Chapter 36: Writing for Medicine and Healthcare

⚠️ Disclaimer: This chapter is for educational purposes only and does not constitute medical advice. Clinical decisions should always be made in consultation with qualified professionals and current clinical guidelines.

"First, do no harm" is a clinical principle. It is also a writing principle. — a maxim attributed to the Hippocratic tradition

Chapter Overview

A nurse reads a discharge instruction at the end of a twelve-hour shift, hands it to a patient who is frightened, in pain, and not listening closely. The patient drives home, reads it again at the kitchen table, and decides what "take as needed for pain, do not exceed recommended dose" means in their own kitchen, with their own bottle, at two in the morning. No clinician is in the room. The writing is the clinician now. If the instruction can be misread, someone will misread it—not because they are careless, but because that is what ambiguity does under stress.

This is what makes medical and healthcare writing different from everything else in this book. Throughout, we have argued that audience is everything (Chapter 2) and that ambiguity is a flaw to be cut (Chapter 3). Here those principles stop being matters of craft and become matters of safety. A buried conclusion in a status report wastes a manager's time; a buried conclusion in a clinical note can mean the next clinician misses the one fact that changes the plan. A vague step in a software tutorial frustrates a developer; a vague step in a medication instruction can put someone in the emergency room. The stakes raise the bar on everything we have already learned, and they add conventions you will find nowhere else: the SOAP note, health literacy, teach-back, informed consent, evidence grading.

And the difficulty compounds, because in healthcare you are almost never writing for one reader. The same clinical fact must often serve three audiences at once—the clinician who needs precision and speed, the patient who needs to understand and act, and the regulator who needs traceability and proof. Chapter 2 taught you to write one finding four ways for four different readers. This chapter teaches you to do it under the constraint that getting the audience wrong is not merely ineffective—it can be dangerous. By the end, you will be able to write a clinical note in SOAP structure, rewrite a jargon-dense instruction into plain language a patient can act on, and look at any healthcare document and ask the question that governs the whole field: can this be misread into an unsafe action?

In this chapter, you will learn to:

• Structure a clinical encounter as a SOAP note, sorting each fact into Subjective, Objective, Assessment, or Plan.
• Write patient-facing material at a 6th–8th-grade reading level and prove it works with a teach-back check.
• Translate one clinical fact for three audiences—clinician, patient, regulator—without altering the fact.
• Diagnose and remove the ambiguities in medication and device instructions that cause real-world harm.
• Read evidence-grading language (GRADE) well enough to write and interpret a recommendation honestly.

📕 Engineering/Science Track: This chapter is core for the science track and essential for anyone heading into clinical research, medical device work, regulatory writing, or health communication. If you are a working clinician, sections 36.1 (SOAP) and 36.6 (medication/device instructions) are your daily bread. If you are a researcher, 36.5 (evidence grading) connects directly to Chapters 14–15. Everyone should read 36.2–36.3 (health literacy and teach-back): the plain-language discipline they teach is the single most transferable skill in the chapter, and it is the most often done badly.

36.1 The SOAP Note: Structure as Clinical Safety

Start with a real-looking clinical reality. A patient comes to a clinic. They say their cough has been getting worse for three days and now their chest hurts when they breathe in. The clinician listens to their lungs, takes their temperature, checks their oxygen, decides what is most likely going on, and writes down a plan. All of that—what the patient said, what the clinician found, what the clinician concluded, and what happens next—has to be written down so that the next person to open this chart can pick up exactly where this clinician left off. That next person might be a covering physician at 3 a.m. who has never met the patient. The note is the handoff.

For decades, the dominant structure for that handoff has been the SOAP note: four sections, always in this order.

• S — Subjective: what the patient reports. Their symptoms, their history, their words. "Patient reports a productive cough for three days, now with sharp left-sided chest pain on inspiration."
• O — Objective: what you measured or observed. Vital signs, exam findings, lab and imaging results. Facts that another clinician would record the same way. "Temp 38.4°C. O₂ sat 94% on room air. Crackles at left lung base. (Values illustrative.)"
• A — Assessment: your clinical interpretation. What you think is going on, and why. The diagnosis or differential. "Findings consistent with a left lower-lobe community-acquired pneumonia; less likely a pulmonary embolism given X and Y."
• P — Plan: what you will do about it. Tests ordered, treatment started, follow-up, patient instructions. "Start oral antibiotics per local guideline. Chest X-ray to confirm. Return if breathing worsens. Follow up in 48 hours."

Why this order, and why so rigid? Because the structure encodes the logic of clinical reasoning, and—this is the part that matters for safety—it forces the writer to keep observation separate from interpretation. The single most dangerous error in a clinical note is to smuggle a conclusion into the data, so that the next reader inherits your interpretation as if it were a measurement. SOAP makes that error structurally awkward. A temperature goes in Objective. The belief that the patient "looks septic" goes in Assessment. The discipline of asking "is this something I observed, or something I concluded?" for every sentence is exactly the discipline Chapter 13 taught for the Results-versus-Discussion boundary in a lab report—except that here the cost of blurring it is measured in patients, not in rejected papers.

🚪 Threshold Concept: In medicine, ambiguity is a safety hazard.

