Moderation of treatment effects by parent-adolescent conflict in a randomised controlled trial of Attachment-Based Family Therapy for adolescent depression
, , and | Sep 03, 2020
About this article
Published Online: Sep 03, 2020
Page range: 110 - 122
DOI: https://doi.org/10.21307/sjcapp-2020-011
Keywords
© 2020 Erling W. Rognli et al., published by sciendo.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
FIGURE 1.
FIGURE 2.
Prior distributions and reasoning for choices of prior
| Parameter | Prior Distribution | Reasoning |
|---|---|---|
| Regression coefficients | Normal (0, 1) | Weakly informative prior as the dependent variable is centred on the median score and scaled by twice the median absolute deviation |
| Error variance | Half-student’s t (3, 0, 1) | Weakly informative prior on the error variance, putting most of the prior weight on errors between 0 and 1, but with heavy tails allowing for a much higher error variance |
| Random intercepts | Hierarchical normal prior, with location 0 and a Half-student’s t (3, 0, 1) prior on the scale. | Defines random intercepts as deviations from the intercept of the whole sample, and estimates the variance of the random intercepts from the data, with a weakly informative hyperprior |
| Latent variables for CBQ IRT model | Multivariate normal (0, 1) with an LKJ ( | Defines the latent variables for parent-adolescent conflict as four correlated Normal (0, 1) variables with a weakly informative prior on the correlation coefficients |
| Item thresholds for CBQ IRT model | Hierarchical normal prior with hyperpriors | Weakly informative hierarchical prior for the item thresholds, as these are interdependent with the defined latent variable |
| Item discrimination for CBQ IRT model | Gamma (2, 0.5) | Places most of the prior weight on discrimination between 1 and 5, which is the most probable range for items of an established instrument, but does not rule out higher or lower values |
Correlation coefficients for latent conflict variables (Posterior means and 90% CI)
| Adolescent on father | Adolescent on mother | Father on adolescent | |
|---|---|---|---|
| Adolescent on mother | 0.20 (-0.06 ; 0.46) | ||
| Father on adolescent | 0.59 (0.39 ; 0.79) | 0.33 (0.04 ; 0.62) | |
| Mother on adolescent | 0.03 (-0.21 ; 0.27) | 0.49 (0.28 ; 0.73) | 0.50 (0.26 ; 0.75) |
Sample characteristics
| Variable | Treatment arm | ||
|---|---|---|---|
| ABFT (n=30) | TAU (n=30) | ||
| Age, years (SE) | 15.03 (1.35) | 14.77 (1.36) | |
| Gender, % (n) | Female | 90 (27) | 83.3 (25) |
| Dropout, % (n) | Excluded | 7 (2) | 3.3 (1) |
| Dropped out | 10 (3) | 13.3 (4) | |
| Ethnicity, % (n) | Norwegian | 100 (30) | 96.7 (30) |
| Scandinavian | 0 (0) | 3.3 (1) | |
| Living with, % (n) | Both parents | 29.6 (8) | 36.7 (11) |
| Two home family | 18.5 (5) | 13.3 (4) | |
| Father (and any new partner) | 18.5 (5) | 13.3 (4) | |
| Mother (and any new partner) | 33.3 (9) | 33.3 (10) | |
| Other | 0 (0) | 3.3 (1) | |
| Psychiatric comorbidity, % (n) | Dysthymia | 3.3 (1) | 0 (0) |
| Any anxiety disorder | 43.3 (13) | 46.7 (14) | |
| Obsessive-Compulsive Disorder | 6.7 (2) | 6.7 (2) | |
| Externalizing disorder | 0 (0) | 13.4 (4) | |
| PTSD | 3.3 (1) | 3.3 (1) | |
| Enuresis | 3.3 (1) | 6.7 (2) | |
| No comorbidity | 53.3 (16) | 46.7 (14) | |
| Depressive symptoms, mean (SD) | BDI-II | 34.23 (7.34) | 36.21 (9.84) |
| GRID-HAMD | 21.87 (4.61) | 21.92 (4.07) | |
Parameter estimates from moderator models
| Conflict report and model parameter | ||||||
|---|---|---|---|---|---|---|
