Longer-term psychotherapies are not always more effective than the initial therapy contact, suggesting that the mere thought that one is being helped counts

The effectiveness of initial therapy contact: A systematic review. Katie Aafjes-van Doorn, Kristen Sweeney. Clinical Psychology Review, November 9 2019, 101786. https://doi.org/10.1016/j.cpr.2019.101786

Highlights
• Few empirical studies have examined the direct effect of initial therapy contacts.
• Initial therapy contact is more effective than no treatment or a waiting list.
• Initial therapy contact might be as effective as 6-session treatment.
• Several effective initial therapy formats and approaches reduce patients' symptoms.

Abstract
‘Initial therapy contacts’, defined as (the first) 3 h or less of face-to-face psychological treatment, encompassing both the early phase of a longer therapy and one-off single session therapies, are seen as a critical phase of treatment. However, little is known about the direct effect of initial therapy contacts on change in common symptoms typically presented by patients in psychological therapy services. Our systematic literature search resulted in 35 identified empirical studies on the effect of initial therapy contacts. These studies were analyzed in three stages: 1) A systematic comparison of study characteristics using the preferred reporting items for systematic reviews and meta-analyses; 2) A domain-based evaluation of methodological rigor of the studies, in line with Cochrane's guidelines on assessing risk of bias; 3) A narrative synthesis of reported findings.

The considerable variability in therapy format (a stand-alone single session, 2 + 1 format, or initial session of multisession therapy) and study design (post/pre-post measurement, with/without control) limited comparability of studies. The quality assessment indicated that the majority of studies had relatively weak methodologies overall. Qualitative synthesis of the effectiveness results suggests that a significant proportion of patients reported benefits, including symptom change. This positive effect is especially clear when compared to no-treatment controls, and appears to be maintained at follow-up. The findings suggest that a broad range of initial therapy formats, could in itself be beneficial to patients in primary care treatment settings, and that further research is warranted.

Keywords: Initial sessionsSingle-sessionEffectSystematicReview

Discussion

Thirty-five empirical studies on the effectiveness of initial therapy contacts were identified. In line with previous reviews (e.g. Hymmen et al., 2013), our quality assessment indicated that the majority of reviewed studies had relatively weak overall methodologies. Studies widely differed in the rigor of their research design and, for example, included four uncontrolled single-case designs, as well as twentyone controlled studies, of which twelve RCTs and three efficacy studies. Findings of the present review support the conclusion that (the first) 3 hours or less of therapy can possibly be an effective intervention in itself for adults with mild to moderate mental health problems. Importantly, reported effects appeared to last over time. All studies that included a follow-up measurement in their design reported that the positive effect of the initial therapy contact had been maintained several months or even years after the intervention, even if no further therapy occurred after the initial therapy contact. Some studies stated that a proportion of patients (e.g. 30% in Abbass et al., 2008) derived sufficient benefit that they did not require further treatment. Although this was not specifically addressed in any of the studies, there appeared to be no difference in outcomes between the effects of initial therapy contact when it consisted of stand-alone sessions (n = 27) or when it consisted of the start of longer therapy (n = 8). For example, a 3-hour stand-alone therapy spread over different sessions (e.g. 2 + 1) appeared to have comparable outcomes to a 3hour session at the start of a longer treatment (Abbass et al., 2008). This suggests that these very brief interventions may be effective in a variety of different formats. Stand-alone session(s). The majority of studies in this review reported on stand-alone session(s). Two specific types of effective stand-alone therapies were identified. First, a “single session therapy”, referred to a planned single-session intervention. The single session may be previously scheduled or provided in a “walkin counselling clinic” (e.g. Hymmen et al., 2013). Another type of effective standalone initial therapy contact has been developed by Barkham and colleagues (Barkham, Shapiro, Hardy, & Rees, 1999). Their “two-plus-one model” (2+1) reflects a very brief three session intervention, comprising of two 1-hour sessions one week apart, followed by a third 1-hour session three months later. The results of this review are in line with other reviews of single-session therapies (e.g. Bloom, 2001; Cameron, 2007; Rockwell & Pinkerton, 1982) and stand-alone therapeutic assessments (Poston & Hanson, 2010), which showed that stand-alone single session(s) in a variety of therapies are effective in reducing symptoms.  Initial session(s) in multisession therapy. In contrast to stand-alone sessions, “initial sessions” refer to the first session(s) of several, or rather, the start to longer therapy. Of the 35 reviewed studies, only 8 studies examined the effectiveness of initial session(s). As mentioned previously, these studies showed comparable outcomes to stand-alone single session(s) intervention. Notably, only one of these eight studies was conducted as RCT (Dunn, Neighbors, & Larimer, 2006), the others used relatively weaker study designs. This means more research comparing the effect of an initial session to overall pre- post treatment change is needed to identify any “first session gains” (Busch, Kanter, Landes, & Kohlenberg, 2006) that set the course of the therapy as a whole (Lambert & Ogles, 2004).  There are several hypotheses as to why initial sessions of multisession therapy might be particularly effective. Frank’s (1974) theory of remoralization suggests that the first hours of therapy are likely to lead to a decrease in symptoms because they help to clarify a patient’s problems, inspire hope and provide experiences of success (see also Howard’s phase model of change; Howard, Lueger, Maling, & Martinovich, 1993; Stulz & Lutz, 2007). In addition to remoralization, others have drawn on goal-setting theory (Locke & Latham, 2006) and the Theory of Planned Behavior (Ajzen, 1988) and emphasized the importance of early changes in the patient’s expectations of therapy (e.g., anticipatory beliefs about what will happen during or because of therapy) (Constantino, 2012; DeFife & Hilsenroth, 2011). According to Ajzen’s (1988) Theory of Planned Behavior, beliefs about expected outcome, self-efficacy concerning necessary ‘patient role’ behaviors and motivation to achieve improvement determine intention to engage and, therefore, the success of therapy. Locke and Latham’s (2006) goal setting theory similarly suggests that an individual’s expectations of therapy may be linked to how much the individual is motivated to engage in working towards their goals and, therefore, achieve symptom change. It has to be noted, however, that these theories of remoralization, goals, and planned behavior not necessarily explain therapeutic benefits of planned singlesession therapy or very brief interventions as stand-alone treatment. Moreover, these theories do not explain why initial therapy contact seemed to be effective irrespective of treatment modality that was used, with vast theoretical and technical differences. Whilst it should be stressed that no statistical comparison between the different models has been attempted, the findings appear to be congruent with the ‘equivalence paradox’ or ‘Dodo-bird verdict’ (Luborsky et al., 2002). These common factors may have an effect on initial therapy contact irrespective of further therapy offered and thus seem to contradict the expectations hypothesis (Constantino, 2012; Greenberg, Constantino & Bruce, 2006). Similarly, the fact that longer-term therapies were not always more effective than the initial therapy contact is inconsistent with the dose-effect literature which identified a relatively larger effect early in therapy, and continuing but diminishing levels of improvement over time. 

