Reporting guidelines for pharmacoepidemiological studies are urgently needed
BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g5511 (Published 17 September 2014) Cite this as: BMJ 2014;349:g5511- Mahyar Etminan, drug safety researcher1
- metminan{at}popi.ubc.ca
Tan and colleagues discuss factors that contribute to inconsistent information on drug adverse reactions but do not mention the lack of reporting guidelines for pharmacoepidemiological studies.1
Most medical journals require authors to complete checklists that are shortened versions of reporting guidelines, such as CONSORT (randomised trials), PRIMSA (systematic reviews), and STROBE (observational studies). These guidelines are intended to provide more transparency in the methodology and design of clinical studies in the hope of improving quality. Currently, pharmacoepidemiological studies fall under the category of observational studies; authors submitting such studies to medical journals are required to complete the STROBE checklist.
Unfortunately, STROBE is a global checklist for observational studies and is not designed to identify methodological and analytical nuances that are unique to pharmacoepidemiological studies, which usually analyse millions of person years of data through large health linked databases. This complexity is augmented by the need to quantify one or more time varying exposures (prescription drugs) over time. These challenges have given rise to biases that are unique to pharmacoepidemiological studies, much like immortal time bias in cohort studies or time window bias in case-control studies.2 To date, the equator network (www.equator-network.org/about-us/who-we-are/)—the group responsible for the creation of CONSORT, PRISMA, and STROBE—has not published a checklist for the reporting of pharmacoepidemiological studies.
Reporting guidelines specifically designed for pharmacoepidemiological studies are urgently needed. Absence of such guidelines will hamper appropriate evaluation of such studies, which may lead to poor quality data with inconsistent findings.
Notes
Cite this as: BMJ 2014;348:g5511
Footnotes
Competing interests: None declared.