In this study, a previously evaluated guided Internet-based cognitive behavior therapy for social anxiety disorder (SAD) was adapted for mobile phone administration (mCBT). The treatment was compared with a guided self-help treatment based on interpersonal psychotherapy (mIPT). The treatment platform could be accessed through smartphones, tablet computers, and standard computers. A total of 52 participants were diagnosed with SAD and randomized to either mCBT (n = 27) or mIPT (n = 25). Measures were collected at pre-treatment, during the treatment, post-treatment and 3-month follow-up. On the primary outcome measure, the Liebowitz Social Anxiety Scale – self-rated, both groups showed statistically significant improvements. However, mCBT performed significantly better than mIPT (between group Cohen's d = 0.64 in favor of mCBT). A larger proportion of the mCBT group was classified as responders at post-treatment (55.6% versus 8.0% in the mIPT group). We conclude that CBT for SAD can be delivered using modern information technology. IPT delivered as a guided self-help treatment may be less effective in this format.
3. Results
3.1. Attrition, user information, and technique
In total, 30 participants (57.7%) completed all treatment modules within the 9-week time frame. Corresponding figures for each group separately were 17 (63.0%) for the mCBT group and 13 (52.0%) for the mIPT group. For attrition see Fig. 1. When accessing the treatment platform, participants used smartphone 42.81%, computer 50.05%, and tablet computer 7.14% of the time. 21 participants of the mCBT program logged into both the anxiety exposure and anxiety hierarchy modules. Out of these 15 worked with both the exposure modules and created anxiety hierarchies. In average these participants entered data from 8.20 exposure exercises (SD = 4.95, min = 1, max = 21) into the platform and put 8.73 items on their anxiety hierarchies (SD = 5.61, min = 4, max = 25). Overall, the instruction to therapists of 15 min of feedback to participants delivered once a week was followed. There were technical problems related to using the new smartphone platform and interface; problems to enter estimated anxiety levels during exposure, complicated log-in codes and the necessity to log out of the smartphone to enter weekly measurements on the primary outcome measure.
3.2. Treatment effects
3.2.1. Primary outcome measure (LSAS-SR)
Results for the primary outcome measure are presented in Table 3. The two groups did not differ significantly on the LSAS-SR at pre-treatment, M = 60.19 (SD = 18.95) for mCBT and M = 65.72 (SD = 27.15) for the mIPT group: t(50) = −0.86, n.s. On the LSAS-SR pre/post, both treatment groups showed significant improvements at the p < .05 level; mCBT: t(24) = 6.18, p = .001; mIPT: t(21) = 2.75, p = .01. Within-group Cohen's d effect sizes were large for mCBT (d = 0.99), and small for mIPT (d = 0.43). The between group effect size was moderate (d = 0.64). Results for the weekly measurements on the LSAS-SR are presented in Fig. 2 for illustration.
Table 3.
Means, SD's, within and between group effect sizes (Cohen's d), confidence intervals, F-values and significance levels for primary and secondary outcome variables.
Measure Group Pre Post-est. values Within group ES Between group ES F-values (1, 49)
n = 52 n = 52 d CI lo CI hi d CI lo CI hi
M (SD) M (SD)
LSAS-SR mCBT 60.19 (18.95) 38.93 (23.18) 0.99 0.58 1.39 0.64 0.06 1.22 5.18*
mIPT 65.72 (27.15) 54.41 (25.19) 0.43 0.09 0.77
SIAS mCBT 45.15 (15.81) 34.30 (14.76) 0.71 0.35 1.07 0.72 0.14 1.30 5.83*
mIPT 50.12 (14.06) 44.54 (13.76) 0.40 0.11 0.70
SPS mCBT 32.59 (11.31) 23.42 (12.15) 0.78 0.30 1.27 0.63 0.05 1.22 2.97
mIPT 37.64 (15.07) 31.30 (12.81) 0.45 0.12 0.77
BAI mCBT 17.30 (9.25) 14.79 (10.97) 0.25 −0.31 0.81 0.46 −0.10 1.03 2.09
mIPT 19.80 (12.12) 20.02 (11.56) −0.02 −0.41 0.37
MADRS-S mCBT 11.70 (5.09) 10.44 (4.77) 0.26 −0.16 0.67 0.88 0.28 1.47 7.25**
mIPT 14.08 (6.30) 14.62 (4.77) −0.10 −0.61 0.42
QOLI mCBT 1.09 (1.58) 1.41 (1.66) −0.20 −0.55 0.16 0.37 −0.25 0.99 0.12
mIPT 0.37 (1.59) 0.82 (1.51) −0.29 −0.72 0.13
Abbreviations: mCBT, mobile-based cognitive behavior therapy; mIPT, mobile-based interpersonal therapy; Pre, pre-treatment; Post, post-treatment; LSAS-SR, Liebowitz Social Anxiety Scale – Self Report; SIAS, Social Interaction Scale; SPS, Social Phobia Scale; MADRS-S, Montgomery-Åsberg Depression Rating Scale-Self Report; BAI, Beck Anxiety Inventory; QOLI, Quality of Life Inventory.
*
p < .05 significance level for effect of group (ANCOVA).
**
p < .01 significance level for effect of group (ANCOVA).
Table options
Full-size image (22 K)
Fig. 2.
Weekly change on LSAS-SR. Scores for both groups at each assessment point; pre-treatment, treatment and post-treatment. Vertical bars denote 95% CI, confidence interval.
Figure options
3.2.2. Secondary outcome measures
Results for the secondary self-report inventories are presented in Table 3. Within group effect sizes on SIAS and SPS were moderate in the mCBT group and small in the mIPT group. Between groups effect sizes for these inventories were moderate, for the SIAS and for the SPS. The inventories for the measurement of general anxiety (BAI) and quality of life (QOLI) revealed small between group effect sizes. The between groups effect size for the measurement of depression, the MADRS-S, was large (in favor of the mCBT group, but within group effects were small for both groups (Table 3)).
3.2.3. Three-months follow-up
In addition to the pre/post analysis a post/follow-up analysis was conducted using paired samples t-tests on the basis of the results on the primary outcome measure at the three-month follow-up (N = 39; with n = 24 in the mCBT group and 15 in the mIPT group). We did not impute the missing data at this stage and focused only on completers of the follow-up. The results were stable for both groups: mCBT, Post M = 38.21 (SD = 24.50), Follow-up M = 39.75 (SD = 24.87), t(21) = −0.74, p = .47; mIPT, Post M = 57.21 (SD = 28.24), Follow-up M = 59.00 (SD = 28.84), t(14) = 1.27, p = .23.
3.2.4. Clinical significance
The number of participants meeting the criteria for clinically significant improvement was calculated for the LSAS-SR. In the mCBT group there were 55.6% (n = 15) who were classified as responders at post-treatment compared to 8.0% (n = 2) in the mIPT group, and this difference was statistically significant χ2(1) = 9.07, p = .04.
