Background The working alliance, or collaborative bond, between client and psychotherapist has been found to be related to outcome in psychotherapy.
Purpose The purpose of this study was to investigate whether the working alliance is related to outcome in physical rehabilitation settings.
Data Sources A sensitive search of 6 databases identified a total of 1,600 titles.
Study Selection Prospective studies of patients undergoing physical rehabilitation were selected for this systematic review.
Data Extraction For each included study, descriptive data regarding participants, interventions, and measures of alliance and outcome—as well as correlation data for alliance and outcomes—were extracted.
Data Synthesis Thirteen studies including patients with brain injury, musculoskeletal conditions, cardiac conditions, or multiple pathologies were retrieved. Various outcomes were measured, including pain, disability, quality of life, depression, adherence, and satisfaction with treatment. The alliance was most commonly measured with the Working Alliance Inventory, which was rated by both patient and therapist during the third or fourth treatment session. The results indicate that the alliance is positively associated with: (1) treatment adherence in patients with brain injury and patients with multiple pathologies seeking physical therapy, (2) depressive symptoms in patients with cardiac conditions and those with brain injury, (3) treatment satisfaction in patients with musculoskeletal conditions, and (4) physical function in geriatric patients and those with chronic low back pain.
Limitations Among homogenous studies, there were insufficient reported data to allow pooling of results.
Conclusions From this review, the alliance between therapist and patient appears to have a positive effect on treatment outcome in physical rehabilitation settings; however, more research is needed to determine the strength of this association.
The relationship between patient and therapist traditionally has been viewed as an important determinant of treatment outcome and is considered central to the therapeutic process.1,2 More recently, this concept has been evaluated in research studies, where it is commonly referred to as the therapeutic alliance, helping alliance, or working alliance.3 For simplicity, this review will refer to this construct as the alliance.
The construct of the alliance in therapeutic situations is derived from theories of transference first outlined by Freud in 1912 and refers to the sense of collaboration, warmth, and support between the client and therapist.4,5 Following on from this concept, Bordin1 in 1979 defined the 3 main components that contribute to the alliance construct as: (1) the therapist-patient agreement on goals of treatment, (2) the therapist-patient agreement on interventions, and (3) the affective bond between patient and therapist. Using this definition, researchers began to measure the alliance in clinical practice and formally assess its impact on treatment outcomes. The majority of this evaluation has been conducted in psychology, counseling, or general medicine settings, where the intervention is typically centered on a one-to-one interaction between the patient and the treating physician or therapist.3,6–12 The research to date has used a variety of different tools to measure the alliance, and there has been some argument that each tool represents conceptually different, although overlapping, constructs. Elvins and Green13 recently conducted an extensive review to investigate the conceptualization and measurement of the alliance. They identified a broad consensus as to the key concepts of the alliance among the various measures, but no single unifying alliance model or a single measure that comprehensively addressed all of the key concepts. The most successfully comprehensive measures of the alliance identified in the review were the Working Alliance Inventory (WAI), the Vanderbilt Scales, and the California Scales.13
Several research studies using the above-mentioned alliance measures have found that a positive alliance is associated with positive health outcomes for variables such as depression,14,15 anxiety,15 mood,16 interpersonal problems,17 and general psychological functioning.17 A meta-analytic review of 68 studies conducted in 2000 indicated that the weighted association of the alliance with overall outcome (including outcomes of mood, anxiety, and global assessment scales) was moderate (r=.22).3 In 2001, a further meta-analysis of the relationship between the alliance and the psychotherapy outcome included 90 independent clinical investigations, from which the author reported that the alliance may account for up to half of the beneficial effects of psychotherapy.7
In the medical profession, trust is seen as a global attribute of treatment relationships, encompassing satisfaction, communication, competency, and privacy,11 and has long been viewed as vital to cooperation with treatment and physician recommendations.18 Several studies attempted to measure how trust affects clinical outcomes and found that the patient's trust in his or her physician is positively correlated with self-reported measures of health status,19 symptom status,20 and overall quality of life.21 A recent high-quality study examined how patients’ trust in their physicians affected both self-report and “objective” measures of health status in 480 patients with diabetes.10 The authors reported that patient trust was positively correlated with stronger outcome expectations (r=.46, P<.01) and self-efficacy (r=.45, P<.01), which, in turn, predicted better treatment adherence, leading to better clinical outcomes of improved body mass index, blood glucose, blood lipids, and diabetes-related complications, as well as improved self-reports of mental and physical health.
