Questions: Selecting Appropriate Epidemiologic Study Designs

Question 1

A researcher wants to investigate whether a rare childhood cancer (roughly 80 new cases per year in a country of 40 million) is associated with residential proximity to industrial chemical plants. Which study design is most appropriate?

• A. Cross-sectional survey of the entire population measuring both cancer prevalence and proximity to chemical plants at one time point
• B. Prospective cohort study following 50,000 children living near plants and 50,000 living elsewhere for 15 years
• C. Case-control study recruiting existing cancer cases and matched controls, then investigating their residential histories
• D. Randomized controlled trial randomly assigning children's residential location relative to chemical plants

Answer: (C)

When the outcome is rare, the case-control design is almost always the right choice. A cohort study would need to follow enormous numbers of children for many years to accumulate even a modest number of cancer cases — inefficient and expensive. The case-control design efficiently recruits existing cases (from hospital registries, cancer databases) and matched controls, then reconstructs exposure history. Option A (cross-sectional) cannot establish temporal order. Option D is ethically impossible. The case-control design's efficiency advantage is greatest precisely when outcomes are rare.

Question 2

Why is the cross-sectional study design poorly suited to establishing causal relationships between an exposure and a health outcome?

• A. Cross-sectional studies cannot calculate any measures of association between exposure and outcome
• B. Cross-sectional studies measure exposure and outcome at the same moment, making it impossible to confirm that exposure preceded the outcome
• C. Cross-sectional data relies entirely on self-report, which is too unreliable for causal inference
• D. Cross-sectional samples are always too small to achieve adequate statistical power

Answer: (B)

The fundamental problem is temporality — the first criterion in any causal framework. A cross-sectional study asks: 'Who is exposed and who has the outcome right now?' It cannot determine whether the exposure came first. Someone with high pollution exposure and asthma may have developed asthma before moving to a polluted area, or their asthma may have caused them to move (reverse causation). Cross-sectional studies are excellent for estimating prevalence and generating hypotheses, but the inability to establish temporal sequence is their core limitation for causal inference.

Question 3

Because observational studies cannot randomize participants to exposures, they can seldom provide credible causal evidence — mainly RCTs can establish causation.

• T. True
• F. False

Answer: False

This overclaims. The Bradford Hill criteria — strength of association, dose-response relationship, temporal sequence, biological plausibility, specificity, consistency across studies — provide a framework for building causal inferences from observational evidence. The causal relationship between cigarette smoking and lung cancer was established through cohort and case-control studies before RCTs were feasible or ethical. Observational studies do face confounding that randomization eliminates, but careful design (matching, stratification, restriction), large samples, dose-response analysis, and replication across study types can yield scientifically credible causal conclusions.

Question 4

Case-control studies are especially efficient for studying rare diseases because they start by recruiting existing disease cases rather than following large populations for extended periods.

• T. True
• F. False

Answer: True

This is the defining efficiency advantage of the case-control design. For a disease affecting 1 in 10,000 people annually, a cohort study would require enrolling tens or hundreds of thousands of participants and following them for years to accumulate enough cases for analysis. A case-control study short-circuits this by starting from confirmed cases (often identified from hospital records, registries, or clinics) and matched controls, then reconstructing exposure history. This is far cheaper and faster for rare outcomes. The tradeoff is that case-control studies cannot measure incidence directly, are limited to odds ratios (not relative risks), and are vulnerable to recall bias.

Question 5

A researcher wants to study whether daily low-dose aspirin use prevents colorectal cancer. What practical and ethical considerations might lead them to choose a cohort study rather than an RCT?

Model answer: Practically: an RCT testing colorectal cancer incidence would need to randomize large numbers of participants to aspirin or placebo and follow them for 10+ years, because colorectal cancer has a long latency. This is enormously expensive and prone to loss of follow-up. A cohort study can leverage existing aspirin users (who take it for cardiovascular prevention) and matched non-users, using medical records for outcome ascertainment. Ethically: if preliminary evidence suggests aspirin may be beneficial, randomizing participants to placebo raises concerns about withholding a potentially useful intervention; if side effects (gastrointestinal bleeding) are a concern, randomizing to aspirin raises different ethical issues. Cohort designs observe naturalistic exposure decisions, avoiding these dilemmas. The tradeoff is confounding — aspirin users may systematically differ from non-users — which must be addressed through careful statistical adjustment.

Design selection is always a set of tradeoffs. The cohort design's practical advantages (leveraging existing exposures, using existing records) and ethical advantages (no allocation of risk or deprivation) can outweigh its inferential limitations relative to the RCT, especially when an RCT is impractical or unethical to conduct.