Before you cross this threshold, you treat ambiguity the way most writers do: as a quality problem. An unclear sentence is "not great," you'd "tighten it if you had time," and a reader who misunderstands has merely been inconvenienced. After you cross it, you see that in a clinical context an ambiguous sentence is a latent failure—a trap that will eventually be sprung by a tired reader, a distracted patient, or a covering clinician who lacks the context you had in your head. You stop asking "is this clear enough?" and start asking "how could this be misread, and what happens to a person if it is?" That question reorders all your other instincts. Brevity yields to it. Elegance yields to it. The goal is no longer prose that can be read correctly; it is prose that cannot be read incorrectly.

The before/after: a note that blurs the boundaries

Here is a fragment of an admission note, written the way an exhausted novice writes—as a flowing narrative, with observation and conclusion fused.

❌ Before:

Patient is a 58-year-old who clearly has a bad pneumonia, looking quite unwell and septic, with a nasty cough for a few days and some chest pain. Temp is up and sats are a bit low. Started on antibiotics, seems to be the right call, will keep an eye on things and reassess.

Read it as the 3 a.m. covering clinician. What is measured and what is believed? "Clearly has a bad pneumonia" is a conclusion presented as if it were a finding—but the X-ray hasn't confirmed it, and "looking septic" is an impression, not a vital sign. "Temp is up" and "sats are a bit low" are facts, but where are the numbers? "A few days," "a nasty cough," "a bit low," "keep an eye on things"—every one of these is a soft phrase that the next reader has to re-interpret without the context the writer had. And "seems to be the right call" buries the actual plan in a hedge. A covering clinician inherits an opinion dressed as a record.

✅ After:

S: 58-year-old reports a productive cough × 3 days, now with sharp left-sided pleuritic chest pain. No prior episodes. No leg swelling or recent travel.

O: Temp 38.4°C, HR 102, O₂ sat 94% on room air, BP 128/76. Crackles at left base; no calf tenderness. (Values illustrative.)

A: Most likely left lower-lobe community-acquired pneumonia. PE considered less likely given no risk factors and absence of leg findings, but not yet excluded.

P: Start empiric oral antibiotics per local guideline. Chest X-ray to confirm. Monitor O₂ sat; escalate if it falls below 92%. Reassess in 4 hours; return precautions given. Re-evaluate PE risk if no improvement.

Why it's better: Every fact now lives where the next reader expects it. Measurements are in Objective with numbers; the diagnostic belief is in Assessment, explicitly labeled as a likelihood and openly flagged as not-yet-excluded; the plan is concrete and includes a named threshold for escalation ("below 92%") rather than "keep an eye on things." The covering clinician can reconstruct the reasoning, see what is fact and what is judgment, and know exactly what to do if the patient deteriorates. The structure did the safety work. (Note: clinical values, thresholds, and treatment choices here are illustrative for teaching the form of the note—real care follows current guidelines and the patient in front of you.)

Assessment is not the same as diagnosis

A subtle point that trips up learners: the A is not just a diagnosis label. The best assessments show the reasoning—what is most likely, what else is on the differential, and what would change your mind. "Pneumonia" is a label. "Most likely pneumonia; PE less likely but not excluded; will reassess if no improvement in 48h" is a thought, and it tells the next clinician where the uncertainty lives. This is the clinical version of a principle from Chapter 14: state your claim, but show the reader the boundaries of what you actually know. An assessment that hides its uncertainty is as dangerous as a results section that overclaims.

🔄 Check Your Understanding A clinician writes: "O: Patient appears anxious and is probably overreacting to mild discomfort. Temp 37.9°C." Which part of this sentence is in the wrong SOAP section, and why is that a safety problem?

Answer

"Appears anxious and is probably overreacting" is an interpretation, so it belongs in Assessment, not Objective. Worse, "probably overreacting" is a conclusion that could anchor the next clinician to dismiss the patient—if the discomfort is later found to be serious, the record has biased everyone downstream. The temperature (37.9°C) is the only true Objective fact in the sentence. Putting a judgment in the Objective slot disguises opinion as measurement: the exact failure SOAP exists to prevent.

A practical note on variants

You will meet relatives of SOAP. Some teams add a fifth element and call it SOAPE (the E is Evaluation—did the plan work?). Inpatient teams use the problem-oriented progress note, where each active problem gets its own A and P. Emergency and procedural settings use their own templates. Don't memorize the zoo. Internalize the principle: separate what the patient said from what you measured from what you concluded from what you'll do—because the next reader needs to know which is which, and a person's safety can depend on it.

[📍 Good stopping point]

36.2 Health Literacy: Writing So a Patient Can Actually Act

Now change audiences entirely. The clinician note above is precise for clinicians. Hand it to the patient and it is worse than useless—it is a wall the patient can't climb. The most consequential audience shift in all of healthcare writing is the one from clinician to patient, and it is the one most often botched, because the curse of knowledge (Chapter 2) is at its absolute strongest here. A clinician has spent a decade making medical vocabulary automatic. The patient has spent zero days. The clinician literally cannot feel which words are hard.

The discipline that addresses this is health literacy: writing health information so that people can find it, understand it, and use it to make decisions. And there is a stubborn, well-documented gap to close. Research on health literacy consistently finds that a large fraction of adults struggle with everyday health materials—and that the average reading level of the materials handed to patients is far above the average reading level of the patients themselves. The widely repeated guidance, echoed by major public-health and medical communication bodies, is to write patient-facing material at roughly a 6th-to-8th-grade reading level.

