RAHelp: An Online Intervention for Individuals With Rheumatoid Arthritis


  • ClinicalTrials.gov identifier: NCT00283855.

  • The views expressed in this manuscript are those of the authors and do not necessarily reflect the position or policy of the United States Department of Veterans Affairs.

University of Missouri Department of Health Psychology, DC116.88, One Hospital Drive, Columbia, MO 65212. E-mail: ShigakiC@health.Missouri.edu



To test an intervention for improving self-management in rheumatoid arthritis (RA) using an online, cognitive–behavioral, self-management group program (RAHelp), with weekly telephone support.


A 2-group, randomized study design was used to compare an intervention for RA versus a waiting-list control condition. The intervention used a secure web site (RAHelp.org) to provide a 10-week program with weekly educational modules for improving self-efficacy in self-management of RA, plus tools for group interaction. Weekly telephone contacts were made to encourage use of program tools and apply newly learned skills. A nationwide convenience sample of 106 adult participants (mean age 50 years, 93% women) was recruited primarily through online advertisements. Main outcome measures included the Arthritis Impact Measurement Scales 2 (affective, physical, role, social, and pain/symptom components), Arthritis Self-Efficacy Scale (ASES), Center for Epidemiologic Studies Depression Scale, Quality of Life Scale (QLS), Rapid Assessment of Disease Activity in Rheumatology, Social Provisions Scale, and University of California, Los Angeles Loneliness Scale 3.


Group differences with large and moderate effect sizes (ES) were found immediately postintervention for self-efficacy (ASES; ES 0.92, P = 0.00001) and quality of life (QLS; ES 0.66, P = 0.003), respectively. At 9 months postintervention, differences in self-efficacy (ASES; ES 0.92, P = 0.00001) and quality of life (QLS; ES 0.71, P = 0.004) remained robust.


RAHelp appears to have beneficial effects in terms of self-efficacy and quality of life among individuals with RA who are willing to use an online service format.


Arthritis and other rheumatic conditions have been the leading cause of disability among US adults for the past 15 years ([1]). Although rheumatoid arthritis (RA) affects relatively few people (e.g., while approximately 1.3 million US adults have RA, approximately 26.9 million have osteoarthritis [OA]), the significance of its impact is reflected in its costs, which have been shown to exceed those of OA as well as other common conditions such as hypertension ([2-4]).

The literature suggests that psychosocial factors affect RA outcome. For example, social support has been associated with mobility and fatigue, and both self-efficacy and pain severity have been shown to influence perceived physical functioning ([5-8]). Self-management interventions for chronic conditions typically emphasize self-efficacy and problem solving ([9]). For RA, self-management interventions focus on psychosocial targets, including social support, with the hope of improving functional status ([10-12]). Historically, self-management interventions were provided face to face over the course of several weeks, with alternative modalities (e.g., telephone, internet) successfully explored in recent years ([13]). Both group and individual formats have been described ([11, 12, 14-16]). Although the benefits of self-management intervention for arthritis conditions appear difficult to sustain over time, characteristics of the most successful programs appear to include being cognitive–behavioral in orientation, lasting at least 6 weeks, being highly protocol driven, and requiring trained leaders ([16]).

To date, only one study has provided comprehensive self-management intervention programming for arthritis, delivered primarily via the internet. Results from this study for participants with RA showed sustained improvement in self-reported global health and activity limitation at 12 months post–entry into the program ([10]). This is especially encouraging given recent reports of widespread internet access. Almost 70% of US adults ages ≥25 years are believed to have access to the internet from some location ([17]). Additionally, the majority (61%) of American adults report having looked online for health information ([18]). Data such as these suggest that increasing numbers of individuals will likely turn to and feel comfortable with internet-based tools and services for health-related concerns.

