INteractive Survivorship Program to Improve Health care REsources [INSPIRE]: A study protocol testing a digital intervention with stepped care telehealth to improve outcomes for adolescent and young adult survivors

Abstract

Background:

Adolescents and young adults with cancer (AYAs, ages 15–39 at the time of diagnosis) experience significant adverse health and psychosocial outcomes. AYAs live with emotional distress and health care demands that exceed those of their healthy peers but can have difficulty accessing care. Digitally delivered interventions are an attractive option for AYA survivors, a population that routinely utilizes online resources when seeking health information and support.

Aim:

By improving access to survivorship resources and support and strengthening health literacy and self-management skills, the INteractive Survivorship Program to Improve Health care REsources [INSPIRE] is designed to improve adherence to AYA health care guidelines and reduce cancer-related distress. We describe the protocol for a two-arm randomized controlled trial (RCT) testing the AYA-adapted INSPIRE program.

Methods/Design:

The intervention includes an interactive mobile app, study website, and social media platforms, adding telehealth for those with continued distress, lower survivorship health care literacy, or poor engagement with the digital program at 6 weeks. Participants are randomized to INSPIRE or an active control. In the active control arm, survivors receive access to a study website with links to existing AYA survivor resources followed by delayed access to the INSPIRE program. Participants are not blinded; study staff not providing telehealth are blinded. The primary outcomes are cancer-related distress and health care adherence specific to second cancer and cardiometabolic screenings.

Discussion:

If effective, the program is positioned for accelerated implementation to improve care for AYA survivors by using a scalable informatics-based administration and largely digital intervention program.

Background

Adolescents and young adults with cancer (AYAs, age 15–39 at diagnosis) experience significant adverse health and psychosocial outcomes.(1–4) AYAs will spend most of their lives as cancer survivors with physical, social and emotional needs that differ from their peers and from those of younger or older cancer survivors. They have significant risks for long-term health complications including secondary malignancies (SM), and accelerated development of usual age-related comorbid conditions such as cardiometabolic dysfunction (CMD), with underdiagnosis and undertreatment of these conditions and major gaps in health care maintenance.(5–15)

By improving survivors’ personalized health literacy, self-efficacy, resource access, and emotion management options through digital delivery of a self-management program, this intervention is designed to improve health care adherence (HCA) to guidelines specific to AYAs and reduce cancer-related distress. We will leverage digital advances by adapting an intervention we developed for hematopoietic cell transplant survivors to needs of AYAs.(16, 17) This INteractive Survivorship Program to Improve Health care REsources (INSPIRE) intervention incorporates an interactive mobile app/website and manualized telehealth stepped care component for those not benefiting from the digital methods alone. Stepped care is a medical model of optimizing resources by adding more intensive care for those who require it after first providing less intensive care.(18) We will test the intervention in an established, national multicenter survivorship research network that has previously completed two AYA survivor studies.(19–22) This study focuses on three areas: cancer-related distress and health care adherence specific to SM and CMD screenings. In a previous trial of AYAs that used the same distress measure that is used in this RCT, 73,5% were classified as having high distress.(23) AYAs experience more distress than older cancer survivors or cancer-free peers.(24)

CMD, characterized by insulin resistance with impaired glucose tolerance, atherogenic dyslipidemia, hypertension, and intra-abdominal adiposity, predisposes individuals to the development of cardiovascular disease (myocardial infarction, heart failure, stroke) and diabetes, and is one of the most significant challenges facing AYA cancer survivors. In a study of over 5600 AYA survivors the risk of adverse CMD outcomes was two-fold greater than >57,000 patients without cancer.(25)

As the leading cause of non-relapse late mortality in AYA survivors, SM, are one of the most significant long-term health risks for AYAs.(26–28) While all survivors have an elevated risk for SM, those exposed to radiation have the highest risk. (29, 30) In a study of secondary breast cancer in young AYAs, almost half of the survivors received radiation as part of their treatment.(31) From a cohort of Danish AYA lymphoma survivors, over half had radiation as part of their treatment.(32)

