Study design

To address the primary aim, this comparative effectiveness study involves a cluster-based randomised trial at four University of Texas MD Anderson Cancer Center Houston Area Location oncology clinics (see figure 1). Two control sites will be randomly assigned to continue with usual care; two intervention sites will receive provider training, access to the Pathways patient decision aid and follow-up telephone counselling for patients to facilitate decision-making and referral to a fertility specialist, if desired. At the end of the study, discussion sessions will be held with the providers at each site regarding their experience and recommendations for intervention improvement.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study design.

This protocol and the overarching programme of research is based on the underlying decision-making and cognitive psychology theories of the Ottawa Decision Support Framework, and follows the quality guidelines of the International Patient Decision Aid Standards (IPDAS) Collaboration.27–32 35–46The core research team includes a reproductive endocrinologist (TLW), women’s health advanced practice provider (DAH), decision scientists (RV, ASH) and research assistant (LCC). A Stakeholder Advisory Panel included three female cancer survivors who had previously considered fertility preservation, two patient advocate leaders and two oncology providers (gynaecological and paediatric).

Eligibility criteria

Women aged 18 to 45 years old who can read, write and speak English; are at risk of cancer-related infertility; and are newly diagnosed with a breast tumour, female genital system tumour, colorectal tumour and/or lymphoma or myeloma are eligible for inclusion in the study. These criteria were chosen to align with the current guidelines for fertility preservation discussions.1 All women will be recruited from The University of Texas MD Anderson Cancer Center Houston Area Location oncology clinics. These clinics were chosen because they serve a large and diverse population, have a centralised electronic health record for tracking referral and treatment utilisation and may be more generalisable to the US population than the MD Anderson main campus. Providers of these clinics will be eligible for inclusion in the poststudy provider discussions.

Randomisation

We will generate a randomisation list for the four oncology clinics using nQuery Advisor (1995–2007, Statistical Solutions) with two study arms (control, multicomponent intervention) and a block size of 4.

Treatment arms

At the two sites randomised to the control condition, oncologists will proceed with usual care, which involves an oncology consultation and offering a referral to the fertility specialist, if desired.

At the two sites randomised to the intervention, three components will be provided—provider seminar, access to Pathways and follow-up telephone counselling. Dr Woodard (a fertility specialist) will present a departmental seminar designed to (1) enable and motivate oncologists to address fertility issues in women at risk of cancer-related infertility and refer them to reproductive endocrinologists, if warranted;  (2) introduce the Pathways patient decision aid; and (3) describe the study procedures so that providers can introduce the study to eligible women.

Second, all participants at the intervention sites will be provided with access to the Pathways decision aid website (v1.0, 1 April 2017) after their initial oncology consultation. Results of the formative studies (provider and patient needs assessments, user-centred design and production and usability/acceptability pilot-testing) and efficacy study are published separately.33 Selected screenshots and the overall architecture of the Pathways website are provided in figure 2; scores on the IPDAS Quality Checklist are provided in online supplementary file A.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Components and features of the Pathways patient decision aid website. *An interactive personalisation activity (eg, open-ended goal-setting questions, Values Clarification Rating Scales, initial treatment leaning items); responses are collected in the My Personal Decision Summary.

Pathways provides women with an introduction to the effects of cancer on fertility; descriptions of fertility preservation and other family-building options; and interactive My Personal Decision features that support women in personalising the medical information, clarifying their decision-making values, comparing the relevant options and preparing for their discussions with their providers and family. Pathways tailors the information to each woman’s cancer type and provides explanations of the oncofertility terminology and procedures in eighth-grade language. Each woman’s My Personal Decision information is provided in a printable summary.

Within the following week, follow-up telephone counselling for participants will be offered to support informed, values-based decision-making as fertility laboratory results and cancer treatment plans become available, and to facilitate navigation and timely referrals to a fertility specialist, if desired.

Outcomes

Table 1 illustrates the study data collection for each objective and time point (baseline, 1 week and 2 month). Online supplementary file B provides the psychometric properties for each measure/instrument. The primary measure is decisional conflict, assessed pre/postintervention using the 16-item Decisional Conflict Scale.47 All measures have been tested during the formative studies and pilot-testing, as well as in other fertility preservation or other patient decision aid research studies.

