This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Depressive symptoms are common in people with type 2 diabetes mellitus (T2DM). Effective depression treatments exist; however, access to psychological support is characteristically low. Web-based cognitive behavioral therapy (CBT) is accessible, nonstigmatizing, and may help address substantial personal and public health impact of comorbid T2DM and depression.
The aim of this study was to evaluate the Web-based CBT program, myCompass, for improving social and occupational functioning in adults with T2DM and mild-to-moderate depressive symptoms. myCompass is a fully automated, self-guided public health treatment program for common mental health problems. The impact of treatment on depressive symptoms, diabetes-related distress, anxiety symptoms, and self-care behavior was also examined.
Participants with T2DM and mild-to-moderate depressive symptoms (N=780) were recruited online via Google and Facebook advertisements targeting adults with T2DM and via community and general practice settings. Screening, consent, and self-report scales were all self-administered online. Participants were randomized using double-blind computerized block randomization to either myCompass (n=391) for 8 weeks plus a 4-week tailing-off period or an active placebo intervention (n=379). At baseline and postintervention (3 months), participants completed the Work and Social Adjustment Scale, the primary outcome measure. Secondary outcome measures included the Patient Health Questionnaire-9 item, Diabetes Distress Scale, Generalized Anxiety Disorder Questionnaire-7 item, and items from the Self-Management Profile for Type 2 Diabetes.
myCompass users logged in an average of 6 times and completed an average of .29 modules. Healthy Lifestyles users logged in an average of 4 times and completed an average of 1.37 modules. At baseline, mean scores on several outcome measures, including the primary outcome of work and social functioning, were near to the normal range, despite an extensive recruitment process. Approximately 61.6% (473/780) of participants completed the postintervention assessment. Intention-to-treat analyses revealed improvement in functioning, depression, anxiety, diabetes distress, and healthy eating over time in both groups. Except for blood glucose monitoring and medication adherence, there were no specific between-group effects. Follow-up analyses suggested the outcomes did not depend on age, morbidity, or treatment engagement.
Improvement in social and occupational functioning and the secondary outcomes was generally no greater for myCompass users than for users of the control program at 3 months postintervention. These findings should be interpreted in light of near-normal mean baseline scores on several variables, the self-selected study sample, and sample attrition. Further attention to factors influencing uptake and engagement with mental health treatments by people with T2DM, and the impact of illness comorbidity on patient conceptualization and experience of mental health symptoms, is essential to reduce the burden of T2DM.
Australian New Zealand Clinical Trials Registry ACTRN12615000931572; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=368109&isReview=true (Archived by WebCite at http://www.webcitation.org/7850eg8pi)
Type 2 diabetes mellitus (T2DM) is a global public health problem, affecting more than 370 million people worldwide [
Depression is frequently comorbid with T2DM, affecting up to 40% of people with the disease [
International diabetes treatment guidelines now recommend regular screening for depressive symptoms, with referral to appropriate psychological treatments as part of standard diabetes care [
Internet delivery of evidence-based psychological therapies is now established as apopular, clinically effective, and cost-efficient means of upscaling access to psychological treatments. In people with T2DM, both diabetes-specific [
myCompass is a broadly available Web-based CBT program available free of charge in Australia. A previous trial reported symptom reduction of mild-to-moderate depression in the general population compared with a placebo condition [
The primary aim of this randomized controlled trial (RCT) was to evaluate the effectiveness of myCompass in improving work and social functioning—a major contributor to the T2DM burdens [
This study is the primary outcomes evaluation of a 2-arm RCT called
The study utilized a broadly-based recruitment strategy, including offline and online recruitment methods, to enroll the required sample size for sufficient statistical power. Recruitment began in September 2015 and continued until November 2017. Offline recruitment occurred via letters from participating general practices in New South Wales (NSW) and Victoria to their patients with T2DM, distribution of promotional materials in general practice settings throughout NSW and Victoria (eg, study flyers and posters), and print advertisements in national diabetes-related publications.