At least half of AYAs do not adhere to recommended survivorship health care guidelines.(33, 34) In qualitative studies, AYAs note many barriers to receipt of guideline-concordant care, including lack of continuity of care or clear provider recommendations, fear of recurrence, wishing to move on with life, competing life responsibilities, (e.g. work and parenting), and not perceiving the need for follow-up due to lack of symptoms.(35) AYA-focused interventions that increase self-efficacy for disease management also improve health-promoting behaviors and physical and mental health, as well as adherence to CMD and cancer screening.(36, 37)

Survivorship care plans (SCPs) are recommended for all cancer survivors and are included in the 2020 Commission on Cancer accreditation mandates.(38, 39) Delivery of a printed SCP, without follow-up to personalize the application of the information to the individual, has not improved health care adherence (HCA), although our studies demonstrate lower distress among survivors who were provided with personalized SCPs in a context of an individualized phone call or the online INSPIRE program, or a clinic visit with follow-up calls for AYAs.(17, 22, 40–48) Personalizing an action plan for the individual through online interaction and telehealth follow up to address barriers to HCA enhance the effectiveness of SCPs (49, 50)

The specific aims of this study are: 1) Determine the efficacy of a self-management program, INSPIRE, delivered by interactive digital cross-device options and stepped care telehealth coaching, compared to an active control arm among AYA survivors with elevated cancer-related distress; 2) Among AYA survivors with inadequate adherence to health care guidelines, determine the efficacy of the INSPIRE digital and telehealth self-management program in improving CMD or SM surveillance compared to the active control arm. 3) Examine mediators associated with intervention efficacy [including engagement with the online program, participation in telehealth contacts, and improvements in self-efficacy and health literacy, AYA-specific impact of cancer, and barriers to health care; and examine moderators of intervention engagement or response including cancer clinical factors, health status, and sociodemographic and environmental factors]. 4) Estimate costs to maintain and deliver INSPIRE program components if delivered in clinical practice. Aims will be accomplished with a two-arm, multicenter, risk-stratified RCT of N=700 AYA survivors of leukemia, lymphoma, melanoma, sarcoma, colorectal, breast or thyroid cancer diagnosed between ages 15–39, and who are 1–5 years post-diagnosis when enrolled. The primary outcome is reduced distress at 3 months; secondary outcomes are improved adherence to screenings for CMD and SM and reduced distress at 12 months.

Methods

Ethics Approval

This study was approved by the Institutional Review Board (IRB) of Fred Hutchinson Cancer Center. Because of the multi-site nature of the study, the IRBs of Dana Farber Cancer Institute, Memorial Sloan Kettering Cancer Center, University of California at Los Angeles and the University of Pennsylvania agreed to rely on the Fred Hutchinson Cancer Center’s IRB. This study is registered through ClinicalTrials.gov (NCT04593277). The Data Safety and Monitoring Board (DSMB) consists of four members unassociated with the study who meet annually to review enrollment and adverse events.

Study Design

The study is an RCT with an active control condition. Participants complete baseline assessments; those with elevated distress or non-adherence to CMD or SM screenings are randomized to control vs. the INSPIRE program intervention. The intervention includes a website, mobile app, optional private social media groups, and a personalized treatment summary and survivorship care plan which is loaded to the participant’s INSPIRE website and mobile app along with specific individualized goals to address gaps in health care adherence and distress management. At 6 weeks, intervention participants randomized to the INSPIRE program complete a brief assessment to determine their need for stepped care telehealth. Four telehealth sessions use manualized Problem-Solving Therapy (PST) to facilitate participant engagement with the digital INSPIRE Program, and support goal setting related to CMD and SM screening adherence or distress management. These calls are spaced two weeks apart. The intervention lasts 12 months. The study schema is shown in Figure 1. Inclusion and exclusion criteria are presented in Table 1.

