Abstract 1332
Background
Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) statement was published in 2015 to enhance the quality of reporting. Limited data exist regarding adherence to this guideline in publications of oncology journals. The purpose of this study was to evaluate the abstracts in oncology journals from the view of the adherence of TRIPOD statement.
Methods
We searched the top 100 impact factor oncology journals by the Journal Citation Reports 2013 via PubMed. The search periods were from 2012 to 2016. We applied search filter developed by Mallet 2010. Two independent authors reviewed titles, abstracts, and fulltexts if necessary to select. One review author extracted data from all included studies, another review author confirmed the results. We estimated Clopper-Pearson confidence intervals for each proportion. We used chi-squared test for the comparison. P < 0.05 was considered significant.
Results
A total of 12044 abstracts were searched. We randomly sampled 250 abstracts in each year. From 1250 abstracts we found 316 prediction model articles (proportion: 25%). We randomly selected 200 full articles and evaluated the reporting quality focusing on the abstracts. Higher impact factor ranking (≥50) was significantly associated with the non-reporting of study designs (risk ratio (RR): 0.79; 95%CI 0.66 to 0.95) and settings (RR: 0.68; 0.46 to 1.00). Published year (≥2016) was not associated with any reporting characteristics.
Conclusions
Our findings showed that some methodological aspects, especially in study designs, settings and statistics, were poorly reported in abstracts. Given the importance of reporting quality to use prediction models, authors should be encouraged to adhere to the TRIPOD statement.
Editorial acknowledgement
Clinical trial identification
CRD42015020565.
Legal entity responsible for the study
Yuki Kataoka.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.