Background
Anorexia nervosa (AN) has been recognized as a severe disorder that is challenging to treat. This is partly reflected by the often long and severe course of the disorder, which in many patients is characterized by recurrent treatment attempts and dropouts, rehospitalization, high rates of relapses and chronicity [
1‐
3]. The comparably large group of patients with long-lasting problems and a chronic course has been identified as being particularly difficult to treat and having poor treatment outcomes [
4]. Generally, evidence on treatment approaches for long-standing AN is very limited [
4] and more knowledge on predictors of treatment response and resistance in AN patients could inform the further development of such approaches. Therefore, the aim of the present study was to investigate the association of the motivational stages of the patient with treatment outcome variables and therapeutic alliance in chronic AN patients.
One major challenge that has been identified in AN treatment is the ambivalent motivation for change and treatment in affected patients. This ambivalence has been associated with the pronounced egosyntonicity of AN symptoms, including severe underweight, as a remarkable feature of the disorder, which constitutes a major barrier to behavioral change [
5]. Two approaches targeting motivation in therapeutic change are the Transtheoretical Model (TTM) by Prochaska & DiClemente [
6,
7], as a more theoretical framework, and Motivational Interviewing [
8], as a treatment approach specifically targeting the enhancement of motivation. Both approaches have recently been adopted in the field of eating disorders, including AN [
9]. The TTM suggests that therapeutic interventions should be adapted to the motivational stage of the patient. It distinguishes four empirically validated stages of change: In the
precontemplation stage, patients have no intention for a therapeutic change. In the
contemplation stage, patients ambivalently think about therapeutic change. In the
action stage, active work on therapeutic change is central and in the
maintenance stage, patients focus on relapse prevention. One central assumption of the TTM is that patients move through the stages of change in a spiral pattern. Hence, relapses are regarded as integral parts in the change cycle [
10]. The most widely used dimensional instrument to measure the stages of change is the University of Rhode Island Change Assessment (URICA, [
11]). Recently, a short version with excellent psychometric properties has been constructed [
12]. A meta-analysis found robust effect size of d = .46 for the association between URICA stages of change and outcome [
13]. The URICA comprises several items that clearly reflect relapse struggle [
12,
14]. Hence, it is an instrument that specifically identifies patients at relapse risk. This is of importance concerning AN research, because, as we have outlined above, high rates of relapses and chronicity are critical elements in AN treatment [
15].
In their recent review, Dray & Wade [
9] point out that while there is first evidence that the TTM is applicable to eating disorders and the stages of change are associated with treatment outcome in affected patients, only few studies have been conducted and more research is needed. The investigation of different patient samples (e.g. age groups, patients with a chronic course), a broader focus on different relevant treatment outcomes, and the consideration of the therapeutic alliance have been identified as major research gaps [
9]. To our knowledge, there are currently only three studies published on the predictive value of stage of change on treatment outcome in AN patients [
16‐
18], two of which were conducted in adolescent samples. All of them report on a global score of stages of change as an indicator of motivation to change and identified this global score as a positive predictor of weight gain, improvement of eating disorder pathology, and composite improvement outcome at discharge [
17,
18] as well as of weight maintenance after discharge [
16].
In the present study, we have investigated the predictive value of stages of change in an adult sample of patients receiving inpatient treatment for AN. As outlined above, relapses and chronicity are major concerns in AN treatment. Hence, we investigated adult patients who suffered at least one year from full syndromal AN, and predominantly focused on chronic patients. Complementing earlier evidence and addressing current research gaps outlined by the recent review by Dray & Wade [
9], we have focused on different relevant outcomes in AN treatment, including weight gain, general psychopathology, and therapeutic alliance. In-line with earlier evidence on the role of stages of change in the therapeutic process of inpatient psychotherapy by our work group [
12], we hypothesized that lower ratings on precontemplation as well as higher ratings on contemplation and action are associated with a better outcome in adult AN inpatients in terms of weight gain and improvement of general psychopathology. Further, we presumed that higher ratings on the maintenance scale are associated with a more negative outcome, because the maintenance score of the URICA-S reflects the struggle with relapses [
12]. Turning to the therapeutic alliance, we hypothesized in-line with earlier evidence [
19] that contemplation at an earlier stage of therapy is positively associated with the alliance at a later stage of therapy. We further hypothesized that precontemplation and maintenance at earlier stages of therapy would be negatively associated while action would be positively associated with therapeutic alliance at later stages of therapy.
Discussion
The aim of the present study was to investigate the association of stages of change according to the TTM [
6,
7] with different treatment outcome variables in a sample of predominantly chronic full syndromal AN patients. Addressing current research gaps outlined by a recent review [
9], we investigated the predictive value of stages of change for weight gain as assessed by changes in BMI, general psychopathology as assessed by the SCL-90, and therapeutic alliance as assessed by the SACiP [
22]. As we investigated a mainly chronic sample and because high relapse rates are known to be one of the core problems of AN treatment, we applied the URICA-S as a measure of stages of change because its
maintenance scale reflects relapse threat [
12,
14]. (See Appendix for detailed wording of the items of the maintenance scale).
To our knowledge, our study is the first analyzing outcome profiles in AN patients for discrete stages of change. The few earlier studies available predicted outcome by applying a global score [
16‐
18]. We were especially interested in therapy processes in patients treated for chronic AN, and our sample therefore shows a high average duration of illness of about 9 years.
