Urgent care is the specialty where MDM complexity is hardest to capture quickly. You see a sprained ankle and a chest pain work-up in the same hour, at 4-5 patients per hour, with a disposition clock running the whole time. The 2023 CMS E&M guidelines made medical-decision-making the primary code selector. Good news for clinicians who think carefully. The catch: the quality of your MDM documentation now carries all the weight that history and physical exam used to.
AI Medical Scribe by Patient Square is an ambient AI medical scribe that listens during the visit and hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft, ready to review and sign about two minutes after the visit.
Key takeaways
- Urgent care uses office/outpatient E&M codes 99202-99215, determined since 2023 by medical decision making or total time, not by history and physical exam documentation.
- MDM has three elements (problem complexity, data reviewed, risk); you need two of three to meet a level. Capturing that reasoning is where documentation gaps happen.
- A 2025 Annals of Emergency Medicine study found ambient AI scribes cut on-shift documentation time by about 28% in an ED setting (2:45 vs 3:50 per encounter), the closest published analog to urgent care throughput.
- At a 4-5 patient/hour UC pace, even 1 minute saved per encounter adds up to 4-5 minutes per hour back in your shift.
reduction in on-shift documentation time with ambient AI scribe in a 2025 Annals EM study (2:45 vs 3:50 per encounter)
urgent care physician throughput benchmark (AAUCM); documentation savings compound at this pace
MDM elements required under 2023 CMS E&M guidelines to establish coding level — problem complexity, data reviewed, risk
What actually changed about urgent care E&M in 2023
The shift started in 2021 and was clarified for 2023: urgent care centers now use office and other outpatient E&M codes 99202 through 99215, with code selection driven by MDM or total time, not by whether you documented a 10-point review of systems or a full organ-system physical exam. History and exam still happen. You still document them. But they no longer determine the code.
That sounds like a documentation simplification. It is, partially. But it moved the load: now the three MDM elements carry all the weight. If your note reflects sound clinical reasoning, you code correctly. If the note is thin on reasoning, even if your actual decision-making was thorough, you are at risk for downcoding on audit.
The three MDM elements, under the current framework:
- Number and complexity of problems addressed. “Acute” means recent or short-term, not necessarily new. A return visit for an ankle you splinted last week is still an acute problem. Self-limited problems (common cold, minor bruise) are the bottom tier.
- Amount and complexity of data reviewed. Ordering or reviewing each unique test counts. Discussing a case with an external provider (not your own group) counts. An independent historian counts. A translator does not.
- Risk of complications, morbidity, or mortality. This is about treatment risk, not inherent patient risk. Gastroenteritis treated with OTC recommendations is low risk. Social factors like homelessness or food insecurity now count toward moderate risk under the 2023 rules. The decision to forgo a test or treatment also earns MDM credit, if you document the reasoning.
To bill a given level, two of those three elements must meet that level’s threshold.
The acuity range problem in urgent care
Primary care deals with familiar patients and mostly chronic disease. The ED deals with higher average acuity and has more clinical support. Urgent care sits between them and sees the full spread: a 9-year-old with an ear infection at 9am, then a 58-year-old with hypertensive urgency at 10am, then a worker with a hand laceration at 10:15am.
That range is not a documentation problem when you have time. At 4-5 patients an hour, it is. The ear infection is a 10-minute encounter that codes moderate MDM if there is a prior antibiotic failure to work through, but you have to document the reasoning. The hypertensive urgency may code high MDM, but the note has to capture the risk discussion and the disposition logic. Type between patients or dictate in the hallway and those elements get thin.
A 2025 study in Annals of Emergency Medicine tracked 8,740 eligible encounters at an academic ED after ambient AI scribe deployment. On-shift documentation time dropped from 3:50 per encounter to 2:45, a 28% reduction. Total EHR time per encounter dropped 16% (10:21 to 8:39). This is the closest published evidence to a UC setting; no randomized trial has run exclusively in urgent care yet. The direction is consistent with other ambulatory studies, including a multi-site study of 1,800 clinicians that found scribe adopters saved 16 minutes of documentation time per 8-hour care block.
A 2026 STAT analysis comparing AI scribes to human scribes found AI saved about 1.6 minutes per patient in an ED context, versus 3.3 minutes for a human scribe. The AI tool costs a fraction of a human scribe’s cost.
What the MDM-to-documentation gap looks like in practice
Here is a typical urgent care scenario where the gap shows up. A 45-year-old comes in with right-sided chest pain, onset two hours ago. You take the history, do an ECG, review troponin, calculate a HEART score, and decide the probability is low enough to discharge with return precautions and next-day cardiology referral. That decision involves at least moderate MDM: acute problem with chronic illness interaction, data reviewed, moderate risk from the decision not to admit.
