Chapter 9

Sample Size Calculator

Continuous outcomes · Binary proportions · Survival events · Power & minimum detectable effect · Attrition & Bonferroni correction

Sample Size — Two Means

The Sample Size Decision Hierarchy

Step 1: Define the MCID

What is the smallest difference that would change practice? This is a clinical judgement. Setting it too small gives an impossibly large n. Setting it too large gives an underpowered study for the true effect.

Step 2: Estimate SD or event rate

For continuous outcomes, SD comes from published data or pilot studies. For binary outcomes, the control event rate determines variance. Conduct sensitivity analysis for ±20%.

Step 3: Choose power and alpha honestly

80% is the convention. 90% is more appropriate for pivotal confirmatory trials. Alpha 0.05 two-tailed is standard. If multiple primary outcomes, apply Bonferroni before calculating n.

Step 4: Inflate for attrition

Enrol more than you calculate to end up with enough analysable patients. Be honest about dropout: 5% is rarely realistic for complex interventions over 3 years.

Disclaimer: This calculator is intended for educational purposes only. Results are provided to support statistical learning and should not be used as the sole basis for clinical decision-making. Always interpret statistical outputs in the context of study design, clinical relevance, and professional judgement. An Evidence Integration Lab Initiative.