Biology • Year 12 • Module 8 • Lesson 13
Analysing Epidemiological Data — Pattern Recognition and Risk Factor Quantification
Lock in the core vocabulary and formulas for risk measures, the evidence hierarchy, and survival curve reading before moving to application questions.
1. Term–definition match
Match each term to its definition. In the right-hand column write the matching term from this list: absolute risk, relative risk, absolute risk reduction (ARR), relative risk reduction (RRR), number needed to treat (NNT), randomised controlled trial (RCT), cohort study, double-blind trial, placebo, systematic review. 10 marks
| # | Definition (shuffled) | Matching term |
|---|---|---|
| 1.1 | The actual probability of an event occurring in a group, expressed as a proportion (e.g. 5 in 100 people develop disease X over 5 years). | |
| 1.2 | The ratio of the risk in the exposed group to the risk in the unexposed group; a value of 1 means no difference between groups. | |
| 1.3 | The arithmetic difference in risk between the control group and the treatment group; calculated as risk (control) minus risk (treatment). | |
| 1.4 | The proportional reduction in risk expressed relative to the baseline (control group) risk; calculated as ARR ÷ risk in control group, expressed as a percentage. | |
| 1.5 | The number of patients who must be treated for a specified period to prevent one additional adverse outcome compared to control; calculated as 1 ÷ ARR. | |
| 1.6 | A study design in which participants are randomly assigned to a treatment or control group, providing the strongest single-study design for establishing causation. | |
| 1.7 | A prospective observational study that follows a group sharing a common characteristic over time to track the development of disease and identify risk factors. | |
| 1.8 | A study in which neither the participants nor the researchers assessing outcomes know which group received active treatment — minimising both performance and assessment bias. | |
| 1.9 | An inert treatment identical in appearance to the active intervention, given to the control group to control for the psychological effect of receiving treatment. | |
| 1.10 | A type of literature review that uses pre-specified, reproducible methods to identify and critically appraise all relevant studies on a question, placed at the top of the evidence hierarchy. |
2. Fill in the blank — risk measure formulas in context
Complete the passage using words from the word bank. Each word is used once. 8 marks
Word bank: absolute risk reduction · relative risk reduction · number needed to treat · clinically · statistically · smaller · baseline · misleading
When a treatment's benefit is reported, the most important distinction is between relative and absolute measures. The (ARR) tells you the actual difference in event rates between groups, while the (RRR) expresses the same reduction as a fraction of the control group's risk. When the risk is very low, even a large RRR produces a very ARR — which is why quoting only RRR can be . The (NNT) translates the ARR into a clinically meaningful statement: how many patients must be treated to prevent one outcome. A result can be significant (p < 0.05) while remaining unimportant when the ARR is tiny.
3. True or false — with correction
For each statement, circle T or F. If false, write the corrected version. 10 marks (1 for T/F, 1 for each correction where needed)
3.1 A p-value below 0.05 proves that a treatment is clinically important. T / F
3.2 In a Kaplan-Meier survival curve, the y-axis represents the proportion of participants who have not yet experienced the primary outcome, and it starts at 1.0 (100%). T / F
3.3 A double-blind trial design means that participants do not know which group they are in, but the assessing researchers do know. T / F
3.4 A systematic review is considered higher on the evidence hierarchy than a single well-designed RCT because it pools results from multiple high-quality studies, increasing statistical power and minimising selection bias. T / F
3.5 A Kaplan-Meier curve that reaches zero by the end of the study period proves that all patients eventually recover from the disease. T / F
4. Function recall — what does each risk measure tell us?
Answer each in 1–2 sentences using precise terms from the lesson. 8 marks (2 each)
