The POPH192 final is worth 70 percent of the paper, after the two progress tests at 15 percent each. Unlike the progress tests, it covers the whole semester, so it's not only epidemiology. Health, illness and society, Māori and Pacific health, and public health practice like screening and prevention are all in scope.
That spread is where people lose marks. Most students revise the calculations hard, because they feel like "real" study, and leave the social science content until the night before because it feels easy to read. Here's how we'd split the time instead.
What the final covers
The paper falls into roughly six areas. Write them down and score yourself out of 5 on each before you start, based on whether you could explain it to someone without notes.
- Introduction to epidemiology: prevalence, incidence proportion, incidence rate and person-time.
- Study designs and measures: cross-sectional, ecological, cohort, case-control and randomised controlled trials, plus relative risk, odds ratio, risk difference and number needed to treat.
- Error, bias and causation: chance, confidence intervals and p-values, selection and information bias, confounding, and judging causation.
- Public health applications: screening (sensitivity, specificity and predictive values), levels of prevention, the population and high-risk approaches, and outbreaks.
- Health, illness and society: disease and illness as different things, illness narratives, stigma, power, and the limits of biomedicine.
- Indigenous and Pacific health: colonisation and its effect on health, and models like Te Whare Tapa Whā, Te Pae Mahutonga and Fonofale.
The lowest scores are where your first revision sessions go.
The calculations: fast marks once they're automatic
Calculation questions are reliable marks, and you can get quick at them in a few sessions. The trick is working out the study design first, because that decides which measures you can calculate at all. Our post on working out the study type before the maths covers that in detail. Here are three worked examples to check yourself against.
Relative risk and risk difference
A cohort study follows 1,000 exposed people and 2,000 unexposed people. 40 exposed and 20 unexposed people develop the disease.
- Risk in exposed: 40 / 1,000 = 0.04 (4 percent).
- Risk in unexposed: 20 / 2,000 = 0.01 (1 percent).
- Relative risk: 0.04 / 0.01 = 4. The exposed group has four times the risk.
- Risk difference: 0.04 - 0.01 = 0.03, or 3 extra cases per 100 exposed people.
If you're shaky on what each of those means, what halving your risk means goes through it.
Number needed to treat
A treatment lowers the risk of an outcome from 10 percent to 6 percent. The absolute risk reduction is 4 percent, or 0.04. The number needed to treat is 1 / 0.04 = 25, so 25 people need treating for one to benefit.
Screening and positive predictive value
A test with 90 percent sensitivity and 95 percent specificity is used on 10,000 people, and 1 percent of them have the disease.
- 100 people have the disease. The test picks up 90 of them (true positives) and misses 10.
- 9,900 people don't have the disease. 95 percent test negative, which leaves 495 false positives.
- Positive predictive value: 90 / (90 + 495) = about 15 percent.
So even with a good test, only about 15 in every 100 positive results are real cases when the disease is rare. Questions love this, and it's worth being able to explain in a sentence why PPV rises when prevalence rises.
Bias and causation: say the direction
For bias and confounding, recognising the name isn't enough. Strong answers say how the bias got in, and which way it pushes the result. For example, if people with a disease remember their past exposures more carefully than healthy controls, that's recall bias. It tends to make the exposure look more strongly linked to the disease than it really is.
For confounding, practise naming a confounder for a given association and checking it against the three properties: it's linked to the exposure, it's a risk factor for the outcome, and it isn't on the causal pathway between them.
The social science content: practise writing it
This part is easy to read and harder to write well. When a question asks about a concept like stigma, illness narratives or a Māori health model, a strong answer usually does three things:
- Defines the concept accurately, using the course's terms.
- Applies it to the specific example or person in the question.
- Links it to a health outcome or to what a health professional should do differently.
Most weak answers stop after step 1. For example, describing Te Whare Tapa Whā's four dimensions (taha tinana, taha hinengaro, taha whānau and taha wairua) is a start. Explaining how a care plan that only looks at taha tinana might fail a particular patient is what earns the marks.
For the determinants of health and equity content, our post on memorising the social determinants of health has a method that works for this kind of material.
A five-session plan
- Measures and study designs. Work 15 to 20 mixed calculation questions, design first.
- Bias, confounding and causation. For each type, write one example and the direction it pushes the result.
- Screening, prevention and outbreaks. Do two PPV calculations from scratch and explain the levels of prevention with examples.
- Health, illness and society, and Indigenous and Pacific health. Write three short answers under time using the define, apply, link structure.
- Mixed practice across all six areas, then your error list, then rest.
FAQ
Is the POPH192 final harder than the progress tests?
It's broader, not necessarily harder. It covers the whole paper, so the content you skipped for the progress tests needs to be covered now.
Should I memorise the formulas?
Know them well enough that you don't have to think. More important is knowing which study designs let you use which measure. A case-control study, for example, can't give you a risk, so it can't give you a relative risk.
How much should I study the Māori and Pacific health content?
As much as any other area. It's examinable, and answers that just list the parts of a model without applying them don't score well.
For POPH192 practice questions with worked explanations, see how Cutline HSFY works.
Cutline is not affiliated with the University of Otago and doesn't hold or reproduce University exam papers. Assessment details are from the POPH192 paper information, checked October 2026. Check your own course outline for this year's exam format.
