Core Revision
AKT statistics questions: what actually comes up
Dr Victoria HammondCo-Founder & Lead for Educational Design · 3 min read
Written August 2026
Evidence interpretation is 10% of the AKT — about sixteen marks. It is also the part candidates most often decide to leave until last, and the part examiners most often flag as poorly answered. Those two facts are related.
The good news is that this section is small, bounded and repetitive in a way the clinical 80% is not. There are only so many things it can ask you.
What the College says it is testing
The RCGP is unusually clear here. The AKT assesses basic understanding and principles — interpreting graphs and charts, and simple calculations. It is not asking you to derive anything. It publishes a list of data interpretation topics that have appeared in recent papers, and it commissioned a full set of video tutorials on the subject after candidates asked for them.
Read that as an instruction. You are not expected to become a statistician. You are expected to read a result and say what it means for a patient in front of you.
The honest shortcut
If you understand what sensitivity and specificity actually do, why NNT and relative risk reduction can describe the same trial and feel completely different, and what a confidence interval crossing 1 tells you — you have covered most of what this section reliably asks.
Sitting the AKT? Try twenty questions free.
Twenty original AKT questions in the exam's 80/10/10 split: sixteen clinical, two on evidence-based practice and two on organisation and management, with the reasoning after every one and a marked report. Free, no card and the first five are open to all.
The topics that keep coming back
Diagnostic test statistics
Sensitivity is the proportion of people with the condition the test correctly picks up. Specificity is the proportion without it that the test correctly clears. The one that catches people is positive predictive value, because it depends on how common the condition is — the same test performs very differently in a screening population and in a symptomatic one. That dependence is the point of the question, almost every time.
Absolute versus relative risk
A treatment that cuts risk from 2 in 1,000 to 1 in 1,000 is a 50% relative risk reduction and a 0.1% absolute risk reduction. Both are true; only one is useful in a consultation. Expect to be asked to convert between them, and to work out number needed to treat — which is simply one divided by the absolute risk reduction.
Confidence intervals and p values
For a ratio — odds ratio, relative risk, hazard ratio — a 95% confidence interval that crosses 1 means the result is compatible with no effect. For a difference between means, the equivalent line is zero. A narrow interval means a more precise estimate, usually from a larger study. A p value below 0.05 is the conventional threshold for statistical significance, which is not the same thing as clinical significance, and questions do test that distinction.
Study designs and their plots
Know what a randomised controlled trial, cohort study, case-control study and systematic review can and cannot tell you, and which is vulnerable to which bias. Forest plots and funnel plots turn up regularly — a forest plot for the pooled result and the spread of individual studies, a funnel plot for publication bias.
Do these questions in short sessions spread across your revision rather than in one block near the end. Sixteen marks is worth an hour a week for six weeks; it is not worth a panicked weekend.
— Dr Victoria Hammond, Co-Founder & Lead for Educational Design
If you want to work through this properly rather than in fragments, our AKT Statistics Guide covers the whole of it in AKT-sized pieces, and we have a longer piece on relative versus absolute risk because it is the one that causes the most trouble in real consultations as well as in exams.