Figure 1. Study Schema.

^a CTXD = Cancer and Treatment Distress Measure; HCA = Health Care Adherence to recommended surveillance guidelines

^b INSPIRE Program includes: mobile app, website, social media private sites, plus telehealth if needed

* Stratified based on their baseline CTXD and HCA scores

Table 1.

Inclusion and Exclusion Criteria

Inclusion Criteria
1.	Diagnosed with a first invasive malignancy of leukemia, lymphoma, melanoma, sarcoma, breast, thyroid or colorectal cancer (stage 1–3 for solid tumors) between the ages of 15–39 years
2.	Current age ≥18 when approached
3.	Currently within 1 to 5 years from the time of diagnosis
4.	Completed active treatment for disease ≥ 6 months previously
5.	Received a therapeutic intervention (with curative intent if advanced stage disease) that included any of the following modalities: surgery, cytotoxic chemotherapy, biological or targeted agents, radiation therapy
6.	English proficiency adequate to complete assessments
7.	Access to email and smartphone mobile app and/or internet
Exclusion Criteria
1.	Diagnosed with a subsequent invasive malignancy other than non-melanoma skin cancer after the first invasive malignancy
2.	Received hematopoietic stem cell transplant
3.	Health issues prohibiting computer use or ability to comply with study procedures
4.	Residing in an institution or other living situation where health care decisions are not made by the participant (e.g., hospitalized, prisoners, living in a rehabilitation facility)
5.	Does not complete baseline Patient Reported Outcome (PRO) assessment items required to determine stratification or whether the survivor meets inclusion and exclusion criteria

Each participating center obtains a list of all potentially eligible participants from their tumor registry. The study statistician has developed a program for random selection of potentially eligible participants with oversampling of racial and ethnic minorities (initially all eligible minority cases are approached) until a minimum of 35% of those enrolled are members of a racially or ethnically minoritized group. From randomly ordered recruitment lists, up to two study invitation letters and/or emails and a culturally sensitive brochure describing the study are sent from the survivor’s treatment center. The opt-in or out response page can be returned to the center or participants can go directly to the study URL to register, consent and begin participation. Centers contact survivors by phone or email if they have not responded or express interest on the returned opt-in page and send interested participants the registration link. Follow-up calls and emails continue for three contacts, leaving up to three messages until a survivor either declines participation, is determined to be ineligible, or expresses interest and is screened as eligible. At the discretion of the site investigator, participating site staff may approach and/or follow-up with survivors in person if they are scheduled for a clinic visit. Survivors who are not reached and/or do not respond after three contacts are logged as passive refusal one month after the final contact. All approach and response tracking at network centers are entered into the study tracking logs in the digital study database located at the coordinating center. Approached non-participants’ data are de-identified but tracked for summarizing enrollment, eligibility and reasons not consented.

Interested participants who are eligible for the study are sent individual log-on information for secure online consent and assessment. Survivors choose a username and passcode, register, consent, and can complete the baseline assessment in a single online session or return as often as needed to complete their survey. Full information about the purpose, goals and risks of the study, and the option to rescind consent at any time, without any impact on their care, is provided in the consent and available at all times on the study website. Consent to collect self-report data is obtained by electronic documentation of online agreement to consent and date of consent. The consent also informs participants that they will be contacted by a study investigator if their self-reported assessments indicate severe depression.

