Guided programs

One approach to delivering eHealth interventions combines technology with the active involvement of a professional who interacts with individuals completing the program. The backgrounds of these professionals vary from clinical psychologists with extensive training and experience delivering behavioral interventions to trainees (e.g., a postdoctoral fellow or graduate student) or laypeople. We refer to them as “therapists” for brevity. In these programs, therapists may deliver much of the behavioral intervention, but more typically their activities complement Internet-based multi-media training in pain management. For instance, therapists may help program “users” learn to apply new skills to their daily life, assist in problem solving, deliver motivational messages or encouragement, provide feedback, or follow-up on users to maximize adherence. Thus, this type of intervention may be considered to be “minimal contact” because much of the online content and eHealth resources replace what therapists traditionally provided during face-to-face interventions. Communications between the therapist and users may be asynchronous (e.g., through emails or “threaded” bulletin boards) or synchronous (e.g., in a “chat room”).

An early example of a therapist-guided eHealth intervention was a Swedish program for low back pain patients [48]. The program website included six training modules designed to be completed over 6 weeks. Training was based on cognitive-behavioral treatment techniques (i.e., goal setting, relaxation, exercise and stretching, cognitive restructuring, activity pacing, and relapse prevention). Users completed pain diaries and had structured telephone calls with a therapist each week. Calls lasted about 10 min and involved reviewing homework, answering questions, and providing reminders and motivational messages. Additional email contact with therapists and technical support staff was available as needed. Therapists were trained graduate students supervised by a clinical psychologist and a physiotherapist with expertise in pain management.

A small (N=56) randomized controlled trial compared this program with a waitlist control. At the post-intervention assessment, the intervention group significantly improved in several aspects of pain coping (catastrophizing, control over pain, ability to decrease pain), but not on outcomes such as pain, pain interference, anxiety, or depression. By the 3-month follow-up, the waitlist control group had completed the intervention, so analyses examined change from pre-intervention to follow-up in the entire sample and found significant improvements in aspects of pain coping (praying and hoping, catastrophizing, control of pain, ability to decrease pain), perceived life control, punishing responses from significant others, pain-related attitudes, and depression. Besides lack of a real control group at follow-up, a limitation of the study is that it failed to compare the behavioral intervention to another intervention that controlled for therapist attention. Although the program appeared to have been well-accepted by users (e.g., there was minimal dropout), we note that all users had pre-existing Internet access and that recruitment methods (newspaper and online advertisements) probably ensured a motivated sample. More recently, a version of this intervention was tested in which telephone contact was replaced with email contact [49], and the researchers concluded that the two modes of communication worked equally well despite a failure to replicate effects of the intervention on perceived control over pain and ability to decrease pain.

Another example of a guided program is offered by the Expert Patients Programme (EPP) (https://selfmanage.org/BetterHealth/SignUp, http://www.expertpatients.co.uk/), a 6-week chronic disease self-management intervention [50] based on a face-to-face intervention developed by Lorig [51, 52]. It is relatively information-focused, providing extensive content in four parts. The first part, the Learning Center, offers interactive training in pain-relevant topics including individualized exercise programs, cognitive symptom management (e.g., relaxation, visualization, self-talk), negative emotion management, medications, physician–patient communication, healthy eating, fatigue management, action planning, and problem solving. The second part, the Communication Center, offers four interactive bulletin boards for discussion of action planning, problems, difficult emotions, and celebrations. Users join the program simultaneously in 20- to 30-person groups to help ensure the boards are populated. In addition to threaded (asynchronous) discussions begun in the boards, users’ responses to exercises in the Learning Center are posted. A third part, My Tools, provides online tools for developing an action plan, an exercise log, and a medication log, plus access to audio relaxation exercises and links to health websites. The fourth part, a Help section, allows users to email trained peer moderators or program administrators (who can also be telephoned). Peer moderators are layperson tutors who have gone through the program and received 1 day of online moderator training. They use email to remind users to log in to the program, model action planning and problem solving, offer encouragement, and answer questions. They also post to the bulletin boards and monitor bulletin board content daily. Their role is to assist in tailoring and supporting users’ interactions with the program.

