How to use these questions
These nine questions practise common epidemiological calculations and interpretations. They are not tied to a particular examination format. For each one, choose the single best answer; the explanation then works through the calculation or reasoning and addresses a tempting alternative. For numerical questions, keep the denominator and the measure being asked for in view: confusing people, person-time and odds is a frequent source of avoidable errors.
1. Incidence rate
This tests whether you can distinguish a rate based on person-time from a proportion of people. A MySummaries question on the distinction looks like this:
A cohort contributes 600 person-years of follow-up, during which 12 participants develop a disease for the first time. What is the incidence rate?
Incidence rate = new cases ÷ person-time = 12 ÷ 600 = 0.02 per person-year, or 20 per 1,000 person-years. The tempting 2% is a proportion and does not use the person-time denominator; it cannot be inferred without knowing the number of people at risk.
The units are part of the answer. Writing “20” without “per 1,000 person-years” hides whether you have calculated a rate or a proportion.
2. Point prevalence
This checks whether you can calculate the proportion of a population with a condition at one specified time. A MySummaries question puts the date and denominator into the stem:
On 1 June, a survey finds that 40 of 800 residents have asthma. What is the point prevalence on that date?
Point prevalence = existing cases ÷ population at that time = 40 ÷ 800 = 0.05, or 5%. The 0.5% option is a decimal-place error. A person-time rate is inappropriate because the survey gives a cross-section, not follow-up time.
Prevalence describes how widespread a condition is at a point or over a period. It is not the same as the number of new cases arising during follow-up.
3. Risk ratio in a cohort
This tests calculation and interpretation of a risk ratio. In MySummaries, the working is shown alongside the answer:
In a cohort study, 30 of 300 exposed people and 20 of 400 unexposed people develop an outcome during follow-up. What is the risk ratio for exposed versus unexposed people?
Risk in the exposed group is 30/300 = 0.10; risk in the unexposed group is 20/400 = 0.05. Risk ratio = 0.10/0.05 = 2. The value 0.5 reverses the comparison (unexposed versus exposed); 1.5 is not the ratio of the two risks.
State the direction of comparison when you interpret a ratio. Here, the observed risk in the exposed group is twice that in the unexposed group; this calculation alone does not establish that exposure caused the difference.
4. Odds ratio in a case-control study
This question tests the cross-product calculation and recognises why a case-control study commonly uses an odds ratio. A MySummaries version keeps the four cells visible:
A case-control study includes 60 exposed and 40 unexposed cases, and 30 exposed and 70 unexposed controls. What is the odds ratio for exposure among cases compared with controls?
The cross-product odds ratio is (60 × 70)/(40 × 30) = 4,200/1,200 = 3.5. The value 2.0 is the ratio 60/30 of exposed counts, which ignores the unexposed cells. The case-control sampling design does not directly provide the population risks needed for a risk ratio.
For a two-by-two table, label the cells before using the cross-product formula. This reduces the risk of swapping exposure and outcome or calculating a ratio of counts instead of odds.
5. Confounding
This tests whether you can distinguish a confounder from a mediator. In MySummaries, the key issue is the variable’s role in the causal pathway:
A study reports an association between coffee drinking and pancreatic cancer. Smoking is associated with coffee drinking and is an independent cause of pancreatic cancer; it is not caused by coffee drinking. What is the best description of smoking in this analysis?
Smoking is associated with the exposure, is an independent risk factor for the outcome and is not on the proposed causal pathway from coffee to cancer. It is therefore a potential confounder to address in the analysis. A mediator would lie on the pathway; effect modification is a different phenomenon in which the exposure association differs across levels of another variable.
Adjustment can reduce confounding when the relevant variables are measured and the analysis is appropriate. It does not automatically remove bias from measurement error, selection or unmeasured confounders.
6. Sensitivity
This tests the denominator for sensitivity: people who truly have the condition, not everyone tested. A MySummaries question makes the four cell counts explicit:
A test is evaluated against a reference standard. It gives 90 true-positive results, 10 false-negative results, 20 false-positive results and 80 true-negative results. What is its sensitivity?
Sensitivity = true positives ÷ all people with the condition = 90/(90 + 10) = 90%. The 81.8% option is 90/(90 + 20), which uses all positive test results and calculates positive predictive value instead.
A useful check is to say the denominator aloud: sensitivity asks, “Of those who have the condition, what proportion test positive?” Specificity instead starts with those who do not have it.
7. Positive predictive value
This checks how prevalence affects interpretation of a positive result. A MySummaries question supplies a hypothetical population so the arithmetic is transparent:
A test has 90% sensitivity and 90% specificity. In a population of 10,000 where prevalence is 1%, approximately what is the positive predictive value?
About 100 people have the condition: 90 test positive. Of the 9,900 without it, 10%—990—test positive. PPV = 90/(90 + 990) ≈ 8.3%. The 90% option confuses sensitivity with PPV. Even a reasonably accurate test can have a low PPV when prevalence is low.
When explaining a screening result, do not report sensitivity as the chance that a person with a positive result has the disease. That probability is PPV and depends on prevalence as well as test performance.
8. Number needed to treat
This tests the conversion from absolute risk reduction to number needed to treat. MySummaries presents the two risks so the difference is easy to check:
Over the same follow-up period, an outcome occurs in 12% of the control group and 8% of the treatment group. What is the number needed to treat to prevent one outcome over that period?
Absolute risk reduction = 0.12 − 0.08 = 0.04. NNT = 1/0.04 = 25 people treated over the stated follow-up period to prevent one outcome, on these estimates. The value 4 is the percentage-point reduction, not the NNT; 12 is the control risk expressed as a percentage.
Always attach the time period and outcome to an NNT. It is derived from absolute risk reduction, so it can differ between populations with different baseline risks even when relative effects are similar.
9. Confidence interval
This tests what a confidence interval around a ratio says about statistical uncertainty. A MySummaries question asks for interpretation, not just the point estimate:
A study estimates a risk ratio of 1.4 with a 95% confidence interval of 0.9 to 2.1. Using the conventional two-sided 5% significance threshold, which interpretation is best?
For a ratio measure, the null value is 1. Because the interval includes 1, the result is not statistically significant at the conventional two-sided 5% threshold. The point estimate alone does not determine significance. The interval does not prove no association, and the frequentist interpretation is not a 95% probability statement about this fixed interval containing the true value.
A non-significant result is not proof of no effect. Consider the estimate’s precision, the range of effects compatible with the interval, the study design and the possibility of bias.
What to revisit
Use errors to choose the next topic, rather than simply repeating questions you already answer reliably. A MySummaries review summary for a candidate part-way through revision might look like this:
If you missed a question twice, turn the specific distinction into a short retrieval prompt. For example, a remediation card could ask:
You lost this mark twice: for sensitivity, what is the denominator, and how does it differ from the denominator for positive predictive value?
In your next session, answer the prompt from memory and explain the denominator, not just the formula. Then try a fresh example with different numbers. If the reasoning holds when the values change, you are more likely to have learnt the concept rather than memorised one calculation.
How MySummaries helps
Build an epidemiology revision board from your own notes, then use it to generate question practice and cards for the measures or study designs you need to revisit. Open MySummaries to work from your material.