Recently, FDA provided guidance documents on “Best Practices for Conducting and Reporting Pharmacoepidemiologic Safety Studies Using Electronic Healthcare Data Sets” The guidance was limited to pharmacoepidemiology safety studies using electronic healthcare data sets and Medical Devices were not within the scope of this guidance.
Though FDA in its guidance has not recommended any particular study design, FDA strongly discourage the use of “one size fits all design.” In “one size fits all design”, the investigator does not address the study question of interest while designing the study. There were suggestions from expertise like Biotechnology Health Organization (BIO) to highlight the importance of conducting preliminary pilot studies and data assessment. This might help the sponsor to understand the study population, determine the sample size, identify confounding variable and define the study outcomes before finalizing the protocol.
To identify the confounders in the study, FDA have stated to use particular methods that addresses the confounding variable. In general, if a confounding variable is not identified, it might cause damage to the internal validity of the experiment. BIO in its response has commented that FDA should clarify on whether validation on confounding variables should also be conducted.
Additionally, the study design should include the dosage information, the GAP therapy involved, and the data source used in the studies.
Regarding analysis plan, FDA mentions to address pre-specify plan while performing preliminary analysis. Also, other methods like the sub groups method and ad hoc analysis, if used should be described. International Society for Pharmacoepidemiology (IPSE) has specifically commented on ad hoc analysis stating that the FDA has not acknowledged the recent controversies surrounding the ad hoc analysis. Also BIO comments that the primary and secondary analysis should be clearly stated. Additionally, FDA has provided guidance on the QA/QC procedures in the construction of analysis data set and analysis of data.
For additional information and guidances on pharmacoepidemiology studies you can get information from ISPE Guidelines on Good Pharmacoepidemiologic Practices and The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidance. There are also ISPE guidelines for Quality Conduct in DatabaseResearch and ISPOR guidance on conducting and reviewing database studies.
Image reference
