The Efficacy of Web-Based Cognitive Behavioral Interventions for Chronic Pain: A Systematic Review and Meta-Analysis

The Journal of Pain, Vol 11, No 10 (October), 2010: pp 917-929 Available online at www.sciencedirect.com

Critical Review The Efficacy of Web-Based Cognitive Behavioral Interventions for Chronic Pain: A Systematic Review and Meta-Analysis Debora Duarte Macea,*,y Krzysztof Gajos,z Yasser Armynd Daglia Calil,x and Felipe Fregni*,y * Laboratory of Neuromodulation, Physical Medicine and Rehabilitation Department, Spaulding Rehabilitation Hospital, Harvard Medical School, Boston, Massachusetts. y Department of Neurology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts. z School of Engineering and Applied Sciences Health, Harvard University, Boston, Massachusetts. x School of Public Health Harvard University, Boston, Massachusetts.

Abstract: Our objective was to conduct a systematic review and meta-analysis to quantify the efficacy of web-based cognitive behavioral interventions for the treatment of patients with chronic pain. MEDLINE and other databases were searched as data sources. Reference lists were examined for other relevant articles. We included 11 studies that evaluated the effects of web-based interventions on chronic pain using specific scales of pain. The pooled effect size (standardized mean difference between intervention versus waiting-list group means) from a random effects model was .285 (95% confidence interval: .145–.424), favoring the web-based intervention compared with the waiting-list group, although the effect was small. In addition, these results were not driven by any particular study, as shown by sensitivity analysis. Results from funnel plot argue against publication bias. Finally, the average dropout rate was 26.6%. In our meta-analysis, we demonstrate a small effect of web-based interventions, when using pain scale as the main outcome. Despite the minor effects and high dropout rates, the decreased costs and minor risk of adverse effects compared with pharmacological treatments support additional studies in chronic pain patients using webbased interventions. Further studies will be important to confirm the effects and determine the best responders to this intervention. Perspective: Our findings suggest that web-based interventions for chronic pain result in small pain reductions in the intervention group compared with waiting-list control groups. These results advance the field of web-based cognitive behavioral interventions as a potential therapeutic tool for chronic pain and can potentially help clinicians and patients with chronic pain by decreasing treatment costs and side effects. ª 2010 by the American Pain Society Key words: Web-based interventions, chronic pain, meta-analysis, cognitive behavior intervention, systematic review.

C

hronic pain costs more than $200 billion annually in United States.3 In fact, pain management is generally expensive due to the extensive need of longterm rehabilitation in multidisciplinary treatments.37 At least 40 million Americans suffer chronic recurrent headaches,39 and 15% of the adult U.S. population has had

Address reprint requests to F. Fregni, MD, PhD, 125 Nashua Street, 7th floor, Laboratory of Neuromodulation, Spaulding Rehabilitation Hospital, Boston, MA 02114-1198. E-mail: [email protected]. edu 1526-5900/$36.00 ª 2010 by the American Pain Society doi:10.1016/j.jpain.2010.06.005

persistent low back pain.40 Back pain is one of the most common medical problems in the United States, affecting 8 of 10 people, and is also the most widespread cause of job-related disability.40,41,48 Therefore, there is a significant global burden of chronic pain that directly influences quality of life.1 In this context, novel interventions to treat this condition are desirable. Cognitive behavior therapy (CBT) is one of the most well-known treatments for chronic pain and is a widely used and effective form of therapy for many psychological disorders, including depression and anxiety disorders.20,26 Behavioral and cognitive 917

918

The Journal of Pain

treatments that are designed to ameliorate pain, distress, and disability were first introduced over 40 years ago and are now well-established treatments for depression, anxiety disorders, and chronic pain.6,25,26 The main principles of CBT for pain management are based on helping the patient to understand how cognition and behavior can affect the pain experience, coping-skills training, and cognitive restructuring. During treatment, patients are encouraged to apply their coping skills to a progressively wider range of daily situations.12,27,45 One disadvantage of treatment with CBT, mainly in developing countries, is its high costs and difficult accessibility,24,42 often being an unavailable treatment. As internet access has grown exponentially, CBT can be administered through internet-based interventions.49,50 Internet use continues to increase, including among groups that were not always familiar with this technology, such as women, elderly persons, and minority groups.31,43,44 The widespread use of the internet offers new treatment opportunities, adapting the suitability of cognitive and behavioral techniques to a computer-based format.15,50 Web-based intervention has advantages over traditional cognitive and behavioral therapy (CBT)50 for both clients and health care professionals.52 For instance, the patient’s anonymity and free accessibility removes the stigma that is incurred by seeing a therapist.20,47 In addition, the treatment can be administered at any time and place, at its own pace, and the material can be reviewed as often as desired. There are some limitations with this method as well, such as increased difficulty in providing feedback to patients, lack of motivation, and discipline to follow the treatment. In addition, the lack of personalized treatment creates a barrier to the web-based CBT as it results in decreased capacity to modify and change behaviors associated with chronic pain.6,43 These factors decrease the enthusiasm for the web format of CBT. A large number of randomized controlled trials (RCTs) testing internet interventions has taken place, most of them being published during the past half decade, focusing primarily on depression and anxiety-related disorders and ameliorating symptoms in these patients.5,6,25 Although the evidence for internet-based interventions with regard to anxiety and depression is robust, the evidence of its efficacy in chronic pain is mixed and uncertain.7,9,16,23,46 Therefore, we conducted a systematic review and meta-analysis to assess the results of randomized controlled trials of internet-based cognitive behavioral programs, with or without minimal therapist contact for patients in treatment of chronic pain.