People hear "6th-grade level" and flinch—it sounds like dumbing down. It is not. It is the same lesson as Chapter 3, stated for the highest-stakes audience there is: clarity is not the enemy of precision; jargon is. A patient who cannot understand "take this medication on an empty stomach" cannot follow it, no matter how precise the phrase feels to the prescriber. Writing simply for a patient is not condescension. It is the only version of the instruction that has a chance of working. Recall the threshold concept: an instruction the patient can't parse is not merely unhelpful—it is unsafe.

What "6th–8th-grade level" actually means in practice

It is not a vocabulary list you memorize. It is a set of concrete moves:

• Common words over clinical words. "High blood pressure," not "hypertension." "Heart attack," not "myocardial infarction." "Trouble breathing," not "dyspnea." "Swelling," not "edema." When you must use a medical term (because the patient will see it on a label or hear it from another clinician), define it the first time in plain words.
• Short sentences, one idea each. A 40-word sentence with three subordinate clauses is a comprehension test, not an instruction. Break it.
• Active voice, direct address. "Take one pill in the morning," not "One pill is to be taken in the morning." Tell the reader what they do—this is Chapter 3's "free the verb" and Chapter 22's imperative step, now load-bearing for safety.
• Concrete actions, not abstractions. "Drink 8 glasses of water a day," not "maintain adequate hydration." "Maintain adequate hydration" is invisible; it names a goal, not an action a person can perform.
• Numbers and quantities a person can picture. "Take this 1 hour before you eat" beats "administer on an empty stomach." "About the size of a golf ball" beats a measurement in millimeters for a layperson.
• Define the "when to worry." Every patient instruction should answer, in the patient's own likely words: what do I do, how do I know it's working, and when do I call for help? The third one is most often missing and most often the one that matters.

🧩 Productive Struggle Before reading on, try to rewrite this sentence for a patient at a 6th-grade level. Don't peek at the rewrite first—attempt it yourself.

"Maintain NPO status for a minimum of 8 hours prior to the scheduled procedure; failure to comply may necessitate cancellation owing to aspiration risk."

Write your version, then ask: did you remove every clinical term? Did you say what the patient does, and why it matters to them?

One good rewrite

"Do not eat or drink anything for 8 hours before your procedure. (Water, coffee, gum, and candy all count.) This keeps your stomach empty so the procedure is safe. If you eat or drink during this time, we may have to cancel and reschedule."

Notice what changed: "NPO" (Latin nil per os, "nothing by mouth") became "do not eat or drink anything," and we defined the edge cases a patient gets wrong—people think water or gum doesn't count. We said why ("so the procedure is safe"), because a reason that the patient owns is followed far more reliably than a rule imposed without one. And we replaced the clinical threat ("aspiration risk") with the consequence the patient actually cares about ("we may have to cancel"). The fact did not change. The chance the patient acts on it went way up.

The before/after: a discharge instruction no one can follow

This is the chapter's anchor. A patient leaves the emergency department after a small wound was stitched. Here is the discharge instruction they're handed—real in style, anonymized in content.

❌ Before (as written by the clinical team):

Wound care: Keep the laceration site clean and dry. Apply topical antibiotic ointment BID and re-dress with sterile gauze. Monitor for signs of infection including erythema, purulent discharge, increasing edema, or febrile episodes. Suture removal in 10–14 days. Return for any signs of dehiscence or systemic symptoms. Resume activity as tolerated.

Count the barriers. "Laceration site"—a patient says cut. "BID" is clinical shorthand (Latin bis in die, twice a day) that a patient has no reason to know. "Erythema," "purulent discharge," "edema," "febrile," "dehiscence," "systemic symptoms"—six clinical terms, each naming something the patient can absolutely recognize if you describe it, and absolutely cannot act on as written. "Resume activity as tolerated" sounds permissive but tells the patient nothing about what they may actually do. The instruction is precise to the team and a closed door to the patient. And remember who is in the room when they read it: no one.

✅ After (plain language, with teach-back built in):

How to care for your cut at home

Keep it clean and dry. - For the first 24 hours, keep the bandage on and dry. - After that, put a thin layer of the ointment we gave you on the cut 2 times a day (morning and night). Then cover it with a clean bandage.

Watch for signs of infection. Call us or see a doctor if you notice any of these: - The skin around the cut gets red, warm, or more swollen. - Yellow or green fluid (pus) comes from the cut. - The cut opens back up. - You get a fever (temperature over 38°C / 100.4°F) or feel sick all over.

Come back to have the stitches taken out in 10 to 14 days.

What you can do: You can go back to your normal activities when you feel up to it. Avoid heavy lifting or anything that pulls on the cut until the stitches are out.

Questions? Call [clinic number]. (Before you leave, the nurse will ask you to tell them, in your own words, how you'll care for your cut and when you'll call us.)

Why it's better: Every clinical term is gone or translated in place: "erythema" → "red," "purulent discharge" → "yellow or green fluid (pus)," "febrile" → "fever (over 38°C / 100.4°F)," "dehiscence" → "the cut opens back up." The vague "as tolerated" became a concrete boundary. The instruction is organized for a frightened reader who scans (Chapter 4): bolded headers, short lines, a clear "when to call." And it ends with the move that closes the loop—teach-back—which is the subject of the next section. The fact content is identical to the "before." The difference is that this version can be acted on by the person it's for.

