Data Interpretation

Data interpretation questions are common in medical school interviews. They test your ability to analyse scientific information, apply logical reasoning, and show that you understand the biological and clinical principles behind the numbers. Medical schools use them to see whether you can think like a scientist and a doctor: reading data accurately, explaining it clearly, and drawing conclusions that the evidence actually supports.

These questions may involve graphs, tables, calculations or the results of a study, usually set in physiology, pharmacology or a clinical scenario. The most common types are graph interpretation and drug dose and dilution calculations. Both should feel familiar from A Level Biology, Chemistry and Maths, and you should approach them the same way you would in an exam, except that you have to say your working out loud instead of writing it.

This lesson covers what interviewers are looking for, a four-step method you can apply to any graph or table, the skills behind reading graphs and doing drug calculations, and the background science you'll need to explain what you see.

Doctors work with data every day: blood results, observation charts, drug charts, and the research that shapes treatment decisions. Misreading a value or slipping a decimal place in a drug calculation can directly harm a patient. Interviewers want to see that you handle numbers carefully and think critically about what they show.

Nobody expects you to know advanced pharmacology. Interviewers are looking for a set of habits that every good doctor uses every day:

Accurate data handling

:

Reading values carefully, quoting numbers with units, noticing scales and labels. Doctors read blood results, observation charts and drug charts constantly, and misreading one can harm a patient.

Scientific understanding

:

Linking a pattern to a mechanism, such as explaining slower glucose clearance by impaired insulin action. Clinical decisions rest on understanding why something is happening.

Numeracy and safety

:

Methodical calculations, unit conversions and sanity checks. Drug calculation errors are a well-recognised cause of patient harm.

Clear communication

:

A structured, signposted answer with your working said out loud, just as you would at a handover, when explaining results to a patient, or when presenting to a senior.

Critical thinking

:

Spotting limitations, avoiding overclaiming, and asking what else you need to know. Evidence-based medicine means judging how far data can be trusted.

Composure and honesty

:

Staying calm when unsure and correcting your own mistakes openly. Recognising and flagging errors is central to safe practice.

Formats vary between universities, so always check each school's own interview guidance. Data interpretation usually appears in one of three ways:

MMI station

:

You are handed a graph, table or scenario (sometimes with a short reading period outside the room) and then have around 5–10 minutes with an interviewer who asks a sequence of questions that get progressively harder.

Panel interview

:

A graph or short data set is placed in front of you partway through a traditional interview and you are asked to talk about it.

Calculation task

:

A short clinical scenario (for example, preparing a drug) where you work out a dose, volume or rate. A calculator is often not provided, so the numbers are usually chosen to be manageable by hand.

Within a station, questions typically build in layers and get progressively harder: reading the data ("What does this graph show?"), comparing groups ("How does patient B differ from patient A?"), explaining ("What condition might patient B have?"), calculating ("What volume of stock would you draw up?"), and extending ("What else would you consider before choosing this drug?").

Use the same four steps for every graph, table or data set. Having a fixed method stops you from freezing, keeps your answer organised, and makes sure you hit the points on the mark scheme in a logical order. Before you start, read the whole question: stations often pack several tasks into one prompt, so break it into parts and answer them in order.

1. 1.Orient: Say what the data is showing before you interpret it.
2. 2.Describe and compare: Overall trend, key features with numbers, direct comparisons. This is the bulk of your answer.
3. 3.Explain: Link the pattern to biology or clinical reasoning, with appropriate caution.
4. 4.Extend: Limitations, what else you would want to know, clinical implications.

Step 1: Orient

Before interpreting anything, work out exactly what you are looking at. Then say one orienting sentence out loud. This shows the interviewer you are methodical and gives you thinking time. Check:

• Title / context: what experiment or clinical situation is this?
• Axes: what is on each axis, and in what units? (mg vs micrograms, mmol/L, hours vs minutes)
• Scale: linear or logarithmic? Does the y-axis start at zero? A truncated axis exaggerates differences.
• Key / legend: how many groups? Is there a control?
• Graph type: is it a normal line graph, a cumulative graph, a bar chart, a survival curve?
• Uncertainty: error bars? Sample size (n)? Individual patients or group averages?

