Methods for analysing cardiovascular studies with repeated measures

Area of Science:

• Cardiovascular research
• Biostatistics
• Clinical trial analysis

Background:

• Repeated measurements within subjects are more similar than measurements between subjects.
• Standard analyses of repeated data can underestimate treatment effects.

Purpose of the Study:

• To review statistical methods suitable for analyzing cardiovascular studies with repeated measures.
• To highlight the importance of accounting for the nature of repeated measures in data analysis.

Main Methods:

• For between-subjects comparisons: summary measures (e.g., area under the curve, maximal values) and random-effects mixed-linear models.
• For within-subjects comparisons: repeated-measures ANOVAs, allowing inclusion of subgroup factors like gender and age.
• For non-Gaussian data: Wilcoxon's and Friedman's tests; for binary data with two observations: McNemar's tests.

Main Results:

• Random-effects mixed-linear models offer better precision for between-subjects comparisons than simple summary measures.
• Repeated-measures ANOVAs are suitable for within-subjects comparisons and can incorporate subgroup analyses.
• Existing methods for non-Gaussian and binary repeated measures have limitations, especially for more than two observations.

Conclusions:

• Appropriate statistical methods are crucial for accurate analysis of cardiovascular studies with repeated measures.
• Random-effects models and repeated-measures ANOVAs are recommended for between- and within-subjects analyses, respectively.
• Further development of standard methods is needed for complex repeated measures, particularly for non-Gaussian and binary data.