This study reports the results of a randomized trial of an internet-based, multimodal, cognitive–behavioral intervention for RA (RAHelp). The RAHelp web site (RAHelp.org) and program were developed to replicate a face-to-face, professionally delivered, and empirically validated intervention for RA developed by Parker and colleagues ([11]). Our study sought to demonstrate the value of internet-based delivery methods. In this regard, it is similar to the 2008 online arthritis intervention study by Lorig et al, which employed their well-known, peer-led, chronic illness self-management program model ([10]). Our intervention was novel, however, in that we employed a clinician leader to administer and monitor the web site and facilitate success through weekly phone check-ins with participants during the intervention phase. Additionally, our program focused specifically on RA and did not include participants who have only OA. We hypothesized that, compared to a waiting-list control group, participants with RA randomized to the RAHelp intervention would experience significantly improved self-efficacy, quality of life, health status, and pain.

Box 1. Significance & Innovations

  • This study is significant for its robust effect sizes for improved self-efficacy and quality of life, using online delivery of an empirically-based psychosocial intervention for self-management in rheumatoid arthritis (RA).
  • RAHelp offers an online program delivery model that has potential for scalability and integration with health care systems such as “medical homes.” Given increased public comfort using the internet, clinician burden and costs currently associated with RAHelp can likely be reduced by decreasing or eliminating phone contacts. Integration of self-management support can be achieved by leveraging the capabilities of the electronic medical record.
  • Intensive and accessible behavioral self-management support, delivered through a model such as ours, will significantly reduce the information and communication gaps between physicians and patients.


This research was approved by and conducted under the auspices of the Health Sciences Institutional Review Board of the University of Missouri. The findings presented in this study reflect the primary analyses from a randomized, waiting-list controlled study designed to evaluate the effects of participation in RAHelp, an online self-management program for RA.


RAHelp.org was primarily delivered through a secure web site that is fully compliant with the Health Insurance Portability and Accountability Act. During the study, the web site was not open to the public; only affiliated health professionals (called leaders), consented participants with RA (called members), and affiliated technical support staff had access to the site. The web environment supports an educational program, social networking applications, and assessment tools for members, as well as monitoring tools and a database for leaders. The theoretical underpinnings for the intervention were based on Social Learning Theory and a cognitive–behavioral approach to enhance self-efficacy, positive coping, and proactive behavior change ([19, 20]). The objectives of the program were to increase self-efficacy and skills needed for coping with the effects of RA, including garnering social support in an online environment, in hopes of reducing pain and improving functional status.

The online environment allowed for integration of the intervention content material with peer support and a variety of psychoeducational resources. Although the content and distribution of material were the same, the RAHelp online delivery differed from the empirically validated parent program. Online, the intervention content material was presented in visual (slideshow) format, whereas previously, content was delivered didactically by a clinician who also served to engage participants and facilitate learning ([11]). At the time this trial was conceptualized, there was no known precedent for safety, feasibility, or acceptability of interventions delivered completely online, without real-time clinician contacts. RAHelp therefore included weekly one-to-one phone contacts between members and clinician leaders. The leader for this study was a masters-prepared counselor with training in cognitive–behavioral group interventions.

Some features of RAHelp.org were designed for individualized use, whereas others were designed for group use and community building. Individual features included 10 educational modules that were adapted from the clinician-administered, in-person program; these encourage positive coping strategies for enhancing self-efficacy ([11]). Similar to the in-person program, one module was completed each week in a prescribed sequence. Topics covered included Overview and Rationale, RA Stressors, Effective Coping, Life Goals, Pain Management, Emotional Responses, Managing Change, Self-Esteem, Relationships, and Community Participation. In contrast with its predecessor, however, real-time attendance was not required.

Other individual features included a personalized to-do list (a personalized display of tasks and appointments that popped up after each log-in, used to guide participants in learning about the web site applications and to complete lessons and assessments), a news feature (providing current community and health-related updates), and a resource library (information from online sources posted by the clinician leader, including sources forwarded by participants). Self-monitoring as a component of self-management was encouraged throughout the program. A “homework” journal was provided with self-monitoring tools (graphic user interfaces with radio buttons, numbered 0–10) for members to track pain and stress (“How much pain [stress] are you feeling today?”). Text boxes were also provided, where pleasant events and weekly challenges could be described.

