## Cumulative Incidence (Risk) Calculation **Key Point:** Cumulative incidence (risk) is calculated as the number of new cases divided by the **initial at-risk population, adjusted for those who did not complete follow-up** (losses to follow-up and deaths from unrelated causes). This is the standard epidemiological approach when not all participants complete the full observation period. ### Formula for Cumulative Incidence $$\text{Cumulative Incidence (Risk)} = \frac{\text{Number of new cases during follow-up period}}{\text{Number at risk at baseline} - \text{Withdrawals (losses + deaths)}}$$ ### Calculation for This Cohort | Group | Count | |-------|-------| | Original cohort | 2,000 | | Lost to follow-up | 40 | | Deaths from unrelated causes | 20 | | **Adjusted denominator** | **2,000 − 40 − 20 = 1,940** | $$\text{Cumulative Incidence} = \frac{160}{1940} \times 100 \approx 8.25\%$$ **High-Yield:** When participants are **censored** (lost to follow-up or die from competing causes before the end of the study), the standard epidemiological practice is to subtract these individuals from the denominator when calculating cumulative incidence. They did not complete the at-risk period and therefore cannot be counted as having been fully observed for the outcome. ### Why Withdrawals ARE Subtracted | Scenario | Included in Denominator? | Reason | |----------|--------------------------|--------| | **Losses to follow-up (40)** | No — subtracted | Did not complete follow-up; outcome status unknown | | **Deaths from unrelated causes (20)** | No — subtracted | Competing risk; removed from the at-risk pool before study end | | **Incident cases (T2DM, 160)** | No (numerator only) | They are the outcome being measured | | **Remaining completers** | Yes | Fully observed for the entire follow-up period | **Clinical Pearl:** The distinction between **cumulative incidence** (a proportion) and **incidence rate** (person-time denominator) is critical. When follow-up is incomplete, cumulative incidence adjusts the denominator by removing censored individuals, while incidence rate uses person-years of observation. Both approaches account for incomplete follow-up, but in different ways. (Park's Textbook of Preventive and Social Medicine, 26th ed., Ch. 3) **Common Pitfall:** Using the full original cohort (2,000) as the denominator ignores the fact that 60 individuals (40 lost + 20 deceased) did not complete the observation period and should not be counted as having been fully at risk for the entire 5 years. [cite:Park 26e Ch 3; Gordis Epidemiology 6e Ch 4]
Sign up free to access AI-powered MCQ practice with detailed explanations and adaptive learning.