Step 2: Describe and compare

This is where most marks are won or lost. Work from the big picture down to the detail:

1. 1.Overall trend in one sentence (rises then falls; decreases steadily; stepwise increase).
2. 2.Key features with values: starting value / baseline, peak (value and time), where the gradient changes, plateau, end value, return (or not) to baseline, anything anomalous.
3. 3.Direct comparisons between groups: which is higher or faster, and by how much (a difference or a ratio).

Attach a number and a unit to everything you describe. Compare these two descriptions of the same graph:

Weak

:

"Patient 2's drug level goes higher and stays up longer."

Strong

:

"Patient 2 peaks at about 9.5 mg/L at around 2.5 hours, compared with about 7 mg/L at 1 hour for patient 1. Patient 2's level is still around 2.3 mg/L at 24 hours, whereas patient 1 falls below the effective level by about 7 hours."

The strong version earns marks for peak value, time to peak, comparison and return to baseline in two sentences.

Step 3: Explain

Now link the pattern to the science. Good explanations:

• Use a mechanism, not just a label. "Patient B is diabetic" is a label; "patient B's glucose stays high because insulin secretion or insulin sensitivity is impaired, so glucose is not taken up into cells as quickly" is an explanation.
• Use cautious language that matches the strength of the evidence: "this is consistent with…", "this suggests…", "one possible explanation is…". Avoid "this proves".
• Offer more than one possibility where appropriate, and say what would help you tell them apart (e.g. "it could be type 1 or type 2 diabetes; the patient's age, history and antibody tests would help distinguish them").

Step 4: Extend

Finishing with a short, thoughtful extension is what separates a good answer from an excellent one. Pick one or two that are relevant, rather than reciting a list:

Limitations of the data

:

Only one patient per group? Small sample? No error bars? Short time frame? Lab data vs real patients?

What else you would want to know

:

Patient's age, weight, kidney/liver function, other medicines, symptoms, repeat measurements.

Clinical implications

:

Does this change treatment or dosing? Is monitoring needed? Is it safe?

Practical and ethical factors

:

Side effects, cost, route of administration, patient preference, access, consent.

Correlation vs causation

:

Could a confounding factor explain the link?

Logarithmic scales

Dose–response and concentration graphs very often use a log scale on the x-axis, because drug effects happen across a huge range of concentrations. Many candidates misread these, so noticing one and saying so immediately shows good data awareness.

• Each major gridline is 10 times the previous one (0.1 → 1 → 10 → 100), not an equal step.
• Halfway between 1 and 10 on a log axis is about 3 (more precisely 3.16), not 5.5.
• Roughly, between 1 and 10: about 30% of the way along is ≈ 2, about 50% is ≈ 3, about 70% is ≈ 5, and about 85% is ≈ 7.
• Something that looks like a small shift sideways can be a large difference: one gridline to the left means 10 times more potent.

To estimate a value on a log axis: identify the two labelled gridlines either side of the point (e.g. 1 nM and 10 nM), judge roughly what fraction of the gap the point sits at, then convert using the guide above and give an approximate answer ("roughly 5–6 nM"). Always say you are estimating: "reading from the log scale, the IC50 is approximately…"

Gradients, rates and cumulative graphs

• The gradient of a line tells you the rate. Steeper means faster. A flattening curve means the rate is slowing.
• A cumulative graph shows a running total (e.g. total dose given so far). It can only go up or stay flat. The gradient is the rate of delivery, and the final value is the total.
• Average rate = total change ÷ time taken. E.g. a cumulative dose that reaches 3000 micrograms over 24 hours averages 125 micrograms per hour.
• A stepwise (staircase) cumulative graph means the drug was given in separate doses (boluses) rather than as a continuous infusion, which would give a smooth line. Step height = size of each dose. Step width = time between doses.
• If the step heights change, the dose changed. Say so, because it suggests the dose was titrated (adjusted according to the patient's response).

Error bars, sample size and uncertainty

• Error bars show variability or uncertainty around a mean. Always check what they represent (standard deviation, standard error or a 95% confidence interval).
• For 95% confidence intervals: if two intervals do not overlap, the difference is very likely to be statistically significant. If they overlap, the difference may or may not be significant, and a formal statistical test is needed. Avoid saying "overlapping bars prove there is no difference".
• Small samples give wide error bars. A large effect in a small study is less certain than a modest effect in a large one.
• Statistical significance is not the same as clinical significance. A tiny benefit can be "significant" in a huge trial but not worth the side effects or cost.