In addition to online features, each member was provided with 1:1 leader support through weekly phone contacts, typically lasting between 15 and 30 minutes. The goals of leader–member phone contacts were to address any questions, to accommodate individuals who learn better through auditory versus visual presentation of educational materials, and to facilitate the integration of new concepts with learning from previously assigned material. Of note, leader contacts were not used to introduce new material, nor were they used as individual psychotherapy sessions per se. Leaders most typically used phone sessions to encourage participants to observe patterns (using self-monitoring tools) and apply newly learned skills (from lessons).

Members had access to several “community” features and activities. Each member created a structured profile and selected an avatar, which are then made available to other members in the “RAHelp Village” area. This allowed for increased personal disclosure among members, with low risk for inadvertent identity disclosure. Interactive features include a discussion board, biweekly scheduled chats, and a secured messaging system that allowed communication among individual members. In order to encourage community development and group support, the educational modules directed members to try each feature in the online environment. Additionally, during weekly phone contacts, leaders encouraged community engagement as a means for skill building and developing positive coping. Additional information about the web environment has been published in a report on its development ([21]).

Participant engagement

This study focused on the survey outcomes endorsed by participants at scheduled assessment timeframes. We have previously published findings on engagement from participants in the RAHelp active treatment condition during the 10-week treatment phase ([21, 22]). In an analysis focusing on social interactions, we reported a median of 32 log-ins with a mean of 23 minutes per log-in, results that are approximately comparable to findings from the online arthritis intervention study by Lorig and colleagues ([10]). In another study, we discussed the RAHelp program and environment development and reported that participants made 9,676 clicks of activity to read the 10 learning modules (module length range 20–37 pages) ([21]). Asynchronous community-based tools were deemed well received. The median number of times the discussion board was accessed was 150, with a median total use of 150 minutes. Also, 466 messages were sent to 1,331 recipients, with 5,211 clicks of associated activity to read, send, and save messages ([21]). Qualitative review of discussion board posts suggested that participants valued both the social support facilitated by the RAHelp online environment and the program content ([22]).

Subject recruitment

A nationwide convenience sample of adults with RA was recruited using a predominantly passive online recruitment approach. This included running Google ads, requesting volunteers on craigslist, and posting announcements on RA-relevant discussion boards. Rheumatologists at local clinics were also informed of the study by letter and asked to provide interested patients with contact information.

As volunteers were identified, a telephone screening interview was used to confirm inclusion criteria. These inclusion criteria were as follows: must be age ≥18 years, must have RA diagnosed by a rheumatologist, must have an RA medication regimen stable for 3 months, must have no previous exposure to self-management interventions, must have no uncontrolled psychiatric diagnoses, must have no uncontrolled medical comorbidities (e.g., active cancer), and must be willing to participate in the activities required by the program. Written confirmation of RA diagnosis was required from the participant's board-certified rheumatologist. Consent was obtained from participants using a waiver of documentation process following the telephone screening. Participants meeting all inclusion criteria and documentation requirements were enrolled consecutively and randomized into either the active treatment group or a waiting-list control group.

Outcome measures

Arthritis Impact Measurement Scales 2 (AIMS2).

AIMS2 is a self-report questionnaire designed to measure health status and well-being in the rheumatic diseases. In previous studies, evidence has been provided for the usefulness of the original AIMS in clinical research applications ([23]). The revised AIMS2 consists of 57 core items organized into 12 scales. Respondents are asked to rate each item using 5-point Guttman scales. Test–retest correlation coefficients for the scales have ranged from 0.78–0.94 over a 1-week period. Factor analysis of the AIMS2 has resulted in a 5-factor model with components reflecting functioning in the following domains: physical, affective, symptom, social interactions, and role (i.e., work role). For the purpose of this study, we used the more intuitive name of “pain” rather than “symptom” for this component because it is comprised of 2 items that ask about pain and interference due to pain “in the past 4 weeks.” Scores for the components range from 0–10, with higher scores indicating poorer health status ([23]).

Arthritis Self-Efficacy Scale (ASES).