Online recruitment involved a range of techniques targeting health professionals and individuals. Health professionals were targeted through member associations such as the Australian Association of Practice Managers, the Australian Diabetes Educators Association, the Australian Association of Practice Nurses and the Australian Primary Health Care Nurses Association. Contact was via electronic direct mail, informing members of the study and inviting them to refer appropriate candidates to the SpringboarD website for screening.
Individuals were recruited via Google and Facebook advertisements targeting an Australia-wide audience aged 18 years and over, with interest in
People were eligible to take part if they were aged 18 to 75 years, diagnosed with T2DM by a physician, screened positive for depression on the self-report 2-item Patient Health Questionnaire (PHQ-2; ie, ≥2) [
Eligible participants received immediate feedback of their suitability for the study via the website and were provided onscreen instructions for completion of the baseline assessment. Approximately 25 min were required to complete the screening and baseline assessment. All data captured by the study website during the screening, baseline, and postintervention assessments were stored in password-protected files on secure servers that comply with UNSW HREC privacy regulations regarding online data collection.
Computerized block randomization with blocks of 8 was undertaken to assign participants to the intervention and control condition at a 1:1 allocation ratio. Randomization was initiated immediately after the completion of the baseline assessment by the automated randomization system built into Black Dog Institute’s study management software. Allocation was concealed from participants and researchers. Though participants were ineligible if they had previously used myCompass, some participants allocated to the intervention condition may have independently accessed public information about myCompass over the course of the trial and therefore become aware that they were allocated to an existing Web-based therapy program.
myCompass (mycompass.org.au) is a public health fully automated, self-guided CBT intervention that users complete in their own time and at their own pace on their computer and/or mobile phone (see
Screenshot of the myCompass landing page.
Screenshot of the myCompass self-monitoring page.
The program offers a personalized treatment plan based on an assessment of user symptoms, either at registration or at any time during program use [
Participants randomized to the myCompass arm were provided access to the full program for 8 weeks. The program recommended that users complete 3 CBT modules and self-monitor up to 3 symptoms or behaviors. A 4-week tailing-off period followed, in which only the symptom monitoring function was accessible. Studies have shown that the use of Web-based interventions is improved when users receive program feedback that is personalized in its content [
myCompass user privacy is managed by a password-protected log-on and by ensuring that user-generated data (eg, self-monitoring ratings) are not stored on the user’s device but are instead transferred via the internet using secure sockets layer protocols (which encrypt transmitted data by rendering data unreadable to anyone other than the intended recipient) and by storing the data in secure servers. Participant user data were identified in myCompass using email details provided during study registration. Once extracted, data were deidentified and stored in a password-protected file.
The placebo control program,
A summary of the measures obtained from participants at baseline and 3 months postintervention is presented in
A comparison of the core features of the myCompass and Healthy Lifestyles programs.
Feature | MyCompass | Healthy Lifestyles |
Modality | Website | Website |
Symptom tracking | Yes: track up to 3 symptoms | Not available |
Usage reminders | Yes: automated short message service text message or email | Yes: automated email |
Number of modules | 12: including interactive learning activities | 12: including interactive learning activities |
Module content | Cognitive and behavioral strategies for mood, anxiety, stress, and mental well-being | Generic health literacy information |
Measures obtained at each assessment phase.
Measure | Time 1 | Time 2 | |||
Demographic data (eg, age and gender) | X | —a | |||
Disease-relevant data (eg, age at diagnosis and treatment) | X | X | |||
Mental health history (eg, previous diagnoses and help-seeking) | X | — | |||
Work and Social Adjustment Scale [ |
X | X | |||
Patient Health Questionnaire-9 [ |
X | X | |||
Diabetes Distress Scale [ |
X | X | |||
Generalized Anxiety Disorder Scale [ |
X | X | |||
Self-management Profile for Type 2 Diabetes Scale (behavior items only) [ |
X | X | |||
Glycosylated hemoglobin | X | — | |||
Days out of role [ |
X | X | |||
Health service utilization | X | X |
aNot applicable.