Methods Literature Review The first step of our meta-analysis review was to perform a literature search using the following databases: MEDLINE, Cochrane, Psychoinfo, and Scielo. In addition, we examined reference lists of the retrieved papers. We

Web-Based Therapy for Chronic Pain: A Meta-Analysis used the key search terms ‘‘web-based,‘‘ ‘‘internetbased,‘‘ ‘‘online intervention,‘‘ ‘‘web-based interventions,‘‘ ‘‘distance cognitive therapy,‘‘ ‘‘online treatment,‘‘ ‘‘online intervention,‘‘ ‘‘internet delivered,‘‘ ‘‘distance cognitive therapy,‘‘ ‘‘web interventions,’’ and ‘‘internetbased interventions,’’ all of which were matched with the keywords ‘‘chronic pain.’’ We did not use the specific term ‘‘cognitive behavioral therapy’’ alone or with these other terms as we observed that using this term did not yield us any additional articles. Therefore we excluded it from our search strategy. We found 164 articles, 40 of which were related to using the internet for chronic pain. Subsequently, we checked each article according to our inclusion criteria.

Selection Criteria We included prospective studies that evaluated the effects of web-based intervention for treating chronic pain. We adopted the following inclusion criteria: 1) article written in English; 2) web-based interventions for the treatment of chronic pain; 3) randomized controlled trials; 4) measurement of pain as 1 of the outcomes; 5) studies published in a book, journal, proceeding, or indexed abstraction; 6) studies reporting the pain scale before and after the treatment; 7) study reporting a comparison between an intervention group and a waiting-list group; and 8) treatments that included cognitive and behavioral treatment principles as the main strategy to treat chronic pain. Lorig et al34 included patients with chronic disease, such as hypertension, lung disease, and arthritis, but 1 of the primary outcomes was pain improvement. Consequently, we included this study because it matched all of our inclusion criteria.

Data Extraction The data were extracted by 1 author (DDM), using a structured form, and checked by another author (FF). The following variables were extracted: 1) mean and standard deviation of pain scales before and after treatment and at the follow-up (when available) for the active and waiting-list groups; 2) demographic, clinical, and treatment characteristics (eg, number of patients in the control and treatment groups, age, gender, type of chronic pain, type of previous treatments); 3) dropout rate; 4) methods of assessment; 5) evaluation model; and 6) type of intervention. When a study measured mean pain at 2 or more time points, the mean pain at the shortest time point was taken; if not available, we deduced this value from other parameters or via author contact. We tried to address other outcomes, such as depression and quality of life. However, the outcomes used in each article were too heterogeneous and it was not possible to compare effects between studies for other domains. We analyzed the other outcomes only qualitatively since it was not possible to conduct a meta-analysis with these other outcomes. In Discussion, we include the changes in other outcomes. Two studies by Brattberg8,9 used the same population sample; therefore, only 1 article was included. Andersson

Macea et al 4

et al compared 2 active self-help treatments, adding telephone contacts for just 1 group, with no waitinglist group. Thus, we excluded this study. When a study did not report the standard deviation for pain levels, we deduced them from other parameters34 or we contacted the authors. We asked Schulz et al46 for their pain values and standard deviation after treatment.

Qualitative Analysis Due to the small number of studies in our metaanalysis, we decided to exclude qualitative analysis as a method of inclusion criteria. Nevertheless, we conducted a qualitative analysis of the studies that met our inclusion criteria. We conducted a qualitative analysis, considering the following aspects: 1) randomization (allocation sequence); 2) attrition rate, properly analyzed and described; 3) adjustment for multiple testing; 4) prior power calculation; and 5) adequate data presentation (test statistic, P value, and confidence limits provided; endpoints tabulated).

Quantitative Analysis and Statistical Analysis All analyses were conducted using STATA statistical software, version 8.0 (StataCorp, College Station, TX). We initially calculated the standardized mean difference and the pooled standard deviation for each comparison. Continuous pain outcomes were converted to a 0–100 scale and pooled using a random effects model. We used Cohen’s d as a measure of effect size (effects sizes were computed by comparing the mean changes from pre- to post-treatment ratings of the active versus waiting-list group). Then, we measured the pooled weighted effect size (weighted by the inverse variance of each study), using the random and fixed effects models. The random effect model lends relatively more weight to smaller studies and wider confidence intervals than the fixed effect model. Heterogeneity was evaluated with the Q statistic— that is, a useful test to assess normality of the residuals within a range of an independent variable, thus to give information whether there is a significant variance across the studies’ results. Because there was significant heterogeneity, we used the random effects model. For the studies that provided long-term follow-up, we only used the first measurement after the end of treatment. Furthermore, we also assessed publication bias using the Begg-modified funnel plot in which the standardized mean difference of each plot was plotted on a logarithmic scale against its corresponding standard error for each study and performed the Egger’s test18 to assess whether there was a significant asymmetry. With this test, if smaller studies show effects that differ systematically from larger studies, the regression line will not run through the origin, therefore indicating possible publication bias (in other words, small studies with negative results are not published).18

The Journal of Pain

919

Results Studies Retrieval Keyword searches on MEDLINE, Scielo, PyschINFO, and Cochrane Library yielded 164 citations. Using our inclusion criteria, we selected 15 studies that used webbased interventions for chronic pain patients for the final analysis. Of these studies, 11 met all our inclusion criteria (Fig 1 details the search strategy) and were analyzed in our review. References were excluded primarily due to: 1) reviews; 2) studies that did not assess web-based interventions to treat chronic pain; 3) studies that assessed website performance; and 4) other topics. Fig 1 shows the QUOROM diagram flow and details that were used to identify studies in our meta-analysis.

Demographic Findings Aggregation of participant data demonstrated a total of 2,953 participants (67.5% women). The average age of the participants was 41.32 (range of 7 to 91) years. Demographic findings of these studies are summarized in Table 1.

