Predictors of dropout in a controlled clinical trial of psychotherapy for moderate depression

Abstract

A significant number of psychotherapy clients remain untreated, and dropping out is one of the main reasons. Still, the literature around this subject is incoherent. The present study explores potential pre-treatment predictors of dropout in a sample of clients who took part in a clinical trial designed to test the efficacy of narrative therapy for major depressive disorder compared to cognitive-behavioral therapy. Logistic regression analysis showed that: (1) treatment assignment did not predict dropout, (2) clients taking psychiatric medication at intake were 80% less likely to drop out from therapy, compared to clients who were not taking medication, and (3) clients presenting anxious comorbidity at intake were 82% less likely to dropout compared to those clients not presenting anxious comorbidity. Results suggest that clinicians should pay attention to depressed clients who are not taking psychiatric medication or have no comorbid anxiety. More research is needed in order to understand this relationship.

A high number of psychotherapy clients remain untreated and dropout is one of the main reasons, since rates are high (approx. 20%, up to 74%; Swift & Greenberg, 2012). Few predictors of dropout have emerged recurrently in research, such as low socio-economic status (Reis & Brown, 1999), being married or living with a partner, previous experience with psychotherapy (Werner-Wilson & Winter, 2010), low education (Swift & Greenberg, 2012), older age (Pomp, Fleig, Schwarzer, & Lippke, 2013), being African-American (Lester, Artz, Resick, & Young-Xu, 2010), being female (Shamir, Szor, & Melamed, 2010), suffering from low levels of anxiety and/or depression (Baekeland & Lundwall, 1975), less clinical experience of therapists (Roos and Werbart, 2013, Swift and Greenberg, 2012) and weaker alliance (Sharf, Primavera, & Diener, 2010). Still, most researchers who have investigated psychotherapy dropout agree that there is no clear evidence for a pattern of dropout predictors (Baekeland and Lundwall, 1975, Casares-López et al., 2011, Swift and Greenberg, 2012). The aim of this study is to explore the attrition data of a psychotherapy clinical trial by assessing potential pre-treatment predictors for dropping out of treatment.

Method

Participants, therapists, procedures and treatment conditions are described in more details in the original report of the study (Lopes et al., 2014).

Participants

Sixty-three clients diagnosed with moderate Major Depressive Disorder (MDD) according to the DSM-IV (American Psychiatric Association, 2000) were assigned to either narrative therapy (NT, n = 34) or cognitive-behavioral therapy (CBT, n = 29). No significant treatment group differences were found [i.e., 81% female, mean age of 35.44 years old (SD = 11.51), 79.4% had comorbid anxiety and 60.3% entered treatment taking psychiatric medication]. Ten therapists, all psychologists, with an average of 1.9 years of clinical experience (SD = 2.13) met clients individually.

Dropout rate was 36.50% [NT = 41%, CBT = 31%; χ²₍₁₎ = 0.32, p=.568]. Dropout was defined as the unilateral termination by the client without the therapist's approval or knowledge (Jung, Serralta, Nunes, & Eizirik, 2013).

Treatment conditions

The CBT (Beck, Rush, Shaw, & Emery, 1979) and NT (White, 2007) treatment manuals consisted of 20 sessions and made similar requirements on the clients (e.g., completion of forms, in- and out-of-session assignments).

Measures

• -Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbons, & Williams, 2002) was used to gather clients’ demographic (i.e., age, gender, relationship status, professional status and education) and clinical information (i.e., global assessment of functioning [GAF], presence of anxious co-morbidity, medication at intake, previous hospitalizations, previous suicide attempts and previous psychotherapy).
• -Graffar Index (Graffar, 1956) was used to measure socio-economic status (SES).
• -Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1996) was used to measure baseline severity of depressive symptoms.
• -Outcome Questionnaire (OQ45.2; Lambert et al., 1996) and its subscales were used to measure general psychological distress, interpersonal problems and social role.
• -Working Alliance Inventory (WAI-C; Horvath & Greenberg, 1989) was used at session four to assess the quality of working alliance from the perspective of 40 completers and 15 dropout clients.

Results

Attrition along treatment

Of the 23 clients who eventually dropped out, 48% did so by the end of the fourth session, and 91% left treatment before the 11th session. Mean length of stay in treatment for the dropout group was 6.4 (SD = 4.4).

Prediction of dropout

As shown on Table 1, t tests and chi-square tests were used to compare dropouts and completers on general characteristics (age, gender, years of education, socioeconomic status, marital status, and employment status), clinical characteristics at intake (GAF, anxious comorbidity, being medicated, previous hospitalization, previous suicide attempt, previous psychotherapy and pre-treatment scores on the BDI-II, OQ-45.2 and its subscales), process related variables (treatment assignment and WAI-C score) and a therapist variable (clinical experience). Those at least marginally significant entered a logistic regression model predicting dropout. Only consumption of medication [χ²₍₁₎ = 6.79, p=.009] and anxious comorbidity at intake [χ²₍₁₎ = 3.15, p=.076] met that criterion. When entered in the logistic regression model, these variables showed independent effects on the odds of dropping out of treatment. Medicated clients were 80% less likely to drop out, compared to non-medicated clients [OR = 0.2 (95% CI = 0.063-0.639), p=.007]. Clients presenting anxious comorbidity were 82% less likely to drop out compared to those not presenting anxious comorbidity [OR = 0.182 (95% CI = 0.033-1.016), p=.0520].

