Does Personalised Treatment Matching Improve Outcomes in Psychological Therapy?

Systematic Literature Review - Abstract
Objectives: The review aimed to examine whether personalised treatment is associated with improved mental health outcomes relative to standardised treatment, and to investigate the effectiveness of different approaches to personalisation in psychological therapy.
Methods: This was a systematic review and meta-analysis of randomized controlled trials that compared the outcomes of personalised treatment with standardised treatment and other control groups. Studies were identified through Scopus, PsychINFO and Web of Science. Subgroup analyses were applied to investigate sources of effect size heterogeneity. The review protocol was pre-registered in the Open Science Framework.
Results: Sixteen studies (N = 7003) met inclusion criteria for the review, eight of which (N = 4634) provided sufficient data for inclusion in the primary meta-analysis. A risk of bias assessment indicated that fifteen of the sixteen studies had some concerns or high risk of bias, but there was no significant evidence of publication bias. A small, statistically significant effect size was found in favour of personalised treatment relative to standardised treatment (was d = 0.21 [95% CI = 0.02, 0.41], p = 0.031). When studies with a high risk of bias were removed, this effect size was smaller but remained statistically significant (d = 0.12 [95% CI = 0.06, 0.19], p <0.001).
Conclusion:. The systematic literature review and meta-analysis indicates that personalisation is an effective strategy for improving overall outcomes from psychological therapy.

Empirical Project - Abstract
Objectives: The study aimed to identify specific symptom networks in depression and to evaluate whether these symptom networks are associated with differential response to Cognitive Behaviour Therapy (CBT) and Counselling for Depression (CfD).
Methods: A supervised machine learning method was applied in a routine Improving Access to Psychological Therapies (IAPT) dataset (N = 6363) to identify combinations of symptoms observed at first treatment session predictive of post-treatment outcome in CBT and CfD. This resulted in the development of a Personalised Advantage Index (PAI) model, which was subsequently validated in a statistically independent hold-out dataset (N = 6341).
Results: Distinct symptom networks were identified as important in the prediction of depression outcomes in CBT and CfD. Each symptom network was found to predict the probability of reliable and clinically significant improvement (RCSI) more accurately than chance when applied to the validation dataset (CBT network AUC = 0.56; CfD network AUC = 0.60). Although differential responders assigned to their model-indicated treatment had higher rates of RCSI (42.5%) compared to those assigned to their suboptimal treatment (39.7%), these differences were not statistically significant (Odds Ratio = 1.13 [95% CI = 0.91, 1.34], p = .273).
Conclusion: Therefore, our findings suggest that symptom networks are not sufficient for supporting treatment matching recommendations in depression.