The ASES is a measure of self-efficacy for arthritis and has been shown to relate to functional status and health outcomes ([24]). The original 20-item version of the measure was used. Each item asks the respondent to rate their certainty about whether they can reduce or regulate arthritis symptoms and effects (e.g., “How certain are you that you can decrease your pain quite a bit?”). Respondents use a Likert-type rating scale, with higher scores indicating greater certainty. In the original version, the scale extends from 10–100, and 3 subscales can be calculated (self-efficacy pain, self-efficacy other symptoms, and self-efficacy function). We report a sum of the combined self-efficacy pain and self-efficacy other symptoms subscales as a measure of overall self-efficacy. For the purposes of reporting, we have adjusted the scoring to a 1–10 scale, reflecting current use of the measure ([25]).

Center for Epidemiologic Studies Depression Scale (CES-D).

The CES-D is a 20-item self-report of the frequency of depressive symptoms within the past week (e.g., “I felt lonely” or “I had crying spells”). The frequency is assessed on 4-point response scales (range 0–3, where 0 = rarely or none of the time and 3 = most or all of the time). Items indicating positive mood states are reverse scored and responses to the 20 individual CES-D items are summed to obtain a total scale score ([26]). Therefore, the possible range of scores is 0–60, with higher scores indicating more symptoms ([27]). The CES-D has high internal consistency, adequate test–retest reliability, and good factorial and discriminant validity ([28]). The literature supports the CES-D as a valid measure of depression for individuals with arthritis ([26]).

Quality of Life Scale (QLS).

The QLS is a 15-item instrument that measures 5 domains of quality of life, including material and physical well-being; relationships with other people; social, community, and civic activities; personal development and fulfillment; and recreation ([29]). Patients rate their present level of satisfaction with each item using a 7-point scale called the “delighted-terrible scale.” Responses include “delighted” ([7]), “pleased” ([6]), “mostly satisfied” ([5]), “mixed” ([4]), “mostly dissatisfied” ([3]), “unhappy” ([2]), and “terrible” ([1]). The measure yields a single total score. A higher score indicates a higher quality of life. The literature supports high internal consistency and test–retest reliability ([30]). This non–health-focused instrument is particularly useful for persons with rheumatic diseases, since they view quality of life from a much broader perspective than just factors directly related to their disease. Satisfying relationships, material well-being, and environmental quality are also important ([30]).

Rapid Assessment of Disease Activity in Rheumatology (RADAR).

RADAR is a brief self-report measure of joint pain and tenderness and clinical status ([31]). The standard measure has 6 items: global disease activity in the past 6 months, current disease activity (joint tenderness and swelling), arthritis pain, duration of morning stiffness, functional class, and tender joint list. For the purpose of this study, we used the rating item for “arthritis pain today”; we thought an estimate of symptoms over 6 months would be inappropriate given the 10-week intervention timeframe and because we were primarily interested in the report of pain. As is common in the clinic, the approach requires individuals to rate their pain using a 0–100 scale, with anchors of 0 = not at all and 100 = very severe pain. We adapted the format for computer use by using the prescribed anchors with 21 clickable buttons in 5-point increments from 0–100.

Social Provisions Scale (SPS).

The SPS is a 24-item tool that measures 6 perceived social functions or “provisions” obtained from relationships with others, including guidance, reassurance of worth, social integration, attachment, opportunity for nurturance, and reliable alliance. Examples of items include “I lack a feeling of intimacy with another person” and “There are people who enjoy the same social activities I do.” The SPS uses a 4-point response scale, where 1 = strongly disagree and 4 = strongly agree. After a reversal of negatively-worded items, a total score may be computed by summing all items. Higher scores indicate a greater degree of perceived support. Evidence has supported the reliability and validity of the SPS ([32]).

University of California, Los Angeles Loneliness Scale, version 3 (LS-3).

The LS-3 is a widely used instrument that assesses respondents' feelings of loneliness or social isolation ([33]). The most recent version available (LS-3) has simplified wording compared to the original version ([34]). The instrument consists of 20 items, including 11 negatively worded items and 9 positively worded items. Some examples include “How often do you feel left out?” and “How often do you feel outgoing and friendly?” A 4-point response scale is used, where 1 = never and 4 = always. After a reversal of positively-worded items, possible scores range from 20–80. Higher scores indicate greater degrees of loneliness. The literature supports high internal consistency and test–retest reliability as well as significant convergent and construct validity ([34, 35]).