Pain Measurement and Types of Chronic Pain Changes in pain (mean pain before and after treatment) in the intervention groups ranged from 3.67 to 31 (on a 0-to-100 scale). In Table 2, we show the standardized mean differences of each study (difference between the final pain-scale measurement and the first pain-scale measurement). More than10 types of pain syndromes were studied. As shown in Table 3, 39% of the patients in the control arm and 38% in the intervention arm did not specify the type of chronic pain. Among the etiologies that were reported, the majority was back pain (29%) followed by osteoarthritis (13%).

Web Intervention Designs On review of the selected articles, we observed a notable variability in the design of the web-based intervention. Due to this source of variability, we included in the table individual results from each study. In Table 4, we included not only the posttreatment outcome of pain but also the follow-up assessments. All of the articles reported improvement in healthrelated and behavioral outcomes for participants using the web-based interventions (such as pain awareness, control over pain, health distress, and better work capability) compared with the waiting-list group. Time of intervention ranged from 6 to 20 weeks. A summary of these articles is shown in Table 4. Table 4 also shows the principal characteristics of the 11 studies included in our statistical analysis. Although there was a great variability in the methods across studies, most of them showed improvement in pain-related disorders and quality of life.

920

Web-Based Therapy for Chronic Pain: A Meta-Analysis

The Journal of Pain

Figure 1. Flowchart of the selection process of peer-reviewed articles for mean pain analysis.

Dropout Rates in Short- and Long-Term Follow-Up Studies Because dropouts are common in treatments that are conducted over the internet, we analyzed these data separately, listing all of the results in a separate table, as shown in Table 5. In these studies, the average dropout rate was 26.16% for the intervention and control groups, higher than the dropout rates reported in traditional CBT interventions (14%).38 The mean dropout rate in the waiting-list group was 64%, and 40% in the intervention group, a difference of 24%. The dropouts occurred

Baseline Demographic Characteristics of Control and Treatment Groups in the Beginning of the Study (Before Dropout of Subjects)

Table 1.

Number of patients, total Mean age (6SD) Sex (% female)

CONTROL GROUP

TREATMENT GROUP

1,302 41.23 (8) 67

1,188 41.43 (7,9) 69

at all time points (ie, before the beginning of the treatment, during the treatment, during short-term followup, and during long-term follow-up). The analysis of predictors of dropouts implicated the following factors: 1) longer duration of chronic pain;11 2) less severe disease at baseline;16,49 3) limited knowledge in computing;11 4) young age;35,49 5) higher levels of health distress and limitations of activity;34 and 6) male gender.32

Comparisons of Web-Based Interventions Versus Waiting List for Chronic Pain—Statistical Analysis Most of the studies used small sample sizes, except for those by of Lorig et al,32 Lorig et al,34 and Lorig et al.35 Because the analyzed studies included intervention and waiting-list groups, we performed an analysis to compare web-based intervention versus waiting-list groups. The results from the fixed effects model revealed a significant pooled effect size (.231, 95% CI, .149–.312). The random effects model showed similar results (pooled effect size of .285, 95% CI, .145–.42). For all analysis, we

Macea et al

The Journal of Pain

921

Standardized Mean Differences in VAS From Each Study Comparing After Versus Before Treatment in the Control and Intervention Groups, in a 0-to-100 Scale

Table 2.

Control group Treatment group

SCHULZ ET AL, 200746

LORIG ET AL, 200232

LORIG ET AL, 200634

8

12

.47

0

15

3.6

HICKS ET AL, 200623

BRATTBERG, 20069

STROM ET AL, 200049

4

4.5

.8

14

23.6

10.35

BERMAN ET AL, 20087 7 6.4

LORIG ET AL, 200835

BUHRMAN ET AL, 200411

CONNELLY ET AL, 200613

DEVINENI

.3

18

.9

4.9

6.7

14

6.1

13.2

ET AL

200516

NOTE. Negative results indicate pain reduction.

observed statistically significant evidence for heterogeneity in the primary analysis (Q12, X2 = 18.404, P = .041). Because there was significant heterogeneity, the random effects model was more appropriate (Fig 2). Sensitivity analysis showed that the results would not change even if the study that had the largest effect was removed. Notably, the primary results are not altered if any study is excluded—the pooled risk ratio would have increased slightly if Lorig et al,35 Berman et al,7 Lorig et al,34 or Buhman et al11 (change to .36) were excluded. Therefore, the overall finding of a significant effect of web-based interventions on pain reduction compared with waiting-list groups remains steadfast after the exclusion of any study (Fig 3). Because publication bias is an important issue in a meta-analysis, we constructed a funnel plot for visual assessment. Using the Begg and the Egger test for all 11 trials, we found no evidence of publication bias. In fact, the distribution of studies was symmetrical (Fig 4). Finally, the P value for the Egger test was not significant, therefore suggesting that the results of this metaanalysis do not stem from publication bias. The Q statistic was significant (P < .05); thus, we used the random effects model. We also conducted sensitivity

Types of Disease From the Studies, When Described

Table 3.

TYPE OF DISEASE Rheumatoid arthritis Fibromyalgia Burnout* Whiplashy Myofascial pain syndrome Neuralgia Headachez Osteoarthritis Chronic pain syndrome Abdominal recurrent pain (RAP) RAP1 headache Low back pain Back pain Not specified

CONTROL GROUP

TREATMENT GROUP

72 49 10 2 1

72 49 8 3 1

5 95 158 0 3

5 83 136 3 3

17 299 29 467

16 316 22 394

NOTE. The table shows the main diseases from the group after the intervention. *It includes patients with burnout and burnout 1 fibromyalgia. yIt includes patients with whiplash, whiplash 1 DPT and whiplash 1 fibromyalgia. zIt includes all types of headache.

analysis to measure the influence of each study and in fact this analysis showed that our results were not being driven by any single study.

