The purpose of this study is to evaluate the efficacy of a web-based intervention, ES, against a free, publicly available web-based self-management intervention developed for cancer survivors by the National Cancer Institute (NCI) and the ACS called Springboard Beyond Cancer [102]. A total of 600 patients diagnosed with a first primary oral or oropharyngeal cancer within the past 3 years are being recruited via the New Jersey State Cancer Registry (NJSCR), Cancer Registry of Greater California, and Pennsylvania State Registry and randomly assigned to ES or Springboard Beyond Cancer. The participants complete measures at baseline and at 2 and 6 months after baseline. A process evaluation of ES is completed. The study has 3 primary aims and 1 exploratory aim.

The first primary aim is to evaluate the impact of ES versus Springboard Beyond Cancer on the primary outcomes of self-efficacy in managing care, preparedness for managing survivorship, and head- and neck-specific QOL. We hypothesize that the participants randomized to ES will report significantly greater improvements in self-efficacy for managing their care, preparedness for survivorship, and head- and neck-specific QOL at 6 months.

The second primary aim is to evaluate the mediators (planning, information needs, support needs, activation, and fear of recurrence) and moderators (age, race and ethnicity, HPV status, baseline self-efficacy, and preparedness) of ES’s impact on self-efficacy in managing care, preparedness, and health-related QOL. The primary outcomes are self-efficacy in managing self-care, preparedness for managing survivorship, and health-related QOL at 6 months. We hypothesize that the impact of ES on self-efficacy, preparedness, and health-related QOL will be mediated by improvements in survivorship care planning, reductions in information and support needs, and reductions in the fear of recurrence. Little is known about which survivor subgroups benefit the most from self-management interventions. In addition, we hypothesize that the survivors with fewer resources (survivors who belong to people of color, with lower education, who have not received a written survivorship care plan, with less baseline self-efficacy, and with less baseline preparedness) will benefit more from ES.

The third primary aim is to evaluate the impact of ES versus Springboard Beyond Cancer on the secondary outcomes of engagement in oral self-examinations and head and neck mobility exercises.

We hypothesize that the participants randomized to ES will report more frequent engagement in and comprehensiveness of oral self-examinations and head and neck mobility exercises than the participants randomized to Springboard Beyond Cancer at 6 months.

The exploratory aim is to conduct a process evaluation of ES to inform future implementations.

We hypothesize that the participants randomized to ES will use the intervention more, evaluate it more highly, have greater improvements in self-efficacy, preparedness, and QOL.

The development of original web-based intervention ES was described in the study by Manne et al [65]. Briefly, it had 4 modules: Introduction, Oral Care, Swallowing and Muscle Strength, and Long-term Follow-up Care and Detecting Lesions. These modules are listed in Table 1. Module 1, Introduction, orients participants and reviews modules. Module 2, Oral Care, reviews the causes, contains a symptom self-assessment, and provides information on how to manage dry mouth. Module 3, Swallowing and Muscle Strength, provides information on the symptoms of swallowing difficulty, methods to reduce swallowing difficulty, and methods to improve flexibility and range of motion in the head and neck. Module 4, Long-term Follow-up Care and Detecting Lesions, describes the importance of head and neck self-examinations; provides information on how to conduct a head and neck self-examination, signs to pay attention to, goal setting, and managing barriers to self-examination; and provides a link to a survivorship care plan website that includes follow-up information, assesses alcohol and tobacco use, and reviews the participant’s care team and recommendations. The original ES program had >20 interactive activities to engage participants and foster skill acquisition. An example is symptom assessment, which is followed by goal-setting feature. Participants rate symptom severity, frequency, concern about symptoms, and confidence in and importance of managing symptoms. Next, the participants select a goal related to the symptom; rate the importance of the goal; choose strategies to address the goal from a list of ideas; and rate benefits and barriers to achieving the goal, confidence in achieving the goal, and support needed to achieve the goal. This function is in modules 3 and 4.

