CAP 4: Randomised Controlled Trials

Question 1

What is a randomised controlled trial?

Answer 1

🧪 What is a Randomised Controlled Trial (RCT)?

A randomised controlled trial (RCT) is considered the gold standard (cornerstone) for testing whether a treatment really works.

It is:

• A prospective study → researchers follow people forward in time.
• A closed type of cohort study → the group is defined at the start, and no new participants are added later.
• Participants are assigned to groups by random allocation → every person entering the study has an equal chance of being placed in any group (e.g., treatment or control).

🎲 What does “randomisation” mean in the design?

Randomisation is the process of assigning participants to groups so that:

• Each person has a known and equal chance of being assigned to any group.
• The group assignment cannot be predicted in advance.

This prevents researchers (or participants) from influencing which group someone enters.

🔀 Types of Randomisation

There are two main types:

1️⃣ Simple Randomisation

• Usually based on a single random sequence (like flipping a coin or using a random number generator).
• Each participant is assigned independently of the others.

2️⃣ Restricted Randomisation

• Uses additional rules to maintain balance between groups (for example, ensuring equal group sizes).
• Helps prevent one group from accidentally becoming much larger than the other.

🎲 Why do we randomise?

Randomisation is important because it:

• Spreads confounding factors evenly between groups (reduces confounding bias).
• Creates groups where the only planned difference is the treatment (the intervention of interest).
• Makes blinding easier, so participants and researchers may not know who received what treatment.

In other words, if both groups are similar at the start, and the only difference is the treatment, then any difference in outcomes is likely due to the treatment.

🏥 Clinical Example (1–3 sentences)

Researchers test a new antidepressant by randomly assigning patients to either the new medication or a placebo. Because assignment is random, both groups should have similar ages, illness severity, and other characteristics. If one group improves more, the difference is likely due to the medication.

👶 Explanation for a 10-Year-Old (1–3 sentences)

Imagine flipping a coin to decide who gets a new energy drink and who gets juice. Since it’s random, both teams are fair and similar. If one team runs faster, it’s probably because of the drink.

🧠 Memory Hook (funny, logical, interesting)

“RCT = Random Coin Toss Trial” 🪙 — flip fairly first, then compare the results.

🚨 Key Takeaways

• RCT = gold standard for testing treatment effectiveness.
• Prospective and closed cohort design.
• Participants randomly assigned.
• Randomisation gives each person an equal and unpredictable chance of assignment.
• Two types: simple and restricted randomisation.
• Randomisation balances confounders.
• The only intended difference between groups is the treatment.
• Allows proper blinding.

If you’d like, I can also compare simple vs restricted randomisation in more detail — that’s another common exam question.

Question 2

What is restricted randomisation?

Answer 2

🎲 What is Restricted Randomisation?

Restricted randomisation is a method used in clinical trials to keep the treatment groups balanced in size or characteristics, instead of relying purely on simple coin-flip randomisation.

In simple terms (rewording your image clearly):

> It is any procedure used to control random allocation so that the groups stay similar in numbers or important characteristics (like age or sex).

This helps prevent one group from accidentally becoming much bigger or very different from the other.

🔀 Types of Restricted Randomisation

1️⃣ Blocking

Blocking keeps the number of participants in each group closely balanced throughout the trial.

For example, if the block size is 4, then within every 4 participants, 2 might go to treatment and 2 to control.

• ✅ Advantage: Keeps group numbers very balanced.
• ⚠️ Disadvantage: If researchers figure out the block size, they might predict the next assignment and influence allocation.

2️⃣ Stratification

In small trials, groups may accidentally differ in important baseline characteristics (like age or sex), which can weaken internal validity (increase confounding bias).

Stratification solves this by:

• Dividing participants into subgroups (called strata) based on important characteristics (e.g., male/female, young/old).
• Performing separate randomisation within each subgroup.

This ensures the treatment groups are balanced within each stratum.

3️⃣ Minimisation

Minimisation is another restricted method that balances several patient factors at once.

• It ensures good balance between groups for multiple characteristics.
• Especially useful in small trials.
• Makes small groups closely similar throughout the study.

🧠 Why use restricted randomisation?

Simple randomisation can accidentally create uneven groups, especially in small studies.

Restricted methods help:

• Keep numbers similar.
• Keep important characteristics evenly distributed.
• Improve the study’s internal validity.