🔄 Check Your Understanding A patient leaflet reads: "Adverse reactions may include but are not limited to gastrointestinal disturbances." Name two things wrong with it from a health-literacy standpoint, and rewrite the core message in plain language.

Answer

First, "adverse reactions" and "gastrointestinal disturbances" are clinical terms a patient may not parse—"side effects" and "stomach upset" are the plain equivalents. Second, "may include but is not limited to" is legalistic hedging that tells the patient nothing concrete and signals "this list is incomplete" without telling them what else to watch for. Plain version: "This medicine may upset your stomach. You might feel nauseous or have loose stools. If this gets bad or doesn't go away, call us." It names the likely effect, in plain words, and tells the patient what to do—the part that was missing.

36.3 Teach-Back: The Test for Patient Writing

Chapter 22 gave you the master test for any set of instructions: has someone who has never done this followed them to success, with no help from you? Healthcare has its own version of that test, and it is one of the most powerful tools in patient communication. It is called teach-back (sometimes "show-me" or "closing the loop").

Teach-back is simple to state and humbling to practice. After you explain something, you do not ask "Do you understand?"—because almost everyone says yes, whether they understood or not, especially when they're anxious or embarrassed or trying to seem like a good patient. Instead, you ask the patient to explain it back to you in their own words, framed as a check on your own clarity, not a test of them: "I want to make sure I explained this clearly—can you tell me how you'll take this medicine when you get home?" If they can say it back correctly, the communication worked. If they can't, you didn't fail them—your explanation failed, and now you get to fix it before they leave, not after they're back in the ED.

The connection to this whole book is exact. Teach-back is the curse of knowledge (Chapter 2), defeated by mechanics rather than willpower. You cannot tell from inside your own head whether your explanation landed—your expertise contaminates the judgment. So you externalize the test: you watch the patient reconstruct the message, and the gaps light up. It is the same logic as the "someone who has never done this" test in Chapter 22, the same logic as testing a tutorial on a real beginner in Chapter 26. The reader's reproduction of your meaning is the only evidence that your meaning got through.

What teach-back changes about how you write

Teach-back is usually described as a spoken technique, but it reshapes written patient material too, because written instructions should be designed to be teach-back-able. That means:

• One action per instruction, so the patient can repeat it back as a discrete step (the Chapter 22 rule).
• The "why" attached to the "what," because a patient who understands why can reconstruct the instruction; a patient who memorized a rule can only parrot it.
• A built-in checkpoint, even in writing: "Before you leave, you'll be asked to tell us how you'll take this." This primes the patient that comprehension—not nodding—is the goal.
• The three questions answered explicitly: what do I do, how do I know it's working, when do I get help.

🔍 Why Does This Work? Why does asking a patient to explain instructions back work so much better than asking "Do you understand?" Think about it before reading on.

Answer

Three reasons, all of them about getting past the limits of self-report. (1) Social pressure defeats the yes/no question. "Do you understand?" invites the polite, face-saving "yes"—it costs nothing to say and protects the patient from looking foolish. Reconstruction can't be faked: either the words come out right or they don't. (2) It tests retrieval, not recognition. Recognizing an instruction when you re-read it is easy; producing it from memory is the actual skill the patient needs at home, and it's the same retrieval-practice principle that makes the quizzes in this book work. (3) It relocates the failure. By framing it as "let me check I explained this well," teach-back makes a gap the writer's problem to fix, not the patient's fault to hide—so gaps surface instead of staying buried. The patient's own words become the evidence your writing worked.

✏️ Try This Take any instruction you've written recently—even a non-medical one, like setup steps for a tool. Ask a friend to read it once, then put it away and tell you back what they'd do. Note every place their version diverges from yours. Those divergences are your defects, exactly as a teach-back surfaces a clinician's.

[📍 Good stopping point]

36.4 One Fact, Three Audiences: Clinician, Patient, Regulator

Chapter 2's central exercise was writing one finding four ways for four readers. Healthcare gives that exercise its most demanding form, because three audiences often need the same underlying fact simultaneously, and each will be harmed in a different way by writing for the wrong one. Take a single clinical fact and watch it become three documents.

The fact: A new medication lowered a key risk marker, but caused stomach upset in a meaningful minority of people in the trial, and shouldn't be taken by people with severe kidney problems.

For the clinician (precision and speed; full vocabulary is help, not barrier):

Reduced [marker] vs. placebo in the pivotal trial. GI adverse effects (nausea, dyspepsia) in a notable minority; typically mild and transient. Contraindicated in severe renal impairment. Dose-adjust in moderate impairment.

The clinician needs "contraindicated," "renal impairment," "dyspepsia"—these are precise, shared terms that let them act fast and correctly. Plain language here would slow them down and lose nuance. This is the case from Chapter 3: jargon is precision when the audience shares it.

For the patient (understand and act; the curse of knowledge is the enemy):

This medicine lowers your risk of [outcome]. Some people get an upset stomach when they start it—this is usually mild and gets better. Take it with food if it bothers your stomach. Tell us if you have kidney problems, because this medicine isn't safe for everyone with kidney disease. Call us if the stomach upset is severe or doesn't go away.

