Key Takeaways
- A benefit is not “the treatment works.” It is a defined outcome, for a defined group, over a defined period—and it should matter to the patient.
- Prefer absolute numbers such as “3 in 100” to relative phrases such as “twice the risk.” Use the same denominator and time period for every option [1].
- Show both frames: if 92 in 100 avoid an event, 8 in 100 experience it. Hearing only the reassuring or alarming half can distort a choice [1].
- Separate likelihood from seriousness. A frequent temporary side effect and a rare permanent harm need different conversations.
- “No treatment” is still an option with consequences. Compare it, observation and other active treatments using the same outcomes.
- Plain language is not enough until understanding is checked. Teach-back asks the patient to explain the decision in their own words and tests the explanation, not the person [2].
Content
Medical decisions often fail in the space between a technically correct sentence and what the patient hears. “This reduces risk by 50%” may sound dramatic even when the chance changes from 2 in 1,000 to 1 in 1,000. “Minimally invasive” may be heard as “minor.” “Most patients do well” says nothing about who was studied, what “well” means or how long it lasted.
Good risk communication does not remove uncertainty. It makes the uncertainty visible enough to use. NICE recommends personalizing information, using absolute risk, natural frequencies, consistent denominators and time periods, and showing both positive and negative framing [1]. The aim is not to make a patient choose treatment; it is to make the available choices understandable.
First, turn each option into one complete sentence
For every option, ask the clinician to complete:
For people like me, this option aims to , measured by , over , compared with .
“People like me” matters. Evidence from younger, lower-risk, newly diagnosed patients may not apply cleanly to an older person with prior treatment and several chronic illnesses. “Measured by” prevents a laboratory change from being confused with feeling or living better. “Over” supplies the time horizon. “Compared with” reveals whether the alternative is usual care, another treatment, observation or placebo.
If no reliable number exists, that is not a reason to invent one. Ask for the best available range, the source, why it may or may not fit this patient and what decision will be revisited when new information arrives.
Translate benefits into outcomes the patient can recognize
Benefit can mean very different things:
- living longer;
- avoiding a stroke, fracture or hospital admission;
- curing disease or delaying progression;
- relieving pain, breathlessness or bleeding;
- walking, eating, speaking, working or sleeping better;
- obtaining a diagnosis;
- reducing treatment burden.
Ask whether the outcome is directly experienced by patients or is a surrogate, such as a scan, biomarker or laboratory value. A surrogate may be clinically useful, but its relationship to survival, symptoms or function should be explained rather than assumed.
Replace “responded” with the response definition and duration. Replace “successfully completed” with the outcome after the procedure. Technical success can coexist with persistent pain, loss of function or a need for another treatment.
Put the baseline back into every percentage
Relative risk describes a proportional change; absolute risk shows the actual difference. Consider a purely hypothetical example:
- Without treatment, 4 in 100 people have the event within one year.
- With treatment, 2 in 100 have it within one year.
- The relative reduction is 50%.
- The absolute reduction is 2 in 100.
All four statements can describe the same comparison, but “50% lower” alone hides the starting point. Ask: “Out of 100 people like me, how many would have the outcome with each option?”
NICE advises natural frequencies such as 10 in 100, the same denominator across comparisons and a defined period [1]. Do not compare 1 in 20 with 40 in 1,000 in the same table; rewrite both as 50 in 1,000 and 40 in 1,000. Do not compare a one-month complication with a five-year benefit without clearly labelling the different clocks.
Read the reassuring and alarming versions together
These two statements carry the same information:
- 92 in 100 patients will not experience the complication.
- 8 in 100 patients will experience it.
Seeing both helps counter framing. A visual array of 100 equally sized icons can help, provided the colors, denominator and legend stay consistent. Pictures should clarify numbers, not replace them; oversized red symbols can make a small risk feel larger.
Words such as “common,” “rare,” “likely” and “good chance” are interpreted differently by different people. CDC’s numeracy guidance notes the limitations of relying on vague words alone or numbers alone and supports combining them with explanation [3]. Ask for a number or range whenever one is reasonably available.
Separate chance, severity, timing and reversibility
A risk statement needs four parts:
- Chance: how often it occurs in a relevant group.
- Severity: what actually happens to the patient.
- Timing: during treatment, early recovery or years later.
- Reversibility: temporary, treatable, disabling or permanent.
“Nerve injury is rare” is incomplete. Which nerve function? Temporary numbness or permanent weakness? How often in similar cases? How is it detected and treated?