EPP was tested in a non-randomized prospective longitudinal study of 568 people with chronic diseases (mostly arthritis) [50]. Assessments occurred at baseline and at 6- and 12-months post-treatment. The 12-month follow-up found significant improvement in pain, distress, communication with a physician, health-promoting behaviors (e.g., exercise), satisfaction with health care, and health care utilization. Users logged on to the program 40 times on average and completed 5.2 out of a possible 6 sessions; 79% completed all sessions. Bulletin board usage was high, averaging nearly 45 posts per participant. Cost savings were estimated to be 128 British pound sterling per participant based on observed reduction in health care usage for the intervention group versus controls and including fixed program development costs that would be reduced with wider dissemination. Although promising, the lack of a randomly assigned control condition limits conclusions that can be drawn with confidence.

A unique aspect of EPP (and other versions of it that address different patient populations and countries; [53–56]) is its use of trained laypeople or peer moderators. Because of their personal experience with a disease, peers are thought to be uniquely able to help users develop self-management skills and confidence in their ability to manage their disease. Studies of interventions using trained peers have yielded mixed results, and beneficial effects are often small in size [57, 58]. This research would benefit from greater reporting of details of peer activities and greater attention to best practices for training peers and appropriate roles for peers versus professionals [58]. Moreover, many of these studies have recruited samples consisting primarily of educated White women [57, 59], suggesting that this intervention format may be particularly appealing to that population. Nonetheless, we view this as a promising approach. Research is needed to evaluate the generalizability of the intervention to other populations [58, 59] and to identify subgroups most likely to benefit (e.g., those with low perceived control over their ability to manage their disease) [60].

Unguided (automated) programs

Automated programs allow patient users to complete an intervention in a self-directed manner. With the exception that no therapist contact is used in this approach, much of the content is similar to guided interventions. In terms of function, instead of a therapist, unguided interventions may rely on tailored feedback, interactive problem solving, and automated remediation, reinforcement, and reminders (e.g., to practice skills or to complete additional program components). The training itself may be tailored to users’ needs using their responses to questions, tests, or exercises.

One highly interactive program called painAction (http://www.painaction.com/Members/MyPage.aspx) targets people with chronic low back pain, combining educational content (divided into eight sessions) with multi-media and tailored feedback [78]. Developed with input from back pain patients, pain treatment clinicians, and back pain researchers, a unique feature of the program is that each session’s content and activities are tailored to user characteristics with a “recommendation engine” (computer-implemented algorithms guided by user-reported data collected through online questionnaires). Users are instructed to review the tailored content delivered to them, and they are also allowed to explore the rest of site (a feature intended to simulate self-directed web “surfing”). Content covers topics often featured in cognitive behavioral treatment protocols such as managing thoughts and moods, goal setting, problem solving, relapse prevention, and activities to improve collaborative decision making, sleep, nutrition, stress management, and exercise. The program also includes interactive tools for symptom monitoring and use of skills, online assessments, and a library of articles.

A randomized controlled trial of 209 people with chronic low back pain compared the intervention condition (in which participants completed eight sessions in 4 weeks followed by five monthly booster sessions involving additional visits to the site) with a control condition (in which participants were provided with a standard text-based back pain management guide and asked to read it over 4 weeks). Intervention participants were asked to spend at least 20 min on the website for each session, and their website “dosage” was standardized by having them follow protocols that involved reviewing their tailored online content. Assessments were completed at baseline and at 1-, 3-, and 6-months post-baseline. Significant positive effects of the intervention were found for stress, use of several positive coping strategies, and global perceived improvement, but not for a number of other outcomes. However, intervention group participants recruited online (but not those recruited in pain clinics) reported significant improvements in worst and average pain and in use of coping self-statements compared with control group participants. The investigators noted that compared to people recruited online, pain clinic patients may have more complex or severe problems and may be less adept at using an Internet-based intervention. We note that this study used a more credible control condition than most studies and that the investigators used methods for controlling and tracking the intervention dose. Use of a relatively large sample is also a strength of the study.

