The progressive substitution of hazard ratios for relative risks in biomedical research
In biomedical research, epidemiological measures of risk/effect [effect sizes (ESs)] are predominantly derived from risk (or rate) ratios (RRs), odds ratios (ORs), or hazard ratios (HRs). Using the whole PubMed database, we detailed in this paper a phenomenon not yet described: a massive trend for HRs to be globally used as substitutes for RRs. All PubMed citations were bulk-downloaded and a data mining process led to a comprehensive database of 1,071,584 ES values. The proportion of abstracts containing only HR has exploded since the 2000s, while we observe an inverse trend for abstracts containing only RR. The annual number of abstracts with HR exceeded the number of abstracts with RR for 2006. The average annual growth rate of the number of abstracts with RR only and HR only between 1980 and 2017 was 15.1% and 32.6%, respectively. Training on HRs has become essential in the statistical education of physicians. Since the interpretation of HRs is slightly more difficult than that of ORs or RRs, it is also important to improve day-to-day communication with patients regarding this quite complex entity.
KeywordsData mining Risk Epidemiology Proportional hazards models
Mathematical Subject ClassificationI10 I12
The authors thank Ms. Susan Becker for her assistance with English language editing.
This work was supported by Toulouse University Hospital (CHU de Toulouse), Toulouse University (Université Paul Sabatier), the Midi-Pyrenees region, the research platform of the Toulouse Dental Faculty (PLTRO), and the French National Research Agency (Agence Nationale de la Recherche—ANR—http://dx.doi.org/10.13039/501100001665) under Grant ANR-16-CE18–0019-01.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no competing interests.
- Mantel, N., & Haenszel, W. (1959). Statistical aspects of the analysis of data from retrospective studies of disease. Journal of the National Cancer Institute, 22(4), 719–748.Google Scholar
- Meaney, C., Moineddin, R., Voruganti, T., O’Brien, M. A., Krueger, P., & Sullivan, F. (2016). Text mining describes the use of statistical and epidemiological methods in published medical research. Journal of Clinical Epidemiology, 74, 124–132. https://doi.org/10.1016/j.jclinepi.2015.10.020.CrossRefGoogle Scholar
- Monsarrat, P., & Vergnes, J.-N. (2018b). 2018 database of effect sizes (ES). https://doi.org/10.6084/m9.figshare.7066397.v3.