A carcinogen has a relative risk of 4.0 for a rare cancer affecting 0.01% of the population. A second exposure has a relative risk of 1.5 for a common cancer affecting 30% of the population. Which exposure has the larger population attributable fraction?
AThe first exposure, because a relative risk of 4.0 is more than twice the second exposure's relative risk
BThe second exposure, because prevalence is high enough that even a modest relative risk produces a large attributable fraction
CThey are equal, because relative risk and prevalence cancel out
DCannot be determined without knowing the absolute incidence rates
Population attributable fraction (PAF) depends on BOTH relative risk and the prevalence of the exposure. A very high relative risk applied to a rare exposure produces little preventable burden; a modest relative risk applied to a ubiquitous exposure may dominate. This is why physical inactivity and poor diet often rank ahead of rarer toxins in cancer prevention priority — they are everywhere. Relative risk alone does not answer the policy question.
Question 2 Multiple Choice
A cancer type is rare in Japan but rises to match American rates among Japanese migrants to the United States within one to two generations. What does this pattern most strongly indicate?
AJapanese genetic variants protect against this cancer type in their homeland
BThe genetic background of this population is highly susceptible to American dietary patterns
CEnvironmental or behavioral factors, not genetic factors, are the primary drivers of this cancer risk
DThe Japanese healthcare system underdiagnoses this cancer, creating an artificial rate difference
In migrant studies, the genetic background of the population remains constant while the environment changes. If cancer rates shift within one or two generations to match those of the host country, the change must be driven by modifiable exposures — diet, lifestyle, environmental toxins — not by genes. This logic was foundational to identifying diet and caloric intake as likely contributors to colorectal and breast cancer risk decades before randomized trial evidence was available.
Question 3 True / False
Cross-sectional study designs are poorly suited for cancer epidemiology because of the long latency between exposure and disease onset.
TTrue
FFalse
Answer: True
The latency between a carcinogenic exposure and detectable disease can span 20–40 years. A cross-sectional snapshot therefore captures exposure status and disease status at the same point in time, but the relevant exposure may have occurred decades earlier — making it nearly impossible to establish temporal precedence or to accurately recall past exposures. This is why large prospective cohorts (Nurses' Health Study, UK Biobank) that collect exposure data before disease onset are the backbone of cancer epidemiology.
Question 4 True / False
Demonstrating a strong relative risk between an exposure and a cancer type is sufficient to make that exposure a high-priority cancer prevention target.
TTrue
FFalse
Answer: False
Relative risk establishes that an association is real and causal, but prevention priority requires population attributable fraction (PAF) — which incorporates how prevalent the exposure is. An exposure with a relative risk of 10 but a 0.001% prevalence has a negligible PAF. Prioritizing cancer prevention requires understanding both the strength of the causal relationship and how many people carry the exposure. Without prevalence data, relative risk alone can misdirect prevention resources.
Question 5 Short Answer
Why do major cancer cohort studies collect biological specimens and exposure data prospectively rather than relying on recalled exposure data collected after a cancer diagnosis?
Think about your answer, then reveal below.
Model answer: Recalled exposure data collected after diagnosis is subject to recall bias — participants who develop cancer may remember or report past exposures differently from those who remain healthy. More fundamentally, if the relevant exposures occurred 20–40 years before diagnosis, memory is unreliable and the data inaccurate. Prospective collection captures exposures before any disease develops, eliminating recall bias and allowing biomarkers to objectively document what participants were actually exposed to, rather than what they remember.
This connects two concepts: the latency problem (exposures precede diagnosis by decades) and measurement error (recalled data is systematically biased in case-control settings). Prospective design solves both: exposure data is complete, contemporaneous, and equally accurate across cases and controls because disease status is unknown at collection. This is the core methodological reason why large cohort studies are expensive but irreplaceable in cancer epidemiology.