Another automated program—Living Well with Fibromyalgia—was designed to provide behavioral treatment to rural communities in the northern midwest region of the USA [79]. It includes educational, exercise, and cognitive–behavioral treatment modules that provide information about fibromyalgia, symptom management (i.e., role of medications, exercise, sleep, relaxation, pleasant activities), and lifestyle adaptation (i.e., goal setting, problem solving, graded activation, reframing, communication). The site features multi-media streaming video lectures from experts on each topic, written summaries of lectures, downloadable worksheets, and audio files of relaxation exercises. Thus, it is highly patient-driven and focused on skills training.

To evaluate the efficacy of this website, a randomized controlled trial (N=118) of fibromyalgia patients referred by a physician compared 6 months exposure to this site to standard care. Patients randomized to the intervention arm were instructed to use the modules that seemed most relevant to them, to practice and apply the skills they learned, and to revisit the site as needed. Study staff had no additional contact with participants until a 6-month follow-up assessment, so the trial enabled evaluation of effects of unguided use (i.e., with minimal staff contact). Attrition was modest at 10%, and satisfaction with the program was high (e.g., 91% of website users reported having received the type of service they wanted compared to 67% in the control group). In each month of the program, 89–94% of intervention participants were using at least one skill, and on average participants were using four skills at once. At 6 months, site users had significant improvements in pain, functional status, and global impressions of improvement compared with controls. Improvement in functional status was uniquely associated with use of the exercise, pleasant activity, and problem-solving modules. It is interesting to note that as this trial was being conducted, separate pharmaceutical clinical trials were being conducted to support Food and Drug Administration (FDA) indications for three fibromyalgia medications. The same outcomes that were used in those trials were used in the website’s study. Notably, this “therapist-less” site demonstrated effect sizes (Cohen’s d=.64 for pain and .38 for physical functioning) that were comparable or superior to the new FDA-approved medications for fibromyalgia. In order to achieve a 30% reduction in pain, the numbers needed to treat are 7.2, 19.0, and 8.6 for Duloxetine, Milnacipran, and Pregabalin, respectively [80]; but only 5 for “Living well with fibromyalgia” [79].

An obvious strength of this study was its use of outcomes that were being used in phase 2 and phase 3 pharmacological trials for the same condition, allowing direct comparisons between this behavioral intervention and drug therapy. Potential limitations include the use of a standard care condition rather than another website or a face-to-face intervention that could control for Internet use or modality.

Directions for future research

This review has highlighted the state of the field with respect to Internet-based behavioral medicine interventions for managing chronic pain. There are a variety of issues that could impede their adoption in clinical practice. Issues related to clinical settings are discussed below. In addition, a critical issue is that they must, of course, be shown to improve the health and well-being of patients. Accordingly, there is a pressing need for a larger number of rigorous and methodologically sophisticated clinical trials that use adequate sample sizes and compare the efficacy of behavioral interventions to other credible interventions (e.g., Internet-based information-focused educational interventions). Currently, research on therapist-guided interventions is most well developed—albeit still in critical need of greater rigor—followed by unguided interventions and then mobile applications. A particular concern is that outcome studies usually report multiple outcomes with benefits apparent in some but often not others. Researchers need to be clearer with regard to which outcome they hypothesize to be the primary outcome their interventions are designed to change (e.g., pain reduction or improvement in physical or psychological function). Increased attention to the primary outcome of treatments would also help in tailoring and customizing treatments so they are best able to achieve the desired effect. For instance, in cases of long-standing intractable pain conditions, improvements in pain per se are unlikely to be achievable, whereas improvements in the level and range of patients’ daily activity may be an achievable and key target.

There are a variety of important but unresolved research questions, including questions related to pathways through which these interventions influence outcomes. Consideration of multiple outcomes related to pain and impairment is important (with the above-noted caveat that a primary outcome must be specified), as is consideration of effects on healthcare usage, the patient–provider relationship, and disparities in pain management [97]. Studies that dismantle components of interventions to determine which are most critical to outcomes would be useful. Researchers should also consider alternatives to traditional randomized control trials such as practical trials [98–100] that test interventions in representative settings and populations and use clinically relevant control groups to address questions of interest to clinicians (e.g., relating to clinical benefit and patient adherence and satisfaction).

