Correction

This article refers to:

Implementation of transdiagnostic treatment for emotional disorders in residential eating disorder programs: A preliminary pre-post evaluation

Article title: Implementation of transdiagnostic treatment for emotional disorders in residential eating disorder programs: A preliminary pre-post evaluation

Authors: Thompson-Brenner, H., Boswell, J. F., Espel- Huynh, H., Brooks, G. E., & Lowe, M. R.

Journal: Psychotherapy Research

DOI: https://doi.org/10.1080/10503307.2018.1446563

When the above article was published online, it contained errors. The authors have re-run the affected analyses with the correct data and tables II and IV have been amended accordingly. The revised tables and corrected text are listed below.

Result in Abstract:

‘and more favorable outcomes on ED symptom severity, depression, and experiential avoidance at 6-month follow up (ps ≤ .0001)’

to

‘Relative to patients who were treated during the pre-implementation phase, patients in the post-implementation phase experienced more favorable outcomes on ED symptom severity, depression, mindfulness, and anxiety sensitivity at 6-month-follow-up (ps ≤ .001). A similar result was observed for experiential avoidance, yet this interaction effect was no longer statistically significant (p = .10) when the time x length of stay effect/covariate was added to the model.’

Data Analysis:

‘Time was scaled in weeks and square-root transformed’

to

‘Time was scaled in weeks and both linear and quadratic time effects were tested’

Effect of Implementation on Patient Outcomes at 6-month Follow Up (6MFU)

Full text has been updated in this section as below.

‘Full 6-month follow-up model results can be found in Table IV. Significant fixed effects of linear time on experiential avoidance, mindfulness, and anxiety sensitivity were detected, indicating improvement on these variables across both implementation groups between admission and 6 month-follow-up. The linear time effect was not statistically significant for ED and depression symptoms. With the exception of anxiety sensitivity, the quadratic time effect was statistically significant. Length of stay had a significant effect on all outcomes; longer stays were associated with greater reductions in symptoms, experiential avoidance, and anxiety sensitivity, as well as improvements in mindfulness. A significant fixed effect of site was observed for anxiety sensitivity alone, suggesting that rates of improvement differed significantly between the two sites for this measure alone. With the exception of experiential avoidance, the fixed effect of time × implementation phase was statistically significant. In each case, patients in the post-implementation group experienced better outcomes. This pattern of results was replicated for the fixed effect of quadratic time × implementation phase. The effect was statistically significant for all variables, with the exception of experiential avoidance when the time × LOS covariate was included in the models. For mindfulness, patients in the post-implementation phase continued to experience a steady improvement between discharge and 6MFU, while patients in the pre-implementation phase experienced a reversal of improvements evidenced at discharge. For ED and depressive symptoms, patients in the pre-implementation group experienced a more pronounced deceleration in their gains made by discharge. The effect for anxiety sensitivity seemed to be largely accounted for by the steeper rate of change between admission and discharge assessments.’ Among those admitted underweight (BMI < 18.5), BMI change did not differ significantly pre- vs. post-implementation.

Reliable and clinically significant change on ED symptom severity from admission to 6MFU were as follows for the pre-implementation group: 9.0% deteriorated reliably, 30.3% had no change from baseline admission scores, 27.6% improved reliably but did not reach clinical significance, and 33.1% achieved clinically significant improvement. For post-implementation, 8.3% deteriorated, 34.1% had no change, 17.4% had reliable but not clinically significant improvement, and 40.2% experienced clinically significant improvement. This difference in frequencies of outcome was not statistically significant (χ2(3) = 4.41, p = .22).

Assessment of Risk of Bias from Attrition

‘Results from pattern mixture modelling indicated that joint addition of the dropout terms (including interactions with time and implementation phase) significantly improved model fit for all outcomes (ps ≤ .04).’

to

‘Results from pattern mixture modelling indicated that joint addition of the dropout terms (including interactions with time and implementation) significantly improved model fit for the mindfulness, depression, and ED symptom outcomes (ps ≤ .001), yet did not improve fit for either experiential avoidance or anxiety sensitivity.’

‘This pattern held for EDE-Q scores, CES-D scores, experiential avoidance, and mindfulness (p values comparing drop outs vs. completers ranged from <.0001 to .04); p values for interactions between this effect and implementation group all >.05). For anxiety sensitivity, neither the fixed intercept effect of dropout nor its interactions with time or implementation phase were significant (ps > .14).’

to

‘Across the three measures where dropout terms improved model fit, the interactions with implementation group were not statistically significant (ps > .14).’

Discussion

‘At follow-up, significant differences in improvement in experiential avoidance were still observed between groups.’ to ‘However, this effect was no longer statistically significant after controlling for time × length of stay in the model.’

‘At follow-up, a significantly greater proportion of patients treated in the post-implementation period (64.5%) experienced clinically significant improvement on ED symptom severity.’ to ‘At follow-up, the proportion of patients experiencing clinically significant improvement did not differ to a statistically significant degree between implementation groups.’

