Epidemiology — Incidence, Prevalence, Mortality and Study Design

Activity 1 — Calculations and Interpretation

1. Bowel cancer calculations. (a) Incidence rate = 450 ÷ 5,000,000 × 100,000 = 9 per 100,000 per year. (b) Prevalence = 8,500 ÷ 5,000,000 × 100 = 0.17%. (c) Case fatality rate = 90 ÷ 8,500 × 100 = 1.06% per year. This means approximately 1 in 100 existing bowel cancer patients dies from the disease each year — reflecting that many patients are diagnosed early (stage I–II) and survive for many years, while a smaller proportion with advanced disease contribute most deaths.

2. CVD country comparison. From crude rates: Country A has a higher crude CVD mortality rate (200 vs 150 per 100,000), suggesting more CVD deaths per person in Country A. However, from age-standardised rates: Country B actually has a higher age-standardised CVD mortality rate (190 vs 145 per 100,000). This reversal — where Country A has a higher crude rate but lower age-standardised rate than Country B — indicates that Country A has an older population. The large proportion of elderly people in Country A elevates the crude rate even though the underlying disease risk at each age is lower than in Country B. Age-standardised rates are more valid for comparing the underlying disease burden between populations because they remove the confounding effect of different age structures. If you wanted to know which country has riskier cardiovascular conditions for its residents, Country B's higher age-standardised rate indicates it has the greater underlying risk, despite having fewer total deaths per 100,000 in the raw data.

Activity 2 — Study Design Evaluation

1. New drug for T2D. Best design: Randomised Controlled Trial (RCT). Justification: the drug has been safety-tested and is believed to be beneficial — it is ethical to assign participants to the drug vs placebo. RCT randomisation eliminates confounding — both groups will have similar baseline characteristics (age, diet, activity, genetics) by chance, so any difference in T2D progression can be attributed to the drug. Double-blinding (neither participant nor researcher knows which group they are in) eliminates measurement and assessment bias. Limitation: the trial population may not represent all T2D patients (often excludes very old, pregnant, or multi-morbid patients), limiting generalisability to these groups. Also, the trial duration may be insufficient to detect longer-term effects.

2. Childhood sun exposure and melanoma. Best design: Case-control study. Justification: participants are adults aged 40–60 — we cannot follow children prospectively for 30–40 years to study adult melanoma (too slow, too expensive, too much loss to follow-up). Case-control design efficiently recruits adults who already have melanoma (cases) and compares them to adults without melanoma (controls), asking both groups to recall their childhood sun exposure history. This is retrospective — looking back at exposure rather than following forward. Limitation: recall bias — melanoma patients may more carefully recall and report sun exposure in childhood than controls (who have no particular reason to think carefully about their childhood sun habits). This asymmetric recall can artificially inflate the apparent association between sun exposure and melanoma. This can be partially controlled by using objective measures (e.g. geographical records of sun exposure at birth location) rather than self-report.

3. Red wine and CVD. Confounder 1: socioeconomic status (SES). People who regularly drink moderate amounts of red wine tend to have higher SES. Higher SES is independently associated with lower CVD risk (better healthcare access, healthier diet, more physical activity, lower stress). SES is therefore associated with both the exposure (wine drinking) and the outcome (lower CVD risk) — a classic confounder. Confounder 2: diet quality. People who drink red wine moderately often follow a Mediterranean-style diet (high in vegetables, fish, olive oil, whole grains) which independently reduces CVD risk. Wine drinking may be a marker for this overall dietary pattern rather than a cause of reduced CVD risk. Why the study cannot establish causation: the study is observational — it shows that wine drinkers have lower CVD rates, but cannot determine whether the wine is causal or whether the association is entirely explained by confounders like SES and diet. Without controlling for these confounders (through statistical adjustment, matching, or preferably an RCT of moderate wine consumption), the headline claim is unjustified. An RCT randomising people to drink red wine vs not drink wine would be needed to establish causation — but such trials face practical challenges (compliance, ethics). Existing RCT evidence from polyphenol supplementation (the proposed active ingredient in red wine) does not support a strong protective effect.

Multiple Choice

1. B — Prevalence ≈ incidence × duration: 50 × 10 = 500 per 100,000. A chronic disease accumulates cases over its duration, so prevalence is much higher than incidence. Option A would only be correct if duration were 1 year (very short, acute). Option C is wrong — chronic diseases have high prevalence because people live with them long-term. Option D is wrong — while measured differently, the relationship between incidence and prevalence is a key concept.

2. C — Cohort study is most appropriate: it establishes temporal sequence (pollution exposure measured before COPD develops); can measure cumulative exposure; calculates incidence rates in exposed vs unexposed groups. Option A (cross-sectional) cannot establish which came first. Option B (case-control) is less able to establish temporal sequence and is better for rare diseases. Option D (RCT) is unethical for harmful exposures.

3. D — Ecological correlation (country-level association) cannot establish individual-level causation. Wealthier countries have both more mobile phones and better cancer detection; many confounders exist. Option A is wrong — correlation strength alone does not establish causation. Option B is wrong — observational studies can provide useful information, just not proof of causation. Option C is wrong — replication of an ecological association still does not establish individual-level causation without controlling for confounders.

