Electronic databases were searched for relevant citations between January 2000 and September 2014, including MEDLINE, PUBMED, and EMBASE. Search terms included: “Internet based,” “Internet-based,” “Internet Delivered,” “Internet-Delivered,” “Web based,” “Web-based,” “World Wide Web,” “Online,” “Telemedicine,” “Tele-medicine,” “Email,” “E-mail,” “Mobile,” “Phone,” “Smartphone,” “Tablet,” “intervention,” “treatment,” “therapy,” “communication,” “counseling,” “education,” “educational,” “instruction,” “management,” “self management,” “chronic,” “recurrent,” “duration greater than 3 months,” “low back pain,” “low-back pain,” “lower back pain,” “lumbar,” “lumbosacral region,” “mechanical,” “degenerative disc disease,” “sciatica,” “myofascial back pain,” “nonspecific back pain,” and “adult”. Publication types and study designs of interest including systematic reviews, meta-analyses, practice guidelines, RCTs, and controlled clinical trials. Bibliographies of eligible articles were also searched for relevant studies. Selected journals were also searched individually for any relevant publications.

Articles were eligible for inclusion in this review if they were RCTs studying Web-based interventions directed at adults with chronic low back pain. Retrospective studies, narrative reviews, nonrandomized trials, and observational studies were excluded. However, references listed in these publications were reviewed to look for any studies that may match inclusion criteria for this review. Trials including children or trials including patients with acute pain were also excluded. RCTs studying interventions aimed at prophylaxis or other types of chronic pain were excluded. Studies published in languages other than English were excluded.

After the literature search identified potentially relevant articles, the articles were screened based on titles and abstract. Articles were excluded if they were not RCTs, the patient population was unsuitable for this review, the intervention was not Internet-based, or for other reasons (Figure 1). After this stage, the full text for the remaining articles was reviewed and nine were included for the purposes of this review.

Citations identified by the literature search strategy were screened for eligibility by two of the authors (SG, DG) and discrepancies were resolved using the opinion of the other authors. Information regarding the patient characteristics, intervention, duration, study characteristics, study design, and outcome measures was extracted from each eligible trial by one author (SG) and then reviewed independently by the other authors. Information required to assess the characteristics of studies was reviewed, including method of randomization, whether statistical analyses were performed by intention-to-treat, and allocation concealment [10-18].

The search revealed nine RCTs published between 2002 and 2014. Table 1 describes the characteristics of each study; six of the studies performed intention-to-treat analysis, three of the studies did not describe randomization, and allocation concealment was documented in only four of the published studies.

Trial design and details regarding the interventions used in the studies are presented in Table 2. Studies randomized 51 to 580 participants [10-18]. Study durations lasted from 6 weeks to 1 year. Three of the studies were waitlist controlled.

Patient characteristics including demographics are listed in Table 3. The majority of participants in the studies using online CBT were females. The mean age of participants in the studies ranged between 42 and 52 years.

A variety of diverse outcome measures were used; outcome measures used by each study are available in Table 4. Studies using CBT reported catastrophization as an outcome measure. Most studies reported patient empowerment and pain levels as outcome measures. Disability was reported by only five studies. Only one study assessed impact of intervention on health care utilization.

The studies were presented in two subsections: studies using online cognitive behavioral therapy (CBT) and studies using Web-based approaches to improve knowledge (with an interactive component to provide coping support).

The following trials were registered: Burhman et al [13], Krein et al [16], and Riva et al [17].

Psychological factors, such as depressed mood, negative beliefs, and somatization, have been shown to affect chronicity of pain and disability related to the pain [19,20]. This review identified four RCTs published between 2004 and 2012 that examined the effectiveness of Internet-based CBT as part of the treatment strategy for chronic back pain. The number of participants randomized in each study varied between 54 and 228. The majority of participants in all four studies were women (62.5%-83%).