Statistical analyses

Prior to testing the main hypotheses, normality tests were performed that indicated non-normal distribution of the outcome data. Therefore, nonparametric methods of analysis were used. The Wilcoxon rank sum test was used to compare differences between the intervention and control groups on baseline values. For postintervention effects, nonparametric analysis of covariance was conducted to compare the treatment and control groups immediately following treatment and 9 months posttreatment. In these analyses, corresponding baseline variables were used as covariates. To account for multiple comparisons (11 variables tested simultaneously) and avoid increased risk for Type I error, Bonferroni correction was applied and a criterion of α = 0.004 was used, rather than 0.05, to determine statistical significance.



One hundred eight subjects completed baseline measures after consenting and being enrolled in the study. Two subjects were removed from the study and study analyses due to development of severe psychiatric conditions, which met the exclusion criteria for the study. This left 106 subjects included in the final analyses (see Figure 1 for recruitment and analysis flow chart). At baseline, there were no significant differences among participants in the treatment (n = 54) versus control (n = 52) groups on demographic variables (i.e., age, sex, ethnicity, level of education, income, employment status, marital status, duration of RA diagnosis, or number of comorbid conditions) or on baseline values of the outcome variables. Information about the demographic characteristics for the intervention and waiting-list control groups is provided in Table 1.

Figure 1.

Recruitment and analysis flow chart.

Table 1. Demographic characteristics at baseline*
 Intervention group (n = 54)aControl group (n = 52)a
  1. Values are the number (percentage) unless otherwise indicated. There were no significant differences between groups at baseline for any of the demographic variables. RA = rheumatoid arthritis.
  2. aTotals may not add up because participants were allowed to opt out of answering items. Also, percentages have been rounded.
Age, mean ± SD years50.3 ± 11.649.3 ± 12.3
Female sex50 (93)48 (92)
White ethnicity50 (93)50 (96)
Educational level  
Graduate school16 (30)14 (27)
At least some college28 (53)33 (64)
High school diploma9 (17)4 (8)
Less than high school00
Married33 (61)34 (65)
Employment status  
Employed33 (62)36 (69)
Not employed20 (38)16 (31)
Family income  
$0–34,99915 (29)9 (18)
$35,000–74,99923 (44)23 (44)
≥$75,00014 (27)20 (38)
Duration of RA, mean ± SD years7.4 ± 8.68.5 ± 10.3


Of the 108 participants who started the study (i.e., completed baseline measures), 12 dropped out during the intervention phase. Of these, 9 had been randomized to the treatment group and 3 had been randomized to the waiting-list control group. Wilcoxon's 2-sample test was used to compare baseline measures among participants who dropped out versus those who completed the study. The results of this analysis showed that participants who dropped out had reported poorer AIMS2 physical functioning (P = 0.04) and AIMS2 role functioning (P = 0.04) at baseline.


We first compared the intervention group to the waiting-list control group at the end of the intervention phase. At this time point, significant differences were found for self-efficacy and quality of life (Table 2). Differences were in the expected direction, favoring the treatment group. The effect size for self-efficacy was large, whereas the effect size for quality of life was in the medium range.