Same fact. "Contraindicated in severe renal impairment" became "isn't safe for everyone with kidney disease—tell us if you have kidney problems," because the patient's action is to disclose, not to self-diagnose their renal function. "GI adverse effects" became "upset stomach," with a concrete coping action ("take it with food") and a clear escalation trigger.

For the regulator (traceability and proof; every claim must be evidenced and exact):

In the pivotal trial (n = [N]), the treatment arm showed a reduction in [marker] relative to placebo (effect size and confidence interval as reported). GI adverse events were reported at a higher rate in the treatment arm than placebo, predominantly mild, with [X]% leading to discontinuation. Use is contraindicated in patients with severe renal impairment (defined as [criteria]); dosing guidance for moderate impairment per the approved label.

The regulator's version trades warmth and brevity for defensibility. Every claim is tied to its evidence, numbers are exact, "severe renal impairment" carries its precise definition, and there is no rounding, no reassurance, no editorial gloss. This is closest to the engineering and scientific writing of Chapters 33 and 35: language that exists to be audited.

The lesson is the same one this book has made all along, now with the volume turned up: the fact is constant; the document is a function of the reader. Get the reader wrong here and you don't merely communicate poorly—you write something dangerous. Give the patient the regulator's version and they drown; give the regulator the patient's version and the claim is unverifiable; give the clinician the patient's version and you've slowed down care with vagueness. (All figures bracketed above are placeholders—real labeling carries the actual trial data and approved-label language.)

🔄 Check Your Understanding You have one fact: "The device must be calibrated before each use or readings may be inaccurate." Sketch (one sentence each) how this differs for a biomedical engineer writing the device manual's technical section, a nurse using a quick-reference card at the bedside, and a regulatory submission.

Answer

Engineer (technical manual): the procedure and the tolerance—"Calibrate per §4.2 against the reference standard before each measurement session; calibration drift beyond ±[X] invalidates results." Nurse (quick-reference card): the action and the consequence, scannable—"Calibrate before every patient. Skipping this can give a wrong reading." Regulator (submission): the requirement, the evidence, and the risk control—"Pre-use calibration is a required step (IFU §[X]); failure mode 'inaccurate reading' is mitigated by mandatory calibration and an on-screen prompt; verified in validation report [ref]." Same fact, three jobs: do it, don't skip it, prove it's controlled.

36.5 Evidence Grading: Saying How Sure You Are

Clinical and health writing constantly makes recommendations—"do this," "prefer that," "avoid the other." A recommendation is only as trustworthy as the evidence under it, and a core convention of medical writing is to make the strength of that evidence explicit rather than letting "recommended" do double duty for both "strongly proven" and "expert hunch." The most widely used framework for this is GRADE (Grading of Recommendations Assessment, Development, and Evaluation), developed by an international working group whose method is now used by guideline-makers worldwide.

You don't need the machinery to benefit from the idea. GRADE's central insight, which every technical writer should steal, is that it cleanly separates two things people constantly confuse:

1. The certainty of the evidence — how confident we are in the underlying estimate (graded, roughly, from high down through moderate and low to very low). Many randomized trials pointing the same way is high certainty; a single small observational study is low.
2. The strength of the recommendation — how forcefully we advise the action (often "strong" vs. "conditional/weak"), which depends on the evidence and on the balance of benefits, harms, costs, and patient values.

The reason to keep these separate is that they come apart in important ways. You can have a strong recommendation on low-certainty evidence (the situation is dire, the intervention is cheap and safe, waiting for perfect data would cost lives) and a conditional recommendation on high-certainty evidence (we're very sure of the small effect, but whether it's worth it depends on what the patient values). Collapsing the two—writing "strongly recommended" when you mean "we're certain of a tiny effect," or "weakly recommended" when you mean "the evidence is thin"—misleads the reader about a distinction that should drive their decision.

For the writer, the practical takeaway is about honest hedging, which is exactly the territory Chapter 7 mapped for scientific writing. The language you choose should match the certainty you actually have:

❌ Before: "Studies show that X improves outcomes, so X is recommended for all patients."

✅ After: "Moderate-certainty evidence from several trials suggests X improves [outcome]. We make a conditional recommendation for X, because the benefit is modest and depends on patient priorities; shared decision-making is appropriate."

Why it's better: "Studies show" and "recommended for all" are the overclaiming Chapter 14 warned against—they flatten a nuanced evidence base into a slogan. The revision names the certainty ("moderate"), names the recommendation strength ("conditional"), and—crucially—tells the reader why it's conditional and what to do about it (involve the patient). It is more honest and, paradoxically, more useful: a reader can act appropriately on a calibrated claim and is misled by an inflated one. (The certainty and strength labels here are illustrative of the form; a real guideline assigns them through GRADE's structured process.)

🔍 Why Does This Work? Why is it sometimes correct to give a strong recommendation when the evidence is only low-certainty? Doesn't strength require strong proof?