Baseline characteristics will include clinical (Reproductive Concerns Scale, Brief Symptom Inventory), decision-making (Decisional Conflict Scale, Intolerance of Uncertainty Scale) and sociodemographic factors, including the Single Item Literacy Scale.20 47–51 Across time points, this study will assess women’s decision-making processes using the Decision Self-efficacy Scale, Preparation for Decision Making Scale and open-ended questions assessing other decision-making factors (eg, three primary influences on their decision, role of spouse/partner in decision-making, etc).52 53 Decision quality will be assessed using the Fertility Preservation Knowledge Scale, Values Clarification Leaning Scale and Strength of Preference for Referral/Treatment(s) Scales, as well as an assessment of the concordance of participants’ preferences with subsequent treatments scheduled or completed by 2 months.21 54 55

In preparation for future planned dissemination and implementation studies, exploratory measures include the Patient Decision Aid Acceptability Scale (ie, Leaning Scales rating the length, ease of use, clarity, comprehensiveness and meaningfulness of the decision aid), Client Satisfaction Questionnaire, system usage (eg, preferences for viewing at home or at the clinic, time spent on the website, error rates, etc) and preliminary testing of potential downstream measures (eg, Decisional Regret Scale).56–58 At the conclusion of the study, semi-structured discussions with clinicians at the intervention sites will assess clinician perspectives about the usefulness of the multicomponent intervention and suggestions for improvement. These exploratory measures will inform and guide the design of future longitudinal studies.

Adverse events

The risk of adverse events are low. However, it is possible that discussion of fertility issues can cause or increase emotional distress. If a participant is identified as being significantly distressed (ie, by notifying the study coordinator and/or scoring >63 on the Brief Symptom Inventory), they will be reminded that they can end their participation at any time, and the principal investigator or research study coordinator will refer the participant to the appropriate psychosocial support resources.50 51 An external Data Monitoring Committee is not commissioned for this protocol.

Data management

Study data will be collected and managed using REDCap (Research Electronic Data Capture, www.project-redcap.org) electronic data capture tools hosted at The MD Anderson Cancer Center.59 All protected health information (PHI) will be removed from the data when it is exported from REDCap for analysis. All dates for a given patient will be shifted by a randomly generated number between 0 and 364, thus preserving the distance between dates. A different randomly generated number will be used for each patient.

Sample size and rationale

The primary outcome measure is the percent of patients who score <25 on the Decisional Conflict Scale, as lower scores are associated with making decisions (ie, less uncertainty, anxiety or distress).47 60 We will compare the two study arms (usual care, intervention) with respect to the change from baseline in the percent of patients who score <25 on the Decisional Conflict Scale. In a review of 31 cluster-based studies in primary care, Adams et al found that the median unadjusted intracluster correlation was 0.011.61 Assuming a similar intracluster correlation, 50 patients on each study arm (25 at each oncology clinic) will provide an approximately 80% power with a two-sided significance level of 0.05 to detect a difference of 30% between study arms with respect to the change from baseline in the percent of patients who score <25 on the Decisional Conflict Scale. This sample size calculation was performed using Number Cruncher Statistical Systems Trial and Power Analysis and Sample Size Software 2005 (Number Cruncher Statistical Systems, Kaysville, Utah; www.ncss.com).

The four MD Anderson Houston Area Location oncology clinics see an estimated 250–300 potentially eligible patients per year (21–25/month), and observe a socioeconomically, racially/ethnically and clinically diverse population. Conservatively assuming a 50% participation rate, we anticipate enrolling 10–12 women/month from 1 September 2017 to 30 May 2018. Participants will be provided with $25 gift cards at 2 months postenrolment. If additional recruitment is needed, the MD Anderson main campus oncology clinics may be added, where previous studies in this programme of research have observed a 75%–90% participation rate.

The four oncology clinics will be assigned site identification numbers in alphabetical order, then randomised using nQuery Advisor 7.0 (1995–2007, Statistical Solutions). The randomisation list will be saved in a separate list by the study statistical team.

Statistical analysis plan

The data analytic plan begins with descriptive statistics and boxplots to summarise patients’ characteristics and scores on each of the survey instruments at each assessment time point for each study arm.

