Background
International emphasis is growing on involving patients in their own treatment as a key to health behavior change and improved self-management of chronic diseases [
1,
2]. The central concepts are that patients should take an active role in choosing and implementing their own care and that health care systems should actively support and honor patients’ self-determination [
3‐
5].
A related line of research in alcoholism treatment pertains to patient-treatment matching. On the basis of the idea that different kinds of patients may respond differently to different treatment approaches, the U.S. National Institute on Alcohol Abuse and Alcoholism funded the largest randomized trial of alcoholism treatment methods ever conducted: Project MATCH [
6]. The primary purpose of this project was to test matching hypotheses regarding which patients would respond optimally to three very different treatments: cognitive behavioral therapy, 12-step facilitation therapy, and motivational enhancement therapy. The principal investigators of this trial, who were prominent alcoholism treatment researchers in the United States, generated two dozen
a priori matching hypotheses. The project included replication of nearly every matching effect that had been reported in prior clinical trials [
7] and represented top experts’ predictions regarding which patients should be assigned to which treatments. The stunning finding of Project MATCH was that even with a very large sample size (
N = 1726), very few of the hypotheses were confirmed as statistically significant [
8,
9]. Nearly all previously reported matching effects failed to be confirmed, and several emerged in a direction opposite to prediction. In a partial replication of the MATCH study, the UK Alcohol Treatment Trial (UKATT) research team conducted a pragmatic randomized trial with two different treatment options [
10]. They similarly found improvement in alcohol consumption with all treatment methods, with no difference between the groups or confirmation of matching predictions. In other words, the best “guesses” of some of the field’s most experienced alcoholism treatment researchers in two nations were little better than chance when it came to choosing the best treatment approaches for patients.
Another major attempt at expert matching was a set of patient placement criteria promulgated by the American Society of Addiction Medicine [
11]. As it became clear that the outcomes of inpatient programs were no different, on average, from those of less costly outpatient options [
12], insurance companies dramatically reduced reimbursement for inpatient treatment. In response, a group of treatment program directors in northern Ohio developed a consensus set of decision rules, known as the
Cleveland criteria [
13], to defend the allocation of patients to particular levels of care. These rules were subsequently published by the American Society of Addiction Medicine [
14] and were then revised into even more complex decision systems [
15,
16]. Six types of data from an extensive evaluation are used to recommend which of six levels of care a patient should receive.
If an expert matching system such as this one is valid, then patients who are matched to treatment on the basis of these criteria should have better outcomes than those who are mismatched. The few studies of these criteria to date have provided little evidence for their efficacy. One early study [
17] demonstrated no significant improvement in 6-month outcomes for matched versus mismatched cases given outpatient treatment (level II, Cleveland criteria). Similarly, only one of several potential matches was statistically significant in another study of inpatient treatment [
18]. These studies were limited to evaluation of one level of care. Four trials with patients receiving different levels of care also found little support for this expert matching system. Two of these reported only one “significant” match of the many tested, with no Bonferroni correction for the number of hypotheses tested [
19,
20]. The third, a random assignment trial, found no significant matches at all [
21]. The fourth, a multicenter observational follow-up study, found that only 24.4% of patients were matched properly to the treatment planned [
22]. One naturalistic study in Norway [
23] comparing only two levels of care found that cases given less intensive treatment than recommended showed higher attrition, less improvement in severity, and no significant reduction in substance use when compared with matched cases who were offered the recommended level of care. In summary, the findings are quite mixed, and once again, a complex expert system for placing patients into the best treatment for them appears to be little better than chance.
The difficulties experts have in matching patients to treatment and developing useful algorithms have led many researchers to question whether matching based on patient characteristics is optimal for improving treatment outcome [
24,
25]. Are there other aspects of treatment and treatment allocation that are more important? If experts are not particularly good at deciding which alcoholism treatment is best for patients, how else might matching be done?
One option is, for instance, to allow patients to match themselves, to make an informed choice from among a menu of evidence-based treatment options. This may not be feasible in smaller programs that may have few staff members; however, in larger systems where more than one treatment option can be provided, it would be feasible to allow patients to choose for themselves. Patient preference is increasingly being considered as good practice in health care, such as when there are several cancer treatment options available with similar overall evidence of efficacy. Patients can be given a fair description of the options open to them and permitted to make an informed choice of which treatment they prefer.
