TRACE: Meta-analysis of learning and memory in PTSD

Step 1: Prepare data

Merge datasets

• script: Merge data S1, S2, S3 to one file with prepare_merge.rmd.
• input: search 1 TRACE data_collection_search1.xlsx, search 2 TRACE data_collection_search2.xlsx, search 3 TRACE_data_collection_search3.xlsx
• output:TRACEmerged.RDS

Recode data

• script: Recode variables and add method codes with prepare_recode.rmd.
• input:TRACEmerged.RDS, TRACE_method_codes.xlsx
• output: TRACErecoded.RDS

Process QA data

• script: process QA data with prepare_QA.rmd
• input: TRACE_RoBQA_data.xlsx
• output: TRACE_QA_animal.RData, TRACE_QA_human.RData, and RoB.jpeg (optional: RoB_clinical.jpeg, RoB_preclinical.jpeg)

count data in meta-analyes: readRDS("processed_data/clinical.data.metaregression.RDS") %>%

• distinct(PMID) ->clinical.meta.papers

nrow(clinical.meta.papers) [1] 92 readRDS("processed_data/preclinical.data.metaregression.RDS") %>%

• distinct(PMID)->preclinical.meta.papers

nrow(preclinical.meta.papers) [1] 182

Calculate effect sizes

• script: prepare data for analysis in prepare_effect_size_QA.rmd
• input: TRACErecoded.RDS, TRACE_QA_animal.RData, and TRACE_QA_human.RData
• output: TRACEprepared.RData (nb n=1647)

Step 2: Meta-regression Valence x Phase

• script: meta-regression Valence x Phase: meta_regression.rmd. This script uses meta_regression_influentials.r to calculate potential influential case and outliers
• input: TRACEprepared.RData
• output: datasets used in analyses: clinical.data.metaregression.RDS, preclinical.data.metaregression.RDS and results (main, diagnostics, sensitivity, graphs) -- clinical: phase_valence_PTSD_clinical.csv, phase_valence_PTSD_clinical.tiff, funnel.colours.clinical.tiff, phase_valence_PTSD_clinical.FSN.csv, influentials.clinical.rds, sens.clinical.infout.csv, sens.mod.A.csv, sens.mod.B.csv, sens.mod.C.csv -- preclinical: phase_valence_PTSD_preclinical.csv, phase_valence_PTSD_preclinical.tiff, funnel.colours.preclinical.tiff, phase_valence_PTSD_preclinical.FSN.csv, influentials.preclinical.rds, sens.preclinical.infout.csv, sens.mod.E.csv, sens.mod.D.csv, sens.mod.F.csv -- figure: PTSD.clinical.preclinical.tiff

-> 8.3.22 checked tot hier (en code gerund.) -> bekijkn of resultaten overeen komen met paper (in tabellen).. indien ja dan 'comments' uit script halen.

Step 3: MetaForest and MetaCART

• script: meta_forest_meta_cart.rmd
• input: TRACEprepared.RData
• output: data.explore.rds (NB: this data is also used for descriptives table;

file='processed_data/clinical.data.explorative.RDS'); file='processed_data/preclinical.data.explorative.RDS') "processed_data/REtree.P.rds") lts/metaCART.preclinical.tiff" /fitted.clinicalMetaForest.RDS")

="results/metaforest_Clinical_convergence.tiff", results/metaforest_Clinical_varImportance.tiff", /important_variables_clinical_metaforest.csv") "results/metaforest_PD_clinical.tiff", e="results/metaforest_PD_clinical_metaCARTfollowup.tiff",

file="processed_data/fitted.preclinicalMetaForest.RDS") ="results/metaforest_Preclinical_convergence.tiff", h e="results/metaforest_Preclinical_varImportance.tiff", ),],"results/important_variables_preclinical_metaforest.csv") ="results/metaforest_PD_preclinical.tiff", ="results/metaforest_PD_preclinical_metaCARTfollowup.tiff", ("results/VarImp.clinical.preclinical.tiff",

Step 4: Study descriptives

• Overview numbers for screening steps (for flowchart) via Flowchart.rmd

  • input: PMID hits & screening search 1, 2 and 3: Review PTSD Cognition AbstractScreening_Overeenstemming EG & MS v15.11.2016.xlsx, pubmed_result_search2_human 6.1.20.txt, pubmed_result_search2_animal 6.1.20.txt, pmid.human.s3.learn.22.5.20.txt, pmid.animal.s3.learn.22.5.20.txt, Checked_TRACE_screening_search3_MS_EG.xlsx
  • some additional code for screening in search 2
    • step 1: identify inconsistencies in search 2. input (initial screening data): TRACE_screening_search2_SH.xlsx & TRACE_screening_search2_MS.xlsx; output (abstract screening inconsistencies): inconsistencies_human_s2.csv & inconsistencies_animal_s2.csv
    • step 2: identify required full text checks in search 2. input (files with abstract screening feedback): inconsistencies_animal_s2_SH.csv & inconsistencies_human_s2_SH.csv"; output (uploaded to OSF): required_full_checks_human_s2.csv & required_full_checks_animal_s2.csv
    • step 3: get results full text check search 2. input (files with full text screening feedback): required_full_checks_animal_s2_MS.csv" & required_full_checks_human_s2_SH.csv"; output (uploaded to OSF): animal_inclusions_s2.csv & human_inclusions_s2.csv
  • input: Data extraction information: TRACE data_collection_search1.xlsx, TRACE data_collection_search2.xlsx, TRACE_data_collection_search3.xlsx
  • Analyses datasets (NB output analyses scripts): TRACEprepared.RData, clinical.data.metaregression.RDS, clinical.data.explorative.RDS", preclinical.data.metaregression.RDS, preclinical.data.explorative.RDS
  • output: processed_data/unique_screened_articles.csv. en flowchart.tiff
  • optional output (used to select papers for QA): Human.PMIDs.QA.S2.csv, Animal.PMIDs.QA.S2.csv. en Human.PMIDs.QA.S3.csv en Animal.PMIDs.QA.S3.csv

• Characteristics of the included studies via Descriptives.rmd (uses data.explore.rds, created in DataDrivenAnalysis.rmd script) [input data<-readRDS("processed_data/data.explore.rds")]