An Approach to Integrated Safety Analyses from Clinical Studies
- PDF / 10,384,574 Bytes
- 9 Pages / 612 x 792 pts (letter) Page_size
- 74 Downloads / 236 Views
Grrd K. Rosrnkranz, PhD Novartis Phanna AC. Novartis Campus, Basel, Switzerland
Kry Words Integrated safety analyses; Generalized linear mixed effects models; Shrinkage Corrrspondrncr Addrrss Cerd K. Rosenlnanz. PhD, Novartis Phanna AG, Novartis Campus,
CH-4056 Basel. Switzerland (email: gerd.rosenkranz @ novartis.com).
The analysis of safety data of clinical studies poses specific problems that are related to a multiplicity of usually not explicitly specified hypotheses. Furthermore, individual studies are generally not capable of providing a comprehensive picture of the safety profile of a drug.
INTRODUCTION The analysis of safety data from clinical trials poses specific problems that are different from those present in the analysis of efficacy data. In contrast to efficacy studies, which investigate specifically defined objectives, specific hypotheses exist only rarely in the context of safety. When no specific safety concern is suspected (eg, from animal toxicity data or a known class effect for the compound), safety data are collected with the aim to screen them for potential signals. This raises questions of multiplicity, since the screening of hundreds of variables can result in a number of false positive findings. On the other hand, if multiplicity adjustments are implemented to limit, for example, the family-wise type I error rate, there is an increased risk of overlooking potential signals. Safety information from individual clinical trials will always be limited, and rare events cannot be detected because the sample size is too small. The ICH E l guideline (1) requests that for chronic treatment of non-life-threatening diseases, at least 100 patients have to be exposed for at least a year before the drug can be registered. When no event is observed in a 1-year exposure period, this number of patients can provide reasonable assurance that the true cumulative 1-year incidence is not greater than 3%. Safety information arises from all clinical trials conducted during drug development. Unfortunately, the information from these studies is often only combined before registration in
Therefore, data from different studies have to be combined to reach conclusions. Based on a generalized linear mixed effects model, 1 present an approach that can address both aspects under plausible assumptions. The method is applied to a real example.
the clinical summary of safety,that is, late in the drug development process. This article describes methodology that helps us cope with multiplicity and enables us to combine safety information from several clinical studies. I focus on the analysis of adverse events, that is, "any untoward medical occurrence in a patient or clinical investigation subject administered a pharmaceutical product and which does not necessarily have to have a causal relationship with this treatment" (2). Approaches to analyzing laboratory measurements, the other major building block of safety data, have been studied elsewhere (eg, 3-5) and are not considered here. The method described in this articl
Data Loading...