Without good contemporaneous documentation, the note might say: “Chest pain, 45yo, ECG unremarkable, troponin negative, discharged with follow-up.” That reads like a low-MDM encounter on audit. The reasoning that justified moderate coding, the HEART score rationale, the explicit decision to forgo admission, the social history that factored in, is not there.
An ambient scribe running during that encounter captures the conversation: the HEART score discussion, the troponin-negative interpretation, the disposition reasoning you said out loud while wrapping up. The draft note includes it. You review and sign. The MDM documentation reflects the actual complexity of the visit.
A 2025 study comparing ED billing code distributions before and after the 2023 E&M guideline changes found Level 4 visits increased by about 7 percentage points and Level 3 visits dropped by about 8 percentage points across five hospitals. Community hospitals showed the strongest shift. Part of this reflects the guideline change reducing documentation requirements for higher codes. Part of it reflects clinicians now documenting reasoning more clearly rather than defaulting to conservative coding.
The dispo clock and the note
Urgent care throughput runs on dispo speed. That means 12-15 minutes per encounter from door to dispo, which is not a lot of room to take a history, do an exam, order and review tests, and make a disposition call. The note has to happen somewhere.
Most urgent care physicians do one of three things. Type between patients (slows the floor). Hold notes until after the last patient (means reconstructing from memory). Or run some combination that produces thinner notes than the clinical encounters actually warrant.
An ambient scribe running during the visit removes that tradeoff. The note builds from the actual conversation: the patient’s responses to your history questions, the exam findings you narrate, the plan you explain before they leave. By the time you’re rooming the next patient, the draft is ready. You review it on the walk over.
For a 10-hour shift seeing 40-50 patients, those saved minutes add up. But the bigger return is note quality, not time. A note built from what was actually said in the room captures the MDM more completely than one reconstructed at the end of the day from a fading memory of 40 visits.
The scribe reflects your reasoning, it doesn’t generate it
The scribe captures what happens in the room. If you do not verbalize your differential, the risk stratification you ran through in your head, or why you chose observation over discharge, the note will not capture it. Your thinking has to be audible.
The ICD-10 suggestions that come back with the note are exactly that: suggestions. They are not a coding engine, and they do not auto-submit claims. The prescription draft is a draft you review before anything is signed. Nothing leaves your review queue automatically.
The scribe also does not handle documentation outside the encounter conversation, such as prior authorization letters, peer-to-peer calls, or forms that require separate completion. Those stay on your list.
What it does handle: the core visit note, structured as a SOAP note, including the elements that matter most for MDM, which are the problem complexity section, the data reviewed, and the risk reasoning as you stated it. In an urgent care setting where those elements are often the first thing thinned under time pressure, that is where the practical value sits.
Evaluating a scribe for urgent care before you commit
A few questions worth answering before you pick a tool:
Does it handle the acuity range? A tool optimized for 20-minute primary care visits may produce generic notes for a chest-pain work-up or a pediatric respiratory distress visit. Run it on encounters across your acuity range, not just your easiest patients.
How does it handle interpreter-assisted visits? The 2025 ED study found physicians selectively avoided using AI scribes for interpreted encounters. If a meaningful share of your UC population uses interpreter services, test that scenario explicitly.
What is the note turnaround? In primary care, a 2-minute turnaround is fine. In a high-throughput UC setting where you are rooming the next patient before the previous one reaches the checkout, you want the note ready fast and the review to take under 60 seconds for straightforward visits.
What are the BAA terms? Any tool handling PHI from UC patient encounters needs a signed Business Associate Agreement. Confirm it before you start using it for real patient visits.
Our 9-question evaluation scorecard walks through these questions and others in detail, and it is worth running through before you commit to any tool.
Starting a trial in urgent care
The fastest way to validate a scribe for your UC workflow is a real shift, not a demo. Run it on a shift that covers your normal patient mix: some lower-acuity volume and a few higher-complexity encounters. Compare the note quality and your end-of-shift remaining documentation load against your baseline.
The 7-day trial is the right unit. One shift tells you if it works for simple visits. A week tells you if it holds up across the acuity range, handles your patient population, and fits into your pace without adding friction.
Pricing, with no hidden costs: US Solo plan is $89/month. The full breakdown is on the pricing page. There is no feature difference between plans, just the same product at different volume tiers.
If you want to see a note built from a UC visit before running a trial, book a demo with a case from your actual workflow. Chest pain work-up, pediatric fever, laceration repair, bring the encounter type that puts your documentation under the most pressure, and see how the note comes back.
The MDM documentation gap in urgent care is not a knowledge problem. It is a time problem. The scribe addresses the time part.