Drug Evaluation Committee Recommendations for Sensitivity Analysis in Observational Studies Practical Edition
Data Science Subcommittee
April 2019
The Japan Pharmaceutical Manufacturers Association Drug Evaluation Committee, Data Science Subcommittee, Task Force 2, 2016, has created a practical version and the Excel Tool ESATJ for quantitative sensitivity analysis, a sequel to the "Recommendations for Sensitivity Analysis in Observational Studies: An Introductory Chapter" published in 2017.
In recent years, the environment for observational research using medical information databases has been improving in Japan, and pharmaceutical companies have begun to use these databases for safety monitoring and other purposes. In addition, the GPSP ordinance has been revised in post-marketing surveillance, and "comparative use studies" using comparison groups and "post-marketing database studies" using medical information databases are also being conducted.
Although the results of observational studies conducted by companies on their own pharmaceutical products, especially those that involve causal inferences from comparisons, can lead to some decisions about the products, the biases and uncertainties associated with observational studies, especially those due to unmeasured causes, can make decision-making difficult.
The quantitative sensitivity analysis presented in this document provides numerical results that correct for biases caused by this lack of information by providing information from some evidence or assumptions. The simplified method requires no special software and can be done with spreadsheet software.
We hope this document will help you evaluate the robustness and uncertainty of the results of observational studies.