Participants who consent and complete the baseline (T1) assessment are assigned to either Group 0 if they have no impairments in the primary endpoints of cancer-related distress or HCA to CMD or SM, or Group 1 if they are impaired on distress and/or the HCA primary endpoints. We do not know who reports high distress or low HCA CMD or SM until the baseline is completed so we designated group 0 to denote those survivors who will not be randomized. For Group 1, those who have at least one of the impairments of interest, a block randomization process is based on the following stratification factors: baseline PRO screening values of primary endpoints: a) distress as measured by the Cancer and Treatment Distress: (CTXD) tool (scores <0.9 vs. ≥0.9), b) health care adherence for CMD screening tests by patient report such as blood pressure, cholesterol, glucose, triglycerides, need for EKG based on treatment exposures (percent of recommended tests completed in the past 2 years for everything but the EKG which is since the end of treatment, <80% vs ≥80%) and c) health care adherence for SM screening which includes oral exam, skin exam by a health care provider, colonoscopy and sex-specific testing such as mammogram and gynecologic exams. (percent of recommended tests completed, <80% vs.≥80%). Participants in Group 1 are randomly assigned to the control or intervention (INSPIRE) arms. Participants are notified of their randomization through email. Coordinating center staff and investigators who have contact with participants related to their enrollment, retention, and assessments are blinded to whether an individual is in Group 0 or 1 and their randomized assignments. Randomization is kept separate from other data in a protected file. Table 2 describes the possible categories of participants based on their responses to the baseline survey.

Table 2.

Categories of Participants

Group 1	Meets all eligibility criteria and has one or more elevated distress or low HCA SM or CMD and is randomized to one of the two groups, intervention or control.
Group 0	Meets all eligibility criteria and does not have elevated distress or low cardiometabolic or cancer health care adherence. These participants are provided access to the digital INSPIRE program and printed SCP materials for themselves and providers, without telehealth.
Group X	Eligible and consented but not randomized; tracked as ‘withdrawn’ because they did not complete baseline assessment. They are are not given access to the INSPIRE online program but are registered with Data Management because they consented.
Group Y	Does not meet eligibility criteria after consent. Not randomized, but completed baseline assessment (which determine final eligibility criteria), they are given access to the INSPIRE program but do not receive mailed SCP, telehealth calls, or the follow-up assessments.
Group Z	Consents but reports depression score of >=20 on the PHQ-8,(51) The study psychologist calls these participants to screen for major depression and assess for safety concerns. These participants are not randomized but are given access to the intervention similar to Group Y participants.

Intervention Arm Content and Procedures

Integrating Providers and a Personalized Survivorship Care Plan (SCP)

Participants are mailed a personalized SCP. The SCP is generated using patient-specific data regarding treatment exposures, risk modifying factors (age, sex, anthracycline and radiation exposure), and surveillance tests/exams including targeted values for lab test results. They are sent an abbreviated ‘Plan’ (matched to content to their personalized SCP in their mobile app) to share with their health care providers that includes the surveillance needed based on their treatment exposures.

INSPIRE Mobile Application

The INSPIRE mobile app includes: 1) the participant’s personalized SCP, 2) a record of health-related surveillance tests/exams received by self-report in the T1 assessment, 3) ability to track health appointments with a calendar, 4) test results that can be tracked over time, with a focus on SM/CMD screenings, 5) mood/stress tracking, 6) an interactive method for identifying a problem impacting mood/stress and selecting a ‘small step’ to try, such as mindfulness or other stress reduction methods, assessing the success of that small step, and modifying plans based on that success. The app provides up to 3 goals for each day along with mood tracking, and an ‘ask the experts’ quick question option. Using gamification methods, the app allows participants to “earn” a reward (a virtual badge) for completing surveillance tests or health or mood tracking, in order to encourage participant engagement and motivation. Notifications provide reminders for mood and health care goals and informational website links. We have an Advisory Panel for the study consisting of AYAs from the five enrolling centers who looked at the app and gave us feedback on changes from the transplant app that should be made to make it more attractive to AYAs. The INSPIRE app is iPhone and Android compatible, passcode protected, and currently provided in English.