Ask about common burdens, serious harms and material risks—outcomes that may be especially important to this patient because of work, caregiving, fertility, faith, sport or independence. A professional singer may weigh a small voice risk differently; a person living alone may weigh temporary inability to climb stairs heavily.
Also ask about competing risks. Avoiding one event may increase bleeding, infection, kidney injury, cognitive effects or treatment burden. Benefits and harms should not be presented on separate marketing pages with different denominators.
Do not turn a range into a promise
Evidence contains uncertainty from sample size, study design, follow-up, patient selection and differences in how outcomes are measured. Ask:
- Is this a single estimate or a range?
- How many patients and events produced it?
- Was it from a randomized trial, registry, published series or local audit?
- How similar were they to this patient?
- What happened to people who were lost to follow-up?
- Which important outcome was not measured?
A range of 2–8 in 100 should not become “about 2%.” When individual data are unavailable, say so and explain the direction of uncertainty. A risk calculator can help organize known predictors; it is not a crystal ball.
Build one comparison grid, not three sales pitches
Place all reasonable options in columns, including observation or no immediate treatment. Use the same rows:
Question · Option A · Option B · Observe / no treatment now
Main goal
Chance of that benefit and time
Common burdens
Serious or irreversible harms
Visits, hospital days and recovery
Effect on daily life
What happens if it fails
Evidence uncertainty
“No treatment” does not mean “nothing happens.” It may involve monitoring, symptom control and predefined triggers for action; it may also carry a risk of progression. The comparison must describe those consequences without using fear to force consent.
Shared decision making uses evidence and clinical expertise alongside the patient’s goals, preferences and circumstances [4]. After filling the grid, ask which trade-off matters most—not which option sounds most advanced.
Replace medical shorthand with action language
Plain language puts the important message first, uses familiar words and keeps one main idea per sentence [5]. Ask the clinician to translate:
- “positive margin” → abnormal cells reach the cut edge of the removed tissue;
- “watchful waiting” → which symptoms are treated and what triggers reassessment;
- “contraindicated” → this option should not be used, and why;
- “non-inferior” → the study tested whether it was not worse by more than a stated margin;
- “five-year survival” → which patients were alive five years after a defined starting point;
- “minimally invasive” → the access method, not a promise of minimal risk.
Translation must preserve negation, comparison words, units and uncertainty. “Cannot exclude,” “no evidence of” and “not statistically significant” are not interchangeable. Use a qualified medical interpreter; family can support the patient but should not carry sole responsibility for translating a high-stakes decision.
Use teach-back to expose the weak sentence
“Do you understand?” often produces a polite yes. Instead say:
I want to check that we explained this clearly. In your own words, what are the options, the main benefit you hope for, the harm you are most concerned about and what happens if you wait?
AHRQ describes teach-back as asking the patient or caregiver to explain what they need to know or do in their own words; misunderstanding should trigger a different explanation and another check [2]. It is not a memory exam. The patient may refer to the option grid.
For a cross-language decision, the interpreter should remain for teach-back. Check numbers separately: “Out of 100, how many?” Then check the plan: “Which symptom means you need urgent help?” The Joint Commission emphasizes that informed consent depends on understanding the nature, risks, benefits and alternatives—not simply signing a form [6].
The conversation is ready only when the patient can describe the choice without copying the brochure and the clinician can name what matters most to that patient.
FAQ
1. Is relative risk wrong?
No, but it is incomplete without baseline risk. Ask for the absolute numbers with and without each option, using the same denominator and time frame.
2. What if the doctor cannot give an exact percentage?
Ask for a reasonable range, the evidence source, why it may differ for this patient and which new information could narrow the uncertainty. False precision is not better information.
3. Does “rare” mean the risk can be ignored?
No. Likelihood is only one dimension. A rare but permanent or catastrophic harm may be material to the patient and should be discussed.
4. Are pictures always easier than numbers?
Not automatically. Icon arrays can help when they use the same denominator and neutral design, but confusing colors or changing scales can mislead. Use pictures alongside clear numbers [1].
5. Who should perform teach-back when an interpreter is used?
The responsible clinician should ask; the qualified interpreter should translate the full exchange. Teach-back checks the clinician’s explanation and should not be delegated to a family member.
Sources
- National Institute for Health and Care Excellence — Communicating Risks, Benefits and Consequences
- Agency for Healthcare Research and Quality — Use the Teach-Back Method
- U.S. Centers for Disease Control and Prevention — Numeracy and Health Literacy
- Agency for Healthcare Research and Quality — About Shared Decision Making
- U.S. Centers for Disease Control and Prevention — Plain Language Materials and Resources
- The Joint Commission — Informed Consent: More Than Getting a Signature