Answer

No—and seeing why is the whole point of separating the two. Recommendation strength is a decision under uncertainty, and the decision depends on the consequences, not just the evidence. If an action is cheap, safe, and addresses a severe or rapidly worsening problem, then even modest evidence justifies acting forcefully, because the cost of not acting (while waiting for better data) is catastrophic and the cost of acting is trivial. Conversely, high-certainty evidence of a tiny benefit may warrant only a conditional recommendation if the intervention is burdensome or values-dependent. Strength weighs evidence plus the balance of benefits, harms, costs, and values. Writing that conflates them ("strong = well-proven") hides exactly the reasoning the reader needs to make a good decision.

36.6 Medication and Device Instructions: Where a Word Can Hurt Someone

This section is the chapter's sharpest edge, and it is a direct line to Chapter 38 (Ethics), where medical-device and medication documentation appears among history's most consequential communication failures. Here the threshold concept stops being abstract. An ambiguous medication or device instruction is not a style problem awaiting a copy-edit. It is a mechanism by which a real person can be harmed, and the documentation is frequently the last line of defense before that harm.

Consider how an instruction can be technically accurate and still dangerous through ambiguity.

❌ Before: "Take 2 tablets daily."

This looks fine. It is a trap. Does "daily" mean two tablets once a day (both at once) or two tablets spread across the day (one morning, one night)? For some medicines the difference is harmless; for others it changes the peak concentration in the blood and therefore the safety. The instruction is ambiguous, and ambiguity, under stress, resolves toward whichever reading the reader happens to land on.

✅ After: "Take 1 tablet in the morning and 1 tablet in the evening (2 tablets total each day). Take them about 12 hours apart, with food."

Why it's better: It removes the ambiguity by stating the action concretely—when each dose is taken, the total, the spacing, and the condition (with food). There is now exactly one way to read it. Notice the move: we didn't add precision by adding jargon; we added precision by adding concreteness. That is the whole discipline.

Now consider the categories of information a complete medication instruction has to make unambiguous, each a place where vagueness wounds:

• Dose and frequency, stated as concrete actions (above). Spell out numbers; "two" and "2" are both fine, but be consistent and unmistakable.
• Route and method. "By mouth," "under the tongue," "on the skin"—a medicine taken the wrong way can be useless or harmful. State it.
• Timing relative to food, sleep, or other drugs. "On an empty stomach" → "1 hour before eating or 2 hours after."
• Maximum dose and the hard ceiling. "Do not take more than X in 24 hours" is one of the most safety-critical sentences in all of patient writing, because overdose risk lives here. It must be unmissable and unambiguous—and it must define the window ("in 24 hours"), since "do not exceed recommended dose" is circular and means nothing to a patient who doesn't know the recommended dose.
• What to do about a missed dose, because every patient eventually misses one and will otherwise guess—and a common dangerous guess is "double up."
• Indications and contraindications. What it's for (so the patient takes it for the right reason) and who must not take it.
• Adverse effects and the escalation trigger. Common, expected effects (so the patient isn't alarmed) versus the serious signs that mean call now.

Device instructions and the IFU

Medical devices—from a home blood-pressure cuff to an implantable pump—ship with Instructions for Use (IFU), a regulated document. Everything in Chapter 22 (numbered steps, warnings before the step, the "never-done-this" test) applies, intensified by the safety stakes and the regulatory scrutiny. A few healthcare-specific intensifications:

• Warnings precede the action and name the hazard, the cause, and the response (Chapter 22's rule), but in an IFU the placement and wording are often regulated, and a misplaced warning is both a safety failure and a compliance failure.
• The user is frequently not a professional. A home-use device IFU is patient-facing writing (36.2 applies) and a regulated document (36.4's regulator audience applies) at the same time. You are writing for two audiences in one document, and the lower-literacy reader sets the comprehension bar.
• Foreseeable misuse must be addressed, not just correct use. Chapter 22's troubleshooting becomes, in a medical device, a safety obligation: if a user could plausibly do the wrong thing, the instruction has to head it off.

⚠️ Warning The most dangerous phrase in patient instructions is the one that feels clear to the writer and resolves two ways for the reader. "Take as needed." "Apply liberally." "Use as directed." "Do not exceed the recommended dose." Each is a latent failure. Before you ship any instruction that a patient will act on without you present, hunt these down and make each one concrete: as needed for what, how often, up to what maximum?

🧩 Productive Struggle Here is a real-style instruction fragment. Find every ambiguity before reading the analysis, and note what unsafe reading each one permits:

"For pain, take as needed. May repeat. Do not exceed maximum daily dose."

The ambiguities

Four traps, minimum. (1) "Take as needed" — how much? No dose is stated. (2) "As needed" — how often? No minimum interval between doses, so a patient in pain may re-dose every 20 minutes. (3) "May repeat" — repeat when, and how many times? Unbounded. (4) "Do not exceed maximum daily dose" — what is the maximum? It's referenced but never stated, so the ceiling that prevents overdose is invisible. A frightened patient in pain, reading this at 2 a.m., can construct a genuinely dangerous dosing schedule and never violate a single word of the instruction. A safe version states the dose, the minimum interval, the maximum number of doses, and the numeric 24-hour ceiling. (Specific drug names and milligram doses are deliberately omitted here—this is about the form of a safe instruction, not clinical advice.)

[📍 Good stopping point]

📐 Project Checkpoint

Your Communication Portfolio's user-documentation piece—built in Chapter 22 as a set of numbered procedural instructions—gets a healthcare-grade stress test in this chapter. You will not start a new piece; you will subject your existing one to the standard of a field where ambiguity can hurt someone.