It is still not clear if taking patient preferences into account when choosing between treatment options improves treatment outcome. Authors of a systematic review of the literature on shared decision making in treatment of substance use disorder [
26] reported that only 3 of 25 trials revealed a significant effect when treatments were matched to patients’ preferences. The authors stated, however, that the results should be interpreted with caution owing to heterogeneity of the included studies. One study [
27] showed that informed choice improved adherence and reduced the amount of smoked cigarettes in a smoking cessation intervention program for patients in treatment for chemical dependency. On the contrary, a study [
28] of women with alcohol use disorder, who were given the opportunity of choosing between individual therapy or conjoint treatment with their male partner, showed that the number of patients enrolled in treatment increased, but there was no additional improvement in adherence or reduced drinking days. The two groups also differed significantly in sociodemographic variables that could influence outcome.
There are at least two good reasons for offering patients a choice when the treatment goal is behavior change. The first is that patients are likely to have some wisdom about which behavior change approach is most likely to work for them. Who knows them better? They certainly would know which approaches sound more acceptable or attractive to them. Second, there can be an inherent motivational advantage of choosing one’s own course of action. As Miller et al. stated,
When people perceive that they have freely selected, from among options, a product or course of action, they are likely to be more satisfied with it and committed to it. There is evidence that taking active steps toward change increases the likelihood of successful change, no matter what the action happens to be. If what matters is that the client do
something and stick with it, then it makes sense to allow clients to select that to which they will be most committed [
29].
To our knowledge, however, research is limited on the benefit of having patients freely choose their treatment approach from among options, in contrast to the usual practice of their being assigned to a treatment based on expert clinical judgment. Because there is little evidence that clinical judgment is more effective than chance in choosing optimal treatments, it is reasonable to conduct a rigorous test of patient self-matching to determine whether it does indeed improve retention, adherence, and outcome in alcoholism treatment. As part of the RESCueH studies [
30], in the present randomized clinical trial, we will compare the efficacy of patient self-matching versus treatment-as-usual expert matching.
Purpose and hypotheses
The primary purpose of this randomized controlled trial is to determine if patient self-matching to psychotherapy treatment methods improves drinking outcome, compliance, and quality of life for patients being treated for alcohol problems compared with assignment to treatment as usual, which is by means of expert matching.
A priori hypotheses
Our a priori hypotheses are as follows:
1.
Patients who choose their own treatment will show significantly greater reductions in alcohol consumption (measured by number of days with excessive drinking) at follow-up, when compared with patients assigned to treatment by expert matching.
2.
Patients who choose their own treatment method will show significantly better compliance in treatment (measured by retention) when compared with patients assigned to a specific treatment method by expert matching.
Discussion
The debate on matching patients to treatment has been going on for decades. Several studies, including Project MATCH [
6] and UKATT [
10], have shown that matching patients to treatment is little better than chance, and they have not been able to clarify which aspects of matching improve outcome. In the present study, we will investigate the importance of patient
personal choice rather than clinician matching patient to treatment based on patient characteristics. Hence, the present study will cast light on whether patients’ perceived autonomy yields better treatment outcome. In addition, personality traits will be measured to investigate whether they influence the impact of patients’ free choice on treatment outcome.
Challenges
On the basis of a pilot study of 16 patients that showed a higher preference for supportive therapy in the self-matching group than in the expert matching group, we expect that the proportion of patients assigned to the five treatments may differ between the two groups. This will be controlled for in the analysis to clarify whether it is the therapy method or self-matching that causes any observed difference in outcome.
Some would argue that 50% reduction in excessive drinking days is optimistic and therefore would criticize the power calculation. The power calculation is based on clinicians’ estimation of what would constitute a clinically meaningful outcome, however, rather than on statistical significance derived from former studies.
Innovative aspects
All patients who seek treatment at the center will undergo exactly the same intake procedure, including baseline interviews, for this study. This procedure will provide an opportunity to generate a hypothesis of motives to participate or not participate in studies, because the difference between participating or not is reduced to a single 6-month follow-up interview that will take about 45 minutes to complete. Another innovative aspect is that personality traits will be compared with outcomes in both groups and thereby will provide information on any differences in personality traits of those who profit from self-matching and those who profit from expert matching.