4.1 What does a relative risk (RR) of 0.60 mean, in plain language?
4.2 Why is number needed to treat (NNT) more useful for clinical decision-making than relative risk reduction (RRR)?
4.3 In the evidence hierarchy, why does a cohort study rank below a single well-designed RCT?
4.4 What does it mean when two lines on a Kaplan-Meier plot begin to converge after initially diverging?
5. Rank the evidence
The six study descriptions below are shuffled. In the "Level" column, write a rank from 1 (strongest) to 6 (weakest) according to the evidence hierarchy. 6 marks
| Level | Study description |
|---|---|
| A published account of one elderly patient in Canberra who recovered from severe influenza after drinking particular herbal tea — no comparison group, no statistical analysis. | |
| A Cochrane meta-analysis pooling the results of 28 randomised controlled trials examining the effect of aspirin on cardiovascular mortality, using pre-specified inclusion criteria. | |
| The Nurses' Health Study — 121,700 US nurses followed from 1976, tracking dietary habits and incidence of breast cancer over decades. | |
| A large Australian RCT of 5,000 patients randomly assigned to receive either a new anticoagulant or standard warfarin after pulmonary embolism, with double-blinding and a 2-year follow-up. | |
| A national cross-sectional survey measuring the proportion of Australian adults with type 2 diabetes and their reported physical activity levels at a single point in time. | |
| A case-control study of 500 women with cervical cancer and 500 matched controls, comparing past HPV exposure rates using stored cervical swabs. |
6. Build a concept map — risk measures and their relationships
Draw labelled arrows between the five terms below to show how they connect. Each arrow must carry a linking phrase. Aim for at least 5 labelled arrows. 5 marks
Supplied terms: absolute risk reduction (ARR) · number needed to treat (NNT) · relative risk reduction (RRR) · baseline risk (control group) · clinical significance.
Q1 — Term–definition matches
1.1 absolute risk • 1.2 relative risk • 1.3 absolute risk reduction (ARR) • 1.4 relative risk reduction (RRR) • 1.5 number needed to treat (NNT) • 1.6 randomised controlled trial (RCT) • 1.7 cohort study • 1.8 double-blind trial • 1.9 placebo • 1.10 systematic review.
Q2 — Cloze passage
In order of the blanks: absolute risk reduction (ARR) · relative risk reduction (RRR) · baseline · smaller · misleading · number needed to treat (NNT) · statistically · clinically.
Q3 — True / false with correction
3.1 False. Correction: a p-value below 0.05 means the result is unlikely to be due to chance (statistically significant) — it says nothing about clinical importance. Very large samples can produce statistically significant but clinically trivial results.
3.2 True.
3.3 False. Correction: in a double-blind trial, both the participants and the researchers/assessors do not know which group is which — this is what makes it double-blind. If only participants are unaware, it is single-blind.
3.4 True.
3.5 False. Correction: a survival curve reaching zero means all participants experienced the primary outcome (which may be death, relapse, or another event) within the follow-up period — not that they recovered. A curve that plateaus may indicate long-term survivors or that the study ended before all participants experienced the outcome.
Q4.1 — Relative risk of 0.60
RR = 0.60 means the treated (or exposed) group has 60% of the risk of the control (unexposed) group — i.e. the treated group is 40% less likely to experience the outcome compared to the control group. Equivalently, the relative risk reduction (RRR) = 1 − 0.60 = 40%.
Q4.2 — Why NNT is more useful than RRR
RRR expresses the treatment benefit as a proportion of the baseline risk, so the same RRR can represent very different absolute benefits depending on how common the outcome already is. NNT translates the ARR into a concrete statement — "you must treat X patients to prevent one outcome" — making it directly actionable for clinical decision-making and more meaningful to patients and clinicians than a percentage figure alone.
Q4.3 — Why cohort study ranks below RCT
A cohort study is observational — participants are not randomly assigned, so pre-existing differences between exposed and unexposed groups (confounders) cannot be ruled out as explanations for any observed difference in disease rates. Randomisation in an RCT distributes both known and unknown confounders equally, allowing the difference between groups to be attributed to the intervention rather than to pre-existing differences.
Q4.4 — Converging Kaplan-Meier lines
When two survival curves converge after having diverged, it means the initially superior group is now experiencing the primary outcome at a rate similar to the control group — the treatment benefit is diminishing over time. The two groups are becoming more similar in their event rates, suggesting the treatment's protective effect is not persistent or is wearing off.
Q5 — Evidence hierarchy ranking
Level 1 — Cochrane meta-analysis of 28 RCTs (systematic review + meta-analysis). Level 2 — Australian double-blind RCT of 5,000 patients. Level 3 — Nurses' Health Study (large prospective cohort). Level 4 — Case-control study of cervical cancer and HPV. Level 5 — National cross-sectional survey of diabetes and physical activity. Level 6 — Single patient case report of herbal tea and influenza recovery.
Q6 — Sample concept map
Acceptable arrows include:
- ARR — is calculated from the → baseline risk (control group)
- NNT — is calculated as 1 ÷ → ARR
- RRR — is calculated as ARR ÷ → baseline risk (control group)
- NNT — determines → clinical significance
- ARR — is a better indicator than RRR of → clinical significance
- baseline risk — when very low, makes RRR appear large but ARR small — reducing → clinical significance
Award 1 mark per correctly labelled causal arrow (minimum 5, correctly directed).