This chapter's addition: Take the instructions you wrote for the Chapter 22 increment (or, if your portfolio's instructions aren't patient-relevant, write a short patient-facing instruction now: discharge wound care, a medication schedule, or home use of a simple device). Then do three things, in order:

1. Run the ambiguity hunt. Go sentence by sentence and mark every phrase that could resolve two ways—every "as needed," "as tolerated," "if necessary," "regularly," "a small amount." For each, write the unsafe reading it permits. This is the threshold concept turned into a checklist.
2. Rewrite to a 6th–8th-grade level. Translate every term a non-expert wouldn't know, in place. Convert every abstraction ("maintain hydration") to a concrete action ("drink 8 glasses of water a day"). Add the three answers a patient needs: what do I do, how do I know it's working, when do I call for help?
3. Teach-back test it. Hand it to one person who isn't in your field. Have them read it once, set it down, and tell you back what they would do—and, critically, when they'd call for help. Log every divergence. Each one is a defect in your writing, not in your reader (Chapter 22's lesson, Chapter 26's discipline). Revise until the reproduction matches your intent.

Keep both versions—the before, the after, and your annotated ambiguity hunt—in the portfolio. The contrast is itself evidence of skill: it shows you can take expert content and make it safe for a non-expert, which is the exact competency Chapter 40 asks your portfolio to demonstrate.

Looking ahead: Chapter 37 turns to business and policy writing, where the executive summary must stand alone—the same "lead with what the reader must act on" discipline, in a boardroom instead of a clinic.

36.7 Clinical Considerations: Mistakes, Edge Cases, and "It Depends"

The principles are clean; the practice is messy. Here are the failures that recur and the judgment calls that don't have a single answer.

Mistake 1 — Copy-forward and the note that lies. Modern clinical records let you copy yesterday's note into today's. It is a time-saver and a notorious source of error: a stale finding ("no chest pain") gets carried forward into a note for a day when the patient did have chest pain, and the record now contradicts reality. Every note must describe today's patient. The convenience of copy-forward is real; the discipline of editing what you copied is non-negotiable.

Mistake 2 — Abbreviations that kill. Clinical shorthand is fast, but some abbreviations are dangerous because they're misread, and several have been formally discouraged for that reason (the canonical example is trailing zeros and certain unit abbreviations that have caused dosing errors). The principle, which generalizes: an abbreviation that can be misread as a different, plausible value is a safety hazard, not a convenience. When in doubt in a dosing context, write it out.

Mistake 3 — Defensive documentation that buries the signal. Out of fear of liability, clinical notes sometimes swell with boilerplate that protects the writer but hides the one fact the next clinician needs. This is the medical form of the Chapter 4 failure: burying the conclusion. The reader is scanning, often urgently. Say what matters where it can be found.

Mistake 4 — Writing the patient leaflet at the writer's level, then "checking the box." A leaflet stamped "patient education" that is written at a 12th-grade level with untranslated jargon is education in name only. The box is checked; the patient is not informed. Readability is not a formatting afterthought—it's the function.

Mistake 5 — Confusing "informed consent the document" with "informed consent the process." A signed informed-consent form is a record, not a substitute for the conversation. The document should support shared decision-making—laying out the procedure, its benefits, its risks, the alternatives (including doing nothing), in plain language—but a patient who signed a form they couldn't understand was not informed. The writing serves the conversation; it does not replace it. (Specific consent requirements vary by jurisdiction and institution; this is about the writing, not the legal standard.)

The "it depends" calls:

• How much detail in the note? Enough that the next clinician can reconstruct your reasoning and act safely; not so much that the signal drowns. The audience is a busy peer who may inherit the patient cold.
• How simple is too simple for a patient? There is a floor—you must not lose accuracy or omit a safety-critical caveat in the name of plainness. Plain language simplifies expression, never truth. A 6th-grade reading level does not license a 6th-grade omission.
• When does plain language fight precision? Rarely, but it happens: a legal or dosing term may have no exact lay synonym. The move is to use the precise term and gloss it ("hypertension (high blood pressure)"), not to drop the precision. This is Chapter 7's lesson—define on first use—operating where the stakes are highest.

Frequently Asked Questions

What is a SOAP note and why is it structured that way?

A SOAP note is the standard format for clinical documentation, with four sections in fixed order: Subjective (what the patient reports), Objective (what you measured or observed), Assessment (your clinical interpretation), and Plan (what you'll do). The structure exists to force a separation between observation and interpretation, so the next clinician to read the chart can tell which facts are measured and which are concluded—a distinction that protects patient safety, because inheriting an opinion as if it were a measurement is one of the most dangerous documentation errors.

What reading level should patient materials be written at?

The widely repeated guidance from major health-communication and public-health bodies is roughly a 6th-to-8th-grade reading level, because a large share of adults struggle with materials written above that, and the materials handed to patients are typically pitched far higher than the patients can comfortably read. Writing simply is not condescension—it's the only version of the instruction many patients can actually follow. Use common words, short sentences, the active voice, and concrete actions, and define any clinical term you must keep.

What is teach-back and how is it different from asking "Do you understand?"

