Community Medicine - Measurement, Study Design and Programme Decisions
What this domain tests
The Community Medicine paper asks you to move between population measurement and programme action: choosing the right denominator, judging a study design, and translating findings into a decision inside Sri Lankan field structures (MOH area, RDHS, PDHS, Epidemiology Unit, Medical Statistics Unit). Marks are won on correct rates and explicit assumptions, not on memorised lists.
Core rules and concepts
- Every rate needs a numerator and denominator drawn from the same population and period; a ratio such as the maternal mortality ratio uses live births, not total population, as denominator.
- Sensitivity and specificity are properties of the test and are prevalence-independent. Positive and negative predictive values are not; they shift with the prevalence of the condition in the population screened.
- Cohort studies yield incidence and risk ratios; case-control studies yield odds ratios (only approximating risk when the outcome is rare); cross-sectional surveys yield prevalence and cannot establish temporality.
- Confounding is a distortion to be adjusted for; effect modification is real heterogeneity to be reported by stratum.
- Larval surveys: house index = % houses positive, container index = % containers positive, Breteau index = positive containers per 100 houses inspected.
Worked example
A test with sensitivity 90% and specificity 95% is applied to 10,000 people where prevalence is 1%. Of 100 diseased, 90 are true positives; of 9,900 non-diseased, 495 are false positives. PPV = 90/(90+495) = 15.4%. Apply the same test where prevalence is 10%: 900 true positives and 450 false positives, PPV = 66.7%. Identical test accuracy, a seven-fold difference in predictive value - so the same kit is a screening tool in a low-prevalence MOH area but a confirmatory aid only where prevalence is high.
Traps that cost marks
Putting total population in the denominator of infant or maternal mortality; forgetting to convert per 1,000 to per 100,000; quoting an odds ratio as a risk ratio for a common outcome; reading a high standardised mortality ratio as poor service quality without checking age structure and case-mix; and confusing the Breteau index with the house index.