Qualitative Analysis According to our analysis, Schulz et al46 measured merely several items of our quality assessment. Their study did not include group allocation, dropout descriptions, proper statistical tests, P value, confidence limits, or adjustments for multiple tests. The other studies varied in quality but were considered to have adequate quality overall. None of the articles calculated prior power, and only 37,13,35 adjusted for multiple tests. All of the articles described dropout rates, but only Lorig et al,35 Connelly et al,13 Hicks et al,23 and Lorig et al32 used intention-to-treat analysis. Some of the articles11,16,49 had a high attrition rate and used subjects who completed the study for case analysis. As noted above, the sensitivity analysis showed that the results would not change if any of these studies was excluded.

Discussion Our study includes data from 11 randomized clinical trials, assessing 2,953 subjects. The results of this metaanalysis suggest that web-based interventions for chronic pain effect are associated with small reductions in pain in the intervention group compared with waiting-list control groups. The 11 studies showed wide variability in the type of assessments that were used, the study population, the etiology of chronic pain, and time of intervention. Supporting the results of this meta-analysis, all of the included studies analyzed other outcomes in addition to improvement in pain, such as functional changes and medication use. These other outcomes also showed significant changes, favoring the intervention group. Connelly et al13 and Hicks et al23 reported improvement in headache frequency and duration. Brattberg8,9 demonstrated changes in pain intensity-related behavior immediately after treatment, observing a greater capacity for work after a 1-year follow-up. Devineni et al16 and Schulz et al46 reported a reduction in medication use, and Lorig et al32 showed a decline in physician visits. Patients’ attitudes and beliefs, particularly fear-avoidance beliefs and passive coping strategies, are increasingly accepted as having an important role in disability related to pain problems as the management of chronic pain based on psychological, social, and biological

WOMEN (%)

AGE GROUP (AGE RANGE)

STUDY GROUPS

N

INTERVENTION GROUP

CONTACT WITH THERAPIST

DROPOUT (%)

SHORTER ASSESSMENT: POSTTREATMENT

6-MONTHS FOLLOW-UP

90.15 52.35 (22–89) 1- Internet-based arthritis selfmanagement program 2- Usual care

855 Self-management pain; Email relaxation/ cognitive pain; problem-solving; difficult emotions; exercise; healthy eating; medication; depression

6 weeks (logged in a mean of 31.6 times)

24

Not described

Chronic Lorig back pain et al32,x

61.5

45.5 (not 1- Email discussion described) 2- Waiting-list (nonhealthrelated magazine)

580 Closed, moderated, email discussion 1 book and videotape

6 weeks (mean of 8 times emailed)

27

Not described

Chronic Connely headache et al13,z

49

9.9 (7–12)

1 - CD rom program 2- Standard medical care

16

Shorter headaches duration, improvement at headache Index, lower headaches frequency and headache intensity.

Hicks Chronic et al2,z headache and/or abdominal pain

64

11.9 (9–16)

1- Online manual access 2- Standard medical care

32

71% achieved clinically Not described significant improvement, just 19% of control group reach it; improvement at pain intensity and frequency No significance between Not described groups, except awareness of responses to pain. Reductions in mean pain scores at log on and log off (intervention group)

Berman et al7

Chronic pain

87.15 65.8 (55–91)

1- Online intervention providing mindbody exercises 2- Control group

Email

37 Introduction to headache; Telephone 4 weeks treatment of headaches; call (6 sessions) relaxation; coping, thought changing; problem solving; deep breathing; progressive muscle relaxation; pain behavior management 47 Self-pain management; Email or 7 weeks effects of tension telephone (7 sessions) and relaxation; positive call and negative thoughts; benefits of social and physical activities 89 Self-care modules: Email 6 weeks (median abdominal breathing, of 22.5 visits) relaxation, writing positive and difficult experiences, creative visual expression and positive thinking

12.4

COUNTRY

Significant timeUSA randomization interactions: health distress, activity limitation, self-reported global health and pain. Physicians’ visits USA declined 1.5 visits for the treatment group. Mean hospital days declined nearly .20 days for the treated group. Not described USA

Canada

USA

Web-Based Therapy for Chronic Pain: A Meta-Analysis

Rheumatoid Lorig arthritis, et al35,x fibromyalgia and osteoarthritis

The Journal of Pain

STUDY

HEALTH CONDITION

PERIOD OF TREATMENT (NUMBER OF ACCESS)

922

Table 4. Selected Peer-Reviewed Articles Showing the Main Characteristics of Format, Methods, Target Population, and Long-Term Results From Each Study Included in our Meta-Analysis

STUDY

Continued HEALTH CONDITION

WOMEN (%)

AGE GROUP (AGE RANGE)

Diabetes, 71.4 Lorig hypertension, et al34,x lung disease, heart disease, arthritis

57.5 (22–89)

Chronic Strom headache et al49,*

69

36.7 (19–62)

Chronic Devineni headache et al16,y

80.3

41.3 (24–56)

STUDY GROUPS 1- Web-Based chronic disease self-management program 2- Usual care controls

N

INTERVENTION GROUP

958 Cognitive symptom management; exercise programs; self-manage negative thoughts; healthy eating; medications; action planning; methods for solving problems. 1- Web-based group 102 Progressive relaxation; 2- Waiting-list problem solving; control controlled breathing.

CONTACT WITH THERAPIST

PERIOD OF TREATMENT (NUMBER OF ACCESS)

DROPOUT (%)

SHORTER ASSESSMENT: POSTTREATMENT

6-MONTHS FOLLOW-UP

Macea et al

Table 4.