🏥 Clinical Example (1–3 sentences)

A small trial is testing a new ADHD medication. Researchers worry that age and sex could affect outcomes. They stratify participants by age group and sex, then randomise within each subgroup to ensure both treatment groups have similar numbers of boys, girls, younger children, and older children.

👶 Explanation for a 10-Year-Old (1–3 sentences)

Imagine picking teams for a game, but you make sure each team has the same number of tall kids and short kids. You still pick randomly — but you keep things fair.

🧠 Memory Hook (funny, logical, interesting)

“Restricted = Random… but with rules so it doesn’t get ridiculous.” 🎲📏

🚨 Key Takeaways

• Restricted randomisation = controlled randomisation to maintain balance.
• Used to balance group size and important characteristics.
• Types: Blocking, Stratification, Minimisation.
• Especially important in small trials.
• Improves internal validity and reduces confounding.

Question 3

What is a technique used to reduce selection bias?

Answer 3

Allocation concealment.

Selection bias happens when someone unfairly chooses who goes into which group, so the groups aren’t truly equal from the start.

Allocation concealment prevents this by hiding which group the next person will go into, so no one can secretly choose who gets the “better” group.

It’s like putting team names in sealed envelopes so you can’t peek and give your best friend the strongest team. 🎁

Question 4

How does allocation concealment reduce selection bias?

Answer 4

🔒 What is Allocation Concealment?

Allocation concealment is a method used in randomized controlled trials to prevent selection bias by hiding the upcoming group assignments from the people enrolling participants — until the exact moment the patient is assigned.

In simple terms:

> It means the person putting patients into the study does not know in advance which treatment the next patient will receive.

This prevents researchers (consciously or unconsciously) from influencing who goes into which group.

🎯 Why does it reduce selection bias?

Selection bias happens when researchers influence which participants go into certain groups.

Proper allocation concealment ensures:

1️⃣ The allocation sequence is generated properly (truly random).
2️⃣ The sequence is hidden until the participant is officially assigned.

If researchers can’t see future assignments, they can’t manipulate who gets what treatment.

It prevents knowledge of future group assignments.

🧮 Two Key Parts of Proper Randomisation

Randomisation works well only if BOTH are done properly:

1️⃣ Adequate Generation of the Allocation Sequence

✔️ Good methods:

• Computer-generated random numbers
• Table of random numbers
• Drawing lots
• Tossing a coin
• Shuffling cards
• Throwing dice

❌ Poor methods (introduce bias because they can be predicted or linked to prognosis):

• Alternating patients
• Date of birth
• Day/month/year of admission
• Case record numbers

2️⃣ Proper Concealment of the Allocation Sequence

✔️ Good concealment methods:

• Central randomisation (remote site controls allocation)
• Sequentially numbered, sealed, opaque envelopes
• Identical coded drug containers prepared by an independent pharmacy

❌ Poor concealment:

• Alternation
• Unsealed envelopes
• Transparent or non-opaque envelopes

If the next assignment can be guessed, concealment has failed.

⚠️ Also important:
Bias can occur depending on who generates the allocation sequence, who enrolls participants, and who assigns them to groups. If one person controls all steps, bias risk increases.

🏥 Clinical Example (1–3 sentences)

In a trial comparing a new chemotherapy drug to standard treatment, the allocation sequence is stored in a central computer system. When a patient is enrolled, the clinician calls the system to receive the assignment. Because the clinician cannot see the upcoming allocations, they cannot influence which patient receives which treatment.

👶 Explanation for a 10-Year-Old (1–3 sentences)

Imagine putting team names inside sealed envelopes. You’re not allowed to peek before handing one to a player. That way, you can’t secretly give your best friend the better team.

🧠 Memory Hook (funny, logical, interesting)

“Conceal it so you can’t steal it.” 🔒 — if you can’t see the next assignment, you can’t cheat.

🚨 Key Takeaways

• Allocation concealment hides future treatment assignments.
• Prevents selection bias.
• Requires proper random sequence generation AND hiding the sequence.
• Good methods = computer systems, sealed opaque envelopes.
• Bad methods = predictable patterns or transparent envelopes.
• Different from blinding (concealment happens before assignment).

Question 5

What is a technique used to reduce measurement bias?

