Public Health, Epidemiology and Health Services Management
Core concepts
Epidemiology is tested as a reasoning discipline. Know incidence (new cases divided by population at risk over time) versus prevalence (all cases at a point); measures of association (relative risk, odds ratio, attributable risk); study designs (cross-sectional, case-control, cohort, randomised trial); test validity (sensitivity, specificity, predictive values); bias (selection, information, recall); confounding; the steps of an outbreak investigation; and health-system concepts such as the six building blocks, the basic and essential service packages delivered through health posts, health centres and district hospitals in Afghanistan, and the disability-adjusted life year.
Traps
Confusing incidence with prevalence; quoting predictive value without stating prevalence; interpreting an odds ratio as a risk ratio when the disease is common; confusing selection bias with confounding; and forgetting that a screening test performs differently in different populations. Attack rate is incidence in a defined outbreak population, not prevalence.
Worked example
One thousand women are screened and 100 have the disease. Sensitivity is 90% and specificity is 90%. True positives 90, false negatives 10, false positives 90, true negatives 810. Positive predictive value = 90/(90+90) = 50%. If prevalence fell to 1%, positive predictive value would fall to about 8%, even with the same test.
Rule of thumb
Prevalence equals incidence multiplied by average duration; when prevalence is low, even a good test generates many false positives.