Teach-back asks the patient to explain the instructions back to you in their own words—"Tell me how you'll take this when you get home"—rather than answering "yes" to "Do you understand?" The yes/no question invites a polite, face-saving "yes" whether or not the patient understood; reconstruction can't be faked. It also frames any gap as the explainer's problem to fix ("let me make sure I was clear"), so gaps surface before the patient leaves instead of after they're harmed at home.

How do you write for clinicians, patients, and regulators at the same time?

You don't write one document that serves all three—you keep the fact constant and write the document as a function of the reader. The clinician version uses full medical vocabulary for precision and speed; the patient version translates everything into plain language and concrete actions and answers "when do I call for help?"; the regulator version ties every claim to its evidence with exact numbers and precise definitions. Getting the audience wrong here isn't just ineffective—giving a patient the regulator's version, or a regulator the patient's, can be unsafe or unverifiable.

What's the difference between certainty of evidence and strength of a recommendation?

They're two different things that "recommended" often blurs. Certainty of evidence is how confident we are in the underlying estimate (high to very low). Strength of recommendation is how forcefully we advise the action, weighing the evidence plus the balance of benefits, harms, costs, and patient values. They come apart: you can strongly recommend an action on low-certainty evidence (cheap, safe, urgent problem) or only conditionally recommend one on high-certainty evidence (sure of a small effect that depends on what the patient values). Frameworks like GRADE keep them explicitly separate so readers can act appropriately.

Chapter Summary

Key Takeaways

• Medicine raises the stakes on every principle in this book: audience, clarity, and structure stop being matters of quality and become matters of safety. The threshold concept governs the field—in medicine, ambiguity is a safety hazard.
• The SOAP note (Subjective, Objective, Assessment, Plan) structures clinical documentation so that observation stays separate from interpretation, protecting the next reader from inheriting an opinion as a fact.
• Health literacy means writing patient material at roughly a 6th–8th-grade level—not dumbing down, but the only version many patients can act on. Common words, short sentences, active voice, concrete actions, and a clear "when to call for help."
• Teach-back validates patient communication by asking the patient to explain it back, defeating the curse of knowledge with evidence instead of the unreliable "Do you understand?"
• One clinical fact becomes three documents for clinicians, patients, and regulators—the fact constant, the document a function of the reader—and getting the audience wrong here can be dangerous.
• Evidence grading (GRADE) separates the certainty of the evidence from the strength of the recommendation; honest writing matches its hedging to the certainty it actually has.
• Medication and device instructions must be unambiguous because the documentation is often the last defense before harm: state dose, frequency, route, timing, the maximum daily ceiling, and the escalation trigger as concrete, single-reading actions.

Action Items

• For every patient-facing instruction you write, answer the three questions explicitly: what do I do, how do I know it's working, when do I call for help?
• Hunt and destroy two-way phrases ("as needed," "as tolerated," "do not exceed recommended dose") before any instruction reaches a reader who will act on it without you present.
• Teach-back test anything important: have a non-expert read it once and tell you back what they'd do.
• In clinical notes, ask of every sentence: is this something I observed, or something I concluded?—and put it in the right SOAP section.

Common Mistakes

• Smuggling a conclusion into the Objective section, so opinion masquerades as measurement.
• Writing patient material at the writer's reading level and checking the "education" box.
• Leaving the maximum-dose ceiling unstated or circular ("do not exceed recommended dose").
• Conflating "recommended" with "well-proven"—collapsing recommendation strength into evidence certainty.
• Treating a signed consent form as a substitute for the consent conversation.

Decision Framework — Which audience am I writing for, and what does it demand?

🪞 Learning Check-In You're deep in Part VII now, writing for specific fields. Pause and notice something: almost nothing in this chapter was a new writing principle. SOAP is structure (Chapter 4). Health literacy is audience and clarity (Chapters 2–3). Teach-back is the "never-done-this" test (Chapter 22). Evidence grading is honest hedging (Chapter 7). What changed wasn't the principles—it was the cost of getting them wrong. Ask yourself: which principle from earlier in the book do you most often let slide when you're busy? In medicine, that's the one that would hurt someone. That realization is transferable to every field you'll write in.

Spaced Review

A few questions reaching back, to strengthen retention.

1. (From Chapter 22) What is the "someone who has never done this" test, and how is teach-back the clinical version of it?
2. (From Chapter 4) What does it mean to write a document "top-down," and how does the SOAP note's Plan-and-Assessment-up-front logic relate to BLUF?
3. (From Chapter 35, bridging) Chapter 35 separated the certainty of evidence from a paper's claims; this chapter separated certainty from recommendation strength. Why does conflating "well-proven" with "strongly recommended" mislead a reader?

What's Next

Chapter 37 moves from the clinic to the boardroom: writing for business and policy. You'll meet the executive summary that must stand alone, the white paper that builds authority, the policy brief that turns research into a recommendation, and the one-pager that compresses complexity to a single page. The through-line from this chapter is direct—lead with what the reader must act on, and write for the reader you actually have, not the one who'd enjoy your methodology. Dana Whitfield's churn findings, last seen as a data memo in Chapter 27, return there as a one-page brief for leadership.

Practice: Exercises · Quiz Go deeper: Case Study · Case Study 2 Review: Key Takeaways · Further Reading