COUNTRY

6 weeks (logged in an average of 26.5 times)

19

Not described

Changes in self-efficacy

USA

Email

6 weeks (6 sessions)

56

Not described

Sweden

4 weeks (4 sessions)

38.1

Improvement at headache index: frequency of headache days and peak intensity; 50% with clinically significant improvement. Greater decrease in headache activity, 39% of treated group showed clinically significant improvement on headache symptoms. 47% maintaned improvement; treatment had a significant impact on general headache symptoms and headache-related disability. 35% within-group reduction of medication usage.

Not described

USA

1- Web-based group 139 Progressive relaxation; Telephone 2- Symptom limited biofeedback call monitoring waitlist with autogenic training; stress management

The Journal of Pain

Email

923

HEALTH CONDITION

WOMEN (%)

AGE GROUP (AGE RANGE)

89

47 (18–65)

Chronic low Schulz back pain et al46,z

29

62.5

Buhrman et al11

Chronic back pain

*One-month follow-up. yTwo-months follow-up. zThree-months follow-up. xOne-year follow-up.

1- Web-based rehabilitation course 2- Waiting list

N

INTERVENTION GROUP

CONTACT WITH THERAPIST

DROPOUT (%)

60 19 films 1 Socratic dialogue1 written materials 1 discussion

Email and internet meeting

20 weeks (20 meetings)

8

45.26 (18–65) 1- Access to website 2- Control group

35 Gym (videos and description of exercises) 1 forum and chat-rooms 1 talk to physicians 1 tell a story

Telephone

5 months (average of 11.5 visits)

0

44.6 (18–65)

56 Education; cognitive skill acquisition; behavioral rehearsal; generalization and maintenance.

Email and 6 weeks telephone (8 sessions)

9

1- Internet-based CBT 2- Waiting list

SHORTER ASSESSMENT: POSTTREATMENT Improvement in depression, pain, vitality, social function, performance problems involving work or other activities due to physical illness and the presence of stress symptoms. 13 (23) increased work capacity. Suggest a decrease in the intensity of pain; an increase in physical activity; a reduction in medical consultation and use of painkillers and a gain in declarative and procedural knowledge. Improvement in catastrophizing, control over pain and ability to decrease pain, in both groups.

6-MONTHS FOLLOW-UP

COUNTRY

Not described

Sweden

Not described

Switzerland, Germany

Not described

Sweden

Web-Based Therapy for Chronic Pain: A Meta-Analysis

Brattberg9 Chronic pain and/or burnout

STUDY GROUPS

PERIOD OF TREATMENT (NUMBER OF ACCESS)

The Journal of Pain

STUDY

Continued

924

Table 4.

Macea et al Table 5.

The Journal of Pain

925

Dropout Rates in the Analyzed Studies BASELINE

STUDY Lorig et al35 Lorig et al32 Connelly et al13 Hicks et al23 Berman et al7 Lorig et al34 Strom et al49 Devineni et al16

Brattberg G9 Schulz et al46 Buhrman et al11

DROPOUT RATES*

N

40% 22% in the first 6 months, 7% in the last 6 months 8% 32% 12% 19% 56% Treatment and follow-up dropout rates, 38.1% and 64.8% 13% 0 9%: 1 in the treatment group and 3 in the former control group at the 3 months follow-up

855 580

204 159

37 47 89 958 102 139

6 15 11 178 57 53

60 35 56

5 0 5

N

DROPOUTS

*Just some of them described the subjects’ characteristics at the dropout group.

factors.10,19,45 So, patients who undergo cognitive and behavioral therapy present better self-pain management and attitudes towards their pain. Therefore, patients can increase their social activities and work capacity,8,9 resulting in better clinical outcomes, such as decrease of depression and anxiety.45 Long-term follow-up studies (more than 3 months of follow-up)23,32,34,35 reported better improvement to other outcomes (such as work capability) compared with short-term studies, which indicates that the use of these therapies shows better results over longer periods. One explanation is that behavioral changes take longer to occur;10,12 therefore, longer periods of follow-ups might be the best approach to optimize results in such trials. One important consideration of web-based interventions is the treatment model that is adopted. Not all of the studies implemented cognitive-behavioral intervention as proposed by Beck,12 in which the therapist assesses patients’ problems and develops, in a minimum

Figure 3. Assessment of the individual influence of each study. The change in the overall effect size and 95% confidence interval for the meta-analysis after eliminating the indicated study is shown. Effect size and Cohen’s d (standard mean difference), error bars represent the 95% confidence interval. of 12 sessions, strategies of how to deal with pain. Although there is variability between treatment methods and duration, all the studies gave information and taught their patients about self-pain behavior management, relaxation strategies, coping with their pain, thought changing, problem solving, importance of exercise, and importance of controlling depression.12 As the treatment was delivered by internet and computer, the studies had to adapt to the web-based format to obtain best results.50 One example is the study of Lorig et al32 that evaluated the use of email as a communication tool with patients and also gave access to materials on behavioral changes and pain control. Although we observed a small effect, there are several advantages of using the internet to deliver treatments for chronic pain. For example, it has the potential of expanding treatment options for many patients, especially those whose location prohibits access to relevant care.50 It can also be a substantive first-line treatment choice in cases in which patients do not have access to face-to-face treatments.43 Cognitive and behavioral techniques are largely used to treat patients with chronic pain in face-to-face therapies17 and other diseases, such as

Figure 4. Funnel plot (publication bias assessment) of the efFigure 2. Forest plot including all the studies analyzed in this meta-analysis, with the pooled effect size (intervention versus waiting-list group) for studies of web-based interventions effects on chronic pain.

fect sizes (Cohen’s d) according their standard errors. The horizontal solid line is drawn at the pooled effect size, and angled lines represent the expected 95% confidence interval for a given standard error, assuming no between-study heterogeneity.

926

Web-Based Therapy for Chronic Pain: A Meta-Analysis

The Journal of Pain

Table 6.

