Your knee has been grinding for two years. The surgeon pulls up the X-ray, points at a joint space that isn’t there anymore, and says you’re a candidate for a replacement.
Then she asks what you want to do.
Most people say some version of “whatever you think, doctor.” It feels polite. It feels like trust. And it ends the conversation that was supposed to be the most important part of the visit.
That conversation has a name, a federal training program behind it, and 209 randomized trials testing the tools it uses. Almost no patient has ever been told any of that.
What the model is
MedlinePlus, the consumer health service run by the National Library of Medicine, describes shared decision making as a process in which providers and patients “work together to decide the best way to test for and treat health problems.”
MedlinePlus lays out four movements, and they’re worth reading as a checklist of what should happen in the room.
Every option gets named. Not just the recommended one. That includes the options your clinician isn’t suggesting, along with the reason they’re off the table for you specifically.
The benefits and harms of each option get laid out. Side effects. What a test leads to next, since a positive result usually triggers something. What outcomes are realistic, not just possible.
A decision aid may get used. More on those shortly, because they’re the part with the real evidence behind them.
Then a decision gets made jointly, and waiting is on the menu as a legitimate choice rather than a failure to decide.
MedlinePlus notes this model is meant for consequential decisions: starting a long-term medication, agreeing to major surgery, genetic testing, cancer screening. The page was last reviewed in July 2024 by Linda J. Vorvick, MD, a clinical professor in family medicine at the University of Washington School of Medicine.
One distinction to clear up immediately. This is not informed consent. Informed consent is a legal requirement that you be told a procedure’s risks before you sign, and it usually happens after the decision is already made. Shared decision making is the conversation that produces the decision. Different thing, different purpose.
The Agency for Healthcare Research and Quality, the federal agency that studies how care is delivered, frames it as a collaborative process where patients and clinicians decide together using the evidence plus the patient’s own knowledge, values, goals, preferences, and circumstances. That last clause is the one that gets dropped in practice.
Preference-sensitive decisions are the entire point
A preference-sensitive decision is one where two or more options carry genuinely different trade-offs and the evidence doesn’t crown a winner. The right answer depends on which trade-off you’d rather live with. Nobody can calculate that for you.
The clearest proof that this category officially exists sits in the letter grades of the U.S. Preventive Services Task Force, the independent federal panel that evaluates preventive services.
Take PSA-based prostate cancer screening. In its May 2018 recommendation, the Task Force assigned a C grade for men aged 55 to 69. A C is not a thumbs down. It means the service should be offered selectively, based on individual circumstances and what the patient wants. The Task Force writes that “the decision to be screened for prostate cancer should be an individual one,” and adds that clinicians should not screen men who don’t express a preference for it. For men 70 and older the grade is D, and the panel recommends against screening.
Statins work the same way. In its August 2022 recommendation, the Task Force gave a B grade, meaning prescribe, for adults 40 to 75 who have at least one risk factor such as abnormal cholesterol, diabetes, high blood pressure, or smoking, and a 10-year cardiovascular risk of 10% or higher. For the same adults whose 10-year risk falls between 7.5% and just under 10%, the grade drops to C: offer selectively, taking the patient’s values and preferences into account.
Sit with that. A federal panel wrote down, in advance, that a specific slice of patients should get a decision rather than an instruction. The Task Force also notes that it does not factor cost into its grades at all, so the letter reflects benefits and harms only.
Most patients have never been told that letter grades exist, let alone which one applies to them. You can ask.
Other decisions in this category are easy to recognize once you know the shape: when to replace a worn joint rather than giving physical therapy another year, whether to take a daily pill for a risk you can’t feel, how to deliver a baby, whether to treat or watch a small, slow-growing cancer.
This is the detector question. “Is this a decision where reasonable people with the same information choose differently?” If the honest answer is yes, your values belong in the room.
AHRQ’s SHARE Approach and its five steps
The Agency for Healthcare Research and Quality built a training program for clinicians called the SHARE Approach, organized around five essential elements. The agency’s facilitator guide for the training, published in October 2024 as AHRQ Publication No. 25-0005-1-EF, spells them out: seek your patient’s participation, help your patient explore and compare treatment options, assess your patient’s values and preferences, reach a decision with your patient, and evaluate your patient’s decision.
Those are written for clinicians. Translated into what they look like from your chair:
Seek. Somebody invites you into the decision instead of handing you a plan. If nobody does, you can start it yourself with one sentence, and the last section of this article gives you three.