Online Content for INSPIRE

The INSPIRE internet program is built as a responsive design for viewing on PC, Mac, tablet or mobile devices. The site is cross-platform enabled with the mobile app, linking updates in real time. In addition, the website adds the following levels: 1) A greeting home page, with links to each target area, 2) Bust Stress, Boost Mood, describing PST-based approaches to managing stress and emotions, 3) Take Heart focusing on understanding CMD health, 4) Crush Cancer focusing on understanding risks and strategies to reduce SM risks and screen for early detection of SM, 5) My Health Action Plan, focusing on the SCP, 6) Tools & Tips, with content on key issues faced by AYAs, 7) Resources with contact links.

Social Media Options

We include Facebook and Instagram social media sites in the intervention, which allow private groups that can be monitored and moderated by a trained, experienced moderator. The goal is to include a way for the AYAs to connect with one another and to the study. Only participants randomized to the intervention arm have access to these sites as well as those in the nonrandomized group. The sites provide comparable information with 5-day/week postings and encouragement for participants to share challenges and solutions. Topics are not limited, although moderator postings focus on the endpoints of distress and health care management. Intervention participants are not required to join. We will track who joins as well as the metrics for each platform on views.

Orientation Call

Two weeks after randomization participants receive a 15–30 min scripted orientation call to assure that they understand the INSPIRE program goals, can download the mobile app and/or link to the website and social media, to assist participants in deciding how they will use the program components, and address barriers to using INSPIRE. Calls are audio recorded for reviewing fidelity to the script.

Week 6 [T2] Brief Screening to Determine Need for Telehealth Stepped Care

Intervention participants complete a brief (2 minute) screen at 6 weeks after their orientation call to determine telehealth need based on whether any of 3 criteria are met: 1) Elevated distress/depression is >4, or the first 2 items on the PHQ-8 are scored >2 indicating distress/depression symptoms. 2) Low survivorship self-management knowledge on the survivorship knowledge measure (10-items scored 0 [‘my risk is higher’] or 1 [‘my risk is not different’ or ‘not sure’]) is scored >3. 3) Whether the participant has not viewed at least 3 pages of the website or app as assessed by website and/or app metrics.

Stepped Care PST Telehealth Contacts

Group 1 intervention participants who at 6 weeks indicate no use of the online program, lack of knowledge of their CMD or SM risks, or report elevated distress will receive 4 calls via phone. We have included a lack of engagement with the telehealth calls as criteria as we have demonstrated in the first INSPIRE trial that engagement leads to improved outcomes.(52) Calls, using manualized PST methods, are 45 minutes for the first call and 15–30 minutes for subsequent calls. Calls are led by a ‘coach’ trained by the clinical investigators in AYA PST-oriented self-management, with a nursing, social work or psychology degree. The AYAs can choose which problems within the three outcomes to work on during the calls. In particular, the kinds of problems that can cause stress in AYAs such as parenting, fertility and work challenges can be different from older cancer survivors. The first session focuses on gaps indicated by the week 6 screening. Participants define goals to address their distress, adherence or health literacy needs and an action plan for accomplishing goals. The coach elicits potential barriers and solutions for meeting goals. Following calls review progress on action plans, address barriers and revise plans. Telehealth calls are audiotaped, with recorded permission, for process evaluation of content.

Control Arm Content and Procedures

Immediately following randomization, the control arm receives access to a study-specific control website that has annotated links to existing resources for AYA survivors. After completing the 12-month assessment, control participants are given access to the digital INSPIRE intervention program, without telehealth calls, and a mailed SCP as described above.

Primary, Secondary and Exploratory Endpoint Measurements

Study measures (Table 3) have demonstrated sensitivity and specificity in research with AYA survivors. The CTXD includes items such as “not knowing what the future will bring” and “long term effects of treatment” and asks how distressing or worrisome each item has been in the past week. We have used this measure with AYAs in a previous study testing SCPs.(23) Selected measures address study objectives. Common Data Elements were selected to facilitate data sharing and harmonization.

Table 3.

