CBT for depression: a pilot RCT comparing mobile phone vs. computer

A CONSORT-2010 compliant, registered RCT compared two modes of delivery, a mobile versus computer group. Both groups were followed through 3-months follow-up.

Applicants applied to http://​www.​virtualclinic.​org.​au after reading details about the study. Details of the applicant and participant flow are in Figure 1. During recruitment between March and May 2012, 176 individuals applied for this program and 52 applicants met the inclusion criteria: (i) aged over 18, (ii) self identified as suffering from mild or moderate depression, and have PHQ-9 scores and results of telephone diagnostic interview (MINI) consistent with this, (iii) prepared to provide name, phone number and address, and (iv) to provide written informed consent, (v) had access to a mobile phone or iPad, and a computer with a printer and, (iv) had previous experience with downloading a mobile app. Applicants were excluded if they: (i) had psychosis, bipolar disorder, substance abuse or dependence, (ii) severe depression (with a PHQ-9 score greater than 24) and (iii) or current suicidality, as assessed by Question 9 on the PHQ-9: “how frequently over the last two weeks have you been bothered by thoughts that you would be better off dead or of hurting yourself in some way?” If the applicant answered 2 (= more than half the days), with a history of previous suicide attempts or answered 3 (= nearly every day), with or without previous suicide attempts, they were excluded, and contacted by the study clinicians (CT, AG) and advised as to an appropriate course of action.

115 applicants met selection criteria and of these 101 were able to be contacted by phone. The content of the call included a review of the study and confirmation that the applicant had read and understood the information and consent form. Following this, the Mini International Neuropsychiatric Interview Version 5.0.0 (MINI) [21] was used to confirm the applicant met DSM-IV criteria for a Major Depressive Disorder (MDD). 52 applicants met all inclusion criteria and were randomised via a true randomisation process (http://​www.​random.​org) generated by a team member not involved in the study, to either the Mobile Group (n=22) or Computer Group (n=30). However, at baseline we had 15 participants in the Mobile Group and 20 in the Computer Group, due to participants not starting the first lesson. Concealment of allocation was maintained until the applicant met all inclusion criteria and an offer of participation made. The unsuccessful applicants (n=49) were advised about more appropriate treatment options. The study was approved by the Human Research Ethics Committee (HREC) of St Vincent’s Hospital (Sydney, Australia) and the trial was registered as ACTRN 12611001257954.

The Get Happy Program was based on the principles of CBT and is a version of the previously evaluated Sadness Program [11, 12]. The program consisted of 6 lessons conducted over an 8 week period. The lessons read like a comic book and participants follow the story of Jess, a comic character that has depression, and through her story learn how she comes to manage her symptoms, and participants can then apply these principles to their own life. The size of the font was adapted for the mobile version and some minor revisions were made to the content of the program. On completion of each lesson, participants were assigned and encouraged to carry out the relevant homework activities and review the lesson. Additional resources, such as information on assertiveness skills and sleep hygiene, and stories from previous participant’s experiences were also available Figure 2, Figure 3.

Participants received emails/or phone calls from a clinician, until completion of Lesson 2 because our previous research study has found that additional clinician support does not add any further benefit to participants [13]. Following this, the only clinical contact was when the participant initiated contact, or when a clinician instigated contact due to a deterioration in the K-10 score [23]. The K-10 is a measure of psychological distress and was completed prior to beginning each lesson.

The Patient Health Questionnaire-9 (PHQ-9) [24] is a brief 9-item self-report scale that measures each of the DSM-IV criteria for MDD with scores ranging from 0 to 27. Participants rate the frequency of symptoms over the last fortnight on a scale ranging from 0 (not at all) to 3 (nearly every day), where 1 = several days, 2 = more than half of the days. A PHQ-9 score of ≥10 is used as a clinical cut-off for probable DSM-IV diagnosis of MDD [25]. The PHQ-9 has been shown to have good sensitivity and specificity [26] and excellent reliability and validity [24].

The Kessler 10-item Psychological Distress scale (K-10) [23] is a measure of non-specific psychological distress over the past 14 days. Scores can range from 10 to 50, with higher scores indicating higher distress. The K-10 has good psychometric properties and can discriminate between cases and non-cases of DSM-IV affective disorders [27].

Beck’s Depression Inventory Second Edition (BDI-II)[28] consists of 21 items and is a measure of the presence and severity of a MDD based on the DSM-IV diagnostic criteria for depression. Scores range from 0 to 63 with higher scores indicating a greater severity of depression. The BDI-II possesses excellent internal consistency [28].

Credibility/Expectancy Questionnaire (CEQ)[29] is a scale for measuring treatment credibility and expectancy of improvement. At post treatment, 2 items based on the CEQ were asked including, how satisfied were you with the skills that this program has taught you to manage your depression from 1 (not at all satisfied) to 9 (very satisfied); how confident would you be in recommending this treatment to a friend who experiences similar problems from 1 (not at all confident) to 9 (very confident).

Sheehan Disability Scale (SDS)[30] is a measurement of functional disability and impairment due to psychiatric symptoms and is self rated. The 2 items related to productivity were used for this study. The first item used to measure absenteeism asked participants how many days in the last week their symptoms caused them to miss studies or work, or leave them unable to do their normal daily responsibilities. The second item used to measure presenteeism asked participants on how many days in the last week did they feel so impaired by their symptoms, that even though they studied, went to work or worked at home, their productivity was reduced.