Enhancements were selected based on feedback from the pilot study participants as well as the team’s experience with participants during the ES intervention. In module 1, a section containing information about self-care and goal setting to enhance self-care was added to familiarize the participants with the process. In module 2, the ability to set an email reminder to work on the oral care goal was included to optimize continued implementation. In module 3, because the participants liked the videotaped swallowing exercises and indicated that they would like to see more of them, additional exercises were developed with the team’s speech pathologist. The strengthening exercises were originally presented as illustrations but were very highly rated. To foster engagement, the team’s physical therapist developed videotaped demonstrations of exercises. Because lymphedema was mentioned as a survivorship concern in participant feedback from the pilot study, information about it and its management was added along with demonstrations of home exercises to manage lymphedema by the team’s physical therapist. Finally, additional information on how to add caloric content to food was added. In module 4, to bolster the existing content and address common concerns that were not addressed previously, additional resources were added for tobacco cessation, alcohol use, comorbid medical issues, managing hearing loss, and survivorship care planning. More content on interest in quitting tobacco use and the risks of continued use was added. For current smokers, a link to Becomeanex [103] was provided. If the person indicated a desire to continue smoking, material on the increased risk for cancer recurrence and mortality and reduced treatment benefits for smokers and an assessment of the readiness to quit were added by the team member MBS. As for alcohol use, if the participant indicated a desire to continue drinking, a link to Drinkerscheckup [104], an empirically based, web-based treatment, was provided [105]. As comorbid medical issues are common in this survivor population, a section was added on “other medical issues.” This section assesses comorbid medical problems, associated medications, and interest in improving the management of medical conditions and offers tailored input on the barriers to medication adherence based on an assessment of barriers. Finally, the team’s audiologist added additional material to address the management of hearing loss and tinnitus.

A total of 2 new modules, which were originally planned but not developed, were created: module 5, Calm and Connect, and module 6, Maintenance. The goals of Calm and Connect are to describe common psychological concerns patients experience after treatment, increase the awareness of how cancer has affected the participant’s emotional well-being, assist the participant in developing useful methods to deal with emotions, provide free web-based resources on coping with stress, and discuss psychosocial aspects of having HPV-related oropharyngeal cancer. The introduction portion contains a video introduction by a psychologist portrayed by an actor and audiotaped narratives presenting patients’ emotional experiences. In the next 3 sections, the following coping skills are addressed: (1) managing stress with relaxation, (2) increasing the tolerance for upsetting emotions using acceptance-based strategies, and (3) talking it out with others. A fourth section was developed to address the unique concerns experienced by the survivors of HPV-associated cancers, including common worries (eg, transmission to others), managing questions from friends and family about HPV, and feeling embarrassed about HPV, and the unique experiences of young cancer survivors. All sections include interactive activities. For example, the participants can select their key signs of stress, which activates a patient narrative about the signs of stress and narrated educational content. There are also audio relaxation exercises led by a psychologist portrayed by an actor, self-reflection exercises in which the participant types in responses, and links to outside resources such as web-based apps for meditation and mindfulness exercises (Headspace and Calm).

The goal of module 6, Maintenance, is to facilitate ongoing engagement in self-management, evaluate the progress on goals, provide opportunities for interaction with other study participants, and provide new information. This module has 4 features, namely (1) Favorites: the participants choose their favorite content and exercises from screens throughout the program. Across the ES program, 28 pages can be chosen as favorites. These pages are migrated to the Favorites section for future reference, and participants can log in directly to this content; (2) Goals: goals set by the participants in each module are automatically migrated to this section, where the participants can directly access, review, revise, and update them and set up email reminders about them to send to themselves; (3) News: articles of interest from science and medicine sources are posted by study staff on a regular basis; and (4) Message Board: the participants have access to the Message Board throughout their participation in this study. Every month or two, interesting news articles about oral cancer are placed on the board by the project team. When there is a post, the participants receive an email notification that a message has been posted. They can review the post and post messages at that time. They can also access the message board at any time.

Both enhancements to the existing content and new material were developed in collaboration with ITX, incorporated. ITX is a web-development company and the developer of the original ES intervention. Content was reviewed and implemented during a series of regular meetings between the study team and the ITX development team over a 6-month period. The videotaped and narrated content was created in collaboration with Mindful Designs.

The Patient Advisory Board consisted of 3 oral and oropharyngeal cancer survivors. The board members reviewed the Calm and Connect module after it was developed and provided feedback in a web-based meeting. Notes from the meeting were reviewed by the study team. No substantive content changes were made, other than those intended to fix grammatical and navigation issues.

After the new modules were finalized and enhancements were incorporated into the existing modules, the Patient Advisory Board reviewed ES. Board members were sent a link to ES and asked to review the content and note recommended changes and any feedback. Next, web-based 1:1 interviews were conducted to review content and collect feedback. The following key suggestions were given: several modules: correct navigation errors; module 1: provide more clarity with regard to navigating the menu bar, reduce expectations for goal setting by reminding participants that it may be easier to choose small goals, and add definitions of additional key terms; module 3: provide the ability to print the swallowing exercises; and module 5: add content to normalize the desire to get one’s life back to normal. The study team reviewed the interviews to select modifications. All of the listed changes were made. There were several suggested changes that were evaluated as less relevant to most survivors and were not incorporated. Content changes were implemented using ITX. The final content is shown in Table 2. The sample content from modules 6 and 7 are shown in Figures 2-8. There are more than 75 interactive activities in ES, including in-session exercises, live links, self-assessments, and video and audio narratives. Multimedia Appendix 1 [106] outlines these activities.

