Biology • Year 12 • Module 8 • Lesson 12
Epidemiology: Incidence, Prevalence, Mortality and Study Design
Apply incidence, prevalence and mortality calculations to real Australian data; interpret a data graph; analyse confounding variables and justify study design choices.
1. Interpret Australian Type 2 diabetes data
The table below shows Type 2 diabetes data for Australia at three time points. Analyse the data and answer the sub-questions. 7 marks
| Year | Total diagnosed cases | Population (millions) | Crude prevalence (%) | Age-standardised prevalence (%) |
|---|---|---|---|---|
| 2000 | 640,000 | 19.2 | 3.3 | 4.1 |
| 2010 | 970,000 | 22.3 | 4.4 | 4.3 |
| 2022 | 1,300,000 | 25.9 | 5.0 | 4.2 |
Data adapted from Australian Institute of Health and Welfare (AIHW) diabetes reports 2000–2022.
1.1 Describe the trend in total diagnosed cases and in crude prevalence from 2000 to 2022. Quote specific data values. 2 marks
1.2 The age-standardised prevalence changed from 4.1% in 2000 to 4.2% in 2022 — far less than the crude prevalence change (3.3% to 5.0%). What does this tell us about the apparent "diabetes epidemic" in Australia? 3 marks
1.3 Identify two factors other than a genuine increase in disease rate that could explain the rise in total diagnosed cases shown in the table. 2 marks
2. Interpret graph — Australian melanoma incidence and mortality, 1982–2022
The figure below shows age-standardised incidence and mortality rates for melanoma of the skin in Australians aged 15+. 7 marks
Figure 2. Age-standardised melanoma incidence and mortality, Australia 1982–2022. Adapted from Australian Cancer Database (AIHW, 2023).
2.1 Describe the overall trend in melanoma incidence from 1982 to 2022. Estimate the rate at the start and end of the period. 2 marks
2.2 Incidence rose substantially while mortality remained relatively stable across the same period. Propose two reasons that could explain this divergence — one related to diagnosis, one related to treatment. 2 marks
2.3 A journalist reads this graph and writes: "Australia's melanoma crisis is worse than ever — cases have never been higher." Evaluate this claim using the data. What important distinction does the journalist fail to make? 3 marks
3. Confounders — the Doll and Hill British Doctors Study
In 1951, Richard Doll and Austin Bradford Hill recruited approximately 40,000 British doctors and followed them for over 50 years, recording smoking habits and causes of death. Within the first four years, a strong association between cigarette smoking and lung cancer mortality was clear. A critic argued the association could be entirely explained by the fact that doctors who smoked also tended to work in more industrial cities with higher air pollution, and that air pollution — not tobacco — was the true cause of the elevated lung cancer rates. 6 marks
3.1 Identify the confounding variable proposed by the critic. Explain why it could be described as a confounder in this context: state how it is associated with both the exposure AND the outcome. 3 marks
3.2 Describe two features of the British Doctors Study design that helped researchers address this confounding concern and distinguish tobacco from air pollution as the likely causal agent. 2 marks
3.3 The study showed a clear dose-response relationship: doctors who smoked more cigarettes per day had proportionally higher lung cancer mortality. How does a dose-response relationship strengthen the evidence for causation rather than mere correlation? 1 mark
4. Compare study designs — match to the research question
For each research question below, identify the most appropriate study design from the four options (cohort, case-control, cross-sectional, RCT). Justify your choice in 1–2 sentences and identify one limitation of that design for this specific question. 8 marks, 2 per question)
4.1 Researchers want to test whether a new oral medication reduces the rate of cardiovascular events (heart attack, stroke) in patients with elevated LDL cholesterol. The drug has passed Phase 2 safety trials.
Best design: Justification:
One limitation:
4.2 Researchers want to determine the current prevalence of hypertension (high blood pressure) among Indigenous Australians aged 40–60 years across three states.
Best design: Justification:
One limitation:
4.3 Researchers want to investigate whether childhood obesity increases the risk of developing Type 2 diabetes by age 50. They enrol a group of 10-year-olds, classify them as obese or non-obese, and plan to follow them for 40 years.
Best design: Justification:
One limitation:
4.4 Researchers want to investigate whether people who developed mesothelioma had greater lifetime occupational asbestos exposure than people who did not develop mesothelioma. Mesothelioma is rare (2–3 cases per 100,000 per year in Australia).
Best design: Justification:
One limitation:
Q1.1 — Trend description
Total diagnosed cases roughly doubled from 640,000 (2000) to 1,300,000 (2022) — an increase of approximately 103%. Crude prevalence rose from 3.3% to 5.0% over the same period. Both measures show a consistent upward trend across all three time points.