Measures and schedule of administration (T2 for intervention-only participants)

MEASURES	T1 Baseline screening	T2 6-week assessment	T3 3-month outcome	T4 12-month outcome
Primary and Secondary Endpoints (Aims 1 & 2)
Cancer and Treatment Distress (CTXD)(53, 54)	X		X	X
Health care Adherence (HCA), Heart and Diabetes History(55), (SM and CMD adherence included in measure)	X			X
Baseline Covariates/Moderators, Risk Factors (Aim 3)
Clinical Factors
Cancer Diagnoses and Treatments	X
Comorbidities(56), Medications	X
Self-Reported Health Status
PROMIS Global Function(57)	X			X
PRO-CTCAE Symptoms (58) Concentration problems, memory problems, pain, insomnia, fatigue, anxiety, discouragement, sadness	X			X
Personal Factors
Confidence in Survivorship Information(59)	X			X
PHQ-8 Depression (PHQ-8)(51)	X			X
Sociodemographic/Environmental
Background (e.g. age, race, income, rurality)	X
ENRICHD Social Support(60)	X			X
Barriers to Health care Adherence	X			X
Mechanisms/Mediators (Aim 3)
Impact of Cancer–AYA(61)	X		X	X
Health Self-Efficacy(62)	X		X	X
Health Literacy–AYA(63)	X		X	X
Survivorship Knowledge		X		X
Study Evaluation
Evaluation of Program and Participation				X
Intervention Arm Only
Distress (0–10) & PHQ-8 items 1 & 2		X

Statistical Analyses

Sample size was selected to assure adequate power to test the effects of the RCT aims, with primary analyses based on a two group, intent to treat analysis (Table 3). Of 700 enrolled, we expect at least 400 will be eligible for Group 1 randomization and they will be included in one or more of the analyses of the 3 outcomes based on meeting eligibility at baseline for that endpoint. Some will contribute to all endpoint analyses, others to 1 or 2. Randomization is stratified based on the combination of impaired outcomes participants are categorized at baseline . For example, participants who only meet criteria for the distress outcome only are randomized within the group that only has high distress to ensure study arm is balanced within the analytic population for each endpoint and that impairment on other endpoints is also balanced. Based on experience in previous INSPIRE and AYA RCTs, we project a loss of ~28% during the study from consent to 12-month assessment, including 8% from mortality or severe medical problems. With 700 enrolled, this provides sufficient power for Aims 1, 2 (RCT), and 3 analyses.

Sample Size and Power Calculations

With enrollment of 826 subjects, of whom we estimate 700 will be eligible for Group 1 or Group 0. Among those in Group 1, we expect to attain the 219, 264, and 264 participants needed to meet eligibility criteria for inclusion in the intervention vs. control arm analyses of the dichotomous CTXD, CMD and SM adherence outcomes, respectively (Table 4). Among these, we estimate that 158, 190, and 190 will be available for the 12-month analyses. Assuming control arm rates of events similar to those observed in the INSPIRE–HCT study, these numbers will allow us to detect a minimum relative risk of success for the intervention vs. control arms of 1.9, 1.9 and 1.9, for the CTXD, HCA-CMD, HCA-SM endpoints, with 80% power and 2-sided α=0.017 for each comparison (α=0.05/3 to conserve overall α=0.05). For evaluation of the continuous outcomes, we will have at least 80% power to detect a minimum effect size of 0.40, 0.43 and 0.38 standard deviations for the CTXD, HCA-CMD and HCA-SM outcomes (α=0.017).

Table 4.

