Data of Systematic Review and Meta-analysis of PTSD Symptom Trajectories Following Traumatic Events for the FORAS Project

Project

Background and Preparation The FORAS project established a rigorous methodological foundation for AI-aided systematic reviews before conducting the main analyses. Funded by the Dutch Research Council (Grant No. 406.22.GO.048) and preregistered on PROSPERO (CRD42023494027). Prepared through preliminary methodological studies evaluating active learning, stopping heuristics, and data storage (e.g., Boetje & van de Schoot, 2024; Byrne et al., 2024).

Supported by open-source dataset composition scripts and LLM classification tools (Bron & van de Schoot, 2025). Dataset Lineage 2017-2023 Originated from an initial 2017 systematic search to obtain informative priors for a Bayesian Growth Mixture Modeling PTSD dataset (Winter & van de Schoot, 2017). The resulting dataset was subsequently relabeled to match FORAS inclusion criteria (Van De Schoot et al., 2024).

The database was subsequently updated via direct replication of previous search strings used in the abovementioned search, comprehensive database expansions, and snowballing techniques (Coimbra et al., 2023a; Coimbra et al., 2023b). The search was subsequently enhanced using vectorized OpenAlex data and semantic search scripts to locate elusive papers (Lombaers & van de Schoot, 2024). Publication "The hunt for the last relevant paper: blending the best of humans and AI": Accepted article available online (10.1080/20008066.2025.2546214).

This initial paper details the methodology of blending human screening with AI for the systematic review. Dataset (up to 2023): Hosted on DataverseNL containing the labels and codebooks (doi:10.34894/CRE6ZC). Analysis Scripts: Accessible via Zenodo for full reproducibility (Rens van de Schoot, 2025).

PTSD Trajectories following Potentially Traumatic Events Meta-Analysis (Updated Dataset up to 2025) This updated dataset contains the final dataset as used in the systematic review and meta-analysis of prototypical PTSD symptom trajectories following potential trauma exposure. To arrive at this updated dataset, the 2023 search was updated in 2025 and study-specific in- and exclusion criteria were applied. Publication: "PTSD Symptom Trajectories in the Wake of Potentially Traumatic Events: A Systematic Review and Meta-Analysis of 20 Years of Research", accepted for publication in Clinical Psychological Science.

Preprint: The most recent draft is available on OSF (osf.io/preprints/osf/fkjb2_v3). Dataset (up to 2025): Hosted on this DataverseNL page, with screening conducted according to the Screening Logbook; including the final FT_Screening_and_Selection.csv log (doi:10.34894/XVYG52; Coimbra et al., 2025). Study Quality assessment GRoLTS Scores: the scripts regarding the LLM-generated GRoLTS scores (GROLTS_scores.csv) can be found here:GROLTS scores GitHub.

See also the related preprint: doi.org/10.48550/arXiv.2608.20385 Statistical Analyses: All statistical analysis scripts related to the current manuscript and dataset are hosted on GitHub. Updated Dataset 2025 Repository Files The dataset contains the following files: FT_Screening_and_Selection.csv – literature screening outcomes, with screening conducted according to the Screening Logbook; includes data for the following variables: From which search phase the paper originated (1 = Yes, 0 = No; categorical): Inclusion_via_Synergy; Search_2023; Update_search_2025 Inclusion status in "The Hunt for the Last Relevant Paper: Blending the best of humans and AI" (1 = Included, 0 = Excluded; categorical): FT_final_label_paper_hunt-for-the-last-relevant-paper (Original search update from 2016-2023, DataverseNL repository: "Updated PTSS dataset for the FORAS project") If applicable, exclusion according to exclusion criteria: Clinical_study (1 = Yes); observational_study (1 = Yes, 0 = No; categorical); non_daily_measures (1 = Yes, 0 = No; categorical); no_joint_trajectory_or_PTSDcluster (1 = Yes, 0 = No; categorical); prospective_study (1 = Yes); conditional_model (1 = Yes); poor_quality (1 = Yes) Duplicate samples: cohort_duplicate (0 = not a duplicate; samples with the same number use the same sample for trajectory analyses) Final inclusion status of papers in the current dataset: FT_final_label_paper_prevalences (1 = Included, 0 = Excluded; categorical) PRISMA_Flowchart.pdf – flowchart detailing detailing identification of studies via databases, screening, and final inclusions Dealing_with_Duplicate_or_Overlapping_Samples.pdf – document detailing decisions and rationale on the selection of which paper to include when multiple papers used the same samples for trajectory analyses GROLTS_scores.csv – final quality score of each included sample (for details on the calculation, see the corresponding Github repository) Data_extraction_general_data.csv – extracted data of general descriptives and background variables / moderators of all included samples for the meta-analysis of PTSS trajectory prevalences Data_extraction_model_results.csv – extracted data of model results and methodological details of all included samples for the meta-analysis of PTSS trajectory prevalences Codebook_general_data.pdf – description of all columns of the Data_extraction_general_data.csv dataset Codebook_model_results.pdf – description of all columns of the Data_extraction_model_results.csv dataset