Different methods to analyze clinical experiments with multiple endpoints: a comparison on real data