The aim of the investigation is to be able to provide well-founded empirical information about the interconnection between process factors and therapy outcomes – and especially about their sustainability. The intention in this study is to closely examine the working alliance as a central process factor. In empirical psychotherapy research, the working alliance is usually conceptualised as a predictor (Barber, Khalsa & Sharpless, 2010) and is regarded in the meantime as that construct which has been investigated the best. There is broad consensus that it is defined as the ability of the patient and therapist to become engaged in the tasks of the therapy and reach shared goals on the basis of an affective relationship. The statistical correlation between the working alliance and the treatment outcome has proved to be a robust finding in meta analyses, with r = .21 (Horvath & Bedi, 2002) and r = .28 (Horvath, Del Re, Flückiger & Symonds, 2011) However, the effects are minor and explain less than 7% of the outcome variance (Beutler, 2009). Yet against the background that when clarifying the outcome variance one is more likely to be dealing with many predictors – such as therapist variables or the therapeutic technique (Lambert & Barley, 2002) – which seldom explain more than 15% of the variance, the determining capacity of the working alliance certainly seems to be significant. Thus, in the complex structure of specific and unspecific effective factors, the working alliance is accorded a central role, which it is intended to comprehend and research in the study as an intervening variable (Flückiger, Wampold, Del Re, Symonds, & Horvath, 2012).
The investigation has been conceived as an additional study of a process-outcome study on the interconnection between emotional processing and the therapy outcome (Huber, Seybert, Ratzek, Zimmermann, Klug). Both investigations are based on the preliminary work and data from the Munich Psychotherapy Study (MPS) (Huber, Henrich, Clarkin & Klug, 2013; Huber, Henrich, Gastner & Klug, 2012; Huber, Zimmermann, Henrich & Klug, 2012). In a mediator analysis, the therapeutic working relationship proved to be the most important mediator in this study (OR=3.84; Huber, Henrich, & Klug, 2013).
The intention is to base the investigation on a computer sample as it should concern lege artis conducted, i.e. not broken off, therapies and the long-term outcome, and not the special processes which lead to an early ending of the therapy. Of the 100 patients in total in the intend-to-treat random sample from the MPS, 85 patients (32 AP, 27 DP, 26 CBT) ended their therapy in a regular manner. Following an assessment of the recording quality of the hours in the mid phase, 13 cases (6 AP, 3 DP, 4 CBT) had to be excluded, so that the random sample includes 72 patients (26 AP, 24 DP, 22 CBT).
The following outcome measurement instruments are used for the study:
Beck Depression Inventory (BDI; Hautzinger, Bailer, Worall & Keller, 1994) and Inventory of Interpersonal Problems, short form (IIP; Horowitz, Strauß & Kordy, 2000).
In order to measure the working alliance, the following measurement instruments were selected. Doing so, it was taken into consideration that the process is mapped in a multi-modal manner (patient, therapist, assessor), which is recommended by Stiles and Goldsmith (2010) for the measurement of the working alliance:
1. The Working Alliance Scale – Observer Rating – short form (WAI-O-S; Andrusyna, Tang, DeRubeis & Luborsky, 2001, German version Strauß, Wiltink, Stangier & Nodop, in preparation) is a measurement instrument which, because it has been conceptualised in a trans-theoretical manner (Bordin, 1979), is well suited for investigating therapies from differing therapeutic schools. It is based on the concept from Bordin (1975, 1979, 1980) and measures three components of the working alliance:
a) the affective bond between the patient and therapist
b) the agreement on shared therapy goals, and
c) the agreement on the therapeutic tasks.
Consequently, the WAI-O-S has three sub-scales (bond, goals, tasks), each of which is recorded with four items. In order to apply the WAI-O-S, rater training with reliability checks and recalibration sessions is required. Studies to date suggest there is satisfactory to good inter-rater reliability with a Pearson correlation coefficient of .67 (Andrusyna et al., 2001) and an intra-class correlation (ICC) of .79 (Strunk, Brotman & DeRubeis, 2010). In order to apply the WAI-O-S, cooperation has been established with the SOPHO-NET Project (Leibing, Salzer, & Leichsenring, 2008). The intention is for the WAI-O-S to be assessed by trained raters at the process measurement time points of V1 = 6 months and of V2 = 12 months after the start of the therapy.
2. The Helping Alliance Questionnaire, Patients and Therapists Form (HAQ-P and HAQ-T; Bassler, Potratz & Krauthauser, 1995) is a measurement instrument with 11 items that records two aspects of the working alliance: The perceived helpfulness by the therapists (HA1) and the collaboration and the bonding with the therapists (HA2). In the context of the MPS, the HAQ-P and HAQ-T were already filled out by the therapists and patients at the process measurement time points of V1 = 6 months and of V2 = 12 months after the start of the therapy. The 6-level ratings scale ranges from 1 = completely agree to 6 = completely disagree. In the review from Elvins and Green (Elvins & Green, 2008), the internal consistency and the inter-rater reliability were classified as being acceptable; for the German version, the reliability and validity were classified to be sufficient (Bassler, 2002). In order to avoid any confounding of the working alliance and symptomatology ("latent symptom confound"), the intention is to only include the items in the collaboration and bonding scale (HA2) in the evaluations, as the items in the perceived helpfulness (HA1) include information about the progression of the symptoms (Hatcher & Barends, 1996).
In order to test the questions, multiple regression analyses are calculated. In each case, one of the two outcome variables (BDI, IIP) is applied to one of the three goal time points (therapy end, 1-year catamnesis, 3-year catamnesis) as a criterion. The following measurements are taken into account as predictors:
1) form of therapy
2) respective outcome variable at the start of the therapy
3) Working alliance from a specific measurement perspective (HAQ-P, HAQ-T or WAI-O-S) and at a specific measurement time point (after 6 and 12 months).
The planned study is aligned to a certain extent to the already existing empirical evidence on the significance of the working alliance for the therapy success. Above and beyond this however, it also makes a specific contribution to exploring the significance of the working alliance for long-term psychotherapies (N>100 sessions) and for the stability of the treatment success. The study can also serve as preliminary work for further studies examining the as yet extensively unanswered questions about if and how the working alliance effects change.
Prof. Dr. Dr. Dorothea Huber
International Psychonalytic University
Dr. Melanie Ratzek
Tel. +49 30 300 117-767
Fax +49 30 300 117-769
Dr. Günther Klug
Klinik und Poliklinik für Psychosomatische Medizin und Psychotherapie/Clinic and Polyclinic for Psychosomatic Medicine and Psychotherapy, TU Munich University