Help. More than one option gets described, in terms you can compare. Two options described in different units isn’t a comparison.
Assess. Somebody asks what you’re trying to protect. Your job through a long recovery. Lifting a grandchild. Staying off a daily pill. Avoiding a hospital stay. Fertility. If nobody asks, say it anyway.
Reach. The choice gets made out loud, with the clinician’s own recommendation included. Shared does not mean your doctor refuses to tell you what they’d do.
Evaluate. The decision gets revisited later. Many medical decisions are reversible, and this check-in is the step most often skipped entirely.
The training itself is modest in scale: three videos running about an hour total, plus six group activity sessions of roughly three hours, according to AHRQ’s October 2024 fact sheet. It’s aimed at practice staff who see patients. Which tells you something uncomfortable. The federal government treats this as a skill clinicians have to be taught, not something that happens naturally in a conversation.
Decision aids and what 209 trials found
A patient decision aid is a tool, usually a booklet, video, or web module, that lists the options, gives evidence-based numbers on the benefits and harms of each, and helps you work out which features matter most to you. It is not a brochure and not an advertisement.
The evidence base here is one of the largest in all of patient communication research. The 2024 Cochrane review by Dawn Stacey and colleagues, which searched the literature through March 2022, pooled 209 randomized trials covering 107,698 participants and 71 different decisions. The most studied topics were cardiovascular treatments, colorectal and prostate and breast cancer screening, breast and prostate cancer treatment, mental health treatment, and joint replacement.
Compared with usual care, the review found the following.
Knowledge went up, with high certainty: a mean difference of about 12 points on a 100-point scale, 11.90 with a 95% confidence interval of 10.60 to 13.19, pooled across 107 trials and 25,492 people.
Accurate understanding of personal risk nearly doubled, also high certainty: a risk ratio of 1.94, confidence interval 1.61 to 2.34, across 25 trials and 7,796 people.
Fewer people left the decision entirely to the clinician: a risk ratio of 0.72, confidence interval 0.59 to 0.88, across 21 trials and 4,348 people.
Decisional conflict fell on two measures, feeling uninformed and being unclear about one’s own values.
Choices lined up with informed values more often, though the reviewers rated this one moderate rather than high certainty: a risk ratio of 1.75, confidence interval 1.44 to 2.13, across 21 trials and 9,377 people. The careful word is “probably.”
And here’s the finding most people get wrong: decision regret did not drop. With high certainty, the review found no difference in regret between people who used a decision aid and people who didn’t, a mean difference of -1.23 with a confidence interval of -3.05 to 0.59 across 22 trials and 3,707 people.
That’s worth stating plainly. Decision aids make people better informed, more accurate about their own risk, and more involved. They do not make hard choices stop hurting afterward. Anyone selling them as a cure for regret is overselling.
The time objection also got measured, which almost never happens. When a decision aid was used before the visit, consultations were no longer on average across five trials. When it was used during the visit, consultations ran 1.5 minutes longer, a mean difference of 1.50 minutes with a confidence interval of 0.79 to 2.20 across eight trials and 2,702 people.
One and a half minutes. That’s the price of the thing clinicians most often say they don’t have time for. The reviewers do note open questions, including whether decision aids change whether people stick with the treatment they chose, and what the downstream cost effects are.
Why it still doesn’t happen in most visits
If the tools work this well, why has almost nobody reading this been handed one?
Because getting patients a decision aid and getting clinicians to change how they run a visit are two different problems, and only one of them is solved.
A 2018 Cochrane review led by France Légaré pooled 87 studies covering 45,641 patients and 3,113 health professionals, mostly in the United States, Germany, Canada, and the Netherlands. The studies tested interventions aimed at patients, at clinicians, and at both together. The conclusion was blunt: it is uncertain whether any of these approaches reliably increase clinicians’ use of shared decision making, because the certainty of the evidence is low or very low. The review opens by noting the model has not been widely adopted in practice.
Underneath that statistical verdict are some ordinary human reasons.
Visit length. The decision conversation competes with refills, forms, and the three other things you came in about.
Training and habit. Clinicians are taught to recommend. Recommending is faster, and it reads as competence.
Patients who decline the invitation. “Whatever you think, doctor” closes the door. Some people genuinely want that and should get it. Others say it because nobody told them they were allowed to say anything else.
No tool at hand. Many practices simply don’t stock decision aids for the conditions they treat most.