The primary outcome for the trial was a between-group difference in work and social functioning, measured by the Work and Social Adjustment Scale (WSAS). The WSAS is a psychometrically sound measure of daily functioning across 5 domains, including work, social leisure activities, private leisure activities, home management, and personal relationships [
Symptoms of depression and anxiety were measured by PHQ-9 [
Emotional adjustment to diabetes—or
Diabetes self-care was assessed using a subset of items from the Self-Management Profile for Type 2 Diabetes (SMP-T2D). The SMP-T2D was designed for use in clinical trials to assess level and perceived ease of performance of specific T2DM regimen behaviors. Perceived coping and confidence dealing with diabetes and ease of weight management are also assessed, and all constructs demonstrate appropriate internal consistency, validity, and sensitivity [
At baseline, we collected disease-related (eg, age of onset and treatment regimen), sociodemographic (eg, age, gender, education, and occupation) and mental health history data (eg, service use and previous diagnoses). We also obtained participants’ most recent glycosylated hemoglobin (HbA1c) from their medical records as an indicator of their overall blood glucose management before study commencement. Recent service utilization for physical and mental health problems was assessed at baseline and postintervention, along with days out of role, defined as the number of days in the previous 30 days that a participant was unable to perform work or normal activities because of problems with his/her physical or mental health [
At the conclusion of the intervention period, program engagement data were extracted for myCompass and Healthy Lifestyles, including frequency of log-in, number of modules started and completed, and self-monitoring frequency (myCompass only).
Power calculations indicated that a study sample of 600 was needed to detect a minimum difference of .3 standard deviations between groups in mean change in scores on the WSAS at 3 months post intervention, with power of 80%, 2-tailed alpha=.05, and assuming an attrition rate of 40%. Owing to early indications of a higher attrition rate at postintervention than anticipated, a further 180 people were recruited into the study to ensure sufficient power to test the research hypotheses.
Primary analyses employed an intention-to-treat (ITT) approach using mixed-model repeated measures analyses (MMRM) computed within the Mixed procedure of SPSS version 23 (IBM Corp.). MMRM makes use of all available data to obtain parameter estimates and is widely recognized as an appropriate strategy for analyzing incomplete datasets [
In addition to the ITT analyses, we conducted completer analyses to examine the effects of treatment on those individuals who completed the study, defined as having provided complete data at 3-month follow-up. In these analyses, individuals with any missing data at 3-month follow-up were deleted case wise (myCompass n=175; Healthy Lifestyles n=151), and repeated-measures analyses of variance were conducted for each of the primary and secondary outcomes on the remaining participants. As the sample characteristics and treatment effects in the completer analyses did not differ from those of the ITT analyses, only the ITT results are reported.
Overall sample characteristics are presented in
As shown in
Sample characteristics for myCompass and Healthy Lifestyles groups.
