Wouldn't any statistically significant improvement become clinically significant when applied at a large enough scale, e.g. in global clinical practice?