Summary of the Findings, the Future Directions, and Challenges Shown in this Article ACTUAL FINDINGS

Small improvement in pain

Great improvement in pain-related conditions (such as work disability) Use of cognitive behavioral interventions are effective

FUTURE INVESTIGATION Cost-effectiveness (therapist time, payments to them and investment from patients) Studies including face-to-face treatments and placebo groups Longer follow-up assessments

depression. Using the 20-item Center for Epidemiologic Studies–Depression (CES–D) scale, Mackinnon et al36 showed that participants in both the depression information website and the CBT internet intervention continued to show reductions in depression symptoms after posttest. Also, follow-up results from this study provided some evidence that the reduction in CES–D scores (scale to measure depression) was greater in the treatment group compared with controls and persisted over the 12-month period posttest.36 This study confirms the conclusions from other studies regarding the effectiveness of web-based techniques for depression.5,21,51 The cost-effectiveness of web-based interventions, based on assessment of baseline pain, also remains uncertain. None of the articles in this meta-analysis analyzed this issue in detail, although some concluded that web-based interventions are less costly than face-toface therapy.8,9,11 Lorig et al,33 in a study of an intervention group, was 1 of the few groups that reported costs. They calculated that the cost was approximately £150 (around US$243.27) per person, resulting in a reduction of approximately £272 (around US$441.13) per person compared with standard treatment. Future studies on cost-effectiveness will be helpful in providing clarity to the economical advantages of web-based interventions. It may also be possible to reduce therapist contact time while maintaining efficacy;52 ultimately, the level of therapist involvement can result in no assistance or minimal therapist contact by email or telephone. One problem with internet recruitment is that it can be difficult to generalize the obtained results to the entire population, since all the participants from these studies had access to the internet and thus needed to have some computer skills. Although global internet use is increasing, it is unlikely to occur for chronic condition patients.43 One potential problem of self-help treatment without personal contact is the lack of a proper diagnosis—ie, what is not always remedied through questionnaires—given the issues with therapist experience with these new tools and the small sensitivity and specificity of self-reported questionnaires to give a proper diagnosis.22,51 The high dropout rates that are observed in most of the analyzed studies remain a challenge. Determining whether the remaining subjects present with differences compared with dropouts can aid in the analysis. Buhrnam et al11 reported that 5 patients who dropped out suffered more years of chronic pain, whereas Lorig et al35 had dropouts who had higher levels of health distress and activity limitation. According to Devineni et al16 and Strom et al,49 dropouts had less severe headaches

CHALLENGES High drop-out rates

Access to Internet Best approach to the treatments formats

at baseline. In this instance, the dropouts could have experienced rapid improvement, leaving the study before the final assessment. Devineni et al16 also showed that dropouts perceived less of a benefit from treatment and had fewer years of computing experience than the remaining patients. Adding to these findings, Lorig et al35 showed that noncompleters had fewer years of education. Thus, lack of computer skills can be a barrier to this type of treatment. Two studies enrolled just children and adolescents (age range from 9 to 16),13,23 and 9 studies included adults and elderly (age range from 22 to 91). There were some differences in therapies, mainly in design and content for each population. One of the differences is the inclusion of parents in the child’s treatment since they need to help their children to follow the treatment and do the online exercises.13,14,28 Also, the content is customized for children. One possible issue in dealing with children is the difficulty in explaining the treatment and the assessments tools.14,29 In our meta-analysis, it was not possible to conduct a subgroup analysis to measure differences between different ages on effect sizes because the studies with young populations used small sample sizes. However, in an overall analysis, the benefits gained from all different types of population were the same (similar standardized mean difference in the VAS scale). The placebo effect is another important issue that needs to be addressed. All the articles included in our metaanalysis did not use a placebo group to compare against the intervention groups. Placebo effect must be ruled out in web-based interventions studies, especially in the context of pain that is associated with a large placebo effect.27,37 Considering that these therapies involve constant contact with researchers or therapists, it may present some positive effects over the patients’ chronic pain sensation. Because we compared the results against a waiting-list group, it is possible that the results found were due to a placebo effect. However, the effects observed in these studies also influenced other domains, such as functional domains, and, in addition, there was a long-lasting effect on pain.8,32,35 Nevertheless, randomized controlled trials that explore internet-based treatments are needed, especially those that include a control group, an intervention group with a proven effective therapy, and an internet-based therapy.

Limitations Although the clinical characteristics of patients, when considering each study separately, were fairly similar

Macea et al with regard to age, gender, and years of education, the majority of the studies (8 articles) enrolled a small number of subjects, used different assessments for evaluation, included patients with different types of diseases, and comprised studies that used different models of intervention.7-9,11,13,16,23,46,49 The heterogeneity test that addresses whether effect sizes from different studies are estimates from the same population was significant. Moreover, sensitivity analysis showed that our results were not driven by a particular study, because the exclusion of any study did not change the results. Further, Begg’s funnel plot did not detect a publication bias and showed fairly symmetrical distribution. Another important limitation of our meta-analysis was the number of articles for inclusion. Nevertheless, the goal was to provide an initial summary of results to encourage research in this area, given the relative novelty of internet use for interventions. We show in Table 6 a summary of the findings, and the future directions and challenges shown in this article.