It would appear from the previous research that the alliance is positively associated with treatment outcome and could potentially be used as a predictor of treatment outcome in psychotherapy and general medicine settings. However, the degree to which the alliance relates to outcome in other treatment settings is not clear. Physical rehabilitation, like psychotherapy and general medicine, includes a high level of patient-clinician interaction; however, the characteristics of the patient population, as well as the intervention, are arguably different. It is plausible, therefore, that the relationship between the alliance and the outcome seen in psychotherapy or general medicine settings is not transferable to physical rehabilitation settings. It is thus of great importance to determine whether the alliance of rehabilitation therapists is similar to that of psychotherapists and general practitioners and whether this alliance influences outcome in the physical rehabilitation setting. To our knowledge, there has been no systematic review of the primary research in this area.
The aims of this study were: (1) to identify and summarize studies that have used and analyzed the alliance as a predictor of outcome and adherence in physical rehabilitation settings and (2) to determine whether there is an association between the alliance and the treatment outcome of physical rehabilitation programs. We hypothesized that the patient-therapist alliance would have a positive correlation with treatment outcome.
Data Sources and Searches
An electronic database search using the search strategies outlined in Appendix 1 was conducted for 6 databases (EMBASE, PEDro, PsychINFO, MEDLINE, CINAHL, and LILACS) from the earliest record to February 2009. Citation tracking was performed by manually screening reference lists of eligible trials. Theses and conference proceedings also were included. Additionally, personal communication with content experts in the therapeutic alliance field was conducted. Study inclusion was not restricted by language. The search strategy and exclusion process are illustrated in the Figure.
From the titles identified by the search strategy, original studies were included if they: (1) were prospective, longitudinal studies (randomized controlled trials, controlled trials, or cohort studies); (2) included patients who were managed with physical rehabilitation and there were no restrictions to diagnosis; (3) included at least one measure of therapeutic alliance or therapist-patient interaction/bonding; and (4) used at least one measure of treatment outcome such as pain, disability, physical performance, quality of life, global perceived effect of treatment, and adherence. Physical rehabilitation is defined as an intervention that aims to enhance and restore functional ability and quality of life in those with physical impairments or disabilities. It can include a combination of physical modalities, therapeutic exercise, activities modification, assistive devices, orthoses, and prostheses. The interventions can be delivered by a single therapist or a combination of therapists in a multidisciplinary setting, including physical therapists, occupational therapists, psychologists, chiropractors, speech pathologists, and recreation therapists.22
Data Extraction and Quality Assessment
For each included trial, 2 reviewers independently extracted quantitative data such as change or final scores and standard deviations for all relevant outcomes at all time points used in the study. In addition, correlation or regression coefficients and odds ratios for alliance and outcomes were extracted. For each included study, descriptive data regarding participants, interventions, measures of alliance, and other outcome measures were extracted. If different data were reported by the 2 reviewers, data were rechecked by both reviewers. If disagreement continued, a third author would arbitrate. However, a third author was not necessary, as consensus was reached for all extracted data.
Studies meeting the eligibility criteria were assessed for methodological quality. The methodological quality of the studies was independently assessed by 2 authors using a checklist that comprised 7 criteria: use of a representative sample, having a defined sample, use of blinding, having a follow-up rate greater than 85%, appropriate choice of outcome measures, reporting outcome data at follow-up, and control for confounding via statistical adjustment. These criteria have been used in previous studies,23,24 and their inclusion in checklists for rating methodological quality has been recommended by a recent systematic review of quality assessment tools for observational studies25 and by the STROBE Statement.26 However, this scale was not designed to provide a quality score per se; thus, there is no score allocated to each individual study. Similarly, if different data were reported by the 2 reviewers, data were rechecked by both reviewers. If disagreement continued, a third reviewer was used to arbitrate.
Data Synthesis and Analysis
Studies were grouped according to the study population and outcome measure. Within each study population, meta-analyses were intended to be performed if 2 or more studies used similar measures of alliance and similar measures of outcome. Where there were not multiple studies with sufficient homogeneity, the correlation between alliance and outcome measure of the individual studies was reported.
A total of 1,600 unique titles were identified using multiple databases (ie, EMBASE, CINAHL, MEDLINE, PsychINFO, LILACS and PEDro), citation tracking, and contact with experts in the field. Titles were merged in EndNote X,* and sources included books, theses, abstracts, conference proceedings, and journal articles from both refereed and nonrefereed journals. Following the exclusion process, a total of 14 publications (13 distinct data sets) met the inclusion criteria.27–39 The 2 publications reporting on the same cohort33,34 are treated as 1 study. The 13 studies were published between 1990 and 2009; 10 were from published sources, and 3 were from unpublished doctoral dissertations or master's theses. A detailed description of the methodological quality of each study is presented in Table 1. Considering the possibility of missed articles in this search strategy, readers are encouraged to alert the corresponding author to any papers that have not been cited in this article for future updated reviews of this material.
