Additional Study Notes: Screening Measures and Worked Data Examples

Learning objectives

Calculate prevalence and basic screening-test measures; identify their denominators; and distinguish a screening result from a confirmed diagnosis.

Prevalence and incidence

Prevalence describes existing cases in a defined population at a specified time or during a stated period. Cumulative incidence describes new cases during follow-up among people initially at risk. An incidence rate uses person-time in the denominator. Always state the population, condition definition and time frame.

Original example: if 60 of 300 examined students have the defined condition on the survey date, sample prevalence is 60/300 = 20%. This estimate does not automatically describe all students in the district; the sampling process, participation and measurement method affect generalisability.

An original screening table

Hypothetical results for 1,000 people with a reference assessment: 90 true positives, 10 false negatives, 180 false positives and 720 true negatives. These counts are invented for calculation practice.

Sensitivity = TP/(TP+FN) = 90/100 = 90%. Specificity = TN/(TN+FP) = 720/900 = 80%. Positive predictive value = TP/(TP+FP) = 90/270 ≈ 33.3%. Negative predictive value = TN/(TN+FN) = 720/730 ≈ 98.6%.

The disease prevalence in this example is 100/1,000 = 10%. Despite high sensitivity, only about one-third of positive screens are true positives. Predictive values depend on the population prevalence as well as test performance. A positive screening result needs the appropriate confirmatory pathway.

Avoiding data errors

Keep raw observations, units and coding definitions. Use a distinct code for missing data rather than silently entering zero. Check impossible values, duplicate records and inconsistent eye or patient identifiers. Define whether an analysis is per person or per eye: two eyes from one person are not automatically independent observations.

Report the sample size and denominator with every percentage. For a small or skewed dataset, examine the distribution before choosing summary measures. A median describes the middle ordered observation; a mean uses all values and is sensitive to extreme observations.

Study References

• Source notes: Concepts of Biostatistics; Data Collection; Introduction to Epidemiology
• CDC: Evaluating Public Health Surveillance Systems — predictive value

External references checked 13 September 2026. Worked numerical examples and teaching activities are original.