Table 2.

Examples of worst-case scenarios for statistical analysis in renal denervation trials

Model	Examples of potential statistical errors
ANCOVA	• When the baseline blood pressure (covariate) shows no meaningful variation among the groups, rendering its inclusion unnecessary. • When the sample size is very small, making it difficult to meet the assumptions or obtain reliable estimates. • When the assumption of linearity between the covariates and the dependent variable is severely violated.
T-test	• When the data for the control and renal denervation groups have significant deviations from normality, and transformations do not resolve the issue. • When the sample size is extremely small, t-tests become less reliable and powerful with very few observations. • When there are multiple groups to compare, a t-test is only suitable for comparing two groups.
Fisher’s Model	• When there are very few categorical predictors or when the majority of cells have zero or minimal observations, leading to sparse data. • When the sample size is extremely small relative to the number of predictors, as overfitting may occur. • When the assumptions of the model, such as normality and homoscedasticity, are severely violated.