Sample size calculations for Aim 1 & 2 outcomes, 80% power

	CTXD	HCA-CMD	HCA-SM
Enroll/Consent/complete baseline	700
T1: Eligible for specific endpoint	219	264	264
3 mo. voluntary withdraws or mortality/illness (4%)	[8]	[10]	[10]
3 mo. non-responders, able to respond at 12 mo. (15%)	[32 eligible for T4]	[39 eligible for T4]	[39 eligible for T4]
T3: 3 mo. completers (80%)	175	211	211
T4: 12 mo. potential cohort	207	168	168
12 mo. voluntary withdraw or mortality/illness (8% of T3)	[9]	[8]	[8]
12 mo. non-responders (20% of T3)	[35]	[42]	[42]
12 mo. completers (75% of T3, 72% of T1)	158 [79/arm]	190 [95/arm]	190 [95/arm]

Primary and Secondary Hypotheses Statistical Analyses, Aims 1 & 2

The primary and secondary endpoints to be evaluated are distress score (CTXD) at 3 months (primary) and 12 months, and the proportion of HCA-CMD and HCA-SM reported at 12 months (secondary) and will be defined as binary outcomes. Participants will meet the primary endpoint for absence of elevated distress if they score <0.9 on the CTXD. They will meet each HCA endpoint if their percent of recommended screenings obtained within the past 12 months is ≥80% for HCA-CMD and HCA-SM. The number and pattern of missing values will be summarized and assessed to understand potential biases. If attrition is greater than 10% before the 12-month follow-up, we plan to utilize multiple imputation of the outcome and carry out sensitivity analyses of the impact of missing data on the comparisons between study arms.(64) For hypotheses testing for Aims 1 & 2, the three endpoints will each be compared between study arms as intent to treat analyses using standard logistic regression analyses, adjusted for stratification factors and utilizing a 2-sided α=0.0175 to preserve type I error at 0.05 overall for the three planned endpoints. If other factors appear to be imbalanced between study arms, we will carry out sensitivity analyses adjusting for those factors. Planned secondary analyses will evaluate outcomes adjusting or stratifying based on engagement (viewing 3+ pages of the digital program or completing 4 telehealth calls). Analyses will evaluate continuous variables of the study endpoints using linear regression models.

Aim 3 Sample Size and Statistical Analysis

Although the selected sample size is based on the primary main effects of the intervention, with 150–200 subjects available for analyses of each endpoint, we will be able to detect direct and indirect effect sizes (standardized beta coefficients) of at least 0.39 (moderate) with 80% power and 2-sided alpha=0.05 as illustrated by Fritz and MacKinnon.(65) Analyses for Aim 3 are designed to understand the moderating and mediating relationships between attributes measured pre-intervention (moderators), the intervention, mediator variables (specifically mechanisms) and outcome measures (see Table 3). Moderators will be evaluated by examining predictors (e.g. cancer diagnosis, age at diagnosis, treatment exposures, demographics, health status, and baseline impact of cancer, health self-efficacy and health literacy) of the 12-month binary outcomes for CTXD and HCA-CMD and HCA-SM, and assessing interactions between those factors and the intervention to determine whether the efficacy of the intervention differs by baseline characteristics. We will use a log-link with Poisson error structure and robust sandwich variance estimates to directly estimate the more relative risks (RR).(66) Multivariable models for each outcome will be established using single-variable and step-up and step-down procedures with a liberal p-value for inclusion in the model (p<0.2).

Factors identified will provide information about participant characteristics that might identify AYA survivors who are more or less likely to benefit from the intervention and may be used to tailor future study intervention methods. We will be particularly interested in race/ethnicity, age at diagnosis, and sex as modifying variables. In general, to establish whether a mediation relationship exists, we will follow standard methodology(67–69) by 1) establishing that there is an association between the causal variable (e.g. intervention) and the outcome (e.g. CTXD improvement), 2) establishing the association between the intervention and the mediator (e.g. engagement with mobile app; social media, health self-efficacy), 3) showing that the mediator is associated with the outcome variable in a model with the causal variable included, and 4) in the same model as used to evaluate step 3, determine that the relationship between the causal variable and the outcome is attenuated by inclusion of the mediator in the model. We do not expect to observe complete mediation in these analyses.

