The role of outcome expectancy in therapeutic change across psychotherapy versus pharmacotherapy for depression

Highlights

• Outcome expectancy (OE) improves during psychotherapy and pharmacotherapy for depression.

• Mid-treatment optimistic and pessimistic OE is linked to antidepressant response.

• Pretreatment optimistic and mid-treatment pessimistic OE is linked to psychotherapy response.

• Strategies to enhance optimism during antidepressant treatment may enhance clinical outcomes.

• At pretreatment for CBT, strategies to enhance optimism may enhance clinical outcomes.

Abstract

Background

Patient outcome expectancy - the belief that treatment will lead to an improvement in symptoms – is linked to favourable therapeutic outcomes in major depressive disorder (MDD). The present study extends this literature by investigating the temporal dynamics of expectancy, and by exploring whether expectancy during treatment is linked to differential outcomes across treatment modalities, for both optimistic versus pessimistic expectancy.

Methods

A total of 104 patients with MDD were randomized to receive either cognitive behavioral therapy (CBT) or pharmacotherapy for 16 weeks. Outcome expectancy was measured throughout treatment using the Depression Change Expectancy Scale (DCES). Depression severity was measured using both the Hamilton Depression Rating Scale and Beck Depression Inventory-II.

Results

Latent growth curve models supported improvement in expectancy across both treatments. Cross-lagged panel models revealed that both higher optimistic and lower pessimistic expectancy at mid-treatment predicted greater treatment response in pharmacotherapy. For CBT, the associative patterns between expectancy and depression differed as a function of expectancy type; higher optimistic expectancy at pre-treatment and lower pessimistic expectancy at mid-treatment predicted greater treatment response.

Limitations

The sample size limited statistical power and the complexity of models that could be explored.

Conclusions

Results suggest that outcome expectancy improved during treatment for depression. Whether outcome expectancy represents a specific mechanism for the reduction of depression warrants further investigation.

Introduction

Depression remains the most common mental disorder, with nearly 298 million cases of major depressive disorder (MDD) reported worldwide in 2010 (Ferrari et al., 2013). Although empirically supported treatments for MDD exist, including pharmacological and psychological interventions (Lam et al., 2009, Parikh et al., 2016), it is well recognized that a sizable proportion of patients do not respond to these treatments (Hofmann et al., 2012, Trivedi et al., 2006). Such findings highlight the need for a greater understanding of factors that predict response to treatment in MDD and the mechanisms by which treatment results in therapeutic change.

Patient expectancies of treatment outcome – that is, prognostic beliefs of whether treatment will lead to a change in health status - have been theoretically and empirically linked to both process and outcome in psychotherapy (Constantino et al., 2011, Glass et al., 2001, Greenberg et al., 2006) and pharmacotherapy (Rutherford et al., 2010). Indeed, theoretical models of psychotherapy including Frank's (1961) theory of remoralization and the Snyder et al. (2000) theory of hope posit a central role of patient expectancies in symptom improvement (Constantino et al., 2012, Swift and Derthick, 2013, Westra et al., 2007). Similarly, expectancy theory posits that conscious expectancies of improvement mediate the placebo response seen in pharmacological treatments (Rutherford and Roose, 2013, Stewart-Williams and Podd, 2004). As overall treatment efficacy has been conceptualized as the sum of the placebo effect and the effects specific to the active pharmacotherapy, more optimistic expectancy may contribute to stronger treatment response with active medication (Wampold et al., 2005). This link between outcome expectancy and treatment response has been demonstrated consistently in individuals with a diagnosis of depression (Schulte, 2008, Webb et al., 2013).

The majority of the empirical support for the influence of patient outcome expectancies has been circumscribed to psychotherapy. In a comprehensive meta-analysis of 46 independent studies, outcome expectancy at the beginning of psychotherapy was found to have a small, significant positive association with therapeutic outcomes, such as post-treatment symptom severity (Cohen's d = 0.24; Constantino et al., 2011). More recent research supports the dynamic nature of this construct within psychotherapy treatment in anxious (Brown et al., 2014, Newman and Fisher, 2010) and obsessive-compulsive samples (Vorstenbosch and Laposa, 2015). Change in this construct has also been demonstrated for treatment of mood disorders. In individuals with recurrent major depression with a seasonal pattern, outcome expectancy increased steeply over a 6-week course of group CBT for seasonal affective disorder (SAD) (Meyerhoff and Rohan, 2016). Finally, increased outcome expectancy was recently demonstrated in patients with depression who received a 14-week course of individual CBT (Vîslă et al., 2018).

