Evaluation of a standardized posttraumatic stress disorder treatment framework in routine mental health care: Effectiveness and predictors of treatment outcome in a consecutive sample

Abstract

The primary aim of the present study was to evaluate the effectiveness of standardized care package (CP) treatment for posttraumatic stress disorder (PTSD) in a Danish sample of adult psychiatric outpatients (N = 948). Secondary aims were to identify baseline predictors of treatment outcomes and investigate betweengroup differences in outcomes with regard to sex and treatment modality (i.e., group vs. individual therapy). The naturalistic, nonrandomized study followed a pre-post design. Patient data from five psychiatric outpatient clinics were collected between March 2011 and November 2017. Data were drawn from self-report

A prognostic index to predict symptom and functional outcomes of a coached, web-based intervention for trauma-exposed veterans.

“…To mitigate the possible impact of sample size on our findings, we selected a sample-appropriate ML algorithm (i.e., elastic net regularization), conducted nested cross-validation, and examined a relatively small set of predictors and outcomes with good reliability (i.e., generally high internal consistency for self-report measures). Notably, our findings are consistent with past studies of other PTSD interventions using samples as large as 948 participants (Scharff et al, 2022). Although sample size is only one factor influencing ML findings among others (e.g., measurement quality; Jacobucci & Grimm, 2020; McNamara et al, 2022), it was likely still a limitation of our study.…”

“…In general, however, ML models provided limited additional utility in our context of predicting prognosis for veterans engaging in webSTAIR across the three trials (Bauer et al, 2021; Cloitre et al, 2022; Fletcher et al, 2022). This finding parallels similar predictive studies for PTSD psychotherapy outcomes (Scharff et al, 2022; Stuke et al, 2021) and highlights the importance of comparing ML models to an appropriate benchmark (Aafjes-van Doorn et al, 2020; Rosenbusch et al, 2021). We emphasize that our findings are preliminary and warrant replication, as this study was the first to use ML to develop prognostic indices of outcomes in a digital intervention for trauma survivors.…”

“…For example, examining a sample of women military service members who received either prolonged exposure or present-centered therapy, Stirman et al (2021) identified the following as predicting higher posttreatment PTSD severity: higher baseline PTSD severity, lower perceived treatment credibility, poorer mental and physical functioning, and history of military sexual trauma. However, when testing the utility of a prognostic index, other researchers have found that the ML models do not outperform simpler models such as traditional linear models often with fewer predictors (Scharff et al, 2022; Stuke et al, 2021). For example, in a large Danish sample of adults who received a standardized PTSD care package involving psychotherapy, Scharff et al (2022) found that the additional prognostic predictors identified by the ML model (e.g., occupational functioning, social functioning, hopelessness, and well-being) did not explain a higher percentage of variance in PTSD outcome beyond baseline PTSD severity.…”

“…However, when testing the utility of a prognostic index, other researchers have found that the ML models do not outperform simpler models such as traditional linear models often with fewer predictors (Scharff et al, 2022; Stuke et al, 2021). For example, in a large Danish sample of adults who received a standardized PTSD care package involving psychotherapy, Scharff et al (2022) found that the additional prognostic predictors identified by the ML model (e.g., occupational functioning, social functioning, hopelessness, and well-being) did not explain a higher percentage of variance in PTSD outcome beyond baseline PTSD severity. Additional research is needed to understand the contexts for whether and how prognostic indices are useful in the prediction of outcomes for trauma-related interventions.…”

Scalability, test–retest reliability and validity of the Brief INSPIRE-O measure of personal recovery in psychiatric services

“…The current paper utilized data from the Mental Health Services, Capital Region of Denmark (MHS-CR) which is the largest mental health service in Denmark, covering a catchment area of 1.85 million people with nine psychiatric treatment sites. In Denmark, psychiatric treatment in the secondary health sector is organized in treatment packages that specify relevant evidence-based treatments for specific diagnoses ( 27 , 28 ). To monitor treatment effects, the MHS-CR developed an Internet-based monitoring system (IMS) that collects data pre- and post-treatment for all patients with non-psychotic disorders receiving treatment in a treatment package.…”