Burhman et al [11] used Internet-based CBT in conjunction with telephone support to treat chronic back pain. The study reported that 95 participants would be required for a power of 80%; however, due to lower enrollment the study remained underpowered. The primary outcome measure was catastrophization, defined as the experience of irrationally thinking that something is far worse than it actually may be. This was measured as a subscale of the Coping Strategies Questionnaire (CSQ). The CSQ consisted of measurements of the following parameters: diverting attention, reinterpret pain sensations, coping self-statements, ignore pain sensations, praying or hoping, catastrophizing, increase activity level, control over pain, and ability to decrease pain. Patients were randomized to Web-based pain management or a waitlist control. The intervention group received access to weekly online CBT modules, guidance with physical activity and stretching exercises, and coping strategies over the course of 12 weeks. The intervention group also received weekly telephone calls that included discussion about participant goals, relaxation training advice, exercise guidance, and discussion on coping strategies. These calls occurred during the same period as the online intervention. The treatment group showed lower tendency to catastrophize and also reported better control over pain at 8 weeks. Due to significant follow-up through telephone calls, it is unclear how much of the treatment effect can be attributed to the online modality of treatment as opposed to the telephone-based support.

Burhman et al [13] performed a similar study to the one described previously, but without ongoing telephone support as part of the treatment plan. In all, 54 patients were randomized; it was reported that the study was underpowered to detect differences with conventional levels of confidence. The treatment group in this study also showed improved scores on the catastrophization subscale. Although this study did not include telephone support as part of the intervention, ongoing email support was provided to participants. Therefore, it is not entirely clear how much of the treatment effect can be attributed to the online CBT modules as opposed to the effect of email support. Along with reduction in catastrophization, the participants also reported improved control over pain with the intervention. Because the current paradigm of chronic pain management stresses the importance of maintenance or improvement in patient function, this may be seen as an important initial step toward achieving better self-efficacy and an improvement in the patient’s ability to understand and manage their own pain. However, further research would be useful to clarify whether this may indeed translate into improvements in pain and disability scores.

Carpenter et al [14] also studied an online self-help CBT intervention. The study included 141 participants who had back pain for more than 6 months and were older than 21 years of age. Over the course of three weeks, the treatment group used an online wellness workbook that included elements of CBT; the results reported an improved ability to self-manage pain. The treatment group reported decreased pain catastrophizing and a more positive outlook toward their disability. After week 3, the treatment group reported an improvement in their perceived ability to cope with their pain. Conversely, the participants in the control group were less confident about their ability to manage pain and were more likely to believe that they should avoid exercise. The study then allowed both groups to access the online workbook after the 3-week period and repeated their assessments for all the participants at 6 weeks. It was reported that the differences in the two groups were no longer apparent at 6 weeks, suggesting that access to the workbook successfully affected participants’ pain-related beliefs.

Chaiuzzi et al [12] compared an intervention group with access to a website (painACTION for back pain) designed on CBT self-management principles with a control group of participants provided with a back pain help book. Participants were recruited online and through a specialty pain clinic. The sample size was 228; sample size and power calculations were not reported. The intervention group received access to the CBT website and a weekly chat moderated by a therapist. Posttreatment follow-up at 3 and 6 months was performed. Overall, the intervention group reported reduced stress and improved coping, but pain and physical functioning were not affected significantly. However, in a subgroup of patients recruited online, pain levels did appear to be improved with the intervention compared to the control group.

In summary, four small RCTs reporting the effects of Web-based CBT for chronic back pain have been identified. All studies found reduced catastrophization in patients receiving online CBT.

Each of the trials used different measures to report pain levels. These measures included the Pain and Impairment Relationship Scale (PAIRS), Pain Self-efficacy Scale (SES), a pain diary, and self-reported pain levels for least, average, and worst pains. Of the studies that examined CBT, only Carpenter et al [14] used CBT as an adjunct to opioid therapy. None of the studies reported significant differences in pain severity.

Web-based interventions with interactive features are being increasingly studied for their potential role in the management of chronic diseases. The results from a recent review indicated Web-based interactive interventions for patients with a variety of chronic conditions may have a positive impact on patient empowerment and may facilitate enhanced physical activity [21].