Next Steps Although the results of our study show significant effects toward web-based therapies, the effect size is small. However, the results encourage future investigations to determine the effects of this technique in clinical practice and other issues that are related to the most ideal development for this therapeutic tool. Therefore, some aspects should be discussed when designing such studies. First, it is unclear what the optimal duration of cognitive and behavioral therapies and the duration of its therapeutic effects are. In particular, cognitive and

References 1. American Academy of Orthopaedic Surgeons: International focus on Bone, Joint Decade. Available at: http:// www2.aaos.org/aaos/archives/acadnews/99news/decade. htm. Accessed September 21, 2009 2. American Academy of Orthopaedic Surgeons: Academy focuses research resources on burden of musculoskeletal disease. Available at http://www2.aaos.org/aaos/archives/ acadnews/99news/os1-brd.htm. Accessed July 26, 2009 3. American Academy of Orthopaedic Surgeons: Musculoskeletal conditions in the United States. Available at: http://www2.aaos.org/aaos/archives/bulletin/oct99/musculo. htm. Accessed July 10, 2009 4. Andersson G, Lundstrom P, Strom L: Internet-based treatment of headache: Does telephone contact add anything? Headache 43:353-361, 2003 5. Baker CD, Kamke H, O’Hara MW, Stuart S: Web-based training for implementing evidence-based management of postpartum depression. J Am Board Fam Med 22:588-589, 2009 6. Bee PE, Bower P, Lovell K, Gilbody S, Richards D, Gask L, Roach P: Psychotherapy mediated by remote communication technologies: A meta-analytic review. BMC Psychiatry 8:60, 2008

The Journal of Pain

927

behavioral interventions are long-lasting, but only few studies have performed long-term follow-up evaluations.23,32,34,35 All of them showed that the effects are maintained over 12 months after the end of therapy. Second, the optimal treatment model needs to be reviewed. The studies in this meta-analysis used different methods to assess and treat patients, thus not allowing a conclusion to be made on the best cognitive and behavioral model to use in patients with chronic pain. The third and final issue is effect size of the web-based intervention compared with face-to-face interventions. In traditional CBT interventions for chronic pain, Morley et al38 showed in their meta-analysis that cognitivebehavioral treatments were associated with significant effect sizes (median effect size across domains = .5) on all domains of measurement when compared with waiting-list groups. Kiropoulos et al30 showed that both therapies, web-based and traditional CBT, had similar effects on panic disorder and agoraphobia problems. Because there are no consistent results that compare the efficacy between web-based and face-to-face chronic pain therapies, it is difficult to crown the more effective technique without proper studies. The results of this meta-analysis advance the field of web-based cognitive behavioral therapies as it summarizes the results to date of studies that have been conducted. Because novel technological techniques of self-management of pain have been developed, an important question for future studies is which type of web-based interventions works best for each group of pain sufferers and also whether and when these techniques can be a complementary treatment for chronic pain.

7. Berman RL, Iris MA, Bode R, Drengenberg C: The effectiveness of an online mind-body intervention for older adults with chronic pain. J Pain 10:68-79, 2009 8. Brattberg G: Internet-based rehabilitation for individuals with chronic pain and burnout II: A long-term follow-up. Int J Rehabil Res 30:231-234, 2007 9. Brattberg G: Internet-based rehabilitation for individuals with chronic pain and burnout: A randomized trial. Int J Rehabil Res 29:221-227, 2006 10. Buchbinder R, Jolley D, Wyatt M: Population-based intervention to change back pain beliefs and disability: Three part evaluation. BMJ 322:1516-1520, 2001 11. Buhrman M, Faltenhag S, Strom L, Andersson G: Controlled trial of Internet-based treatment with telephone support for chronic back pain. Pain 111:368-377, 2004 12. Cognitive Therapy of Depression, Beck AT, Rush AJ, Shaw BJ, Emery G (eds), 1st ed. New York, NY, Guilford Press, 1979 13. Connelly M, Rapoff MA, Thompson N, Connelly W: Headstrong: A pilot study of a CD-ROM intervention for recurrent pediatric headache. J Pediatr Psychol 31:737-747, 2006 14. Christophersen ER, Mortweet SL: Treatments that work: Empirically supported strategies for managing childhood problems, 4th ed. Washington, DC, APA Books, 2005

928

The Journal of Pain

15. de Groen PC, Barry JA, Schaller WJ: Applying World Wide Web technology to the study of patients with rare diseases. Ann Intern Med 129:107-113, 1998 16. Devineni T, Blanchard EB: A randomized controlled trial of an internet-based treatment for chronic headache. Behav Res Ther 43:277-292, 2005 17. Eccleston C, Williams AC, Morley S: Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane Database Syst Rev:CD007407, 2009 18. Egger M, Davey Smith G, Schneider M: Minder C: Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629-634, 1997

Web-Based Therapy for Chronic Pain: A Meta-Analysis 33. Lorig KR, Ritter PL, Dost A, Plant K, Laurent DD, McNeil I: The Expert Patients Programme online, a 1-year study of an Internet-based self-management programme for people with long-term conditions. Chronic Illn 4: 247-256, 2008 34. Lorig KR, Ritter PL, Laurent DD, Plant K: Internet-based chronic disease self-management: A randomized trial. Med Care 44:964-971, 2006 35. Lorig KR, Ritter PL, Laurent DD, Plant K: The internetbased arthritis self-management program: A one-year randomized trial for patients with arthritis or fibromyalgia. Arthritis Rheum 59:1009-1017, 2008

19. Fordyce WE, Fowler RS Jr, Lehmann JF, DeLateur BJ: Some implications of learning in problems of chronic pain. J Chronic Dis 21:12, 1968

36. Mackinnon A, Griffiths KM, Christensen H: Comparative randomised trial of online cognitive-behavioural therapy and an information website for depression: 12-month outcomes. Br J Psychiatry 192:130-134, 2008

20. Gega L, Marks I, Mataix-Cols D: Computer-aided CBTself-help for anxiety and depressive disorders: Experience of a London clinic and future directions. J Clin Psychol 60:147-157, 2004

37. Mackintosh C, Elson S: Chronic pain: Clinical features, assessment and treatment. Nurs Stand 23:48-56; quiz 58, 60, 2008

