On the debate over methods for estimating the clinically important difference