The fourteen-second problem: how agenda-setting changes clinic visits

AA patient sits down. The clinician asks what brings her in. She begins to answer. Fourteen seconds later, the clinician interrupts with a focused question. The patient never finishes her opening statement, and the third concern she came in to discuss, the one she was working her way up to, never gets mentioned at all. Forty minutes later, with one foot in the doorway, she says, almost as an afterthought, by the way, my chest has been bothering me. That sequence is so familiar to anyone who has worked in primary care that it is easy to forget it was documented in a landmark study. The Marvel 1999 paper is the origin of the agenda-setting clinical interview literature, and it is the reason most modern communication training opens with a single question at the door.

Kim Marvel and colleagues audio-recorded 264 primary care visits, looking at how often physicians solicited the patient’s full list of concerns at the start. The headline finding most often quoted, that patients were interrupted within roughly eighteen to twenty-three seconds on average, was that physicians redirected patients before they had finished their opening statement in the large majority of visits. In only about a quarter of visits did the physician explicitly ask whether there was something else. When patients were not interrupted, their complete opening statement took, on average, only thirty-two seconds. The interruption was not buying time. It was costing it.

“The interruption was not buying time. It was costing it. Concerns surfaced at the doorway instead of at the door.”

When patients are interrupted at the start, the unaddressed concerns do not disappear. They reappear at the worst possible point in the visit, the close, when the clinician is reaching for the doorknob. This is the so-called doorknob phenomenon, and it is structurally created by the failure to set an agenda. A patient with three concerns, interrupted after the first, will hold the other two in reserve and try to introduce them when there is no time left to address them properly. The result is a visit that runs late, a clinician who feels ambushed, and a patient who feels rushed. Setting the agenda at the start solves all three problems at once.

An agenda-setting clinical interview opens with a specific sequence. The clinician asks what brings the patient in today. After the patient answers, the clinician asks, before going deeper, what else is on your mind today. They keep asking *what else* until the patient says nothing. Only then does the clinician negotiate priorities, which two or three of these can we tackle today, and which should we plan to address next visit. The whole sequence takes roughly forty-five to ninety seconds. Compared to the cost of patient interruption time at the doorway, it is one of the better trades in clinical medicine. Studies that followed Marvel 1999 found that visits with full agenda elicitation did not run longer overall; they ran the same length, but with the right concerns addressed.

Three habits sabotage agenda-setting. The first is the diagnostic instinct: clinicians latch onto the first concern because it sounds like a workable chief complaint and start asking focused questions before the patient has finished. The second is the time-anxiety instinct: clinicians believe that asking what else will surface ten new concerns and blow up the schedule. In practice, patients almost never list more than three. The third is the language of the prompt itself. Anything else is a closed question that subtly invites a no. What else is on your mind today is open and invites the actual list. The difference between these two prompts is small in print and large in practice.

The Marvel 1999 findings have been replicated in specialty clinics, in emergency departments, and in telehealth visits. The pattern is robust. Whenever clinicians are under time pressure, they interrupt earlier and miss more. The cost is borne by the patient in the form of unaddressed concerns and by the clinician in the form of return visits, portal messages, and the doorknob complaint that ends up being the real reason the patient came. Agenda-setting is one of the few communication habits that pays back its time cost within the same visit.

The Empathy in Medicine Initiative builds agenda-setting into the first hour of its ambassador curriculum because the evidence is unusually clean and the behavior change is unusually portable. Students practice the sequence in role-play, watch themselves on video, and learn to time their own openings until what else is on your mind today becomes automatic. EIMI ambassadors then teach the practice to their peers on the wards, where it is most often forgotten. Students who want the free EIMI training, including the agenda-setting module and the Marvel 1999 evidence brief, can find both at /Training and apply to become ambassadors at /Apply.

The fourteen-second problem is not a problem of rude clinicians. It is a problem of habit, and habits respond to structure. A single open question at the start of the visit reorganizes the entire encounter. The data that began with Marvel 1999 still hold: the time you spend eliciting the full agenda is bought back from the time you would otherwise spend chasing the concern that was raised at the door.

A decade after Marvel 1999, John Heritage and Jeffrey Robinson published a small but consequential trial showing that the specific word choice in the agenda-elicitation prompt changed how many concerns surfaced. Asking is there something else you want to address today produced more disclosed concerns than asking is there anything else. The difference between something and anything is grammatically trivial and linguistically substantial. Any leans negative; it primes a no. Some leans neutral; it leaves the door open. The Heritage and Robinson result is a useful reminder that agenda-setting is not just about whether the clinician asks, but how. Communication trials of this kind are why the field moved from general exhortations about listening toward specific, testable verbal moves.

A second refinement from the post-Marvel literature concerns the negotiation step that follows the elicitation. When patients raise three or four concerns, the clinician cannot address them all in a fifteen-minute slot, and pretending otherwise sets up the same time crunch the elicitation was meant to prevent. The behavioral move that closes the loop is explicit prioritization: of these three, which one feels most important to you today, and which can we plan to come back to. Patients almost always choose well when asked, and the clinician gains permission to defer the others without the relational cost of dismissing them. The doorknob concern becomes a planned follow-up rather than an ambush.

A useful exercise for trainees is to audio-record one of their own visits with consent and time, to the second, how long the patient was allowed to speak before the first redirection. The number is almost always lower than the trainee predicts. Most clinicians believe they are good listeners. The recordings tend to disagree, and the gap between self-perception and recorded behavior is where the training has the most room to work. Marvel 1999 is, in this sense, less a paper about other clinicians than a paper about us.