University of Birmingham · FACULTY OF PSYCHOLOGY

13344 Chap.7 Cohort Studies, Attributable Risk and Hazard Ratios

- one subject, every graph, every model, every mark
6 Chapters4-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 7 of 10 · 13344

Cohort Studies, Attributable Risk and Hazard Ratios

Following exposure forward

The chapter explains how cohort studies select people on exposure and follow them to see who develops an outcome, and compares prospective and retrospective cohorts using the module's example of obstetric complications and psychosis.

Measuring effect and cause

It works through relative risk and the population attributable fraction, reads hazard ratios from a Cox model, and sets out confounding and the eight questions the teaching asks before a causal claim.

Calculations to practise

A worked cohort table computes relative risk and the attributable fraction and checks the fraction against expected cases, and two survival curves show what a constant hazard ratio looks like.

Practise moving between the three measures with the same data.

In this chapter

What this chapter covers

  • 01

    Selecting on exposure

  • 02

    Prospective and retrospective cohorts

  • 03

    Incidence and relative risk

  • 04

    Population attributable fraction

  • 05

    Cox regression and hazard ratios

  • 06

    Confounding and causal criteria

Worked example · free

Read a hazard ratio from two relapse curves

Q [4 marks]. AskSia-authored practice. Discharged patients receiving peer support relapse at a constant hazard of 0.06 a year and those receiving usual care at 0.10 a year. Find the hazard ratio and the share still relapse-free at five years in each group, using S = e raised to minus hazard times years. The marks shown are an AskSia study allocation, not a University marking scheme.
Two groups with a constant hazard ratio.Two groups with a constant hazard ratio. Each line shows the share of discharged patients still free of relapse over five invented years; the peer support group's hazard is 0.06 a year against 0.10 for usual care, a hazard ratio of 0.60, so its line falls more slowly throughout. 100% 75% 50% 0 5 years Time since discharge peer support usual care hazard ratio 0.60
Two groups with a constant hazard ratio. Each line shows the share of discharged patients still free of relapse over five invented years; the peer support group's hazard is 0.06 a year against 0.10 for usual care, a hazard ratio of 0.60, so its line falls more slowly throughout.
  • 1Hazard ratio = 0.06 ÷ 0.10 = 0.60.
  • 1Peer support at five years: e^−0.3 ≈ 0.74, so about 74% relapse-free.
  • 1Usual care at five years: e^−0.5 ≈ 0.61, so about 61% relapse-free.
  • 1Interpretation: at any time, the peer support group's relapse hazard is 40% lower.
The hazard ratio is 0.60, a protective relationship, and after five years about 74% of the peer support group and 61% of the usual care group remain relapse-free. The ratio is constant over time, which is the proportional hazards assumption. Writing the five-year percentages beside the hazard ratio helps a reader see that a 40% lower hazard translates into roughly 13 more people in every 100 staying well.
Sia tip — Say which group is the reference before reading a hazard ratio as higher or lower.
Glossary

Key terms

Incidence
Incidence is the number or rate of new cases in a previously disease-free group over a set period.
Attributable Fraction
An Attributable Fraction is the share of cases in a population that could be prevented by removing an exposure, assuming causation.
Confounding
Confounding is the effect of an additional variable that is responsible for an observed association.
Proportional Hazards
Proportional Hazards is the Cox model's assumption that the hazard ratio between groups stays constant over time.
FAQ

Cohort Studies, Attributable Risk and Hazard Ratios FAQ

Why is a prospective cohort good for causal questions?

Exposure is recorded before the outcome develops and participants start free of the outcome, so temporal order is built into the design and reverse causation is excluded. The module recommends it when a question uses the word cause.

What does a population attributable fraction assume?

It assumes the exposure causes the outcome, and it depends on how common the exposure is. If exposure becomes more widespread, the fraction rises even when the relative risk is unchanged, so it varies between places and over time.

Is a hazard ratio the same as a relative risk?

Not technically. A relative risk counts events by the end of follow-up, whereas a hazard ratio compares instantaneous event rates throughout follow-up and assumes their ratio stays constant. The two are often read in similar ways, but they answer slightly different questions.

What is the difference between incidence and prevalence?

Incidence is the number or rate of new cases in a group that was free of the condition at the start of a period. Prevalence is the number with the condition at a given time and reflects both incidence and how long the condition lasts, like water in a bath with taps and a drain.

What questions should I ask before claiming a cause?

The teaching lists temporality, plausibility, consistency, strength, dose and response, reversibility, study design and the number of lines of evidence. A strong essay applies several of them to the evidence rather than treating any one association as proof.

When is a retrospective cohort a sensible choice?

When the outcome takes years to appear and good records already exist, a retrospective cohort looks back at past exposure and later outcomes far more cheaply than following people forward. Its weaknesses are gaps and inconsistencies in records and confounders that were never measured.

Study strategy

Assessment move

Practise moving between the three measures with one cohort table: compute relative risk, then the attributable fraction, then explain how a time-to-event analysis would change the question. Keep the eight causality questions on a card. Read the reference group before every hazard ratio and relative risk you meet, and rewrite the result as a sentence a service user could follow.

Working through Cohort Studies, Attributable Risk and Hazard Ratios in 13344? Sia is AskSia’s AI Psychology tutor — ask any 13344 Cohort Studies, Attributable Risk and Hazard Ratios question and get a clear, step-by-step explanation grounded in how 13344 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 17 of your subjects at University of Birmingham, plus 1,000+ more Bibles across the AskSia library.
Sia - your 13344 tutor, unlimited, worked the way the exam marks it
The full 4-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works