Modern Approaches to the Interpretation of the Results of Large Randomized Clinical Trials

Modern approaches to interpreting the results of randomized trials, which can reduce the distortion of information when they are presented to a wide range of doctors, are discussed in the article. Data on the role of taking into account multiple comparisons in the planning and analysis of large rand...

Full description

Saved in:
Bibliographic Details
Main Author: S. R. Gilyarevsky
Format: Article
Language:English
Published: Столичная издательская компания 2022-07-01
Series:Рациональная фармакотерапия в кардиологии
Subjects:
Online Access:https://www.rpcardio.online/jour/article/view/2754
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Modern approaches to interpreting the results of randomized trials, which can reduce the distortion of information when they are presented to a wide range of doctors, are discussed in the article. Data on the role of taking into account multiple comparisons in the planning and analysis of large randomized clinical trials are given. The validity and approaches to using a hierarchical approach to assessing the statistical significance of indicators (“endpoints”) in cases where, in accordance with a pre-adopted study protocol, several indicators are supposed to be evaluated are considered in detail. Approaches to interpreting the p value as well as 95% confidence intervals are considered. Particular attention is paid to the interpretation of the components of the main combined indicator, since when interpreting just such data, the presentation of the results of the study and the manipulation of doctors' opinions may be distorted. This is especially true for data on mortality and mortality from complications of cardiovascular diseases. Modern approaches to the analysis of research results using methods other than the usual Kaplan-Meier analysis, in particular, the method based on the calculation of the "win ratio" are discussed. Data obtained in modern clinical studies are given as examples.
ISSN:1819-6446
2225-3653