With respect to the primary outcome of the dichotomised decisional conflict score, the statistical team will first tabulate the counts and frequencies. To assess the effect of the proposed multicomponent oncofertility DS intervention compared with usual care, they will use generalised linear mixed models (GLMMs) with one covariate of the study arm and a random effect that takes into account the between-cluster variation. In analysing cluster randomised trials, Klar and Darlington62 showed that there could be considerable gains in power when covariates were adjusted, and a similar phenomenon was also observed in a peer-reviewed research study that investigated covariate adjustment in randomised controlled trials with dichotomous outcomes.62 63 Indeed, covariate adjustment is often recommended to achieve unbiased estimates and to improve the precision, thus increasing the power.64 As such, our statistical team will evaluate the intervention effect via GLMM with adjustment of baseline covariates that may further explain the variation in the primary outcome, hence resulting in potentially improved power. Covariate variables considered in this analysis include age, religion, relationship status, insurance type/coverage, median household income, gravidity, parity, Reproductive Concerns Scale and Fertility Preservation Knowledge Scale. Our selection of this list is based on the conceptual framework underlying patient decision-making, which includes patients’ sociodemographic, clinical and decision-making characteristics as contextual factors that may potentially impact decision-making outcomes.

The statistical team will also model the logit of the probability of achieving a Decisional Conflict Scale score <25 as a function of study arm, assessment time and patient nested within provider using GLMM. The aforementioned covariate list will be considered first in this analysis. After a scientifically reasonable and mathematically stable model is constructed, the statistical team will further investigate other potentially important covariates including race/ethnicity, employment, education, Intolerance of Uncertainty Scale and Brief Symptom Inventory. Last but not least, if distributions allow, additional models will explore the probability of achieving other score cut-offs (eg, <37.5, since scores higher than 37.5 are associated with delaying decisions) and the differences in the change in score pre/postdecision aid.

For the other instruments, the statistical team will use GLMM methods to model scores as a function of study arm, assessment time, patients nested within provider and patients’ characteristics to address the secondary outcomes of decision-making process (eg, preparation for decision-making, decision self-efficacy and satisfaction) and decision quality (eg, fertility preservation knowledge, clarification of patients’ values and congruence of preferences with the decision about whether to accept fertility preservation and/or fertility preservation treatment).

GLMM methods are designed to handle missing data and give unbiased estimates of effects provided that the probability of having missing data depends only on the covariates in the model (or data are missing at random). However, in the case that data are not missing at random, to avoid the bias due to the informative dropout, analyses will compare baseline information and reasons for dropout to examine whether the dropouts are systematically different from non-dropouts. In addition, a non-ignorable model, such as the pattern mixture model, will also be used to fit the data to account for possible informative missing data. As a sensitivity analysis, the results from the non-ignorable model will be compared with those from the standard mixed model.65

Finally, exploratory analyses will tabulate site usage and research feasibility outcomes (eg, time on website, rates of completion of all data collection items, etc). The research team will review open-ended responses and notes from the poststudy discussions with providers to identify any suggested improvements to the decision aid or future implementation.

Ethics and dissemination

Online supplementary file C provides an example of the approved informed consent document (note: this article refers to Part II activities). In addition, the MD Anderson Cancer Network Protocol Review, Integration, and Strategic Management (PRISM) provided initial approval and ongoing review for this study at the four Houston Area Location oncology clinics. Any amendments to this protocol will be reviewed and approved by both boards, and communicated to the collaborating sites, participants and journals, if appropriate. All eligible women who volunteer to participate will be asked to provide informed consent and will be registered in MD Anderson’s Clinical Oncology Research System, and periodic audits may be performed to ensure adherence to protocol.

A Data Monitoring Committee is not required for this study due to low risk of adverse events; however, as a conservative measure, automatic notifications are sent to the core research team for any women who score high (indicating depressed feelings) on the Brief Symptom Inventory.50 If that should occur, the clinical team will be notified and appropriate social services will be provided in accordance with clinical policies. The principal investigator maintains the authority to suspend or terminate the study at any time.

Results of the formative developmental studies (provider and patient needs assessments, user-centred design and production, and usability/acceptability pilot-testing) and efficacy study are published separately.33 These results were also peer reviewed and presented at the scientific meetings of the American Society for Reproductive Medicine and the Society for Medical Decision Making.

Results of this comparative-effectiveness cluster randomised trial will be published in a peer-reviewed journal. Manuscripts will also be prepared for any significant findings for the secondary aim, as appropriate, in peer-reviewed journals. These results will be submitted for peer review for presentation at the scientific meetings of the American Society for Reproductive Medicine and the Society for Medical Decision Making.

Online supplementary file D provides the Standard Protocol Items: Recommendations for Interventional Trials checklist. On completion of the trial and publication of the primary manuscript, requests for access to the Pathways patient decision aid website and database may be made to the corresponding author.