Correlation, causation and other traps

• Correlation does not prove causation. Look for confounding factors (e.g. coffee drinking linked to lung cancer, where smoking is the confounder).
• Don't extrapolate beyond the data range without saying it is an assumption.
• Averages hide individuals. A group mean can conceal patients who responded very differently.
• Check the y-axis starts at zero. If it doesn't, small differences can look dramatic.
• Outliers: mention them, suggest a reason (measurement error, a genuinely different patient), and don't build your whole conclusion on them.

Graph types that commonly come up

You don't need to have seen a graph before to interpret it. The method works on anything. But recognising these common types will make you faster and more confident:

Glucose tolerance / hormone response curve

:

Blood glucose (or a hormone) over time after a challenge, e.g. a 75 g glucose drink. Comment on the fasting baseline, peak value and timing, rate of fall, whether it returns to baseline, and the comparison with control.

Dose–response curve (often log x-axis)

:

Effect (or % cell survival) against drug concentration. Comment on the IC50 / EC50 (concentration for half the maximum effect), potency (how far left the curve sits) and efficacy (maximum height).

Plasma concentration–time curve

:

Drug level in the blood after a dose. Comment on the peak (Cmax) and time to peak, half-life, and time within the therapeutic window (between effective and toxic levels).

Cumulative dose graph

:

Running total of drug given. Comment on the overall rate (gradient), smooth vs stepwise (infusion vs boluses), and total vs maximum safe dose.

Bar chart comparing treatments

:

Mean outcome in each group. Comment on the size of the difference vs placebo, error bars, sample sizes and clinical relevance.

Trend over time (epidemiology)

:

Cases or rates of a disease per year. Comment on the overall trend, turning points, and possible causes (vaccination, screening, changes in diagnosis or reporting).

Calculation questions test whether you can be trusted with numbers that affect patients. Interviewers care far more about a safe, methodical process than about speed. A correct answer reached through clear, checked steps is ideal. A wrong answer that you spot and correct yourself still shows the right instincts.

Working through a calculation

1. 1.Write down what you know, with units, before calculating anything.
2. 2.Convert everything to the same units first (usually mg and mL).
3. 3.Estimate a rough answer so you know what "about right" looks like.
4. 4.Carry units through every step. If the units don't cancel to what you want, the method is wrong.
5. 5.Sanity-check the answer. Is a diluted solution weaker than the stock? Is the volume something you could actually draw up in a syringe?
6. 6.State the answer clearly with units, and mention that in practice you would double-check it with a colleague or pharmacist.

Units and conversions

• 1 g = 1000 mg
• 1 mg = 1000 micrograms (mcg or µg)
• 1 microgram = 1000 nanograms (ng)
• 1 L = 1000 mL

Key formulas

These formulas come up frequently in drug calculation questions:

In the dilution formula, C1 and V1 are the stock concentration and the volume of stock used; C2 and V2 are the final concentration and the final total volume. All of these concepts are covered in A Level Biology and Chemistry, so you will have met them before. The challenge in an interview is applying them calmly while explaining your working out loud.

Worked calculations

Example A: weight-based dose from a liquid

A child weighing 16 kg needs paracetamol at 15 mg/kg. The oral suspension contains 120 mg in 5 mL. What volume should be given?

Knowns: weight 16 kg, dose 15 mg/kg, suspension 120 mg / 5 mL.

1. 1.Dose: 15 mg/kg × 16 kg = 240 mg.
2. 2.Concentration: 120 mg ÷ 5 mL = 24 mg/mL.
3. 3.Volume: 240 mg ÷ 24 mg/mL = 10 mL.

Check: 5 mL contains 120 mg, so 10 mL contains 240 mg. ✓

Example B: percentage strength and a maximum dose

Lidocaine 1% is used as a local anaesthetic. The maximum safe dose in this scenario is 3 mg/kg. What is the largest volume that can be used for a 60 kg adult?

1. 1.Convert the percentage: 1% = 1 g per 100 mL = 1000 mg per 100 mL = 10 mg/mL.
2. 2.Maximum dose: 3 mg/kg × 60 kg = 180 mg.
3. 3.Maximum volume: 180 mg ÷ 10 mg/mL = 18 mL.

Extend: "I'd stay comfortably below the maximum, and I'd want to know whether any other local anaesthetic had already been given, because doses add up."

The "Explain" step needs some baseline knowledge. Nothing here goes beyond a strong A Level student plus some interview-level reading, but being fluent in it lets you explain rather than just describe. You don't need to know the reference values; they are included for context only.