The studies discussed in this section target knowledge about chronic low back pain by providing online resources and also provide support for coping through Web-based interactive features.

Lorig et al [10] performed an RCT to examine the impact of participation in email discussion groups; the outcomes of interest were health status and health care utilization. Study duration was 1 year and 580 participants were randomly assigned to treatment and control groups. The intervention group was enrolled in an email discussion group where various aspects of back pain were discussed with input from content experts. The content experts included a physician, physical therapist, and psychologist. This study used moderated email discussion; however, the topics for email discussion were mostly driven by participants and no specific predesigned content was provided to the participants. Further, the intervention group also received a back pain help book and a videotape modeling active living with back pain. No particular physical activity routine or exercise was suggested; rather, the email discussion answered general questions raised by the participants. The control group did not receive any specific back pain treatment or advice. The study included a 6-month and 1-year follow-up. At 1 year, improvements in pain, disability, role function, and health distress were reported with the intervention. The study was powered to detect these differences with a significance of P<.05. Health care utilization was reduced in the treatment group, but not to a statistically significant degree. The number of physician visits were decreased in the treatment group. Further, the mean number of hospital days (back-related days of hospitalization) were reduced by 0.25 days for the intervention group as compared to an increase of 0.04 days for the control group. Self-care orientation was improved with treatment. The study also reported that older age was associated with greater disability. The authors indicate and recognize that there are multiple factors affecting pain levels and health care utilization due to chronic back pain; consequently, it is unclear how the results can be attributed to the various parts of the intervention (discussion group, back pain help book, and videotape).

Moessner et al [15] studied aftercare intervention for patients who had already received multidisciplinary therapy for back pain. The study randomized 75 patients; low power was reported due to small sample size. Participants received an Internet-based aftercare intervention lasting 15 weeks or treatment as usual. The aftercare program included an individualized online self-monitoring module, where participants answered questions about their compliance with appropriate health behaviors. Also, the aftercare consisted of a 90-minute weekly text-based chat for a period of 15 weeks. The chat was moderated by an experienced group therapist; session topics were decided by the therapist. A physician or physiotherapist were not included as moderators for this chat. The results reported improvements in disability with the intervention. No significant difference in depression or anxiety was reported. Despite the positive results, there is an important caveat: a significant amount of data was lost because only 34 of 75 patients completed all three assessments. Moreover, the authors did not report the components of the multidisciplinary rehabilitation; as such, there is no way of knowing whether—and the degree to which—results of their Internet-based intervention were affected by components of rehabilitation. The authors theorized that patient beliefs about chronic pain may have impacted follow-up.

Krein et al [16] conducted a study that focused on improving the activity level of the participants by providing them with pedometers that gave online feedback regarding their daily activity. The pedometer feedback was used in combination with an e-community social support group. The researchers randomized 229 patients into a control group and a treatment group. The study was designed to detect a clinically meaningful difference (0.4 standard deviation or 2-point difference) in Roland-Morris Disability Questionnaire score and sought to enroll 130 participants in each group to account for potential of 25% attrition; high rates of participant follow-up were achieved for this study allowing for detection of differences in primary outcome. The treatment group received pedometers along with access to online feedback including the number of steps and individual goals to promote improvement. The treatment group was also provided with access to an online social support group. In contrast, the control group received pedometers but did not receive online feedback or social support. Assessments were performed at 6- and 12-month time points. Most participants were males. Significant improvement was reported for the treatment group compared to the control group for back pain disability at 6 months, but the difference was no longer statistically significant at 12 months. No difference was reported between the groups in terms of Fear-Avoidance Beliefs Questionnaire Physical Activity subscale. Physical activity measured by step counts was increased in the intervention group at 6 months; however, this difference was less marked when measured at 12 months. Exercise self-efficacy scores were similar between the two groups at 12 months.

