Cancer Data Analysis, Research, and Public Health Reporting
Domain 4 - Cancer Data Analysis, Research, and Public Health Reporting
This domain tests how registry data become valid information: rates, survival, comparisons, and responsible release for research and public health.
Core rules
- Incidence is new cases in a period divided by the population at risk, usually per 100,000. Mortality uses deaths; prevalence counts people alive with a history of cancer.
- Crude rates reflect actual burden but are distorted by age structure. Age-adjusted rates standardize to a reference population, often the 2000 U.S. standard, so groups can be compared.
- Survival: observed survival counts all deaths; relative survival compares observed survival with expected survival of a matched general population, approximating cancer-specific survival without cause-of-death data.
- Small numbers produce unstable rates. Suppress or flag cells with few cases and report confidence intervals.
- Reporting to NPCR, state registries, and NCDB follows data-use agreements; research releases require IRB review and de-identification or a limited data set.
Worked example
A registry reports 5-year observed survival of 70% for a cohort; the matched general population's expected survival is 85%. Relative survival is 70 divided by 85, or 0.824, meaning 82.4%. Reasoning: the cohort's survival is 82.4% of what would be expected from general-population mortality alone, suggesting excess mortality from the cancer. Reporting only 70% would overstate the cancer effect in an older cohort.
Traps
- Comparing crude rates across states with different age distributions.
- Treating relative survival as identical to cause-specific survival.
- Publishing a rate based on 3 cases without a confidence interval.
- Releasing identified data for research without IRB approval or an agreement.