Presentation Of Population Analyses For Regulatory Submissions

Janet R. Wade

Medical Products Agency, Sweden

The description of the implementation of the population approach for data analysis, either in a protocol or in a final report is important from a regulatory perspective. A poor description may result in the analysis being disregarded. This presentation will focus on several items that should be included in protocols and reports of population analyses so that the regulatory reviewer can properly assess the results.

The Medical Products Agency (MPA) requires that all protocols for clinical trials that will be performed within Sweden should be submitted to the MPA for prior assessment. In a four month period five protocols have been submitted in which it was stated that population analysis was to be performed. The detail in all of these protocols was sparse to say the least, usually just a one line comment that a population analysis would be performed (or attempted) and possibly mentioning the variables to be analysed. Without more detail it is not possible to assess if the stated aim is obtainable or not (Aarons et al, 1996).

Within the same four month period seven reports of population analyses have been assessed, four of which were received via the central European submission procedure, the remaining three were part of national submissions.

The reports have varied in quality from poor to excellent. The poor reports have been categorised as such either because the presentation of the analysis was deficient in a number of aspects and proper assessment could not be made, or, that the presentation was adequate but that the final model was unacceptable, i.e., did not properly describe the data in the assessors opinion.

In order to properly assess a report the following items could be included. A short description of an overview of the analysis is very useful to let the regulatory assessor get a quick feel for what is often a very complex analysis. A run record, which doesnít have to be very detailed, is an informative mechanism for relaying what aspects have been tested during the model building procedure. One of the most important things to include are adequate goodness of fit plots. These are fundamental to assess whether the model is adequately describing the data. Finally , if the population model is to be used to answer a specific question, then confidence intervals for the effect (or lack of it) are desirable.

Reference
Aarons L, Balant LP, Mentre F, Morselli PL, Rowland M, Steimer J-L and Vozeh S. Practical experience and issues in designing and performing population pharmacokinetic/ pharmacodynamic studies. Eur. J. Clin. Pharmacol., 49, 251-254, 1996.

Reference: PAGE 5 (1996) Abstr 556 [www.page-meeting.org/?abstract=556]

Poster: oral presentation