Clinical significance versus statistical significance