Blood glucose control and diabetes

Insulin (made by β-cells in the pancreas) lowers blood glucose by increasing uptake into cells and storage as glycogen. Glucagon raises it. Together they keep blood glucose within a narrow range by negative feedback.

Type 1 diabetes: autoimmune destruction of β-cells, so little or no insulin is produced. Type 2 diabetes: cells respond poorly to insulin (insulin resistance), often with reduced insulin secretion too. Both cause hyperglycaemia, and a single glucose curve can't tell you which type a patient has.

Oral glucose tolerance test (OGTT): the patient fasts overnight, a fasting blood sample is taken, they drink 75 g of glucose, and blood glucose is measured at intervals afterwards. In diabetes, glucose starts higher, peaks higher and falls back more slowly.

For context, the WHO criteria for diagnosing diabetes are fasting glucose ≥ 7.0 mmol/L or 2-hour OGTT value ≥ 11.1 mmol/L.

Potency vs efficacy

IC50 / EC50

:

the concentration of drug needed to produce half of that drug's own maximum effect (IC50 for inhibiting a process, EC50 for producing a response).

Potency

:

how much drug is needed for an effect. A lower IC50/EC50 means a more potent drug, and its curve sits further left.

Efficacy

:

the maximum effect a drug can produce (the height of the plateau).

The most potent drug is not automatically the best: efficacy, side effects, therapeutic window, route, interactions and cost all matter.

How the body handles drugs

Half-life

:

time for plasma concentration to fall by half. After about 4–5 half-lives a drug is mostly eliminated.

Therapeutic window

:

the range between the minimum effective and minimum toxic concentration. Narrow windows need careful dosing and sometimes blood-level monitoring.

Bolus vs infusion

:

a bolus is a single dose given in one go; an infusion is given continuously over time. On a cumulative dose graph, boluses appear as steps and an infusion as a smooth rise.

Drugs are mainly eliminated by the kidneys and liver. If either organ isn't working well, the drug is cleared more slowly and can build up in the blood, increasing the risk of toxicity.

Basic epidemiology

Incidence = new cases in a period. Prevalence = total existing cases at a point in time.

Vaccination, screening, new diagnostic tests and changes in how cases are recorded can all change reported rates.

Normal observations (adult, approximate)

You are not expected to know these values. They are included for reference only, in case a question shows a patient's observations.

Heart rate 60–100 bpm; blood pressure around 120/80 mmHg; respiratory rate 12–20 breaths/min; oxygen saturation 94–98%; temperature about 36.5–37.5 °C.

• Think aloud, but structure it. The interviewer can only credit reasoning they can hear. Talk through what you are doing in organised chunks rather than a stream of consciousness.
• Signpost your structure. Before you start, tell the interviewer how you'll tackle the question, for example by describing the graph first, then comparing the groups, and then suggesting what might explain the difference. This makes you sound organised and makes it easy for the interviewer to tick off mark-scheme points as you go.
• Ask for time if you need it. Asking for a few seconds to look at the graph before you answer is completely acceptable and looks composed.
• Correct mistakes openly. Spotting and fixing your own error is exactly the behaviour expected of a safe doctor.
• Never guess a number wildly. Giving an approximate value and saying you would want to check it more carefully is better than false confidence.

Describing without numbers

:

"It goes up then down" earns little. Quote values and units.

Explaining before describing

:

Answer the question that was asked, in the order it was asked.

Missing the log scale

:

If you don't notice it, you will misread every value on the x-axis.

Ignoring the control group

:

Comparisons with the control are often the whole point.

Overclaiming

:

"This proves patient B has type 2 diabetes" goes beyond the data. Use "suggests" and "consistent with".

Forgetting units

:

Dropping units, or mixing up mg and micrograms, can cause 1000-fold errors.

Using the distractor

:

Not every number given is needed (e.g. height in a weight-based calculation).

Not sanity-checking

:

A calculated answer that is obviously too big or too small should be caught before you say it.

Answering only part of a multi-part question

:

Tick off each part as you go.

Stopping at description

:

Always finish with a brief explanation and a limitation or next step.

Reading about data interpretation will only take you so far. The skill comes from working through real questions out loud, so that the method feels automatic by the time you reach the interview.

The Med Prep Partner question bank contains data interpretation questions from real medical school interviews, each with a mark scheme and a model answer.