4. A — Crude rate 220, age-standardised 145: the crude rate is higher than the age-standardised rate, meaning after accounting for age structure, the underlying rate is lower. This indicates Country X has an older-than-average population — the age-standardisation reference population is younger on average, so applying the standard removes the contribution of the large elderly population. Option B confuses standardisation with accuracy of recording. Option C reverses the logic. Option D is wrong — age-standardisation adjusts for age distribution, not removes cases from older groups.

5. C — The cohort study provides strong observational evidence: temporal sequence established, large sample, dose-response likely present, consistent with biological plausibility (carcinogens in processed meat). However, 'proves' overstates what any observational study can establish — residual confounding is always possible. The finding is consistent with causation and would be appropriately described as 'provides strong evidence for' or 'is consistent with a causal role for processed meat in bowel cancer.' Option A incorrectly uses 'proof.' Option B is wrong — cohort studies are valuable evidence. Option D is wrong — a 40% relative risk increase is clinically and epidemiologically significant.

Short Answer Model Answers

Q6 (4 marks): Incidence is the rate of new cases of a disease arising in a defined population over a specified time period, calculated as: (number of new cases ÷ population at risk) × 100,000. It measures how fast disease is developing. Prevalence is the total proportion of a population with a disease at a given time, calculated as: (number of existing cases ÷ total population) × 100. It measures how much disease exists in the community [2 marks]. Why effective treatment raises prevalence despite falling incidence: prevalence = incidence × average disease duration. Effective treatment extends survival — patients live longer with the disease, so they remain in the 'existing cases' pool for longer. Even if incidence (new cases per year) falls because of prevention programs, the total pool of people living with the disease grows as survival improves. Prevalence thus rises [1 mark]. Example: HIV in high-income countries. Antiretroviral therapy introduced in the mid-1990s dramatically extended life expectancy for HIV-positive people. Annual new infections (incidence) fell due to prevention programs. But people lived with HIV for decades rather than dying within years — so the total number living with HIV (prevalence) rose substantially through the 2000s despite falling incidence. Similar patterns are seen with Type 2 diabetes (better treatment → longer survival with disease → rising prevalence despite stable incidence) [1 mark — 4 marks total].

Q7 (5 marks): Cohort: recruit a large sample (e.g. 50,000+) of adults aged 35–65 who do NOT currently have Type 2 diabetes and who are willing to be followed for 15–20 years. Diversity in physical activity levels is important for adequate comparison groups [1 mark]. Exposure variable: measure physical activity level at baseline and at regular intervals (e.g. every 2 years) — using standardised questionnaires or accelerometers — recording type, duration, intensity, and frequency of exercise per week. Classify participants into physical activity categories (e.g. sedentary, moderate, high) [1 mark]. Outcome variable: development of Type 2 diabetes, defined as fasting blood glucose ≥7.0 mmol/L, HbA1c ≥48 mmol/mol, or physician diagnosis. Measured at each follow-up visit [1 mark]. Evidence of association: calculate and compare annual T2D incidence rates in high-activity vs low-activity groups. Calculate relative risk (incidence in high-activity ÷ incidence in low-activity) — a relative risk significantly less than 1.0 would support a protective effect of physical activity. Test for dose-response: does increasing activity further decrease T2D risk? [1 mark]. Confounding variable: dietary habits (healthier eaters tend to exercise more AND have lower T2D risk through independent mechanisms). Control: collect detailed dietary data at baseline and follow-up; statistically adjust for dietary quality index in the analysis. Alternatively, restrict the analysis to participants with similar dietary patterns [1 mark — 5 marks total].

Q8 (5 marks): Why RCTs are the gold standard: randomisation distributes known and unknown confounders equally between groups by chance — the only method that truly controls for variables not yet identified. Double-blinding prevents measurement bias. RCTs establish causation because the only systematic difference between groups is the intervention [1 mark]. Why RCTs cannot be required for environmental exposures: it is unethical to randomly assign people to harmful exposures — decades of smoking, asbestos inhalation, high UV exposure, or toxic chemical exposure cannot be deliberately assigned to participants. An ethical review board would never approve such trials. Requiring RCT evidence before accepting that a harmful exposure causes disease would mean we could never establish causation for any environmental carcinogen or toxin using experimental methods [2 marks]. What observational evidence can establish: the Bradford Hill criteria provide a framework for establishing causation from observational evidence — strength of association, consistency across studies and populations, temporal sequence (exposure precedes disease), dose-response relationship, biological plausibility, and specificity. When multiple criteria are satisfied simultaneously, causal inference is reasonable. The smoking-lung cancer causal link was established entirely through observational studies (primarily cohort studies like Doll and Hill) combined with molecular mechanistic evidence — an RCT was neither possible nor necessary [1 mark]. Conclusion: the claim is partially valid in the sense that RCTs are the ideal study design when ethically possible — for drug trials, vaccination programs, dietary interventions, and preventive strategies, RCT evidence should be sought. But requiring RCT evidence for harmful environmental exposures is an inappropriate standard that would paralise public health action. A more appropriate standard is convergent evidence from multiple study types — observational studies establishing association and temporal sequence, plus mechanistic evidence establishing biological plausibility, plus dose-response data — together satisfying the Bradford Hill criteria. The field of epidemiology has developed precisely because RCTs are often impossible, and its methods are capable of establishing causation to the standard required for public health and clinical decision-making [1 mark — 5 marks total].