21. Griffiths KM, Christensen H, Jorm AF, Evans K, Groves C: Effect of web-based depression literacy and cognitivebehavioural therapy interventions on stigmatizing attitudes to depression: Randomized controlled trial. Br J Psychiatry 185:342-349, 2004

38. Morley S, Eccleston C, Williams A: Systematic review and meta-analysis of randomized controlled trials of cognitive behaviour therapy and behaviour therapy for chronic pain in adults, excluding headache. Pain 80:1-13, 1999

22. Harris G: The new telecare: Emerging Internet-based models for home healthcare. Telemed Today 7:15-16, 1999

39. National Headache Foundation: Migraine prevention summit proceedings. Available at http://www.headaches. org/pdf/Monograph12.pdf Accessed September 2, 2009

23. Hicks CL, von Baeyer CL, McGrath PJ: Online psychological treatment for pediatric recurrent pain: A randomized evaluation. J Pediatr Psychol 31:724-736, 2006 24. Hollon SD, Stewart MO, Strunk D: Enduring effects for cognitive behavior therapy in the treatment of depression and anxiety. Annu Rev Psychol 57:285-315, 2006 25. Hollon S, Munoz RF, Barlow DH, Beardslee WR, Bell CC, Bernal G, Clarke GN, Franciosi LP, Kazdin AE, Kohn L, Linehan MM, Markowitz JC, Persons JB, Niederehe G, Sommers D: Psychosocial intervention development for the prevention and treatment of depression: Promoting innovation and increasing access. Biol Psychiatry 52:610-630, 2002 26. Hubbard PA, Broome ME, Antia LA: Pain, coping, and disability in adolescents and young adults with cystic fibrosis: A Web-based study. Pediatr Nurs 31:82-86, 2005 27. Jensen MP, Hakimian S, Sherlin LH, Fregni F: New insights into neuromodulatory approaches for the treatment of pain. J Pain 9:193-199, 2008 28. Kendall PC, Southam-Gerow M: Long-term follow-up of a cognitive-behavioral therapy for anxiety-disordered youth. J Consult Clin Psychol 64:724-730, 1996 29. Kendall PC: Treating anxiety disorders in children: Results of a randomized clinical trial. J Consult Clin Psychol 62:100-110, 1994 30. Kiropoulos LA, Klein B, Austin DW, Gilson K, Pier C, Mitchell J, Ciechomski L: Is internet-based CBT for panic disorder and agoraphobia as effective as face-to-face CBT? J Anxiety Disord 22:1273-1284, 2008

40. National Institute of Neurological Disorders and Stroke: Low Back Pain Fact Sheet. Available at http://www.ninds. nih.gov/disorders/backpain/detail_backpain.htm. Accessed August 10, 2009 41. National Institute of Neurological Disorders and Stroke: Second Annual Research Symposium for Advances in Pain Research. Available at http://www.ninds.nih.gov/news_and_ events/proceedings/Advances_Pain_Research_Second_Annual. htm. Accessed September 21, 2009 42. Nutting PA, Rost K, Dickinson M, Werner JJ, Dickinson P, Smith JL, Gallovic B: Barriers to initiating depression treatment in primary care practice. J Gen Intern Med 17: 103-111, 2002 43. Pew Internet & American Life Project: E-patients with disability of chronic disease. Available at: http://www. pewinternet.org/Reports/2007/Epatients-With-a-Disabilityor-Chronic-Disease.aspx Accessed at October 12, 2009 44. Pew Internet & American Life Project: How women and men use the internet, Deborah Fallows. Available at: http:// www.pewinternet.org/Reports/2005/How-Women-and-MenUse-the-Internet.aspx. Accessed at 04/08/2010 45. Romano JM, Turner JA: Chronic pain and depression: Does the evidence support a relationship? Psychological Bulletin 97:18-34, 1985 46. Schulz PJ, Rubinell S, Hartung U: An internet-based approach to enhance self-management of chronic low back pain in the Italian-speaking population of Switzerland: Results from a pilot study. Int J Public Health 52:286-294, 2007

31. Loh PK, Flicker L, Horner B: Attitudes toward information and communication technology (ICT) in residential aged care in Western Australia. J Am Med Dir Assoc 10:408-413, 2009

47. Sirey J, Bruce M, Alexopoulos G, Perlick D, Friedman S, Meyers B: Stigma as a barrier to recovery: Perceived stigma and patient-rated severity of illness as predictors of antidepressant drug adherence. Psychiatr Serv 52:1615-1620, 2001

32. Lorig KR, Laurent DD, Deyo RA, Marnell ME, Minor MA, Ritter PL: Can a Back Pain E-mail Discussion Group improve health status and lower health care costs? A randomized study. Arch Intern Med 162:792-796, 2002

48. Straus SE: FY 2002 Hearing on Chronic Diseases. Available at: http://nccam.nih.gov/about/offices/od/directortestimony/ 032801.htm. Accessed July 1, 2009

Macea et al

The Journal of Pain

929

49. Strom L, Pettersson R, Andersson G: A controlled trial of self-help treatment of recurrent headache conducted via the Internet. J Consult Clin Psychol 68:722-727, 2000

51. Warmerdam L, van Straten A, Cuijpers P: Internet-based treatment for adults with depressive symptoms: The protocol of a randomized controlled trial. BMC Psychiatry 7:72, 2007

50. Wantland DJ, Portillo CJ, Holzemer WL, Slaughter R, McGhee EM: The effectiveness of Web-based vs. non-Webbased interventions: A meta-analysis of behavioral change outcomes. J Med Internet Res10:6-4, 2004

52. Wright JH, Wright AS, Albano AM, Basco MR, Goldsmith LJ, Raffield T, Otto MW: Computer-assisted cognitive therapy for depression: Maintaining efficacy while reducing therapist time. Am J Psychiatry 162:1158-1164, 2005