Characteristic | myCompass (N=368) | Healthy Lifestyles (N=355) | ||||
Age, mean (SD) | 57.7 (10.6) | 57.7 (10.0) | ||||
Female (n=465), n (%) | 229 (62) | 236 (66) | ||||
Married (n=387), n (%) | 204 (55) | 183 (52) | ||||
Employed (n=351), n (%) | 173 (47) | 178 (50) | ||||
Secondary school or lower (n=220) | 112 (30) | 108 (30) | ||||
Trade certificate or diploma (n=270) | 133 (36) | 137 (39) | ||||
University undergraduate or more (n=233) | 123 (33) | 110 (31) | ||||
Sought professional support for mental health (n=571) | 296 (80) | 275 (77) | ||||
Received mental health diagnosis (n=300) | 155 (42) | 145 (41) | ||||
Diagnosed with depressive symptoms or major depressive disorder (n=279) | 143 (38) | 136 (38) | ||||
Sought professional support for mental health (n=113) | 65 (18) | 48 (14) | ||||
Taking antidepressant medication (n=241) | 125 (34) | 116 (33) | ||||
Months using antidepressant medication | 97.70 (94.72) | 73.67 (56.21)a | ||||
Age at diagnosis, mean (SD) | 46.6 (11.1) | 47.2 (10.9) | ||||
Healthy eating (n=437) | 230 (63) | 207 (58)a | ||||
Physical activity (n=323) | 176 (48) | 147 (41) | ||||
Oral medication (n=583) | 295 (80) | 288 (81) | ||||
Insulin (n=216) | 113 (31) | 103 (29) | ||||
Exenatide (n=32) | 21 (6) | 11 (3) | ||||
Visited general practitioner for diabetes (n=419) | 218 (59) | 201 (57) | ||||
Frequency of general practitioner visit | 1.31 (.78) | 1.37 (.71) | ||||
Hospitalized for diabetes (n=24) | 13 (4) | 11 (3) | ||||
Frequency of hospitalization for diabetes | 1.46 (1.5) | 1.36 (.9) |
aMeans differ significantly at
Consolidated Standards for Reporting of Trials diagram of participant flow through the SpringboarD trial. PHQ: Patient Health Questionnaire.
Diabetes-related characteristics were also largely similar between the groups. The average age of onset of T2DM for the sample was 47 years (SD 10.84), and for the 60.0% (434/723) of participants for whom HbA1c data were available from medical records, the mean value was 7.5% (SD 1.6).
Baseline scores for key outcome measures for both groups are presented in
The observed scores for all study outcomes postintervention are presented in
For the SMP-T2D, no group or time differences were observed for the Physical Activity Domain, and both groups reported significant improvement in the Healthy Eating Domain. Scores on the Blood Glucose Monitoring Domain and Medication Adherence Domain increased over time for the Healthy Lifestyles group relative to the myCompass group, with Healthy Lifestyles participants increasing time spent each week monitoring blood glucose and taking medication as recommended by approximately .4 to .5 days, compared with their myCompass counterparts.
Level of glycemic control [
Baseline and postintervention means (SDs) on key outcome variables for myCompass and Healthy Lifestyles groups.
Measure | Baseline, mean (SD) | Post (3 months), mean (SD) | ||
myCompass (n=397) | Healthy Lifestyles (n=387) | myCompass (n=232) | Healthy Lifestyles (n=241) | |
WSASa | 13.64b (7.93) | 12.23b (7.77) | 12.24 (9.31) | 10.82 (9.19) |
PHQ-9c | 11.30 (4.03) | 10.86 (4.11) | 8.68 (5.63) | 8.24 (5.54) |
GAD-7d | 7.60 (4.10) | 7.33 (4.20) | 6.66 (4.73) | 6.20 (4.58) |
DDSe | 2.56 (0.97) | 2.56 (0.95) | 2.24 (0.99) | 2.18 (0.95) |
SMP-BGf | 49.61 (39.50) | 45.55 (38.85) | 49.25 (39.33) | 55.83 (40.21) |
SMP-MAg | 87.50 (22.47) | 87.20 (22.82) | 86.34 (24.15) | 92.04 (15.12) |
SMP-HEh | 49.96b (29.16) | 48.26b (28.61) | 53.52 (28.94) | 54.87 (28.25) |
SMP-PAi | 53.44 (33.48) | 48.26 (33.61) | 55.98 (34.64) | 52.44 (34.07) |
bMeans differ significantly at
aWSAS: Work and Social Adjustment Scale.
cPHQ-9: Patient Health Questionnaire.
dGAD-7: Generalized Anxiety Disorder Scale.
eDDS: Diabetes Distress Scale.
fSMP-BG: Self-Management Profile for Type 2 Diabetes—Blood Glucose Monitoring.
gSMP-MA: Self-Management Profile for Type 2 Diabetes—Medication Adherence.
hSMP-HE: Self-Management Profile for Type 2 Diabetes—Healthy Eating.
iSMP-PA: Self-Management Profile for Type 2 Diabetes—Physical Activity.