This is a prospective, randomized controlled trial to evaluate the effect of ES on confidence in managing self-care, preparedness, and health-related QOL and on the secondary outcomes of oral self-examination and engagement in head and neck strengthening exercises. This is evaluated in 2 groups with a total of 600 survivors (ES vs Springboard Beyond Cancer) and by 3 assessments (baseline, 2 months, and 6 months). The participants complete surveys at baseline and 2 months and 6 months after baseline. Figure 9 illustrates the participant flow. A free, publicly available general survivorship-focused web-based program, called Springboard Beyond Cancer, was selected for comparison based on a review of the literature on the design of behavioral interventions [98,107,108]. This review suggested that there are pros and cons to all comparison conditions and that selection should be guided by the phase of the research. We selected a comparison condition that had the same method of delivery (web-based delivery), same frequency of delivery, and similar topics.

The participants are recruited from 2 state registries and through clinic recruitment. The eligibility criteria are as follows: participants should (1) be aged >18 years, (2) be diagnosed with a first primary oral cavity or oropharyngeal cancer in the last 3 years, (3) be currently cancer free (but could have experienced a recurrence), (4) have computer access, and (5) be able to read English. Children are not included because the incidence of these cancers in children is exceedingly rare (0.2 cases/100,000 children).

Among the cancer registries, approximately 1600 patients are diagnosed with a first primary oral or oropharyngeal cancer per year. Our pilot study suggested that 26% of eligible patients will consent and complete the baseline survey. This enrollment rate is consistent with or higher than that of other intervention trials recruiting from cancer registries (18%-30%) [99,100]. Recruitment will be conducted for approximately 27 months. The sample size is based on the power to detect the effects on our primary outcomes. With 1600 patients available each year and 3 years of data collection, we have a large pool: given the 4800 total participants available and 26% acceptance, we have about 1104 potential participants. The power analysis suggests that the study requires 600 participants.

To achieve this, we include pictures of minority patients on our recruitment materials and oversample Black males in our registry recruitment. We have set a goal to ensure that of the total number of participants enrolled, 8% (n=48) are Black men.

Recruitment was primarily conducted using state cancer registries. Recruitment procedures differ between state registries. For the NJSCR, the registry confirms patient eligibility and, if eligible, sends a letter to the patient’s diagnosing physician, making the physician aware that their patient is eligible. If the NJSCR does not receive a disapproval from the physician within 2 weeks, the patient is sent a cover letter and an “agreement to contact” sheet. Potential participants are called by registry staff to discuss the study. The participants have the option of verbally consenting at that time. The participants who consent have their contact information and a copy of verbal consent forwarded to the members of the study staff, who call the patient. Patients deemed eligible are provided with a web-based consent and survey. The staff members contact patients weekly if they do not return the consent. Patients are considered passive refusers if they do not return a survey after repeated call attempts during days, nights, and weekends over a 3-week period after the consent is sent. For the Cancer Registry of Greater California and the Pennsylvania State Registry, the registry provides a list of patients and their contact information to the staff members. The staff members call patients, and those deemed eligible are provided with a web-based consent and survey, and the same procedures for enrollment as described above for the NJSCR are implemented. A maximum of 8 recruitment calls are allowed: 4 calls can be followed with a voicemail message, and 4 calls are counted as missed calls. Of these 8 contacts, 1 can consist of a text message. Once the electronic consent and survey are completed, the participants are automatically randomly assigned to the ES or Springboard condition using a data management system. Separate assignment schemas are set up for each registry by the study’s statistician.

Participants who do not complete the consent and baseline are contacted by phone or email on a weekly basis for 1 month to facilitate survey completion. The participants are automatically linked to one of the web-based interventions (ES or Springboard) as soon as they are randomized upon the completion of the baseline survey.

The participants complete electronic surveys via a web-based data management system (DatStat) at 3 time points: baseline and 2 months and 6 months after baseline. Table 2 presents the timing of the measures. At the 2- and 6-month time points, regardless of the intervention progress, the participants are sent an email with a link to the follow-up survey. To the participants who do not complete the survey within a week, 3 weekly email and phone call reminders will be sent. After 4 weeks, the participants who do not complete the survey will be considered noncompleters for that survey but will be sent a link to the next follow-up survey. The participants will receive a web-based gift certificate worth US $50 for completing each survey. Each survey takes approximately 30 minutes to complete.