Q1.2 — What the age-standardised data reveals
The age-standardised prevalence changed by only 0.1 percentage points (4.1% to 4.2%) while crude prevalence rose by 1.7 percentage points (3.3% to 5.0%). [1] This substantial difference indicates that much of the apparent "epidemic" in crude terms is explained by population ageing — as Australia's population has grown older, more people fall into age groups with higher Type 2 diabetes rates, inflating the crude prevalence. [1] The underlying rate of Type 2 diabetes at each age has remained relatively stable — the disease is not dramatically more common per person; rather, there are more older people in whom it is common. [1]
Q1.3 — Two factors other than a genuine rate increase
1. Population growth: Australia's population grew from 19.2 to 25.9 million — simply more people means more absolute cases even at the same rate. 2. Improved screening and diagnosis: Better screening programs and lower diagnostic thresholds detect cases that previously went undiagnosed — apparent increase in "cases" without a true increase in disease prevalence at each age. (Also acceptable: expanded diagnostic criteria; increased awareness leading to more testing.)
Q2.1 — Melanoma incidence trend
Age-standardised melanoma incidence rose consistently from approximately 32 per 100,000 per year in 1982 to approximately 55 per 100,000 per year in 2022 — an increase of roughly 70% over the 40-year period. The rise was approximately linear, with no plateau evident by 2022. [1 describe trend; 1 quote approximate start and end values]
Q2.2 — Divergence between incidence and mortality
Diagnosis reason: Improved skin surveillance programs and greater public awareness of early warning signs (e.g. SunSmart campaigns) have led to more melanomas being detected at earlier stages — which inflates incidence counts. Treatment reason: Improved surgical treatment of early-stage melanomas and the introduction of targeted therapies and immunotherapy (from approximately 2011 onwards) for advanced melanoma have substantially reduced case fatality rates, decoupling mortality from the rising incidence.
Q2.3 — Evaluating the journalist's claim
The journalist is partially correct that age-standardised incidence has never been higher [1], but the claim overstates the severity by ignoring the mortality trend. Mortality remained approximately stable across the period and may even be declining [1]. The journalist conflates incidence (new cases detected) with overall disease burden or severity — the critical distinction is that incidence can rise while mortality falls, indicating better diagnosis and improved treatment outcomes, not necessarily a worsening crisis [1]. A more accurate headline would note that both detection and survival are improving simultaneously.
Q3.1 — Confounding variable
Confounding variable: Industrial air pollution (particulate matter, industrial carcinogens). [1] It meets both conditions for a confounder: (1) it is associated with the exposure (smoking) — doctors who smoked tended to live in more industrial cities where air pollution was higher, so smoking status correlated with pollution exposure; (2) it is associated with the outcome independently — air pollution is a known lung irritant and contains carcinogens that increase lung cancer risk regardless of smoking status. [2 — one mark per direction of association]
Q3.2 — Study design features addressing confounding
Feature 1 — Stratified analysis: Doll and Hill could compare lung cancer rates among smokers and non-smokers who lived in the same cities (controlling for pollution level), finding that smoking-related risk remained elevated even within areas of similar pollution. Feature 2 — Consistency across subgroups: The smoking–lung cancer association was found across doctors in rural and urban settings, and across doctors with very different levels of city pollution exposure — making it unlikely that pollution alone could explain the consistently elevated lung cancer risk in smokers.
Q3.3 — Dose-response and causation
A dose-response relationship (more exposure → more disease) is difficult to explain by confounding alone — a confounder would have to track perfectly with dose to produce the same pattern. It is consistent with the Bradford Hill criterion of biological gradient: if tobacco genuinely causes lung cancer, then more tobacco exposure should produce more cancer, which is what the data showed. This pattern greatly increases the plausibility of a causal (not merely correlational) relationship.
Q4 — Study design choices
4.1 RCT. The drug has passed safety testing and is believed beneficial — it is ethical to randomise participants to drug vs placebo. Randomisation controls for confounders (age, diet, exercise, genetics) equally between groups, so any difference in cardiovascular event rates can be attributed to the drug. Limitation: the trial population is often restricted (excludes elderly, pregnant, or multi-morbid patients), limiting generalisability to these groups.
4.2 Cross-sectional study. The question asks for the current prevalence at one time point — a cross-sectional study measures exposure and disease simultaneously across the target population, directly calculating prevalence. Limitation: cannot establish which came first — cannot determine whether hypertension developed before or after any risk factors identified in the survey (no temporal sequence).
4.3 Cohort study (prospective). Researchers are following a defined group forward in time — this is the definition of a cohort study. By measuring obesity status at age 10 before T2D develops, temporal sequence is established (exposure precedes outcome). Limitation: 40 years of follow-up is extremely long — high risk of participant dropout (loss to follow-up) which can bias results if the people who drop out differ from those who remain.
4.4 Case-control study. Mesothelioma is very rare (2–3 per 100,000 per year); a cohort study would require hundreds of thousands of participants and decades of follow-up to accumulate enough cases. A case-control study efficiently recruits existing mesothelioma patients (cases) and matched controls, comparing their past asbestos exposure histories retrospectively. Limitation: Recall bias — mesothelioma patients, who know they have the disease, may more carefully recall and report occupational asbestos exposures than healthy controls, potentially inflating the apparent association.