In addition to establishing their effectiveness, an empirical foundation is needed to guide intervention development. Translating in-person therapies for delivery via the Internet presents numerous methodological and technical challenges, and many decisions about intervention function and content are currently made without empirical guidance. One set of questions relates to methods for attracting users to an intervention and ensuring they complete intervention components (adherence or “engagement”) and the intervention as a whole (retention). The concept of “treatment dose” is a related, but often overlooked, issue in these interventions. In some interventions, patients are meant to complete the entire program, whereas in others, they are meant to complete only sections that are relevant to their needs. Patients who fail to complete an intervention as it was intended to be used can be viewed as having received a lower “dose” than those who are perfectly adherent. However, the question of what constitutes “treatment dose” in Internet-based interventions is complex [101–103]. A first step in clarifying its nature and importance is for researchers to report non-usage and attrition data consistently [102]. User responses to an intervention, including non-usage and attrition, may be influenced by features of the intervention itself (e.g., its interface, content, and interactive approach) [47, 104] and user characteristics (e.g., demographics, psychological factors, and skills) [43, 89, 101]. Ritterband and colleagues [101] have developed a theoretical framework to guide development and evaluation of Internet-based interventions; it provides a useful and thorough discussion of issues to be considered.

It is also important to investigate issues related to dissemination and implementation, which have the potential to impede clinical adoption of interventions even if they are shown to improve patient outcomes. A critical area of research concerns use of these interventions in primary care, where they can serve as an adjunct to standard care delivered by physicians who treat patients with chronic pain but who have a relatively limited array of treatment options to prescribe. Particularly in light of limited reimbursements, research is needed to identify the best ways to help medical professionals integrate Internet-based behavioral pain management interventions into their clinical practice. For instance, it will be important to identify what training providers need to implement interventions and to develop methods for engaging them to ensure providers “prescribe” these interventions, follow-up in ways that reinforce their use, and use data provided by patients or the interventions themselves (e.g., patient-reported outcomes delivered directly to EMRs) to improve clinical care. Just as usability testing with patients is needed to ensure interventions meet their needs, research on implementation and practitioner engagement requires close collaboration with medical professionals [105, 106].

Systemic barriers to clinical implementation

Implementing Internet-based interventions in clinical settings presents various challenges. Practices must make choices about staff role and time commitment. Despite potential advantages of including therapists in interventions, automated interventions are likely to proliferate more rapidly than therapist-guided interventions, in part because of their lower cost and also due to practical issues raised by having trained staff monitor or tailor treatment. If staff is to be involved, it is likely to be nursing staff already present in a provider’s office. Acceptance of these interventions by medical professionals and staff is likely to be highest when interventions are shown to benefit patients and/or reduce costs, but also when they are designed to integrate with existing procedures (e.g., ongoing medication management of pain) to make them more efficient and to expand services that are offered [106].

In addition, practices making the commitment to incorporate these interventions must ensure their patients’ acceptance of and adherence to interventions. Ideally, the steps they take to meet this need will be based on research on the benefits of strategies such as using reminders (e.g., emails), incentive programs, and provision of personal contact and feedback. It may also be important to select interventions with components that have been empirically demonstrated to enhance patients’ motivation and interest.

Other implementation challenges stem from the early stage of this field. Many new Internet-based behavioral pain management interventions are being released. Their development is not coordinated, many are redundant, and many lack academic guidance. Healthcare professionals’ ability to adopt these interventions is hindered by the fact that there is little incentive to choose one over another. No intervention is emerging as the gold standard. Furthermore, licensure issues associated with providing advice across state-lines limit therapist-guided interventions, and protection of the underlying database may limit content and functionality of automated interventions. It is also important that a clear and viable business model has not emerged for implementing these interventions in practice. The latter points underlie the reality that most empirically tested interventions are currently not available to the public.