Table 1.

Comparison of completers and dropouts according to continuous and discrete pre-treatment variables.

	Completers	Dropouts	t(61)	p
	M (SD)	M (SD)
Age of Client	36.2 (11.6)	34.1 (11.6)	0.68	.497
Education of client (in years)	13.1 (5.3)	13.7 (3.6)	−0.48	.632
Social Economic Status (Graffar1)	12.7 (3.4)	11.7 (2.6)	1.11	.273
GAF at the time of the intake	59.1 (11.5)	61.1 (8.4)	−0.71	.478
Total Score BDI-II Session 01	31.5 (10.2)	31.0 (11.7)	0.17	.867
Total Score OQ-45 Session 01	95.9 (20.1)	91.9 (17.2)	0.80	.430
Score OQ-45.2 Distress Subscale Session 01	57.8 (12.8)	54.6 (11.5)	−0.99	.325
Score OQ-45.2 Social Role Subscale Session 01	16.7 (4.9)	16.9 (3.5)	0.15	.885
Score OQ-45.2 Interpersonal problems Session 01	21.4 (5.7)	20.4 (5.7)	−0.62	.535
Score WAI-C at session 04	50.9 (5.3)	51.0 (5.0)	−0.07	.944

	n (%)	n (%)	χ2(1,3)	p
Treatment assignment
Narrative Therapy	14 (41.2%)	20 (50%)	0.70	0.404
Cognitive Behavioural Therapy	9 (31%)	20 (50%)
Level of therapist experience
Novice	12 (30.0%)	4 (17.4%)	1.31	.520
Apprentice	1 (2.5%)	1 (4.3%)
Graduate	27 (67.5%)	18 (78.3%)
Gender of the client
Female	33 (82.5%)	18 (78.3%)	0.17	.680
Male	7 (17.5%)	5 (21.7%)
Relationship status
Married	15 (37.5%)	6 (26.1%)	3.41	.333
Single	17 (42.5%)	10 (43.5%)
Divorced	6 (15.0%)	7 (30.4%)
Widowed	2 (5.0%)	0 (0.0%)
Professional status
Employed	18 (45.0%)	14 (60.9%)	3.94	.268
Unemployed	12 (30.0%)	2 (8.7%)
Student	9 (22.5%)	6 (26.1%)
Retired	1 (2.5%)	1 (4.3%)
Previous psychotherapy
No	35 (87.5%)	20 (87.0%)	0.00	.950
Yes	5 (12.5%)	3 (13.0%)
Previous suicide attempts
No	35 (87.5%)	18 (78.3%)	0.93	.334
Yes	5 (12.5%)	5 (21.7%)
Previous hospitalizations
No	36 (90.0%)	20 (87.0%)	χ2(1)= .14	.711
Yes	4 (10.0%)	3 (13.0%)
Psychiatric medication at intake
No	11 (27.5%)	14 (60.9%)	χ2(1) = 6.79	.009
Yes	29 (72.5%)	9 (39.1%)
Anxious comorbidity at intake
No	29 (72.5%)	21 (91.3%)	χ2(1)= 3.15	.076
Yes	11 (27.5%)	2 (8.7%)

¹ Graffar is an international classification for socioeconomic status (SES; Graffar, 1956). The higher is the score, the lower the SES. This sample means indicate high SES.

Note. M = Mean; SD = Standard Deviation; GAF = Global Assessment of Functioning of the DSM-IV (First et al., 2002); BDI-II = Beck Depression Inventory II; OQ-45.2 = Outcome Questionnaire 45.2; OQ-45.2 IR = Outcome Questionnaire 45.2 Interpersonal subscale; WAI-C = Working Alliance Inventory–client version.

Discussion

Variables that are usually found to predict dropping out in the literature (e.g., low SES, low education, being female and therapeutic alliance) were not corroborated by our results. Assignment to treatment not predicting dropout is in line with findings in the literature (Swift & Greenberg, 2012), although it conflicts with the findings showing higher dropout rates in CBT therapies when compared to other therapies (e.g., Peñate Castro et al., 2014). These negative findings should be read with caution due to the small variance, restricting its generalization.

Consumption of medication at intake reduced the odds of dropping out from psychotherapy. Although clinicians and researchers agree that “the alleviation of symptoms with medication would reduce the motivation for gaining the understanding or making lifestyle changes needed to bring about a sustainable ‘cure”’ (Thase & Jindal, 2004, p. 743), they also recognize that pharmacotherapy is “a way to hasten recovery and to help patients make better use of psychotherapy” (p. 744). It appears that the alleviation of acute negative symptoms promoted by medication may have helped the clients adhering to psychotherapy and completing treatment. Also, the fact that medicated clients were assisted simultaneously by a psychiatrist and a psychologist may have decreased the odds of dropout.

The other prominent predictor for treatment dropout was not having a comorbid anxiety disorder or symptoms at intake. It can be suggested that anxiety symptoms would increase activity level and enhance motivation to complete the intervention. Since negative symptoms such as lack of activity and anhedonia characterize MDD, it is possible that depressed clients with no comorbid anxiety would have less motivation for completing treatment.

Results suggest that clinicians should pay extra attention to the clients not taking medication or with no comorbid anxiety, especially on the early stages of psychotherapy, when dropout is most likely to occur.