Answer 5

Blinding/ Masking

Measurement bias happens when the way we measure or judge something is unfair, often because someone knows which group a person is in and that changes how they record results.

Blinding (masking) helps prevent this by keeping people (like patients or researchers) from knowing who got which treatment, so they measure and report things more fairly.

It’s like grading a test without knowing whose paper it is — you’re less likely to give your friend extra points. 📄

Question 6

How does blinding/ masking reduce measurement bias?

Answer 6

🎭 How does Blinding (Masking) reduce measurement bias?

Blinding (masking) means keeping certain people in a clinical trial unaware of which treatment a participant received.

In simple terms (rewording your image clearly):

> It is the practice of keeping trial participants, healthcare providers, data collectors, and people analysing the data unaware of the assigned treatment.

📉 How does this reduce measurement bias?

Measurement bias happens when knowing who received which treatment changes how outcomes are measured, reported, or interpreted.

Blinding reduces this by:

• Preventing performance bias (people behaving differently because they know the treatment).
• Preventing observer/assessment bias (researchers judging outcomes differently based on expectations).

If nobody knows who got what, measurements are more objective and fair.

👀 Types of Blinding

🔓 Open Label

Both the patient and researcher know the treatment given.

👁️ Single Blind

Either the patient or the researcher knows — but not both.

👁️👁️ Double Blind

Both patients and researchers do not know which treatment was given.

👁️👁️👁️ Triple Blind

Patients, researchers, and outcome assessors all do not know the treatment assignment.

In some studies, researchers even test how successful the blinding was by asking participants or investigators which treatment they think was given.

🏥 Clinical Example (1–3 sentences)

In a trial testing a new pain medication, both patients and doctors are blinded to whether the pill is the real drug or a placebo. If doctors don’t know who received the real medication, they are less likely to unintentionally rate one group’s pain as improved just because they expect it to work.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If you’re judging a cookie contest but don’t know who baked which cookie, you’ll taste and score them more fairly. If you knew your best friend made one, you might score it higher without realising.

🧠 Memory Hook (funny, logical, interesting)

“Blind the judge so the score won’t fudge.” 👀🚫

🚨 Key Takeaways

• Blinding keeps participants and/or researchers unaware of treatment assignments.
• Prevents performance bias and observer/assessment bias.
• Types: open label, single, double, triple blind.
• Makes outcome measurement more objective and fair.

Question 7

What is a technique used to reduce attrition bias?

Answer 7

Intention to treat analysis.

Attrition bias happens when more people drop out of one group than the other, which can make the results look unfair or misleading.

Intention-to-treat analysis helps fix this by counting everyone in the group they were originally placed in — even if they quit or didn’t follow the rules.

It’s like keeping players on the scoreboard for their team even if they leave the game early, so the final score stays fair. 🏃‍♂️

Question 8

What is intention-to-treat analysis and how does it reduce attrition bias?

Answer 8

🎯 What is Intention-to-Treat (ITT) Analysis?

Intention-to-treat analysis is a method used in randomized controlled trials (RCTs) where all participants are analysed in the groups they were originally assigned to, even if they:

• Stop the treatment
• Switch treatments
• Drop out
• Don’t follow instructions

In simple terms (rewording your image clearly):

> ITT means everyone stays in the analysis in the group they were allocated to at the start — no one gets removed later.

📉 How does ITT reduce attrition bias?

Attrition bias happens when people drop out unequally between groups, which can distort results.

ITT reduces this by:

• Keeping all participants in the analysis.
• Preventing researchers from only analysing the people who completed the study.
• Maintaining the fairness created by randomisation.

This makes the results more realistic and less biased.

🔎 Common Methods Used in ITT

When participants drop out and outcome data are missing, researchers may estimate outcomes using methods such as:

• Last Observation Carried Forward (LOCF):
  The last recorded result is used as the final result.
• Worst Case Scenario (WCS):
  Dropouts are assumed to have the worst possible outcome (to avoid overestimating benefit).

📊 Types of Outcomes in ITT

Outcomes measured in RCTs can be:

• Continuous outcomes (measured on a scale):
  Example: Change in symptom scores like PANSS or MADRS.
• Dichotomous outcomes (yes/no outcomes):
  Example: Alive vs dead, or 50% improvement vs no improvement.

