Cronbach’s α=0.78-0.92), test-retest reliability (r=0.81), and responsiveness (longitudinal validity) (39).
Decisional satisfaction is assessed at follow-up 2 and, for patients only, follow-up 3 with the 6-item Satisfaction with Decision (SWD) measure. The SWD has shown acceptable internal consistency (Cronbach's α=0.86), discriminant validity and responsiveness (31, 40).
Decisional regret is assessed at follow-up 2 and, for patients only, follow-up 3 with the 5-item Decisional Regret Scale (DRS). The DRS has shown acceptable internal consistency (Cronbach's α =0.81-0.92) and convergent validity with related measures (r =0.31-0.60) and responsiveness (31).
The quality of men’s and partners’ illness communication is assessed at baseline and follow-up 1 with the 4-item Couples’ Illness Communication Scale (CICS). The CICS was adapted for use in prostate cancer populations. The original scale has shown acceptable internal consistency (Cronbach’s α= 0.80) and convergent validity with other measures of dyadic adjustment (41).
Anxiety is assessed at baseline and follow-up 1 with the 7-item PROMIS Emotional distress-Anxiety 7a short-form. The Anxiety 7a is a standardised, valid and precise measure specifically developed for use in clinical oncology research (42).
Prostate cancer-specific quality of life is assessed at baseline and every follow-up with the 26-item Expanded Prostate Cancer Index Composite short-form (EPIC-26). The EPIC-26 comprises four subscales: urinary, bowel, sexual and hormonal. The hormonal subscale is not relevant in this context, so will not be administered. Subscales have shown acceptable internal consistency (all Cronbach’s α >0.70), test-retest reliability (all r>0.69) and responsiveness when used independently (43).
The sample size calculation was based on 80% power, a two-sided α=0.05 test and a difference in uptake of the AS treatment option of 15%. Assuming 65% uptake in the usual care group (3) and 80% uptake in the intervention group (a conservative assumption), the required sample size is 272 patients (136 in each study arm). Our sample size calculation was performed with PASS version 16. Allowing for 10% attrition a total sample size of 304 participants is required.
All analyses will be performed using Stata 16. Prior to formal analysis, descriptive statistics and graphical displays will be used to identify missing and out-of-range values, assess distributional assumptions and identify outliers.
Recruitment bias will be assessed by comparing demographic and clinical characteristics of consenters and non-consenters using t-tests (or Mann-Whitney) and chi-squared (or Fisher’s exact) tests as appropriate. Differential attrition will be assessed by comparing baseline characteristics of drop-outs and continuing participants using the same statistical tests.
The primary analysis will follow the intention-to-treat principle and all study participants will be analysed as part of the study arm to which they were randomised in all outcome analysis. Missing data will be imputed using multiple imputation, redrawing 50 samples. Pearson’s chi squared test will be used to analyse study arm differences in the primary outcome. Regression will be used to analyse study arm differences for preparedness for decision-making (follow-up 1 adjusted for baseline), illness communication (follow-up 1 adjusted for baseline) and anxiety (follow-up 1 adjusted for baseline). Independent-samples t-tests will be used to analyse study arm differences for decisional conflict (follow-up 1), satisfaction (Follow-ups 2 and 3) and regret (follow-ups 2 and 3). Finally, prostate cancer-specific quality of life will be analysed with a random effects mixed model; more specifically, a reference cell model with the mean of the control group at Baseline used as the reference (44). An appropriate covariance matrix will be used and the model will be estimated by maximum likelihood. Effect size estimates for between-group differences on self-report measures will be calculated using recommended procedures (45, 46). Results will be presented with 95% confidence intervals and a p-value less than 0.05 (two-sided) will be deemed to be statistically significant.
Participants will be analysed as part of the study arm to which they were randomized in all exploratory analysis. Regression will be used to analyse study arm differences in the following exploratory outcomes: men’s anxiety (follow-up 1 adjusted for baseline) and partners’ illness communication with men (follow-up 1 adjusted for baseline). Independent-samples t-tests will be used to analyse study arm differences in the following secondary outcomes: partners’ decisional conflict (follow-up 1), satisfaction (follow-up 2 and 3) and regret (follow-up 2 and 3). Again, effect size estimates for between-group differences on self-report measures will be calculated using recommended procedures (45, 46).
The aim of the economic sub-study is to assess the financial burden and cost-effectiveness of the Navigate online decision aid in the Australian context. This will compare the costs and patient outcomes of intervention versus usual care aligning to the goals of the planned RCT. The analysis will take a societal perspective and include Medicare data for all patient participants to assess healthcare costs to government and patient out-of-pocket expenses from participant surveys.
Health utility data will be collected at baseline and follow-ups 2 and 3. The 5-item EuroQol-5D (EQ-5D-5L) will be used, which is a standardised measure of health status developed specifically for economic evaluation and is suitable for cancer patients. It has shown acceptable internal consistency (Cronbach’s α=0.71), test-retest reliability (kappa=0.7), convergent validity (r>0.49), discriminant validity and responsiveness (47, 48).
Patient health care costs will be reported by patients at follow-ups 2 and 3. These include out-of-pocket expenses for all medical services, as well as costs for travel, accommodation and income lost from interrupted employment. The Patient Costs Questionnaire (PCQ) is purpose-built for Australian men with prostate cancer (49) and includes items such as general practice and specialist visits, counselling and support services, as well as sexual and incontinence aids. Follow-up 3 will also include the COST-FACIT questionnaire (50) capturing financial hardship and additional questions used previously by men with prostate cancer relating to early retirement and financial situation.
At the end of the data collection period, Medicare data on all services and medicines for all consenting participants will be accessed. With this data, the types and dollar amounts exchanged for health services, community services and medicines for prostate cancer and other diseases will be assessed, including medical services for privately-insured patients. The main economic outcome will be incremental cost per quality-adjusted life year for the Navigate tool versus usual care. Methods for this economic evaluation will be governed by standardised guidelines in modelling studies and economic evaluations (51, 52).
Web analytics sub-study
Website analytics will be used to indicate: preferred modes of interaction; which aspects users (participants) find most engaging; and how users prefer to access information on the Navigate website. This provides indicative answers to specific questions, such as how comprehensively users consume the information regarding the different management options before submitting preferences on decisions. Analysis of user behaviour may lead to a better understanding of the effectiveness of the decision tool, as well as potentially identifying areas for improvement. Planned data capture includes: click information; pages visited; time spent on each page; information viewed and/or downloaded; and values clarification exercise responses. Since data is associated with user ID, behaviours can be linked to demographic characteristics such as age, sexual orientation, or user type (i.e. patient or partner).
Analysis will be restricted to the intervention group of the randomised study since the focus of this sub-study is the characteristics and effectiveness of the web-based decision tool. Effects to be investigated include informational value of pages (determined by number of visits and time spent), as well as complex behavioural effects such as the amount of information consumed before coming to decisions. Variance across demographic sub-groups will be measured using appropriate statistical metrics (e.g., ANOVA, chi-squared) to determine demographic-based preferences. Data mining techniques (e.g., association-rule mining) will be applied to attempt to determine other associations between behaviour characteristics (e.g., consuming certain information leading to certain decisions and/or outcomes).