| Mean | SD | Median | 66% HDI | 90% HDI | ESS | |
| Intercept | 0.09 | 0.09 | 0.09 | 0.00 ; 0.17 | -0.05 ; 0.24 | 11801 |
| Variance of random intercepts | 0.36 | 0.13 | 0.37 | 0.28 ; 0.50 | 0.14 ; 0.56 | 1764 |
| Variance of errors | 0.53 | 0.08 | 0.53 | 0.45 ; 0.59 | 0.41 ; 0.65 | 2382 |
| Time | -0.53 | 0.15 | -0.53 | -0.65 ; -0.37 | -0.78 ; -0.29 | 10094 |
| Treatment | 0.01 | 0.18 | 0.01 | -0.16 ; 0.18 | -0.29 ; 0.30 | 10608 |
| Parent-adolescent conflict | 0.27 | 0.29 | 0.27 | -0.01 ; 0.55 | -0.21 ; 0.76 | 7372 |
| Treatment x time | 0.07 | 0.11 | 0.07 | -0.04 ; 0.17 | -0.12 ; 0.25 | 9108 |
| Conflict x time | 0.15 | 0.19 | 0.15 | -0.03 ; 0.31 | -0.16 ; 0.47 | 7425 |
| Treatment x conflict x time | -0.69 | 0.34 | -0.70 | -1.01 ; -0.38 | -1.25 ; -0.15 | 5919 |
| Intercept | 0.09 | 0.09 | 0.09 | 0.00 ; 0.18 | -0.07 ; 0.24 | 12959 |
| Variance of random intercepts | 0.36 | 0.13 | 0.38 | 0.27 ; 0.51 | 0.14 ; 0.58 | 2059 |
| Variance of errors | 0.58 | 0.08 | 0.57 | 0.49 ; 0.64 | 0.45 ; 0.69 | 2853 |
| Time | -0.49 | 0.16 | -0.48 | -0.64 ; -0.34 | -0.75 ; -0.23 | 11610 |
| Treatment | 0.00 | 0.19 | 0.00 | -0.19 ; 0.16 | -0.30 ; 0.31 | 11630 |
| Parent-adolescent conflict | 0.20 | 0.31 | 0.20 | -0.07 ; 0.51 | -0.29 ; 0.71 | 11097 |
| Treatment x time | 0.04 | 0.1 | 0.04 | -0.06 ; 0.13 | -0.13 ; 0.21 | 10099 |
| Conflict x time | 0.13 | 0.22 | 0.13 | -0.07 ; 0.34 | -0.24 ; 0.47 | 5198 |
| Treatment x conflict x time | 0.01 | 0.40 | 0.00 | -0.38 ; 0.36 | -0.60 ; 0.70 | 4474 |
| Intercept | 0.09 | 0.09 | 0.09 | 0.00 ; 0.17 | -0.05 ; 0.25 | 12613 |
| Variance of random intercepts | 0.37 | 0.12 | 0.39 | 0.29 ; 0.50 | 0.18 ; 0.58 | 2008 |
| Variance of errors | 0.54 | 0.08 | 0.54 | 0.46 ; 0.60 | 0.41 ; 0.66 | 2488 |
| Time | -0.55 | 0.15 | -0.54 | -0.69 ; -0.4 | -0.80 ; -0.30 | 10341 |
| Treatment | -0.03 | 0.18 | -0.03 | -0.21 ; 0.13 | -0.33 ; 0.26 | 11666 |
| Parent-adolescent conflict | 0.19 | 0.30 | 0.20 | -0.07 ; 0.51 | -0.29 ; 0.70 | 8534 |
| Treatment x time | -0.08 | 0.11 | -0.08 | -0.18 ; 0.02 | -0.26 ; 0.09 | 9297 |
| Conflict x time | -0.02 | 0.19 | -0.02 | -0.19 ; 0.16 | -0.33 ; 0.28 | 6471 |
| Treatment x conflict x time | -0.67 | 0.34 | -0.68 | -1.01 ; -0.38 | -1.22 ; -0.11 | 6593 |
| Intercept | 0.09 | 0.09 | 0.09 | 0 ; 0.17 | -0.05 ; 0.24 | 9540 |
| Variance of random intercepts | 0.37 | 0.12 | 0.38 | 0.29 ; 0.5 | 0.17 ; 0.57 | 1651 |
| Variance of errors | 0.51 | 0.08 | 0.50 | 0.43 ; 0.58 | 0.38 ; 0.64 | 1740 |
| Time | -0.54 | 0.16 | -0.54 | -0.68 ; -0.38 | -0.79 ; -0.27 | 5165 |
| Treatment | -0.01 | 0.18 | -0.01 | -0.18 ; 0.16 | -0.32 ; 0.26 | 10840 |
| Parent-adolescent conflict | 0.30 | 0.30 | 0.30 | 0.01 ; 0.59 | -0.19 ; 0.81 | 5729 |
| Treatment x time | -0.03 | 0.11 | -0.03 | -0.13 ; 0.08 | -0.21 ; 0.16 | 6466 |
| Conflict x time | 0.41 | 0.24 | 0.41 | 0.18 ; 0.61 | 0.03 ; 0.81 | 2780 |
| Treatment x conflict x time | -0.86 | 0.44 | -0.87 | -1.28 ; -0.46 | -1.60 ; -0.17 | 2701 |
Model comparison with leave-one-out cross-validation
| Reporter and models compared | Difference | SE |
|---|---|---|
| Time | 0 | 0 |
| Time | 0 | 0 |
| Moderator | -1.86 | 1.36 |
| Treatment | -1.87 | 0.14 |
| Non-specific predictor | -3.91 | 0.71 |
| Time | 0 | 0 |
| Treatment | -1.87 | 0.14 |
| Non-specific predictor | -5.57 | 0.40 |
| Moderator | -3.74 | 0.31 |
| Time | 0 | 0 |
| Treatment | -1.87 | 0.14 |
| Moderator | -3.02 | 1.32 |
| Non-specific predictor | -4.47 | 0.30 |
| Time | 0 | 0 |
| Treatment | -1.87 | 0.14 |
| Moderator | -2.34 | 1.86 |
| Non-specific predictor | -4.54 | 0.77 |