In addition to the separate stepwise modelling approaches outlined above, we will jointly evaluate relationships using Structural Equation Modelling (SEM) approaches to evaluating both moderation and mediation, which will allow simultaneous evaluation of the multiple pathways and classes of variables together in a single model, with improved power, allowing estimates of direct and indirect effects calculated from each.(70, 71) All effect size estimates will be presented along with associated confidence intervals and 2-sided p-values, both unadjusted and adjusted for multiple comparisons.

Aim 4

Costs will be computed by recording purchase costs of materials and staff time spent in each of the designated delivery of care activities: moderating and maintaining digital programs and participant contacts, mailing SCPs, and conducting telehealth calls. These will be converted to actual salary plus benefits per staff person and standardized to report the cost per participant who reaches at least one of the primary endpoints for the study, specified for each type of cost as well as total costs. Because we will track the costs of the program provided to the control group and will know their outcomes, we will be able to calculate an incremental cost-effectiveness ratio for the internet intervention program compared to the control, similar to Mandelblatt et al.(72, 73)

Process analyses

Use of the digital modalities will be tracked with date, time and page stamped autorecorded data and evaluated for engagement and patterns of views related to subgroups and outcomes. Evaluation of the telehealth calls will include: 1) who requires calls, 2) call completion rates, and 3) fidelity scored from audiotape reviews. The coaches will rate their fidelity for each session in addition to the other coach providing peer supervision. We will conduct descriptive analyses of these results. The fidelity rating scale that will be used by the coaches has been developed and was based on our previous study(17) which measures fidelity to the telehealth manual that was developed for the study. These ratings will be kept and used to describe fidelity for each coach.

Data Quality and Management

We anticipate limited missing data and out of range responses by using the electronic data capture, with phone calls to check with non-responders and mailed abbreviated forms if necessary. We have established a minimum abbreviated set of forms that provide primary outcomes (CTXD, HCA). Assessment programming will set parameters for allowable responses and will return responders to items either missed or outside range, with an option to click a ‘choose not to respond’ box. Validity checks will also be programmed to flag responses that are markedly inconsistent and may be invalid. Outliers will be identified to determine whether clinical response is needed. The PI will be notified and will contact a participant as needed to assess validity or urgent need for care.

Conclusions

This RCT is testing a personalized, sustainable, survivorship informatics system in a risk-stratified intervention program that can reach AYA survivors with unmet needs including those with reduced access to care. This may extend the reach and engagement of AYA survivors by meeting them where they live on their mobile devices or by telehealth. The intervention adds a larger ‘dose’ and enhanced reach to those with greater needs or low engagement/receptivity through the addition of stepped care telehealth. This study will extend our experience in reaching AYAs who are underserved, particularly those who are more distant from specialty care or for whom digital health interventions on their own were not sufficient. One limitation of the study is that the healthcare adherence outcomes are based on patient report and are not substantiated via chart review. These survivors are frequently no longer being seen at the tertiary cancer center so verifying their healthcare adherence outcomes at another health care system would be a significant challenge, but worthwhile to pursue in future work. Another limitation is that the study coaches will rate each other for adherence to the treatment manual as there is no one else on the study team with the expertise required to do this task. A goal of future research will be to corroborate patient report with the EHR. A strength of the study is that by oversampling racial and ethnic minorities, it should provide precise, unbiased estimates of intervention effects in a heterogenous population. Additional targeted late effects can be added if the RCT demonstrates efficacy with these methods. If the RCT demonstrates that this intervention is effective, the program is positioned for accelerated dissemination/implementation to improve care for AYA survivors with its scalable informatics-based administration and largely digital intervention.

Abbreviations

INSPIRE

INteractive Survivorship Program to Improve Health care Resources

AYA

Adolescent or young adult

RCT

Randomized controlled trial

SM

Secondary malignancy

CMD

Cardiometabolic dysfunction

SCP

Survivorship care plan

HCA

Health care adherence

IRB

Institutional review board

DSMB

Data safety and monitoring board

PRO

Patient reported outcomes

Phq-8

Patient health questionnaire-8

PST

Problem solving therapy

RR

Relative risk