The extent to which conclusions can be drawn from the reviewed studies was limited by their respective internal validity (i.e. whether the study results can be attributed to the effect of the initial therapy contact, or whether they might be a result of other factors, such as concurrent treatments, type of control group or therapist effects unrelated to the model of treatment) and external validity (i.e. whether the findings can be legitimately generalized to other people and situations in clinical practice). First, it is possible that we have underrated the overall quality of the reviewed studies in our methodological assessment. For example, two welldesigned efficacy studies (Gellis et al., 2013; Goerling et al., 2014), received an overall weak quality rating, due to their reported low participation rates. Also, attempts to assess risk of bias are often hampered by incomplete reporting of what happened during the conduct of the study (Higgins & Green, 2008). It is therefore possible that the reviewed studies were of higher internal validity than indicated in our systematic review, because important details of the applied study methodologies were left out in the final publications of the studies. Second, publication bias may have impacted the results of this systematic review by alluding to a treatment efficacy when non-published studies failed to replicate such findings (Liberati et al., 2009). Whilst a publication bias should be considered within any systematic review, the existence of negative reporting (e.g. Hutchinson & Krippner, 1988) suggests that it is unlikely a publication bias would have significantly impacted the findings of the current review. Notably, the majority of the more robust study designs were reported in the most recent publications, possibly indicating the current focus in psychotherapy research on evidence-based practice and increased standards of peer-reviewed journals. Third, some of the results of the descriptive studies might have been affected by a response bias, when patients’ inclination to give positive feedback when contacted by the service (Battaglia, Shapiro & Zell, 1996), which could explain why the patients in both the longer-term therapy control groups and the initial therapy contact reported symptom improvement/ recovery in the phone-interviews. Alternatively, in the case of the study by Askevold (1983), the results might also have been affected by patients’ memory of the content, as there was an extraordinary nine years between intervention and follow-up.  Besides these aspects of internal validity, evidence for the external validity of the findings also appears to be mixed. The populations in the studies appear representative of mainstream adult mental health services with respect to age and gender balance (Barkham et al., 2001). However, many relevant patient and therapy
characteristics were not reported. Future research would, for example, benefit from detailed descriptions of treatment setting (e.g., outpatient, crisis clinic), level of therapist training, therapist orientation and location of services (e.g., rural/urban). Moreover, the majority of the studies involved patients with mild to moderate mental health problems, often subclinical, which means that the conclusions are confined to treatments for patients with relatively mild problems, that is, consistent with the severity of problems of patients seen in primary care mental health services (Barkham et al., 2001; Haaga, 2000). However, this also resulted in a small scope for reduction in scores on standardized measures, and might thus understate the potential efficacy of initial therapy contacts. The impact of initial therapy contacts on interpersonal or personality problems was not addressed in the majority of studies reviewed and, therefore, the current studies cannot challenge the existing notion that these difficulties take longer to change (Hardy et al., 1995; Lambert & Ogles, 2004; Merbaum & Butcher, 1982). It may be useful, in future research, to explore the impact of initial therapy contacts for more severe presentations, such as those seen in, for example, secondary and tertiary care services. To know beforehand for whom ultra-brief therapy is enough (e.g. patient factor) would be extremely important knowledge for the cost-effectiveness and efficiency of mental health services.  In order to develop the evidence-base of initial therapy contact further, it is crucial to determine not only the patient and treatment moderators but also the mediators of improvement. An important next step could be to examine the relative benefits of key process elements, such as alliance (Hilsenroth & Cromer, 2007), expectations and hope (Constantino, Arnkoff, Glass, Ametrano, & Smith, 2011) that form early in therapy and are assumed to mediate the effectiveness of initial sessions (Messer & Wampold, 2002). Given that the reviewed empirical studies did not report on moderation or mediation analyses, future researchers should address this gap in the psychotherapy literature.  As next step in future research, the effectiveness as well as relevant moderators and mediators of initial therapy contact should be examined in a metaanalysis. In this review, meta-analytic strategies could not be applied because only a small number of studies (n = 13) (e.g. Abbass et al., 2008; Armento et al., 2012) reported significance levels, and the vast majority of studies did not provide enough information to enable calculation of effect sizes. When researchers ensure that the intervention and treatment effect (or effect size) is reported consistently from one study to the next, meta-analysis can be used to numerically pool the results of the studies and arrive at a summary estimate to identify this effect of initial therapy contact. Subgroup analyses and meta regression can then be conducted to test if there are subsets of research that capture the summary effects. This next step will then help test hypotheses around the relative importance of the length, format or therapy model of the initial therapy contact in determining symptom change.  Further research might illuminate whether patients who substantially improve during the initial sessions of longer therapy would have done so even if treatments were planned to be only ultra-brief (3 hours or less). Building on our findings, future researchers may clarify if initial therapy contacts with varying total treatment dose are equally effective (see dose-effect model; Howard et al., 1986), or if rate of change is related to total dose of therapy (see the good-enough level model of change; Baldwin, Berkeljon, Atkins, Olsen & Nielsen, 2009).


