Medical Content
Medical Accuracy Is Not the Same as Medical Communication
By Miracle Olajuyigbe 5 min read
Here is a sentence that is completely true:
Metformin is associated with gastrointestinal adverse effects in a significant proportion of patients, which are typically transient and dose-dependent. Every word of that is defensible. A pharmacologist would sign it. It would survive any fact-check you threw at it.
It is also, for the person who just got handed a prescription and is reading the leaflet in a car park, close to useless. It does not tell them what will actually happen, when, whether to worry, or what to do about it.
Here is the same information, doing its job:
About one in four people get diarrhoea, nausea, or stomach cramps when they start metformin. It is unpleasant and it is not dangerous. For most people it settles within a couple of weeks as the body adjusts. Taking the tablet with food, and not on an empty stomach, makes a real difference. If it has not eased after a month, tell your doctor. There is a slow-release version that many people tolerate much better, and switching is straightforward. Same facts. Same accuracy. Completely different outcome for the person reading it.
That gap is the thing I want to talk about, because a lot of healthcare content lives on the wrong side of it while believing it has done its job.
Accuracy is the floor, not the goal
Accuracy is necessary. Nobody is arguing otherwise, and getting it wrong in health content is not a small thing.
But accuracy is a property of the text. Communication is a property of what happens in the reader’s head afterwards. You can achieve the first and completely fail the second, and no amount of fact-checking will catch it, because fact-checking asks “is this true” and never asks “did this work.”
The better test is behavioural. After someone reads this, do they know what to do? Do they do it? Are they less frightened than they should be, more frightened than they should be, or about right?
That last one matters more than people think. Calibrating fear is part of the job. Content that terrifies someone about a benign symptom has failed, and so has content that soothes someone away from a symptom that needed attention this week.
The curse of knowing things
Once you understand something, you lose the ability to remember not understanding it. This is not a character flaw. It happens to everyone, and it is the single biggest reason expert-written health content underperforms.
You stop noticing that “contraindicated” is not a normal word. That “presents with” is hospital dialect. That when you write “monitor your levels,” you have a specific picture of what that involves and your reader has none. You skip the step that seems obvious, and it seems obvious only because you already took it, years ago, and cannot see it anymore.
The practical fix is embarrassingly low-tech. Read the draft as though you have never encountered the topic and see where you would stumble. Better still, hand it to somebody who genuinely has not, and watch where they slow down. The place they reread is the place you skipped a step.
Numbers are where good intentions go to die
Most people, including a fair number of clinicians, are poor at probability. Content that ignores this is technically accurate and practically misleading.
“This doubles your risk” is a sentence I would ban if I could. It sounds enormous. It might mean a risk went from two in ten thousand to four in ten thousand, which is a rounding error in most people’s lives, or it might mean something that genuinely changes a decision. The reader has no way to tell, and the phrasing pushes them toward alarm regardless.
Give them the actual numbers, in a form humans can picture. Not “a 0.04 percent absolute risk increase,” which is technically better and still unreadable. Try: “out of every ten thousand people who take this, about two would normally have this problem. On this medication, about four would.” Nobody needs a statistics course to work with that.
The same logic applies to timeframes. “Long-term use” means nothing. “After about five years” means something. Vague quantifiers feel cautious and function as noise.
Structure is not decoration
Where information sits changes whether it gets read at all.
Most people do not read a health page from top to bottom. They scan, they stop where something catches, they leave when they think they have what they came for. If the thing that matters most is in paragraph nine, it might as well not be there.
So put the reassurance early if the answer is reassuring. Put the red flags somewhere they cannot be missed by a scanner, which usually means their own heading and their own short paragraph, not buried in a list of nine bullet points where they carry the same visual weight as the trivial items. If someone needs to act today, say so before you explain the mechanism.
Explaining how something works is often the least urgent thing on the page, and it is very frequently where writers start, because it is the part they find most interesting.
Plain language is not simplification
There is a persistent worry that writing plainly means writing down to people, and that it costs credibility.
It does the opposite. Anyone can hide an incomplete understanding behind technical vocabulary. Explaining something clearly, in ordinary words, without losing the parts that matter, is much harder and demonstrates far more command of the material. If a specialist reads your plain-English explanation and finds nothing to object to, you have proven more than any amount of terminology would.
What plain language does not mean is stripping out nuance. If a treatment works well for one group and poorly for another, say so. If the evidence is genuinely uncertain, say that too, in a way that tells the reader what to do with the uncertainty rather than just transferring it to them.
“We do not know yet” is a fine sentence. “We do not know yet, so the usual approach is to try X first and reassess in six weeks” is a much better one.
The test I use
Before anything goes out, one question: if the person I am picturing reads this once, on a phone, while distracted, will they end up doing the right thing?
Not “is every claim defensible.” Not “would a specialist approve.” Those matter, and they are the entry requirement, not the finish line.
Getting the facts right is the part you can check with a reference. Getting them across is the part that takes actual work, and it is the only part the reader ever experiences.