Mixed-model repeated measures analyses fixed effects for time, group, and time×group on key outcome variables.
Variable and effect | Betaa | SE | 95% CI | ||||||||
Time | 1.344 | 0.48 | 2.800 (544.48) | .005c | 0.401 to 2.287 | ||||||
Group×Time | 0.196 | 0.684 | 0.287 (550.76) | .77 | –1.147 to 1.540 | ||||||
Time | 2.395 | 0.31 | 7.736 (596.32) | <.001c | 1.787 to 3.003 | ||||||
Group×Time | 0.144 | 0.44 | 0.326 (604.61) | .74 | –0.720 to 1.008 | ||||||
Time | 0.302 | 0.047 | 6.361 (518.94) | <.001c | 0.208 to 0.395 | ||||||
Group×Time | –0.062 | 0.068 | –0.919 (523.64) | .36 | –0.194 to 0.070 | ||||||
Time | 1.009 | 0.264 | 3.825 (553.83) | <.001c | 0.490 to 1.526 | ||||||
Group×Time | –0.276 | 0.375 | –0.736 (560.75) | .46 | –1.013 to 0.460 | ||||||
Time | –8.253 | 2.278 | –3.623 (538.46) | <.001c | –12.727 to –3.778 | ||||||
Group×Time | 8.98 | 3.248 | 2.764 (545.46) | .006c | 2.598 to 15.360 | ||||||
Time | –3.126 | 1.24 | –2.520 (490.84) | .01c | –5.562 to –0.688 | ||||||
Group×Time | 5.117 | 1.763 | 2.901 (496.65) | .004c | 1.651 to 8.582 | ||||||
Time | –5.704 | 1.604 | –3.553 (527.04) | <.001c | –8.856 to –2.550 | ||||||
Group×Time | 3.266 | 2.282 | 1.430 (532.61) | .15 | –1.217 to 7.750 | ||||||
Time | –3.440 | 2.228 | –1.543 (559) | .12 | –7.816 to 0.936 | ||||||
Group×Time | 0.083 | 3.165 | 0.541 (566.75) | .98 | –6.301 to 6.134 |
aBeta: unstandardized regression coefficient for the effect holding constant age and sex.
bWSAS: Work and Social Adjustment Scale.
cSignificant at
dPHQ-9: Patient Health Questionnaire-9 item.
eDDS: Diabetes Distress Scale.
fGAD-7: Generalized Anxiety Disorder Scale-7 item.
gSMP-BG: Self-Management Profile for Type 2 Diabetes—Blood Glucose Monitoring.
hSMP-MA: Self-Management Profile for Type 2 Diabetes—Medication Adherence.
iSMP-HE: Self-Management Profile for Type 2 Diabetes—Healthy Eating.
jSMP-PA: Self-Management Profile for Type 2 Diabetes—Physical Activity.