Clinical Implications.

The reported risks of bias in the reviewed studies precludes drawing any strong conclusions about the effectiveness of each specific intervention. However, for clinicians it will be important to know that, in contrast to the review of single-session debriefing after trauma (Bisson, 2010; Rose et al., 2003), there were no indications of any significant harm or distress following the initial therapy contacts applied in general outpatient services. In other words, initial therapy contacts (as a stand-alone intervention or as part of longer treatment) may constitute a beneficial therapeutic intervention in itself. The self-reported patient outcomes suggest that there might be a number of initial therapy formats (singlesession, 2-1, & initial session), which could potentially benefit patients in primary care treatment settings, more than no treatment or being on a waiting list.  Understanding the direct effect of these initial therapy contacts is thus of great significance to clinicians who, in this managed-care era, are under pressure to provide effective relief in the shortest time possible. Clinicians will want to know who gets better and what it is that leads to that improvement, to stratify selection or alter therapy content accordingly. Moreover, by identifying the most important factors associated with change in initial therapy contacts, we may aid the development of therapist training to maximize the beneficial impact of this critical stage in therapy. Ultimately, effective initial treatment sessions might not only benefit patients, but might also aid service providers, when these early outcomes translate into lower non-attendance or drop-out rates in clinical services. Similarly, the very brief 2+1 therapies are likely to harness the effects of elapsed time within the service context and may thus be both cost effective and clinically effective. Service administrators may view single session therapy as a less risky and more long term cost-effective alternative to lengthy waitlists and may argue that one session of therapy may be all that many patients need (Boyhan, 1996; Talmon, 1990). Therefore, the application of these very brief interventions in clinical services also raises ethical concerns that decisions to limit the number of sessions available to patients may be based on budgetary constraints rather than clinical judgements (Campbell, 2012; Hurn, 2005).