Epidemiology and Public Health Surveillance
Epidemiology and Public Health Surveillance
This domain tests whether you can measure disease occurrence, compare groups correctly, recognise the design that produced a number, and run or interpret a surveillance signal. Expect 2×2 tables, attack rates, screening statistics and questions about bias.
Measures
- Incidence proportion (risk) = new cases / population at risk at the start.
- Incidence rate = new cases / person-time.
- Prevalence = existing cases / population; prevalence ≈ incidence × duration at steady state.
- Association: risk ratio (cohort), odds ratio (case-control), rate ratio; attributable risk and population attributable fraction.
- Standardisation: direct (apply study rates to a standard population) and indirect (SMR = observed/expected × 100).
Design and bias
Cohort studies give incidence and temporality; case-control studies are efficient for rare outcomes but yield only an odds ratio; cross-sectional studies give prevalence and cannot establish temporality; ecological studies compare groups and are vulnerable to the ecological fallacy. Confounding is a third variable associated with exposure and outcome and not on the causal pathway; effect modification is a different strength of association across strata and is described, not "controlled away". Selection bias, information/recall bias and loss to follow-up are the usual suspects.
Surveillance
Passive (routine notifiable reporting), active (staff solicit cases), sentinel (selected providers), syndromic (pre-diagnostic indicators). Core attributes: case definition, timeliness, completeness, sensitivity, positive predictive value, simplicity. Outbreak steps: verify the diagnosis, define cases, find cases, describe by person/place/time, generate hypotheses, test with an analytic study, control, communicate. Epidemic curves: point source = single peak within one incubation period; continuous common source = plateau; propagated = successive peaks one incubation period apart.
Screening
Sensitivity = TP/(TP+FN); specificity = TN/(TN+FP); PPV = TP/(TP+FP). PPV rises with prevalence, so a test validated in a high-risk clinic performs poorly in the general population.
Worked example
At a wedding, 120 guests attended; 80 ate the potato salad and 40 did not. Illness occurred in 32 of 80 eaters (40%) and 4 of 40 non-eaters (10%). Risk ratio = 0.40 / 0.10 = 4.0. Attributable fraction in the exposed = (0.40 − 0.10)/0.40 = 75%. Overall attack rate = 36/120 = 30%. Reasoning: because the cohort is defined and followed, a risk ratio is appropriate; 75% means three-quarters of illness among eaters is attributable to the salad if the association is causal.
Traps that cost marks
- Reporting an odds ratio as a risk ratio when the outcome is common (above 10%).
- Confusing incidence with prevalence in a cross-sectional study.
- Ignoring that PPV depends on prevalence.
- Treating a p-value as a measure of effect size.
- Reading a propagated curve as a single-source event.