🏥 Clinical Example (1–3 sentences)

A trial tests a new antidepressant. Some patients in the medication group drop out because of side effects. In ITT analysis, those patients are still counted in the medication group, and their last recorded depression score may be used in the final analysis.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If two teams start a race, you count everyone on the team at the end — even if some kids stop running halfway. That way, the score doesn’t unfairly ignore the kids who quit.

🧠 Memory Hook (funny, logical, interesting)

“Once assigned, you’re signed — you stay on the team!” 📝⚽

🚨 Key Takeaways

• ITT = analyse everyone in the group they were originally assigned to.
• Minimises attrition bias.
• Preserves randomisation.
• Uses methods like LOCF or worst-case scenario if data are missing.
• Reflects real-world effectiveness rather than ideal conditions.

Question 9

What do these terms refer to? (One sentence answer.)
1. Control event rate
2. Experimental event rate
3. Absolute Risk Reduction/ Absolute Benefit Increase
4. Relative Risk
5. Number Needed to Treat/ Number Needed to Harm

Answer 9

These are all important measures of effect.

Question 10

Control event rate (CER) is an important measure of effect. What is it?

Answer 10

📊 What is Control Event Rate (CER)?

Control Event Rate (CER) is the proportion (percentage) of people in the control group who experience the event being studied.

In simple terms (rewording your image clearly):

> It is the percentage of patients in the control group in whom the outcome (event) actually happens.

The “event” could be something good (recovery) or something bad (stroke, relapse, death).

🧠 Why is it important?

CER tells us how common the outcome is without the new treatment.
It acts as the baseline risk, which helps us calculate other important measures like absolute risk reduction (ARR) and number needed to treat (NNT).

🏥 Clinical Example (1–3 sentences)

In a trial studying a new heart medication, 20 out of 100 patients in the control group have a heart attack. The control event rate (CER) is 20%. This tells us the baseline risk of heart attack without the new drug.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If 10 out of 50 kids in the “no helmet” group fall off their bikes, the control event rate is how many fell in that group. It shows what happens when you don’t use the new safety rule.

🧠 Memory Hook (funny, logical, interesting)

“CER = Control Experiences the Result.” 🎯

🚨 Key Takeaways

• CER = proportion of events in the control group.
• Shows baseline risk.
• Used to calculate treatment effect measures.
• Important for understanding how helpful a treatment really is.

Question 11

Experimental event rate (EER) is an important measure of effect. What is it?

Answer 11

📊 What is Experimental Event Rate (EER)?

Experimental Event Rate (EER) is the proportion (percentage) of people in the experimental (treatment) group who experience the event being studied.

In simple terms (rewording your image clearly):

> It is the percentage of patients in the intervention group in whom the outcome (event) occurs.

The “event” can be something good (recovery) or something bad (stroke, relapse, death).

🧠 Why is it important?

EER tells us how often the outcome happens with the new treatment.
We compare it to the Control Event Rate (CER) to see whether the treatment helps, harms, or makes no difference.

🏥 Clinical Example (1–3 sentences)

In a trial of a new cholesterol drug, 10 out of 100 patients in the treatment group have a heart attack. The experimental event rate (EER) is 10%. This can then be compared to the control group’s rate to measure how effective the drug is.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If some kids wear a new kind of helmet and 2 out of 20 fall off their bikes, the EER is how many fell in the helmet group. It shows what happens when you use the new thing.

🧠 Memory Hook (funny, logical, interesting)

“EER = Experiment Experiences the Result.” 🧪🎯

🚨 Key Takeaways

• EER = proportion of events in the experimental (treatment) group.
• Shows what happens with the new treatment.
• Compared with CER to calculate treatment effect.
• Helps determine whether the treatment works or causes harm.

Question 12

Absolute Risk Reduction & Absolute Benefit Increase are important measures of effect. What are they?

Answer 12

📊 What are Absolute Risk Reduction (ARR) and Absolute Benefit Increase (ABI)?

Both ARR and ABI measure the actual difference in event rates between two groups in a study.

They answer the question:
👉 “How much difference does this treatment really make?”

🔴 Absolute Risk Reduction (ARR)

ARR is used when the treatment is meant to reduce bad outcomes (like death, stroke, relapse).

It tells us how much the treatment lowers the risk compared to the control group.

Formula:

ARR = CER – EER

(CER = Control Event Rate, EER = Experimental Event Rate)

So we subtract the event rate in the treatment group from the control group.

