Biostatistics
Core concepts
Biostatistics tests whether observed differences exceed chance. Know descriptive vs inferential statistics, the null hypothesis, type I error (α) and type II error (β), and power (1−β). A p-value is the probability of data this extreme if the null is true-not the probability the null is true. Confidence intervals convey both magnitude and precision; a 95% CI excluding the null value corresponds to p<0.05.
Match tests to data: t-test (means, two groups), ANOVA (means, 3+ groups), chi-square (categorical associations), correlation/regression (continuous relationships). Understand parametric vs nonparametric assumptions. Know standard deviation (spread of data) vs standard error (spread of sample means; SE = SD/√n).
Traps
- Interpreting a p-value as effect size or as the probability of replication.
- Confusing statistical with clinical significance-large samples make trivial differences significant.
- Misreading relative risk reduction vs absolute risk reduction vs number needed to treat (NNT = 1/ARR).
- Ignoring multiple comparisons inflating type I error.
- Mixing up sensitivity analysis with test sensitivity.
Worked example
A drug cuts event rate from 10% to 6%. ARR = 0.10 − 0.06 = 0.04 (4 percentage points). NNT = 1/0.04 = 25. RRR = 0.04/0.10 = 40%. The same trial phrased as "40% reduction" sounds larger than "4 percentage points"-questions test whether you compute both.
Rule of thumb
For a 2×2 table, always compute ARR and NNT alongside RR; report the CI to judge precision.