We identified significant predictive effects of the
maintenance scale on general psychopathology for all three measuring times. The higher the
maintenance score, that is, the higher the struggle of patients with relapses, the higher general psychopathology in the SCL-90. No other stages-of-change scales resulted in additional predictive effects on therapeutic outcome in the regression models. Hence, the
maintenance score of the URICA-S as an indicator of relapse threat obviously is especially important to identify AN patients who improve during inpatient treatment and those who suffer from repeated drawbacks. In many patients, the course of AN is characterized by recurrent relapses, and the first time after hospital discharge seems to be a specifically vulnerable phase [
15,
24]. It might therefore be speculated that those patients who score lower on
maintenance, that is, who struggle less with relapses, have taken a perspective on their disorder that understands recovery as a process incorporating relapses, and have developed strategies to cope with drawbacks and both could facilitate the success of inpatient treatment. This, in turn, would first underline the importance of integrating elements of relapse prevention into inpatient treatment programs for AN patients as, e.g. education and skills training related to drawbacks and relapses, and would second underline the importance of relapse prevention interventions following inpatient treatment. Both approaches might improve the course of the disorder, particularly in patients with long-standing problems, as Long, Fitzgerald and Hollin [
25] recently reported in a prospective follow-up study that those chronic AN patients had a better outcome who received continuity of care, that is, who had outpatient treatment immediately following inpatient care. The URICA
maintenance scale might be applied as one aspect of a relapse identification tool, which could be used to identify subgroups of AN patients who might especially benefit from such relapse-addressing aftercare programs [
9,
26].
In our study, stages of change surprisingly had no predictive effect on BMI. This is in contrast to an earlier study where Castro-Fornieles et al. [
16] identified stages of change as a highly significant predictor of BMI at discharge. Possibly, this may be attributed to the different measures applied to define stages of change and different samples under investigation. While we are reporting data of adult patients with long-lasting full syndromal AN, Castro-Fornieles et al. [
16] investigated an adolescent sample. According to state-of-the- art practice in adolescent therapy, patients were treated to normal-weight status. Further, as we investigated patients who mostly suffered for many years from AN symptoms, smaller weight change has to be expected than in adolescent patients whose eating disorder just emerged. Hence, the pre-post change in BMI was much larger in this adolescent sample (15.5 to 18.4) than in our predominantly chronic adult sample (14.9 to 16.2), which received on average much shorter inpatient treatment. Hence, more potential variance remains to be explained in weight gain in the study by Castro-Fornieles et al., which probably is responsible for the better predictive value of stages of change on BMI. However, it is generally difficult to evaluate our finding on weight gain as evidence on stages of change in adult samples is scarce [
9]. Further research is needed to investigate the predictive value of stages of change on weight gain as a pivotal aim and outcome of AN treatment.
Turning to the predictive effect of stages of change at therapy entrance on therapeutic alliance on discharge, we identified significant predictive effects of the
contemplation scale on both
emotional bond and
agreement on collaboration at discharge. This is in line with earlier evidence [
19], where contemplation at baseline predicted positive therapeutic alliance at session one and session three in a sample of sixty patients with mixed psychiatric diagnoses. Although
precontemplation at therapy entrance did correlate significantly and negatively with emotional bond and agreement on collaboration at discharge, it was not a significant predictor in the regression analyses when controlled for the autoregressor. AN patients often value many aspects of their disorder [
5] and therefore are at least in parts or periodically characterized by
precontemplation with regard to therapeutic change. This constitutes a major difficulty in AN treatment and assigns therapists with the motivational task to challenge
precontemplation and try to induce
contemplation in patients. Techniques of Motivational Interviewing [
8] have been specifically designed to address this aspect and to reach patients in between
precontemplation and
contemplation. Our data shows that
contemplation is a predictor of a positive therapeutic alliance in the course of inpatient treatment. This might lead to the conclusion that early intervention strategies to induce
contemplation in AN inpatients are not only important for therapeutic change, but might also form a basis for a stable therapeutic alliance that even allows for the common cautious exploration of further stages of change in the course of treatment [
8].
Our study implied several limitations as discussed hereafter. First, the URICA is sensitive to identify relapses, but it is not a disorder-specific instrument. Hence, in future studies it might be important to address patients’ relapse threats with a disorder-specific measure. The Anorexia Nervosa Stages of Change Questionnaire (ANSOCQ) [
27] is an instrument measuring stages of change specifically in AN patients. In contrast to the URICA, where the
maintenance score reflects relapse struggle, in the ANSOCQ, this score refers to the ability to maintain progress [
12,
14]. Nevertheless, as we have outlined above, relapse measures are of importance concerning AN research, because high rates of relapses are critical elements in AN treatment [
15]. Therefore, it might be of interest to phrase a new, more differentiated disorder-specific stage-of-change scale, containing a progress maintenance as well as a relapse subscale. Second, our study predominantly included chronic inpatients. Effects of stages of change on outpatients and non-chronic patients might be different. Third, we only applied a measure of general psychopathology as a symptomatic outcome instrument. Future studies should investigate stages of change outcome associations with measures that are more specific to the eating disorder symptoms, which are at the core of AN. Fourth, the duration of inpatient treatment varied across participants. Treatment duration might influence the quality of therapeutic alliance. More specifically, in our sample, patients with longer versus shorter treatment duration expressed slightly higher therapeutic alliance ratings. As sample size in our study was too small to analyze further differences between these two groups, future research should address this aspect in studies with larger sample sizes.
Competing interests
None of the authors declares financial or non-financial competing interests related to the present study.
Authors’ contributions
JM, MT and SZ have contributed to conception and design of the study, JM and KK have performed acquisition and analysis of data, JM and KEG have contributed to interpretation of data and drafted the manuscript, MT, KK and SZ have critically revised the manuscript for important intellectual content. All authors have approved the final version of the manuscript.