Int J Oral Maxillofac Implants 32 (2017), No. 6 21. Nov. 2017
Purpose: Research waste occurs when research is ignored, cannot be found, cannot be used, or is unintentionally repeated. This article aims to investigate how dental survival analyses were indexed and reported, and to discuss whether errors in indexing and writing articles are affecting identification and use of survival articles, contributing to research waste.
Materials and Methods: Articles reporting survival of dental prostheses in humans (also known as time-to-event) were identified by searching 50 dental journals that had the highest Impact Factor in 2008. These journals were hand searched twice (Kappa 0.92), and the articles were assessed by two independent reviewers (Kappa 0.86) to identify dental survival articles ("case" articles, n = 95), likely false positives (active controls, n = 91), and all other true negative articles (passive controls, n = 6,769). This means that the study used a case:control method. Once identified, the different groups of articles were assessed and compared. Allocation of medical subject headings (MeSH) by MEDLINE indexers that related to survival was sought, use of words by authors in the abstract and title that related to survival was identified, and use of words and figures by authors that related to survival in the articles themselves was also sought. Differences were assessed with chi-square and Fisher's Exact statistics. Reporting quality was also assessed. The results were reviewed to discuss their potential impact on research waste.
Results: Allocation of survival-related MeSH index terms across the three article groups was inconsistent and inaccurate. Statistical MeSH had not been allocated to 30% of the dental survival "case" articles and had been incorrectly allocated to 15% of active controls. Additionally, information reported by authors in titles and abstracts varied, with only two-thirds of survival "case" articles mentioning survival "statistics" in the abstract. In the articles themselves, time-to-event statistical methods, survival curves, and life tables were poorly reported or constructed. Overall, the low quality of indexing by indexers and reporting by authors means that these articles will not be readily identifiable through electronic searches, and, even if they are found, the poor reporting quality makes it unnecessarily difficult for readers to understand and use them.
Conclusion: There are substantial problems with the reporting of time-to-event analyses in the dental literature. These problems will adversely impact how these articles can be found and used, thereby contributing to research waste. Changes are needed in the way that authors report these studies and the way indexers classify them.
Keywords: information storage and retrieval, medical subject headings, quality of reporting, research waste, survival analyses