Study design elements in preclinical animal experiments

posted on 22.08.2018 by Annette O'Connor, Sarah C. Totton, Jonah N. Cullen, Mahmood Ramezani, Vijay Kalivarapu, Chaohui Yuan, Stephen B. Gilbert
This study aimed to assess the design elements used in biomedical preclinical experiments

Study population sources
Eligible manuscripts were selected from two topic areas that were chosen for pragmatic reasons: ready access and expected variation. The first group of studies came from The Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies (CAMARADES) group at the University of Edinburgh. The CAMARADES group has expertise in systematic reviews and meta-analyses in animal studies and, in particular, traumatic brain injury (e.g., stroke) and neurological disease. CAMARADES provided a citation dataset of 500 preclinical trials previously included in systematic reviews of stroke. These 500 references (out of 1152 publications, all of which had been screened for MCAO (stroke model)), were randomly selected using the random number generator [1].
The second group of studies was selected from systematic reviews that had been conducted in the field of environmental toxicology. We selected 80 eligible studies from the three most recently conducted reviews listed on the National Toxicology Program website (https://ntp.niehs.nih.gov/pubhealth/hat/noms/index.html) (fluoride, folic acid, and perfluorooctanoic acid (PFOA)/perfluorooctane sulfonate (PFOS)) at the start of the project (Dec 2016), that included animal studies [2-4]. Because these reviews provided an insufficient number of studies, we also selected studies from an EFSA (European Food Safety Authority) mercury review, purposefully chosen because it was recent, included animal studies, and related to toxicology [5].
Study selection
Of the 500 CAMRADES references screened for eligibility, 26 references were excluded because the full text was not available in English (25 were in Chinese, 1 was in Russian), 2 were not primary research, 121 described in vitro and/or ex vivo experiments, and 51 did not assess an intervention. Of the remaining 300 references (all of which were in vivo primary research animal intervention studies in English), in 8 studies the number of experiments conducted was unclear, and another 89 references described more than one experiment, leaving 203 references eligible for evaluation.
During data extraction, an additional 6 studies had to be excluded because the corresponding Portable Document Format files could not be annotated within the AFLEX interface (the tool used to extract the data) because of problems with the files (file size too big, document was a scanned photocopy so that text could not be highlighted). In each case, the study was excluded and the next eligible study, as determined by the random number sequence generator, was selected for data extraction.

1. R Core Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. 2017.
2. National Toxicology Program: Monograph on Immunotoxicity Associated with Exposure to Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). Research Triangle Park, NC: National Toxicology Program. 2016.
3. National Toxicology Program: NTP Monograph: Identifying Research Needs for Assessing Safe Use of High Intakes of Folic Acid. 2015.
4. National Toxicology Program: Systematic Literature Review on the Effects of Fluoride on Learning and Memory in Animal Studies. NTP Research Report 1. Research Triangle Park, NC: National Toxicology Program 2016.
5. Hassauer M, Kaiser E, Schneider K, Schuhmacher-Wolz U: Collate the literature on toxicity data on mercury in experimental animals and humans, Part I. Supporting Publications 2012:EN-297. [360 pp.]. . 2012.

This folder contains four files
1) Digital depository Design elements(raw).xls - an excel spreadsheet with two worksheets. The 1st worksheet contains the extracted data ( See publication) and the 2nd worksheet contains the legend. This is the raw dataset extracted from the PDF to text tool
2) Digital depository Design elements(corrected).xls - an excel spreadsheet with two worksheets. The 1st worksheet contains the extracted data ( See publication) and the 2nd worksheet contains the legend and data dictionary. When PDF files are extracted to text - some characters are usual and these have been replaced or removed in this dataset.
3) Code Book Digital Depository Design Element dataset (corrected). This contains the codebook used to convert the Digital depository Design elements(raw) to Digital depository Design elements(corrected)
4) Citation.txt - This file contains the reference information in RIS format. The identifier that links to the design elements dataset is the LB field. This field matches the field in the File Name variable in the Digital depository Design elements.xls. This can be used to find the reference for each extracted text. Open with a text file reader.

change log: v3 updated to include link to paper.


Iowa State University Presidential Initiative for Data Driven Science