Of the total trial participants, 59.2% (235/397) of the participants from the myCompass program and 63.3% (245/387) of the participants from the Healthy Lifestyles program provided at least one postintervention measure. A multivariate analysis of variance comparing nonresponders (ie, participants who did not provide any postintervention measures) with responders at baseline revealed differences on several key outcome variables. Nonresponders reported more severe depressive symptoms (
Participants in the treatment group logged in to myCompass an average of 6 times (SD 9.01; range 1-71), started a mean of 0.71 modules (SD 1.18; range 0 to 8), fully completed an average of .29 modules (SD .87; range 0-7), and monitored their symptoms an average of 2 times (SD 5.79; range 0-53). Participants in the control group logged in to Healthy Lifestyles an average of 4 times (SD 3.22; range 1 to 17), started a mean of 2.61 (SD 2.78; range 0 to 8), and completed an average of 1.37 modules (SD 2.24; range 0 to 8). Participants who logged in to their assigned Web-based program did not differ significantly from those who did not, except for a slightly higher anxiety score on the GAD-7 among myCompass users who logged in (
This trial examined the effectiveness of a self-help, unguided Web-based CBT program (myCompass) for improving work and social functioning in people with T2DM and mild-to-moderate depressive symptoms compared with an active placebo intervention. Our ITT analyses showed that participants’ work and social functioning improved significantly postintervention, irrespective of the intervention received. Significant improvements were also observed for both groups in depressive and anxiety symptoms, diabetes distress, and most aspects of diabetes self-management. Our Healthy Lifestyles group showed small but significant improvements in blood glucose monitoring and medication adherence over and above those observed for the myCompass group. As morbidity may influence the outcomes of depression treatments [
The absence of differential treatment effects for work and social functioning and depressive symptoms for people with T2DM following treatment with myCompass was surprising in light of previous findings of accelerated symptom gains in myCompass users in the general community [
Our recruitment methods targeted people with T2DM and mild-to-moderate depressive symptoms to investigate a public health rather than a clinical application of Web-based CBT; however, our final sample unexpectedly comprised a group with minimal symptoms of depression (47% of participants in the
Even in the absence of true depression, people with chronic illness typically report poorer daily functioning than those without [
It is important to understand why and how increasing severity of depressive symptoms compromises study involvement in trials of Web-based CBT to inform the take-up of suitable interventions for this patient cohort and to maximize treatment effectiveness. One possibility is that attrition is related to increasing levels of amotivation, concentration difficulties, and behavioral inactivation that are hallmark symptoms of depressive disorders [
Importantly, the myCompass program is a completely self-guided mental health intervention that is generic in content. It is, therefore, lower in treatment intensity than previously studied therapist-guided programs [
Of the set of secondary outcomes, the only consistent group effect was for medication adherence, with a small but significant positive effect for the attention control. This finding was isolated and seemingly counterintuitive. However, it is possible that improvement in medication taking was prompted by the focus paid by the attention control program to healthy lifestyle behaviors.
Recruitment and retention in RCTs targeting comorbid physical and mental illness can be difficult [
Although previous trials of Web-based CBT in people with diabetes have compared the active treatment with either treatment as usual [
In conclusion, this study sought to determine if a broadly available public health Web-based CBT intervention could help individuals with T2DM and mild-to-moderate depressive symptoms improve their daily lives. Functioning and symptom outcomes improved between baseline and postintervention; however, no treatment advantage was observed for the myCompass group. Further research is necessary to identify the factors that impact participant retention in studies of Web-based interventions in T2DM, especially among those in greatest need of psychological support and who stand to benefit most from treatment. At the same time, there is a need to better understand how individuals with diabetes conceptualize mood symptoms in the context of diabetes to ensure that Web-based interventions are both personally and clinically relevant. The personal and societal health burdens posed by comorbid T2DM and depressive symptoms are considerable and will continue to grow. Continued investigation of the potential for Web-based CBT to provide a solution in T2DM is essential.
cognitive behavioral therapy
Diabetes Distress Scale
Generalized Anxiety Disorder scale
glycosylated hemoglobin
Human Research Ethics Committee
intention-to-treat
mixed-model repeated measures analyses
New South Wales
odds ratio
Patient Health Questionnaire
randomized controlled trial
Self-Management Profile for Type 2 Diabetes
short message service
type 2 diabetes mellitus
University of New South Wales Sydney
Volunteer Research Register
Work and Social Adjustment Scale
The authors gratefully acknowledge the dedication and input of all study participants. The authors also acknowledge the support of the Australian Government Department of Health in funding the development and dissemination of the myCompass program and the support of the National Health and Medical Research Council in funding this research trial (NHMRC Grant 1083116). JP, JC, SS, and PB are grateful to Mark Donoghoe of UNSW Stats Central for assistance with MMRM analysis procedures.
None declared.
CONSORT‐EHEALTH checklist (V 1.6.1).