The participants report sex, age, education level, marital status, insurance status, race and ethnicity, income, and financial hardships in the last month (“Did you have adequate financial resources to meet the daily needs or you and your family?”).

At baseline, the time since diagnosis, whether the survivor had been diagnosed with a second cancer, recurrence status, treatments received (radiation, chemotherapy, and surgery), and comorbidities (Charlson index [116]) are reported by the participants. At all time points, oncology-related service use is reported by the participants. Tumor location and cancer stage are extracted from the tumor registry data. Tumors are characterized as oral, oropharyngeal, or salivary cancers. The participants also report the receipt of a survivorship care plan, their HPV tumor status, their current tobacco and alcohol use, and their current insurance status.

The Patient Health Questionnaire-9 [122] is a self-report measure of depression that assesses key diagnostic criteria for depression based on the original PRIME-MD. It is a widely used measure and has demonstrated strong reliability and construct validity [123] in nonill populations as well as cancer populations [123]. Scores from 0 to 4 indicate no or mild depression, from 5 to 9 indicate mild depression, from 10 to 14 indicate moderate depression, from 15 to 19 indicate moderately severe depression, and from 20 to 27 indicate severe depression.

In both interventions, the following are assessed: (1) 17 items assess general features and content (use, helpfulness, learning something new, interesting, and valuable) and the degree to which the program improved confidence in managing symptoms and (2) 5 items assess website features (eg, for ES, survivor stories, goal setting, and discussion board and for Springboard, action card decks, symptoms page, stress and mood page, and wellness page). For ES, an additional 19 items assess the content of each module. Open-ended feedback is solicited.

A total of 30 participants from the ES condition participate in a brief web-based or phone interview. The survivors will be contacted after they complete the 6-month follow-up. The questions ask the participants about the best timing for the intervention (during or after treatment or another time), delivery schedule (all at once or one module at a time), best setting to offer the intervention (eg, dental visits or oncology care), and whether the participant would be willing to pay for the intervention.

In year 5 of this project, the team will conduct key informant interviews with 30 oral cancer providers (eg, oncologists and physical therapists) who have contact with this population. Questions will focus on the timing (eg, right after treatment), duration (weekly vs all at once), and settings (eg, during dental visits) that they feel would make ES most useful.

The primary aim is to examine the relative impact of ES versus Springboard Beyond Cancer on our primary outcomes of self-efficacy in managing self-care, preparedness for managing survivorship, and health-related QOL at 6 months. We anticipate that 600 survivors will enroll (300 in each condition), and we conservatively estimate, based our ES pilot, that 80% (n=480) will complete the 1-month follow and 75% (n=450) will complete the 6-month follow-up. However, all available data will be included in the analyses.

Multilevel modeling (MLM) will be used to examine whether there are significant differences between the ES and Springboard Beyond Cancer interventions over time. In our primary analysis, both time and treatment will be treated as categorical. This approach yields a main effect for treatment, a main effect for time, and an interaction between time and treatment. Covariates will include 3 effect-coded variables representing sex, race and ethnicity, marital status, education, age, tumor location, stage, time since diagnosis, whether the cancer recurred over the study period, whether the participant received a survivorship care plan, HPV status, and baseline tobacco and alcohol use. The MLM approach has the advantage that it does not exclude participants with missing data at some time points, as would be the case if a traditional ANOVA were used. Thus, this analysis used all available data.

Power is calculated for the intent-to-treat effect sizes from the pilot study, assuming that 2-sided z tests are conducted at the P<.05 level. With an intent-to-treat analysis and a total sample of 600 participants, there is >99% power to detect significant differences between ES and Springboard Beyond Cancer in our primary outcomes of self-efficacy and preparedness for survivorship. There is 90% power to detect significant differences in health-related QOL at the 6-month follow-up.

This aim evaluates the hypothesized mediators of the expected association between the intervention and outcomes and hypothesized moderators of the intervention effects. Mediators include action planning, information needs, support needs, patient activation, and the fear of recurrence. Moderators include race and ethnicity, education, receipt of a survivorship care plan, baseline self-efficacy, and preparedness. Figure 2 depicts the proposed mediational model in which the intervention impacts the survivors’ planning skills; reduces their information and support needs; and reduces their fear of recurrence, which ultimately impacts their self-efficacy, preparedness, and health-related QOL.

