Participant factors as correlates of patients’ psychotherapy outcome expectation: A meta-analytic and box-count review

ABSTRACT

Objective

More positive pre- or early therapy patient outcome expectation (OE) has consistently correlated with better treatment outcomes. Thus, it is important to identify factors that contribute to patients’ OE, which can inform therapist responsivity to such risk or facilitative markers. With growing research on OE correlates—centered primarily on patient characteristics/treatment factors and, to a lesser extent, therapist factors—a comprehensive synthesis is warranted to elucidate replicated and mixed associations and stimulate further research. Accordingly, we set a pragmatic cutoff of k ≥ 5 for meaningful empirical aggregation of participant factor-OE associations; otherwise, we conducted box counts.

Method

We searched for articles published through March 2022 that included a clinical sample, a measure of patient’s pre- or early treatment OE, and an explicit test of the factor-OE association.

Results

Patient problem severity, problem chronicity, education, age, and quality of life were meta-analyzed. Greater severity correlated with lower/less optimistic OE (r = −0.13, p < .001) and higher QOL correlated with higher/more optimistic OE (r = 0.18, p < .001). Box counts revealed that few variables had consistent associations with OE.

Conclusions

Some factors can help forecast patient OE, though additional research is needed to enhance confidence and clinical meaning.

Keywords:

• patient outcome expectation
• participant factors
• meta-analytic review
• box-count review

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 An asterisk indicates any derivation with this root.

2 Although the selected cut-off of at least 5 studies was a pragmatic decision, we intentionally selected this criterion because it was consistent with the one used in another recent OE meta-analysis (Constantino et al., 2021b).

3 Because this approach (of setting effect sizes reported to be null but without specific statistical data) may be overly conservative, we conducted sensitivity analyses in which we instead dropped effect sizes that were reported to be null but that were missing corresponding effect size data. In all cases, the pattern and significance of the omnibus meta-analytic results remained the same. Contact the first author to obtain the full results of these analyses.

4 Note that although the total number of references included in either the meta-analyses or the box count was 34 (reporting on 38 independent samples), some references were only included in the meta-analyses or in the box count. Specifically, four references were included in the meta-analyses but not in the box count review.

5 Note that one other reference (Warden et al., 2010), which included 3 independent samples, examined one additional clinical characteristic (i.e., patients’ scores on an alcohol screening test) as a correlate of OE but did not describe the measure or the scoring procedures for it. Because we could not be sure as to the precise meaning of these associations, we excluded them from the box count review. However, it may be worth noting that the alcohol screening test scores had a positive and significant association with OE in one sample and a null association in the other two.

6 Note that one of these studies (i.e., Westra et al., 2011) reported a moderately sized positive correlation between OE and treatment credibility (r = .45), but the authors did not explicitly state whether this association was statistically significant. However, because the moderately sized correlation suggested that this relation was clinically meaningful, we counted this as a significant credibility-OE association.