Statistics in Public Health

EPIDEMIOLOGY relies on STATISTICS to find PATTERNS in disease that are invisible at the individual level. Why do some communities have higher cancer rates? Does a new drug actually work? Are vaccines causing rare side effects? Statistical methods turn raw data into evidence and policy. Without statistics, doctors would just be guessing — sometimes guessing well, sometimes catastrophically wrong.

Key concepts. RATES (cases per 100,000): allow comparing populations of different sizes. CORRELATION vs CAUSATION: smoking is correlated with cancer; rigorous study established CAUSATION. RANDOMIZED CONTROLLED TRIALS (RCTs): the gold standard for testing treatments — random assignment ensures comparable groups. CONFIDENCE INTERVALS: express uncertainty; "5% lower risk (95% CI: 2-8%)" is more honest than "5% lower." STATISTICAL POWER: ability to detect real effects; small studies miss real differences. P-VALUES: probability of seeing a result if the null hypothesis were true; controversial in modern science.

Common pitfalls. SELECTION BIAS: skewed samples (testing drugs only in young men, then prescribing to old women). SURVIVORSHIP BIAS: studying only survivors miss what killed others. CONFOUNDERS: hidden variables linked to both cause and effect. SIMPSON'S PARADOX: correlation reverses when subgroups are combined. Modern epidemiology fights these with careful design, peer review, and replication. Reading health news critically requires understanding these — most headlines simplify or distort.

Statistics is the language of evidence-based public health. With it, lives are saved. Without it, harmful interventions hide behind anecdotes. Numerical literacy matters for everyone — patients, voters, parents.