Riva et al [17] randomized 51 patients into two groups. It was reported that the study was designed to achieve power of 80% with 95% confidence, and sufficient numbers were recruited for this purpose. The intervention group received access to a self-management website with interactive components including quizzes, virtual gym, an action plan, and additional online resources. The control group only received access to static features and information on the website. Four- and 8-week assessments were performed. Outcome measures included empowerment, medication misuse, physical exercise, and pain burden. The intervention group was reported to have improved patient empowerment and reduced medication misuse. Pain burden decreased, but to equal measures in both the control and intervention groups. Because pain levels decreased in both groups, it appears that the interactive features available to the intervention group did not make a significant difference to their pain levels. However, participant empowerment was reported to be significantly improved in the intervention group. It appeared that interactivity and feedback through the Internet may improve a sense of control or empowerment in chronic back pain patients.

Weymann et al [18] included participants with chronic low back pain and type 2 diabetes in their study. A total of 561 participants were randomized, of which 382 were enrolled with chronic back pain. The intervention was a tailored interactive health communication app, which provided support to participants with regards to their knowledge and attitudes about their condition. The coping style of participants was assessed prior to intervention; participants in the intervention group were offered tailored content based on their coping style. For chronic low back pain participants, information was based on recent guidelines and Cochrane reviews. Primary outcomes were patient knowledge and patient empowerment.

The study aimed to detect differences with conventional levels of confidence and 80% power; however, due to attrition, only 202 of 382 chronic low back pain participants performed the 3-month follow-up. No significant differences were detected in outcomes with the intention-to-treat analysis.

Two of the studies [10,15] reported reduction in disability. Krein et al [16] reported reduction in disability at 6 months, which was not sustained in further assessments. Lorig et al [10] reported statistically significant reduction in pain, whereas Moessner et al [15] reported improvement with the pain subscale of the 36-item Short-Form Health Survey (SF-36), but not with Numeric Rating Scale (NRS).

Four of the studies did measure empowerment/self-efficacy and mixed results were reported. Empowerment was reported to be improved in one of the studies [17] and another study reported improved self-efficacy [10]. However, no difference was reported in other studies [16,18].

A variety of diverse outcome measures have been used in the studies. Lorig et al [10] did measure health care utilization, which was reported to be decreased in their treatment group. In the context of chronic low back pain, this outcome measure has not been extensively studied in RCTs since this trial; it would be prudent for future researchers to include cost or health care utilization as an outcome measure. Further, based on the study design, it may not be possible to ascertain the individual contribution of each part of the intervention to the reported outcomes. Also, the discussion is not supportive of any particular physical activity intervention and no individual medical advice was provided to participants. This suggests that participant self-efficacy may independently affect outcomes in this population.

Although research on many of the Web-based interventions for back pain reviewed here had mixed results or do not appear to have high external validity, we did find evidence that that there are likely some benefits to online CBT for reduced catastrophization. As such, online interventions may be a useful solution to overcome current limitations of traditional face-to-face CBT because, for example, access to professionals that are able to deliver high-quality CBT remains limited. Second, many patients may not be able to afford access to such professionals or counseling. Third, physical access may also be limited due to the nature of pain, patient comorbidities, or other social factors, and large geographical distances may preclude eligible patients from accessing specialized rehabilitation or chronic pain centers. Fourth, in some cases, there may be a stigma associated with the use of a therapist or counselor. Therefore, online access to CBT may help to alleviate some of the barriers to access and provide patients a convenient alternative to face-to-face visits. Future studies using CBT as an intervention should consider including appropriate numbers of male participants to improve the generalizability of the results.

Further, empowerment/control did show improvement in six of the studies. Three of these used CBT, whereas three of the studies used other forms of Web-based support, such as email/chat or other interactive features. It appears that forms of social support other than formalized counseling or CBT may have some positive effect on the patient’s ability to manage and cope with their chronic condition.

Disability was only assessed in five of the studies and mixed results were reported. Further research in this area with studies having longer follow-up should be a priority. One study reporting health care utilization reported positive effects with the intervention. It would be important for future studies to assess this further because it is important to focus resources on interventions that can reduce use of health care resources. Further research that includes these outcomes could provide insight into future planning for the health care system and implications for clinical practice.