Payment. Nobody bills for the twelve minutes spent explaining active surveillance versus surgery.
How to start it yourself
You can open this conversation. It takes about fifteen seconds, and it works better early in the visit than on the way out the door.
“Is this a close call, or is one option clearly better?” This single question sorts the whole visit. If one option is clearly better, take it and go home. If it’s a close call, you’ve just learned that your values are supposed to be part of the decision.
“Here’s what matters most to me.” Then name something concrete. Not “I want to feel better.” Try: I need to be back at work in three weeks. I don’t want a daily medication. I want to keep running. I can’t be away from my kids for a hospital stay. Clinicians cannot weigh a preference they’ve never heard.
“Is there a decision aid for this?” Ask for it by name. Many common decisions have one, and your clinician may need to go look.
Two follow-ups that earn their keep: “What would you do, and why?” The second half is where the useful part lives. And “How long can I take to decide?” Most decisions carry far more slack than the pace of an appointment implies. Ask for an actual number of days or weeks.
Bring somebody with you. MedlinePlus recommends taking a trusted friend or relative along and having them write things down, and the reason is practical rather than sentimental. A frightened person retains less, and a second set of ears catches the question you forgot to ask. Preparing for an appointment generally is its own skill, and the question technique for drawing out alternatives is a separate subject with its own phrasing.
Where shared decision making is the wrong tool
Two situations, and being honest about both makes the model more credible, not less.
Emergencies. When minutes count, the model steps aside, and it should. Sudden facial drooping, one-sided weakness, or slurred speech; chest pain or pressure with sweating or breathlessness; severe trouble breathing; bleeding that won’t stop; a sudden headache unlike any you’ve had; confusion with a high fever. Those are 911 calls. Nobody should be clarifying values in that window.
Clear winners. When one option is plainly better supported, presenting a menu isn’t respect. It’s abdication. A person with a serious bacterial infection does not need a values exercise about antibiotics. Turning every decision into a buffet is its own failure mode, and it can hand the hardest part of medicine to the most frightened person in the room.
The honest middle ground: a good clinician tells you which decisions are genuinely yours and which aren’t, and gives you a recommendation either way. “I don’t want to influence you” is not shared decision making. It’s an absence.
One more thing deserves saying. Wanting your doctor to decide is a legitimate preference, and the model is supposed to ask how involved you want to be rather than enforce a quota of involvement. The difference between choosing to defer and never being asked is the entire point.
What to take away
One question does most of the work. Ask whether this is a close call.
If the answer is no, you’ve saved everyone time and you can act on the recommendation with more confidence than you had a minute earlier. If the answer is yes, you’ve converted a recommendation back into a decision, and you’ve earned the right to ask for the numbers, for a decision aid, and for a few days to think.
The pooled evidence says that combination will leave you better informed and more accurate about your own risk. It will not make the choice feel easy. The research is clear that nothing does.
Sources
- MedlinePlus Medical Encyclopedia, National Library of Medicine. “Shared decision making.” medlineplus.gov
- MedlinePlus, National Library of Medicine. “Talking With Your Doctor.” medlineplus.gov
- Agency for Healthcare Research and Quality. “The SHARE Approach.” ahrq.gov
- Agency for Healthcare Research and Quality. “The SHARE Approach: A Facilitator's Guide (AHRQ Pub. No. 25-0005-1-EF, October 2024).” ahrq.gov
- Agency for Healthcare Research and Quality. “The SHARE Approach Fact Sheet (AHRQ Pub. No. 25-0005-6-EF, October 2024).” ahrq.gov
- Agency for Healthcare Research and Quality. “Shared Decision Making.” ahrq.gov
- U.S. Preventive Services Task Force. “Prostate Cancer: Screening.” uspreventiveservicestaskforce.org
- U.S. Preventive Services Task Force. “Statin Use for the Primary Prevention of Cardiovascular Disease in Adults.” uspreventiveservicestaskforce.org
- U.S. Preventive Services Task Force. “About the USPSTF.” uspreventiveservicestaskforce.org
- Stacey D, et al. Cochrane Database of Systematic Reviews, 2024. PubMed. DOI: 10.1002/14651858.CD001431.pub6. “Decision aids for people facing health treatment or screening decisions.” doi.org
- Legare F, et al. Cochrane Database of Systematic Reviews, 2018. PubMed. DOI: 10.1002/14651858.CD006732.pub4. “Interventions for increasing the use of shared decision making by healthcare professionals.” doi.org