Table 2. Outcomes at postintervention and 9-month followup for the intervention (RAHelp) group versus the waiting-list control group*
MeasuresPostintervention9 months postintervention
RAHelp (n = 44)Control (n = 49)ESPRAHelp (n = 43)Control (n = 45)ESP
  1. Values are the mean ± SD unless otherwise indicated. ES = effect size; AIMS2 = Arthritis Impact Measurement Scales 2; ASES = Arthritis Self-Efficacy Scale; CES-D = Center for Epidemiologic Studies Depression Scale; RADAR = Rapid Assessment of Disease Activity in Rheumatology; QLS = Quality of Life Scale; SPS = Social Provisions Scale; LS-3 = University of California, Los Angeles Loneliness Scale, version 3.
  2. aGroup difference is significant at P ≤ 0.004 (Bonferroni correction applied to account for 11 variables tested simultaneously).
  3. bFor this variable at postintervention, n = 37 for the treatment group and n = 37 for the control group; at 9-month followup, n = 32 for the treatment group and n = 34 for the control group.
Psychological well-being        
Affective (AIMS2)2.9 ± 1.13.1 ± 1.80.480.123.3 ± 1.43.3 ±
Self-efficacy (ASES)83.9 ± 19.068.5 ± 23.80.920.00001a84.1 ± 16.368.6 ± 23.30.920.00001a
Depression (CES-D)9.8 ± 7.611.9 ± 11.20.440.1410.8 ± 8.213.2 ± 11.20.490.14
Arthritis symptoms        
Pain 4 weeks (AIMS2)4.2 ± 2.14.7 ± 2.50.570.074.1 ± 2.64.3 ± 2.50.310.34
Pain today (RADAR)36.8 ± 28.340.2 ± 31.20.370.2441.4 ± 31.239.2 ±
Quality of life        
Role (AIMS2)b2.2 ± 2.02.0 ± ± 1.92.5 ± 2.10.300.26
Quality of life (QLS)88.4 ± 11.784.9 ± 14.60.660.003a88.0 ± 11.883.1 ± 16.00.710.004a
Overall health        
Physical (AIMS2)1.7 ± 1.61.9 ± 1.70.560.0651.6 ± 1.61.6 ± 1.50.480.16
Social support        
Social interactions (AIMS2)3.6 ± 1.64.2 ± 2.00.450.023.6 ± 1.64.0 ± 1.90.380.06
Social support (SPS)84.7 ± 8.883.1 ± ± 10.081.2 ± 11.70.340.10
Loneliness (LS-3)35.6 ± 9.836.4 ± 11.70.610.0235.2 ± 9.236.1 ± 12.60.490.13

We then compared the intervention group to the waiting-list control group on measures collected at the 9-month postintervention followup. At this time point, differences between the intervention and waiting-list control groups remained significant for self-efficacy and quality of life (Table 2). Effect sizes remained large for self-efficacy and moderate for quality of life.


Proper self-management is essential for chronic conditions such as RA. However, rheumatologists and other physicians who care for these individuals simply do not have adequate time to provide in-depth, psychosocial support and self-management education within the normal workflow of a busy practice. Conversely, many patients also find it difficult to participate in office-based behavioral health services. For individuals with RA, travel may be difficult due to pain or functional limitations. Additionally, for low-prevalence disorders such as RA, face-to-face group meetings can be impractical and internet-based services may serve to increase access. Although the internet clearly holds promise in these respects, only a small number of high-quality, randomized controlled trials have been conducted to date. We address this gap by presenting the main outcomes from a randomized controlled trial of an online self-management intervention designed specifically for individuals with RA.

The primary hypotheses for this study predicted that individuals with RA who participated in RAHelp would experience positive effects on self-efficacy, quality of life, health status, and pain. Our data partially support the hypotheses by demonstrating robust outcomes for self-efficacy and quality of life, favoring the RAHelp intervention group, and by demonstrating that these effects remained significant for at least 9 months following the intervention. Overall examination of the outcomes suggests generally consistent findings, with the intervention group endorsing the same or a better status than the control group. Exceptions included only very small, nonsignificant mean differences in the AIMS2 role (postintervention) and the RADAR pain today (9 months postintervention), favoring the control group.

Our hypotheses regarding health status and pain were not supported. Unfortunately, negative findings for pain and other symptoms are not unusual in the literature. A 2003 meta-analysis of arthritis self-management education programs reported a summary effect size of 0.12 for pain measures, suggesting that programs are able to affect only small, if any, reductions in pain ([36]). We expect that it is particularly difficult to demonstrate the impact of psychosocial interventions on pain in RA, since most participants concurrently take one or more pain medications. Anecdotally, a recurring topic on the RAHelp discussion board was participants' medication changes, suggesting that many individuals rely on pharmacologic approaches. As with many, if not most, self-management programs, our intervention focused more on coping and acceptance rather than diminishing pain per se. Given these difficulties, it may be less optimal to use pain reduction as a measure of success for psychosocial interventions than factors that influence individuals' experiences of quality of life and/or feelings of efficacy in managing their symptoms.