For the PHQ-9, K-10 and BDI-II analyses were conducted using linear mixed-model repeated measures (MMRM) ANOVA with measurement occasion as a within-group factor and intervention as a between-groups factor. These analyses were conducted using the MIXED procedure in SPSS Version 19 with an identity covariance matrix. For the SDS generalised estimating equations (GEE) were used to evaluate reductions in absenteeism and presenteeism across time. GEE is a semi-parametric method used to implement marginal models when the outcome variable is not normally distributed. The GENLIN procedure with a repeated statement was implemented using SPSS Version 19. A poisson distribution with a log link function was specified. An unstructured covariance structure was used to model the within-subject dependencies. Initial models only included measurement occasion, study group and their interaction as fixed effects. For each outcome measure, the two homework questions were included separately in subsequent models to investigate whether any of the outcomes differed based on self-reported homework effort or homework completion.

All participants completed the questionnaire outcome measures (BDI-II, PHQ-9, K-10, CEQ, ERS and SDS) at 1-week post treatment and at 3-months follow up.

The mean age of participants was 41 years (SD= 12.38, range =18 - 63) and 28/35 were female (80% of the sample). Participants reported on average, moderate depression levels using the PHQ-9, severe levels of depression on the BDI –II, and severe levels of psychological distress using the K-10. Participants reported on average 2.14 days of work lost (SD= 2.19) and 4.40 days of work that were underproductive (SD= 2.15) due to the presence of his/her psychiatric symptoms on the SDS (see Table 1).

Independent samples t-tests compared the two groups on baseline demographic characteristics and pre-treatment symptom questionnaires. There were no differences between the groups on age, the BDI-II, PHQ-9, nor K-10.

8.6% (3/35) completed only the first lesson, 2.9% (1/35) completed two lessons, 2.9% (1/35) completed 3 lessons, 5.7% (2/35) completed 4 lessons, 11.4% (4/35) completed 5 lessons and 68.6% (24/35) of participants completed all six lessons. Refer to Figure 1. When comparing the Mobile Group with the Computer Group on adherence, there were no significant differences (t (33) = −.242, P >.05).

Figure 4 displays mean PHQ-9 scores on each measurement occasion as a function of the two conditions.

Statistical tests showed that the benefits of the intervention remained significant in both groups at follow up when using the PHQ-9 (F [3, 73.6] = 28.4, P=<.001), the BDI-II (F [2, 47.09] = 60.1, <.001) and the K-10 (F [7, 1734.5] = 28.4, P=<.001).

Estimated marginal means for PHQ-9 scores estimated under occasion x intervention model).

Results from the MMRM (ANOVA) indicated that the interaction between experimental group and time was not statistically significant when comparing PHQ-9 scores (F [3, 73.6] = .875, P=.458), BDI-II scores (F [2, 47.0] = 1.7, P=.19), or K-10 scores (F [7, 174.91] = .370, P=.919).

In order to investigate differences between each occasion of measurement, a series of pair wise comparisons were conducted. For the PHQ-9 and K-10 the change from baseline to mid-point, post-test, and 3-month follow up was investigated. Significant reductions were found between baseline and all other time points for both the Mobile Group and Computer Group. This pattern was repeated using the BDI-II, when investigating changes from baseline to post-test and 3-months follow up.

The aggregated mean score for homework effort for the Mobile Group was 11.25 (range, 6–17) and 14.46 (range 9–20) for the Computer Group. An aggregated mean score for homework completion for the Mobile Group was 11 (range, 8–16) and 12.4 (range 6–18) for the Computer Group. Results from the MMRM (ANOVA) indicated that there were no significant differences in homework completion or homework effort between the groups on the PHQ-9, the BDI-II, or the K-10.

Large (>.8) within-group effect sizes were found on the BDI-II, PHQ-9 and K-10 measures from pre-treatment to post-treatment. Effect sizes for these measures are included in Table 1.

Statistical tests using the SDS showed a significant reduction in the number of days lost (absenteeism) (Wald Chi-Square =10.31, P = <.05) and in the number of days underproductive (presenteeism) from pre-treatment to post-treatment (Wald Chi-Square =12.33, P = .001). Further results indicated that the interaction between experimental group and time was not statistically significant when comparing the SDS on absenteeism (Wald Chi-Square =.86, P = .35) and presenteeism (Wald Chi-Square =.22, P = .63).

Descriptive data on the environment, level of distraction and ability to concentrate, was collected for participants in the Mobile Group. Over the six lessons, 66.7-92.9% of participants completed the lessons in their home, 46.7-60% of participants completed these lessons when there was slight distraction, and 30.8-53.3% endorsed an ‘okay’ ability to concentrate. Please refer to Table 2.

At post-treatment, only 16/35 (45%) met criteria for depression on the PHQ-9 (see Table 3 for results).

The mean clinician time spent per participant was 4.1 minutes (SD= 4.63) and the mean technician time spent per participant was 6.4 minutes (SD= 5.38). Each time a clinician or technician had contact with a participant the amount of time was recorded to monitor the time required to deliver the program.

Upon completion of the program 54% of Mobile Group and 64% of the Computer Group were very satisfied with the program; with the remaining participants endorsing ‘somewhat satisfied’. 64% of the Mobile Group and 64% of the Computer Group would be very confident in recommending this treatment to a friend; and the remainder endorsed ‘somewhat confident’.