Clinical Significance vs Statistical Significance

This is one of the most important lessons in this entire app. Understanding the distinction between statistical and clinical significance protects you from being misled by headlines and p-values. Statistical significance is a mathematical concept. When p < 0.05, it means there is less than a 5% probability that the observed difference occurred by chance alone (assuming the null hypothesis is true). But p-values say nothing about the SIZE or IMPORTANCE of the effect. Clinical significance is a clinical judgment. It asks: 'Is this difference large enough to matter for patients?' A blood pressure reduction of 1 mmHg might be statistically significant in a 100,000-patient study but is clinically meaningless — no patient benefits from a 1 mmHg drop. Conversely, a study might show a 10 mmHg blood pressure reduction that fails to reach statistical significance (p = 0.08) simply because the sample was too small. The effect is clinically meaningful but the study was underpowered to detect it. This is why modern guidelines emphasize reporting effect sizes (how big the difference is) and confidence intervals (the range of plausible true effects) alongside p-values. Effect sizes tell you the magnitude of the finding. Confidence intervals tell you the precision. P-values alone tell you very little. The minimum clinically important difference (MCID) is the smallest change in an outcome that patients would perceive as beneficial. If a treatment effect is smaller than the MCID, it is statistically significant but clinically irrelevant.