🟢 Absolute Benefit Increase (ABI)

ABI is used when the treatment is meant to increase good outcomes (like recovery or response to treatment).

It also measures the actual difference between groups.

Formula:

ABI = EER – CER

(When the outcome is positive.)

🧠 What’s the difference?

• ARR → used when reducing negative outcomes
• ABI → used when increasing positive outcomes
• Both measure the absolute difference, not a ratio

🏥 Clinical Example (1–3 sentences)

In a trial of a heart medication, 20% of patients in the control group have a heart attack (CER = 20%), and 10% in the treatment group have one (EER = 10%).
ARR = 20% – 10% = 10%.
This means the drug reduces heart attack risk by 10 percentage points.

If instead 30% of treated patients recover compared to 20% in control, ABI = 30% – 20% = 10%.
The drug increases recovery by 10 percentage points.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If 20 out of 100 kids fall without helmets, but only 10 out of 100 fall with helmets, the helmet lowers the number by 10. That “10” is the real difference the helmet makes.

🧠 Memory Hook (funny, logical, interesting)

“Absolute = Actual difference.” 📏 — just subtract and see the real gap.

🚨 Key Takeaways

• ARR = reduces bad outcomes.
• ABI = increases good outcomes.
• Both measure the actual percentage difference.
• Formula is simple subtraction.
• Helps show the real-world impact of a treatment.

Question 13

How do Absolute Risk Reduction & Absolute Benefit Increase relate to the Number Needed to Treat?

Answer 13

🔗 How ARR / ABI Connect to Number Needed to Treat (NNT)

Once you calculate Absolute Risk Reduction (ARR) or Absolute Benefit Increase (ABI), you can turn that difference into something very practical:

> Number Needed to Treat (NNT)

📊 What is NNT?

NNT = 1 ÷ ARR
(or 1 ÷ ABI, if measuring benefit)

⚠️ Important: Convert percentages into decimals before calculating.

NNT tells you:

👉 How many patients need to receive the treatment for one extra person to benefit (or avoid harm)?

🏥 Clinical Example

From earlier:

• CER = 20% (0.20)
• EER = 10% (0.10)
• ARR = 0.10

Now calculate:

NNT = 1 ÷ 0.10 = 10

This means you need to treat 10 patients to prevent 1 additional heart attack.

If it were ABI (for a positive outcome), the calculation is the same.

🧠 Why is this useful?

ARR tells you the size of the difference.
NNT tells you how clinically meaningful that difference is.

• Large ARR → small NNT → very effective treatment
• Small ARR → large NNT → less impactful treatment

👶 Explanation for a 10-Year-Old

If giving helmets reduces falls by 10 out of 100 kids, that means for every 10 kids wearing helmets, 1 fall is prevented. So you need 10 helmets to stop one extra fall.

🧠 Memory Hook

“Flip the risk to know the fix.” 🔄
(Flip ARR to get NNT.)

🚨 Quick Summary Chain

CER & EER → subtract → ARR/ABI → flip it → NNT

That’s the full exam flow.

Question 14

The Number Needed to Treat (NNT) and the Number Needed to Harm (NNH) are both measures of effect. What are they?

Answer 14

📊 What are Number Needed to Treat (NNT) and Number Needed to Harm (NNH)?

Both NNT and NNH tell us how many people need to receive a treatment for one additional outcome to happen — either a good outcome or a bad one.

🟢 Number Needed to Treat (NNT)

NNT is the number of people who need to be treated for one extra person to benefit.

It is calculated using:

NNT = 1 ÷ Absolute Benefit Increase (ABI)
(or 1 ÷ Absolute Risk Reduction (ARR), when reducing bad outcomes)

It tells us how effective a treatment is in practical terms.

🔴 Number Needed to Harm (NNH)

NNH is the number of people who need to receive a treatment for one extra person to experience harm.

It is calculated using:

NNH = 1 ÷ Absolute Risk Increase (ARI)

It tells us how risky a treatment might be.

🧠 What do they mean in real life?

• Small NNT = very effective treatment
• Large NNH = safer treatment
• Ideally, we want low NNT and high NNH

🏥 Clinical Example (1–3 sentences)

In a depression trial, 20% recover in the control group, and 30% recover in the treatment group (ABI = 10% or 0.10).
NNT = 1 ÷ 0.10 = 10.
So you need to treat 10 patients for one extra person to recover (for example, achieving HDRS < 9 at completion).