Few studies in the extant literature examine self-management interventions for RA apart from other forms of arthritis. In small-group formats, Riemsma et al demonstrated improved self-efficacy and fatigue for a subgroup of individuals with RA, and Hammond and colleagues demonstrated significant improvement in self-efficacy, pain, perceived control, and health behavior ([37, 38]). Lorig et al demonstrated the efficacy of an online, peer-led, self-management intervention for a mixed arthritis group, and in segregated analyses, demonstrated improved global health and activity limitation at 1 year postintervention among participants with RA ([10]). Our results are in agreement with these trials and we concur with Lorig and colleagues that continued work is needed to define the populations for whom internet-based self-management interventions are most effective.

A limitation of our findings is that our results are difficult to parcel into what is attributable to the didactic materials and peer support (delivered online) versus what is attributable to the leader–member phone contacts. At the time this trial was conceptualized, there was no known precedent for safety, feasibility, or acceptability of an intervention of this scale delivered completely online. Since the inception of our trial, however, the public has become increasingly comfortable with online life, as have we. We expect that a future version of RAHelp could be conducted safely and effectively without scheduled phone contacts, and that most, if not all, interactions could be conducted online. This will be important to establish formally, however, since phone contacts in and of themselves have been shown to have a therapeutic effect for those with OA ([39, 40]).

Other study limitations include that our participants were not blinded to the intervention and may have benefited from an attention effect. Given that we have demonstrated improvements at 9 months postintervention, however, we think it is unlikely that our findings are primarily due to attention. Further, caution should be used in generalizing our results to a broader population, since this sample of early adopters of internet tools and services reported a high level of education and income. On the other hand, our study informs clinical practice because our participants are likely representative of persons who accept and are comfortable with, or even perhaps prefer, receiving health care services online.

Although self-management support is recognized as a core component of patient-centered services for chronic illness care, truly integrated self-management interventions remain elusive in practice. Online programming such as RAHelp has the potential to provide comprehensive health care delivery models (e.g., “medical homes”) with a more easily scalable service compared to office-based programs. Such services could be administered independently and serve multiple organizations. Further, electronic medical record (EMR) capability may be leveraged to improve delivery and integration ([41]). At a minimum, EMR-based delivery would require a high-quality patient portal, usable and meaningful feedback from the self-management intervention to the physician, and secure and usable systems for information flow between patients and providers.

It should be noted, however, that although cost savings may be realized in scaled applications, online interventions are not without substantial costs. Human resources for technical monitoring and maintenance are needed (in addition to web site hosting fees, all of which can vary), as well as some type of clinician or trained individual to facilitate the intervention and/or monitor for participant safety. While difficult to parcel clinical from research efforts in this study, one full-time equivalent masters-level clinician easily managed clinical aspects for more than 50 active participants and more than 50 control subjects, in addition to data collection for research purposes. At this point, we expect our program could be modified to increase scalability by significantly reducing the amount of clinician phone contact time without introducing untoward risk or consumer rejection. Costs would likely then be comparable to the model by Lorig et al, in which 2 nominally paid peer facilitators are typically used, with associated costs incurred from intensive facilitator training, the need to identify facilitator replacements as new groups are formed, and skilled oversight.

At this time, there is no clear and consistent way to recoup costs associated with online self-management programming. However, evolving changes in the Medicare reimbursement structure are predicated on the theory that cost offset can be realized if attention is paid to important behavioral and psychosocial factors associated with chronic illness self-management ([41]). Testing the veracity of these assumptions in real-world settings will be an important step in chronic illness care research.

In summary, a professionally-led self-management intervention for RA appears to be well accepted and effective in increasing self-efficacy for managing RA and quality of life, with effects persisting at least 9 months following the intervention. The internet may be an especially useful way of providing self-management intervention, especially for individuals with low-frequency chronic conditions such as RA. The ultimate challenge for truly effective internet-based health care services, however, will likely lie in finding ways to contain costs while integrating these services with users' own health care systems and providers.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Shigaki had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Smarr, Siva, Musser, Johnson.

Acquisition of data. Smarr, Musser.

Analysis and interpretation of data. Shigaki, Smarr, Siva, Ge.


The authors would like to thank those who helped to make this article possible: Kathryn Burks, Kathy Donovan Hanson, James Laffey, and Jerry Parker.