The patient population's diagnoses varied among the studies, including brain injury (3/13), musculoskeletal conditions (6/13), cardiac conditions (1/13), and multiple pathologies such as systemic diseases, trauma and postoperative conditions, back pain, and neck and shoulder pain (3/13).
Interventions and Treatment Outcome
The length of treatment was reported in 7 of the 13 studies and varied from 4 to 16 weeks. In 9 of the 13 studies, the interventions were delivered by a single therapist, predominantly a registered physical therapist (8/9). The other 4 studies used multidisciplinary interventions administered by multiple therapists; the alliance was based on the relationship with the client's primary therapist, who was not specified. Various outcome measures were assessed in each study, and a detailed description of the measurement tool is provided in the descriptive summary for each study (Tab. 2).
Alliance Measurement Tools
In the 13 studies, multiple instruments were used to measure the alliance between therapist and patient. The short-form WAI was used most often in the included studies (6/13). Five studies27,28,37–40 used alliance scales that are not commonly referred to in the literature. These scales either were created by the researchers for the specific study or were subscales within more general treatment questionnaires. In the sample of studies, patients were the most common raters of the alliance (12/13), followed by therapists (8/13) and observers (2/13).
Alliance Score Predictor of Outcome
Of the included studies, there was a wide range of patient diagnoses. Included studies were summarized in terms of diagnoses. Within the specific diagnostic groups, there was insufficient homogeneity between measurement of alliance and measurement of outcome to warrant pooling of data. The association between alliance and outcome, therefore, is described as reported in the individual studies. A summary of the included studies, including study characteristics and correlations (if stated), is reported in Table 2. A further detailed description of each included trial is presented in Appendix 2.
Three of the 13 studies included patients who were participating in brain injury rehabilitation programs. The rehabilitation program was similar among trials and commonly referred to as the postacute brain injury rehabilitation program (PABIR). It consisted of a multidisciplinary team working with the patient on achieving goals of improved physical, cognitive, and social function. The results from these studies are inconsistent. Two studies conducted by Schonberger and colleagues33,–,35 found significant positive associations between alliance and adherence, employment, physical training, depression, and therapeutic success. The study by Sherer et al36 found a positive correlation between alliance and program attendance but not between alliance and disability, productivity, or depression (Tab. 2).
Six of the 13 studies included patients with a diagnosis that falls under the category of musculoskeletal pain conditions, including chronic low back pain (3/6), chronic neck pain (1/6), and multiple diagnoses of musculoskeletal conditions (2/6). Various outcomes were measured in all studies. Significant positive associations were found between the alliance and the patient's global perceived effect of treatment,30,38,39 change in pain,32,39 physical function,30,31 patient satisfaction with treatment,28 depression,32 and general health status.32
Each of the remaining 4 studies investigated the alliance in mixed populations, comprising patients with a variety of different conditions. Among these studies, 2 included correlation data, which found that the alliance was significantly positively associated with physical function and depression in geriatric patients with various physical function deficits27 and that a change in alliance was associated with a change in treatment adherence for patients with cardiac conditions.29
Influence of Alliance on Treatment Outcome
The findings of this study suggest that the alliance between patient and therapist positively correlates with treatment outcome for people in physical rehabilitation settings, lending support to this study's hypothesis. The outcomes included in this review are: (1) ability to perform activities of daily living, (2) pain, (3) specific physical function tasks, (4) depression, (5) global assessment of physical health, (6) treatment adherence, and (7) treatment satisfaction. Unfortunately, a meta-analysis was not possible, and we are unable to provide a more precise estimate of the magnitude of association between the alliance and relevant treatment outcomes.
The included studies recruited patients with a mix of diagnoses. Six of the 14 studies included patients with musculoskeletal pain conditions who were undergoing physical therapy or physical conditioning programs. These studies showed a consistent pattern of positive correlations between alliance and outcome. This positive correlation pattern also was seen for patients with other conditions, including cardiovascular disease, geriatric disability conditions, and general chronic pain conditions. However, this pattern was not consistent for patients diagnosed with brain injury, as one study36 reported that as client ratings and therapist ratings of alliance improved, outcomes of physical function, productivity, and depression declined. The authors suggested that this paradoxical effect, in comparison with the other studies, may have been due, in part, to the difference in the time at which the alliance was measured. The study measured alliance in the first 2 weeks of treatment, whereas the other 2 brain injury studies measured alliance either after the treatment program35 or at multiple points during the program33 and then used a mean score for correlation analysis. In both studies that found positive correlations, there was a longer time in which the feelings of bonding and perceptions of tasks and goals of treatments could be formed.