Corrected Table II  

Implementation phase	Admission (n = 616)	Discharge (n = 440)		6MFU (n = 281)
	M (SD)	M (SD)	d	M (SD)	d
Eating Disorder Examination-Questionnaire
Pre-Implementation	4.14 (1.31)	2.50 (1.31)	1.25	3.23 (1.58)	0.63
	n = 331	n = 230		n = 145
Post-Implementation	4.14 (1.35)	2.54 (1.23)	1.24	2.93 (1.57)	0.83
	n = 285	n = 210		n = 132
Center for Epidemiological Studies—Depression Scale
Pre-Implementation	37.12 (12.36)	25.84 (12.36)	0.91	27.77 (13.10)	0.73
	n = 302	n = 256		n = 146
Post-Implementation	37.72 (12.08)	24.34 (11.67)	1.13	24.78 (12.90)	1.04
	n = 262	n = 206		n = 135
Multidimensional Experiential Avoidance Questionnaire
Pre-Implementation	229.21 (38.75)	212.15 (42.09)	0.43	215.77 (41.66)	0.33
	n = 214	n = 239		n = 118
Post-Implementation	231.81 (40.36)	198.56 (46.38)	0.76	205.90 (47.90)	0.58
	n = 249	n = 204		n = 123
Southampton Mindfulness Questionnaire
Pre-Implementation	30.97 (15.79)	37.02 (15.89)	0.38	35.37 (18.24)	0.26
	n = 250	n = 236		n = 137
Post-Implementation	30.19 (15.89)	42.58 (17.51)	0.74	40.53 (18.24)	0.6
	n = 259	n = 203		n = 133
Anxiety Sensitivity Index
Pre-Implementation	32.27 (12.51)	30.58 (13.11)	0.13	29.16 (12.21)	0.25
	n = 251	n = 239		n = 131
Post-Implementation	32.06 (12.82)	27.54 (13.33)	0.35	27.71 (11.97)	0.35
	n = 262	n = 204		n = 129

Corrected Table IV Fixed parameter estimates for outcomes from admission to 6-month follow-up.

Parameter	Experiential avoidance			Mindfulness			Anxiety sensitivity			Depression			Global eating pathology
	b	SEb	p	b	SEb	p	b	SEb	p	b	SEb	p	b	SEb	p
Intercept	206.85	9.11	<.0001	38.60	3.26	<.0001	29.81	2.63	<.0001	25.19	2.25	<.0001	2.62	0.26	<.0001
Age	−0.20	0.17	0.23	-0.05	0.06	.44	−0.07	0.05	.13	−0.03	0.04	.41	−0.05	0.10	.31
Comorbidity^a	11.74	3.42	<.001	−5.21	1.27	<.0001	2.81	1.01	.006	5.30	0.86	<.0001	0.30	0.10	.003
Admission BMI	0.37	0.24	0.13	−0.05	0.09	.57	0.12	0.007	.09	0.10	0.06	.10	0.02	0.007	.03
LOS (days)	0.77	0.17	<.0001	−0.13	0.05	.01	0.09	0.04	.05	0.21	0.04	<.0001	0.03	0.005	<.0001
Site^b	−3.19	3.65	.38	1.97	1.34	.14	−3.55	1.08	.001	−1.12	0.91	.22	−0.21	0.11	.05
Implementation phase (post)	−5.18	4.17	.21	−0.55	1.50	0.71	−0.70	1.16	.55	3.29	1.09	.003	0.46	0.12	<.001
Time^c	−3.42	0.63	<.0001	0.35	0.13	.007	−0.12	0.05	.02	0.007	0.09	.94	0.005	0.01	.59
Time^2	0.14	0.02	<.0001	−0.007	0.003	.05	0.007	0.009	.42	0.003	0.001	.01	0.0003	0.0001	.03
Time * LOS	−0.04	0.007	<.0001	–	–	–	–	–	–	−0.01	0.002	<.0001	−0.001	0.0003	<.0001
Time * implementation phase (post)	−1.68	0.95	.08	1.92	0.33	<.0001	−0.71	0.21	<.001	−2.24	0.27	<.0001	−0.29	0.03	<.0001
Time^2 * implementation phase (post)	0.05	0.03	.10	−0.06	0.01	<.0001	0.02	0.006	<.001	0.07	0.008	<.0001	0.009	0.001	<.0001

^aRefers to total number of comorbid non-ED diagnoses; variable was square-root transformed and centered prior to entry into model.

^bRepresents effect for patients treated at the southeastern US facility, relative to northeast US facility.

^cScaled in weeks. BMI = body mass index (kg/m²); LOS = length of stay; ED = eating disorder. Fixed slope effects of covariates (i.e., interactions with time) were only included in final models if they significantly improved model fit. Convergence issues occurred in the mindfulness and anxiety sensitivity models that included the time × LOS covariate, so these specific results are not reported for these models. Statistically significant effects are printed in bold-face for ease of reading.

The online version has been corrected.