If 5% more patients on the drug develop nausea compared to placebo (ARI = 5% or 0.05),
NNH = 1 ÷ 0.05 = 20.
So one extra person experiences nausea for every 20 treated.

👶 Explanation for a 10-Year-Old (1–3 sentences)

If giving 10 kids a new medicine helps one extra kid feel better, the NNT is 10.
If one kid out of every 20 gets a tummy ache because of it, the NNH is 20.

🧠 Memory Hook (funny, logical, interesting)

“Treat to meet, Harm to alarm.” 📈🚨

🚨 Key Takeaways

• NNT = how many to treat for one extra good outcome.
• NNH = how many treated before one extra harm occurs.
• NNT = 1 ÷ ABI (or ARR).
• NNH = 1 ÷ ARI.
• Best treatments have low NNT and high NNH.

Question 15

Interpret this image in light of what you have learned above.

Answer 15

The image is showing how to calculate treatment effect from a simple clinical trial comparing:

• Fluoxetine + folic acid (the intervention / experimental group)
  vs
• Fluoxetine + placebo (the control group)

The outcome they care about is depression recovery.

Step-by-step: what each part means

1) Start with the raw counts (the “2×2” idea)

• In the treatment group, 33 recovered and 18 did not recover.
  So total treated = 33 + 18 = 51.
• In the control group, 28 recovered and 30 did not recover.
  So total control = 28 + 30 = 58.

The letters in brackets are just labels:

• a = 33, b = 18, c = 28, d = 30

2) Calculate EER (Experimental Event Rate)

EER means: “What proportion improved in the treatment group?”

• EER = a / (a + b)
• EER = 33 / 51
• That’s about 0.65 = 65% (the image rounds to 64%)

So: About 64–65% recovered with fluoxetine + folic acid.

3) Calculate CER (Control Event Rate)

CER means: “What proportion improved in the control group?”

• CER = c / (c + d)
• CER = 28 / 58
• That’s about 0.48 = 48%

So: About 48% recovered with fluoxetine + placebo.

4) Calculate ABI (Absolute Benefit Increase)

Because “recovery” is a good outcome, we use ABI (benefit increase).

ABI means: “How many more people recover because of the treatment?” (in absolute percentage points)

• ABI = EER − CER
• ≈ 64% − 48% = 16%

So: Adding folic acid increases recovery by about 16 percentage points.

5) Calculate NNT (Number Needed to Treat)

NNT means: “How many people need this treatment for 1 extra person to benefit?”

• NNT = 1 / ABI
  (but ABI must be a decimal: 16% = 0.16)
• NNT = 1 / 0.16 ≈ 6.25

So it’s about 6 people (the image says 6).
Tiny exam nuance: many guidelines say round NNT up to the next whole number → 7, but lots of teaching slides round to the nearest whole number.

Meaning: Treat ~6–7 people with folic acid + fluoxetine to get 1 extra recovery compared with placebo.

6) Calculate RBI (Relative Benefit Increase)

RBI means: “How much better is the treatment group compared to control, relative to control?”

• RBI = (EER − CER) / CER × 100
• ≈ (0.64 − 0.48) / 0.48 × 100
• ≈ 0.16 / 0.48 × 100 ≈ 33%

Meaning: Recovery is ~33% higher (relatively) with folic acid added, compared with placebo.

(Absolute difference was 16%, relative difference is ~33%—both can be true.)

🏥 Clinical example (1–3 sentences)

You run an RCT in depression: everyone gets fluoxetine, but half also get folic acid. Recovery is 64% with folic acid vs 48% with placebo, so ABI is 16% and NNT is about 6–7—meaning for every ~6–7 people you add folic acid to, one additional person recovers.

👶 Explanation for a 10-year-old (1–3 sentences)

One group gets medicine plus a “bonus vitamin,” and the other group gets medicine plus a fake pill. More kids get better in the bonus vitamin group, and the math shows you need to give the bonus vitamin to about 6 kids to help one extra kid get better.

🧠 Memory hook (1 sentence)

“EER is the ‘Extra-help’ group, CER is the ‘Control’ group—subtract for the real-life boost, then flip it to get NNT.” 🔁