Analyses will be conducted using 3 separate mediational models for the 3 primary outcomes. The indirect effects of ES on outcomes will be tested using changes in outcomes from baseline to 6 months as dependent variables, changes in the mediators from baseline to 2 months, and changes in mediators from baseline to 6 months. The SPSS macro (IBM Corp) will generate 5000 bootstrapped samples using random sampling with replacement to generate estimates and CIs around the indirect effects [126,127]. The mean of the 5000 indirect effect estimates based on the bootstrapped samples will be examined. Within a multiple mediator model, the indirect effect of each mediator is its unique effect on the outcome when controlling for the other mediators included in the model. We propose that ES will have stronger mediating effects than the Springboard Beyond Cancer intervention.

Treatment moderators will also be evaluated. Very little is known about which survivor subgroups benefit the most from self-management interventions. The available research suggests that survivors with fewer resources (eg, lower education) benefit more [128-132]. We will adopt a similar hypothesis and examine the role of low resources, people of color, lower education, not having received a written survivorship care plan, less baseline self-efficacy, and less baseline preparedness, in ES’s effects. For moderation analyses with continuous or dichotomous (eg, care plan receipt; White race or people of color) moderators, we will simplify the analysis by using MLM with a regression approach in which condition will be effect-coded (ES=1 and generic=−1) and time will be treated as a quantitative variable that was coded in months (0, 2, and 6) and then grand-mean centered. Moderator variables will be grand-mean centered (eg, baseline self-efficacy) or effect-coded (eg, care plan receipt). Moderation analyses will include the same set of covariates used in the primary analyses described for the primary aims.

On the basis of the simulations and corrected bootstrap method outlined in Fritz and Mackinnon [130], there is at least a power of 0.85 to detect small effects (Cohen d=0.26) from both the independent variable to the mediators and the mediators to the dependent variable.

This aim focuses on 3 recommended self-care tasks: oral self-examinations and the level of comprehensiveness of self-examinations, engagement in swallowing exercises, and engagement in head and neck mobility exercises.

We will use the same approach as that outlined in the analysis of the primary aim.

Assuming an intent-to-treat analysis and a total sample of 600 participants, we have 99% power to detect significant changes in oral self-examination and comprehensiveness of oral self-examination and 85% power to detect improvements in performance of swallowing exercises. Owing to low power, the performance of head and neck exercises is exploratory.

The broad implementation of a survivorship-focused intervention involves a community-based implementation. It is premature to broadly implement ES before determining its efficacy. However, as a first step in this process, we will conduct a process evaluation. First, we will use participant treatment evaluations and use data from ES. Systematic reviews have found that the average adherence to web-based health interventions is 50% to 53% [132,133]. ES use was higher in the pilot study in that 75% of the participants viewed 3 to 4 modules [65]. Most web-based interventions are designed to be used at least weekly for several weeks, but many participants access the interventions only once [132-134]. We will examine the predictors of ES use and evaluate the association of ES use with the primary outcomes. This analysis assists in identifying subgroups for targeting to improve the use of the intervention. Second, we interview a randomly selected subset of 30 ES participants to provide feedback upon the completion of the last follow-up. Third, we conduct key informant interviews with 30 oral cancer providers (eg, oncologists and physical therapists) who have contact with this population to identify potential settings for implementation.

As for the participants, we record whether, how frequently, and for how long the participants logged into ES and how many modules are completed. In the analyses, we (1) examine whether these variables are associated with the primary outcomes, (2) evaluate the association of demographic factors (race and ethnicity, alcohol use, and tobacco use) with ES use, and (3) use an evaluation survey and web-based use data from ES.

For the survivor and provider interviews, interview data are transcribed. The analysis explores feedback on barriers to ES use in clinical practice and the preferred timing of ES (eg, should ES be introduced at a follow-up visit, in the waiting room, and by a navigator or health coach?). Analysis of interview transcriptions follow the “constant comparison” analytic approach [135], a method of explanation building in which the findings of an initial case are compared with a provisional category and revised as needed with new cases. This process is continued until an area of interest is fully explicated, reaching theoretical saturation [135]. Central to this process is “thematic” coding. We will develop an initial set of codes. For each code, we will develop ≥1 “secondary codes” that represent more specific aspects of the phenomenon. The schema organizes and assimilates data. The transcripts will be coded by at least 2 coders. Discrepancies will be resolved, and the process will be repeated until an acceptable level of intercoder reliability is achieved (using chance-corrected statistics such as κ for nominal data) [136]. Coded transcripts will be analyzed using the ATLAS.ti qualitative software (ATLAS.ti Scientific Software Development GmbH) [137].

This information will be used to create a website that serves as a storage house for all the study materials. On this website, the team will publish a free manual that details how to access ES and the methods to enhance ES use and contains information on hard-to-reach survivors and our publications and presentations.