This growing literature of dynamic changes in patient expectancy highlights the importance of evaluating outcome expectancy not only at pre-treatment, but also during treatment course. The demonstration of an association between expectancy during treatment and subsequent symptom improvement may substantiate the mechanistic role of expectancy. Indeed, change in expectancy during psychotherapy has predicted symptom severity and functioning level at post-treatment in several studies (Brown et al., 2014, Newman and Fisher, 2010, Vorstenbosch and Laposa, 2015), whereas point estimates of outcome expectancy during treatment were linked to clinical outcomes in others (Meyerhoff and Rohan, 2016, Visla et al., 2016).

Nevertheless, important questions remain in the literature. First, the causal significance of this construct across different treatment modalities is presently unclear (Meyerhoff and Rohan, 2016). Comparing the impact of outcome expectancy across different classes of treatment can help to clarify whether expectancy of change is a putative common mechanism for improvement, or whether it is a possible mechanism specific to treatment change. To this end, exploring the role of outcome expectancy across cognitive behavioral therapy (CBT) and pharmacotherapy, two treatments for depression for which there are comparable outcomes and among the strongest evidence base (Cuijpers et al., 2013), is warranted. Like CBT, there is considerable evidence for the predictive power of pre-treatment expectancy on therapeutic response in pharmacotherapy (Meyer et al., 2002, Rutherford et al., 2017, Rutherford et al., 2010, Sotsky et al., 1991), although expectancy during pharmacotherapy treatment has received limited investigation.

Second, the measurement of patient outcome expectancy is consistently recognized as a limitation in the literature to date. As reviewed by Rutherford et al. (2010), the majority of self-report scales developed for the assessment of outcome expectancy were purpose-built for a specific investigation, with limited psychometric validation beyond those investigations. A further important consideration is that expectancies can be directed to the likelihood of positive outcomes (i.e., optimistic expectancy) as well as negative outcomes (i.e., pessimistic expectancy; Schulte, 2008), although existing instruments largely assess only the former. The existence of optimistic and pessimistic expectancies is highly relevant to MDD in particular; both optimistic and pessimistic cognitions are seen in this disorder in the form of certainty in the absence of positive future events (i.e., low optimistic expectancy) and the presence of negative future events (i.e., high pessimistic expectancy; Miranda et al., 2008), each proposed to stem from distinct underlying affective-motivational systems (MacLeod, 1996). In the context of depression, there is thus a particular need to ascertain both optimistic and pessimistic expectations of outcomes (Dozois and Westra, 2005, Eddington et al., 2014, Schulte, 2008).

The current study adds to the growing expectancy literature by first investigating the temporal dynamics of outcome expectancy during the course of different treatments for MDD. Specifically, growth is compared across CBT and pharmacotherapy as these treatments represent empirically supported interventions for MDD with theoretically distinct mechanisms of action. Second, we investigate whether outcome expectancy both before and during treatment is predictive of depression outcome in each treatment. Finally, this study improves upon the current literature by measuring outcome expectancy using a more refined scale (the Depression Change Expectancy Scale (DCES) (Eddington et al., 2014)) designed to evaluate both pessimistic and optimistic expectancies specific to depression.

Theoretically, CBT improves depression by targeting cognitions via both cognitive and behavioral strategies. As reviewed by Lorenzo-Luaces et al. (2015), there is some evidence that depressive cognitions undergo the greatest improvements as a result of procedures specifically aimed at cognitive change such as those used in CBT. Although cognitive change can also occur in pharmacotherapy (see Garratt et al., 2007, for review), such change likely stems from other mechanisms (see Lorenzo-Luaces et al., 2015); the mechanisms of action of antidepressants have largely implicated neurobiological targets (Rot et al., 2009). As such, cognitive changes in pharmacotherapy may represent “downstream effects”, or may be a reflection of symptom change itself (DeRubeis et al., 1990). Based on the conceptualization of outcome expectancy as a cognitive variable, we hypothesize that this variable will undergo greater change over treatment in the CBT arm relative to pharmacotherapy. We also hypothesize that the association between outcome expectancy during treatment and symptom improvement will be stronger in the CBT arm relative to pharmacotherapy treatment. This hypothesis is first supported by the centrality of positive expectations to theories of depression improvement in psychotherapy (Frank, 1961). Moreover, there is some empirical support that cognitive change is causally linked to symptom change within the context of therapies utilizing cognitive-specific procedures (i.e., CBT) and less so in therapies using non-cognitive strategies (DeRubeis et al., 1990, Evans et al., 2013, Lorenzo-Luaces et al., 2015). Due to limited empirical evidence, no specific hypotheses were made with respect to the differing role of optimistic versus pessimistic expectancy.