Measurement of Alliance in Rehabilitation Settings
It is clear from this review that the alliance has not been systematically investigated in the physical rehabilitation setting, as evidenced by the lack of consensus regarding the methods of measurement. Although 6 of the 13 studies used the WAI to measure alliance, overall 7 different tools were used across the 13 studies. To date, 3 of these measures have been validated in psychotherapy settings,13 and none have been validated for patients undergoing physical rehabilitation. Without appropriate clinimetric testing, it is difficult to assess whether each tool is measuring the same construct. However, because the tool used does not appear to influence the magnitude and direction of the correlation in different musculoskeletal samples, we would suggest there is some indirect evidence that the tools may be equally valid.
There were some similarities in the methodological approach of the studies. The timing of alliance assessment was relatively consistent among studies, with 7 of 13 studies measuring the alliance during the second to fifth treatment sessions. This finding may be due, in part, to recommendations by Horvath that the alliance measured between the first and fifth treatment sessions or within the first third of treatment shows a stronger alliance-outcome association.7 Additionally, 12 of 13 studies included patient ratings of the alliance, 8 chose therapist ratings, and 2 chose observer ratings. This choice also may be due to conclusions from a previous meta-analysis that patients’ ratings of the alliance had a stronger correlation with outcome than therapists’ ratings in psychotherapy settings.3 However, based on the available data, we are unable to determine whether this is the case in physical rehabilitation settings.
The results of this study suggest a positive alliance is associated with improved outcome. Although a few studies27,41 have attempted to identify the factors that influence the alliance, there is no conclusive evidence as to which factors are most important. The limited data would suggest that providing positive feedback, answering the patient's questions, and providing clear instructions for home practice are positively correlated with a good working alliance and satisfaction with treatment.
The WAI was the most frequently used tool among the studies included in this review. There is some evidence that the WAI is appropriate for most research projects because it is well-triangulated measure with good validity data.13 These clinimetric properties, however, are based on its use in different populations undergoing psychotherapy, and further clinimetric testing of this questionnaire is needed to support its use in the physical rehabilitation setting.
The alliance has been previously shown to play a key role in influencing adherence to treatment advice as well as improving treatment outcome in psychotherapy and general medicine. Our review indicates that there are also several studies investigating the alliance in a physical rehabilitation setting, the majority of which include patients with musculoskeletal pain conditions. Although a meta-analysis could not be conducted, the results indicate a consistent positive correlation between the alliance and treatment outcomes of pain, disability, physical and mental health, and satisfaction with treatment. The findings also indicate that instruments used to measure the alliance have been developed for assessment in the psychotherapy setting. There is, therefore, an urgent need to develop a measure of the alliance construct that investigates the factors underlying the alliance in the physical rehabilitation setting before meaningful research regarding prediction of treatment outcome can be undertaken. Once appropriate measurement has been established, further prospective longitudinal studies in which the alliance is systematically measured are needed to obtain a more conclusive understanding of the relationship between the alliance and its effect on treatment outcome.
The Bottom Line
What do we already know about this topic?
The therapeutic alliance between a patient and a treatment provider is positively correlated with treatment adherence and outcome in both general medicine and psychotherapy settings.
What new information does this study offer?
This systematic review found that a positive therapeutic alliance also consistently correlated with improved pain, disability, and treatment satisfaction in physical rehabilitation. However, the retrieved studies used a variety of alliance measures that were developed for use in psychotherapy and have not been tested for reliability and validity in physical rehabilitation. Development of measures validated for physical therapy settings have the potential not only to increase our understanding of interventions but also to increase their effectiveness.
If you're a patient, what might these findings mean for you?
In order to maximize the benefits of physical therapy, a patient-centered approach is recommended as the basis for the development of a good working relationship between physical therapist and patient, with enhanced effectiveness of communication regarding specific tasks required to achieve treatment goals.
All authors provided concept/idea/research design. Ms Hall, Dr P.H. Ferreira, and Professor Maher provided writing. Ms Hall, Dr P.H. Ferreira, Professor Maher, and Dr M.L. Ferreira provided data collection and analysis. Professor Maher provided project management. Dr Latimer provided facilities/equipment. Dr P.H. Ferreira, Professor Maher, Dr Latimer, and Dr M.L. Ferreira provided consultation (including review of manuscript before submission).
The abstract of this article was presented orally at the Australian Physiotherapy Association Annual Meeting; October 1–5, 2009; Sydney, New South Wales, Australia.
↵* Thomson Reuters, 2141 Palomar Airport Rd, Suite 350, Carlsbad, CA 92011.
- Received July 27, 2009.
- Accepted April 27, 2010.
- © 2010 American Physical Therapy Association