Clinical Documentation Guide: Faster, Better Notes

A clinical documentation guide is only worth reading if it changes what happens at 9:40 at night. The short version: most physicians can take real time out of their notes by fixing three things, what they document (much of what you were trained to record no longer affects your billing), when they document (inside the visit rather than after it), and who documents (your team and your tools, not you alone at the kitchen table). Everything below is the evidence and the mechanics behind those three levers.

I have watched a second year resident spend eleven minutes writing an HPI for a straightforward cellulitis admission. It was a lovely piece of writing. Chronological, complete, faintly literary. It also did nothing that the next physician needed, changed nothing about the billing, and cost her eleven minutes she did not have at 6 p.m. on a call day. That gap, between the note we were taught to write and the note the work actually requires, is where most documentation time disappears.

Why your notes take so long, and why it is not a personal failing

Start with the scale of the problem, because physicians routinely assume they are slow when they are simply doing more work than the day contains.

The AMA's Organizational Biopsy data, drawn from nearly 18,000 physician responses across 43 states and more than 100 health systems, found that physicians worked an average of 57.8 hours per week in 2024. Of those, 27.2 hours went to direct patient care. Another 13 hours went to indirect patient care and 7.3 hours to administrative tasks [1]. Roughly a third of the working week never touches a patient.

Off hours work has not improved along with the rest. In that same 2024 dataset, 22.5 percent of physicians reported spending eight or more hours per week in the EHR outside of 5:30 p.m. to 7 a.m., up from 20.9 percent the year before [1][2]. Burnout fell from 48.2 percent in 2023 to 43.2 percent in 2024 [1][2]. Pajama time did not follow it down. The wellness programs moved the survey score. They did not move the inbox.

When physicians are asked directly what is burning them out, paperwork wins. In the Medscape Physician Burnout and Depression Report released in January 2024, 62 percent of burned out physicians named administrative work, charting, and paperwork as a contributor, ahead of long hours at 41 percent and inadequate compensation at 38 percent [3].

None of this is a motivation problem. It is a volume problem with a design problem sitting on top of it.

What a clinical documentation guide should actually fix

Most documentation advice on the internet is written by companies selling documentation products. The advice is not always wrong, but it skips the unglamorous part: a large fraction of physician note time is spent producing content that no longer serves any reader, any payer, or any medicolegal purpose.

So this clinical documentation guide organizes around a single question you can ask of every line you type. Who reads this, and what do they do differently because of it? If the answer is nobody and nothing, the line is costing you money in the only currency you cannot earn back.

Three categories of waste account for most of it:

Ceremonial content. Ten point review of systems on a patient with an ankle sprain. A physical exam that documents organ systems you did not examine. Since January 2021, none of this determines your office visit code.

Duplicated content. The labs are already in the labs tab. The med list is already in the med list. Pulling them into the note creates a second, immediately stale copy that the next physician has to reconcile against the first.

Deferred content. The note you write six hours after the encounter takes longer than the note you write during it, because you are rebuilding the encounter from memory before you can write a word about it.

Cut those three and the note gets shorter, faster, and more useful at the same time. That is the whole argument, and the rest of this article is how to execute it.

The rule that changed in 2021, and the habit that did not

This is the single most expensive piece of outdated knowledge in American medicine.

Effective January 1, 2021, the AMA's revised guidelines for office and other outpatient evaluation and management services eliminated history and physical exam as components of code selection. Level of service is now chosen by medical decision making alone or by total time on the date of the encounter. Documentation must still be medically appropriate, but the bulleted history and exam elements no longer drive the billing level [4].

Read that again if you trained before 2021. The review of systems marathon, the exam checkboxes, the padded HPI, all of it was a response to a rule that no longer exists for these visit types.

The habit outlived the rule. Epic Research analyzed 1.7 billion outpatient clinical notes and found that average note length rose 8.1 percent between May 2020 and April 2023, from 4,628 characters to 5,002 characters, despite the coding overhaul that was supposed to shrink them. Increased use of copy and paste and of auto populating SmartTools tracked with the growth [5].

We were handed permission to write less and we wrote more, because the templates did not change and nobody told us we could stop.

Where the minutes actually go

You cannot fix a time problem you have not measured. The most cited direct measurement remains Arndt and colleagues in Annals of Family Medicine, who combined EHR event log data with time motion observation across 142 family physicians in a southern Wisconsin system [6]. The distribution is the useful part.

Source: Arndt et al., Annals of Family Medicine, September 2017. Figures describe a single system's family medicine physicians over an 11.4 hour average workday and will differ by specialty and EHR build [6].

Two things jump out. First, documentation is the largest clerical block but it is not the majority of it. If you only optimize note writing and ignore orders and inbox, you are attacking about a quarter of the problem. Second, 86 minutes a day happen after clinic ends. That is the number your family actually experiences, and it is the one worth targeting first. We break that down separately in our guide to reducing pajama time (/clinical/reduce-pajama-time).

Inbox deserves its own mention because it is quietly enormous and getting worse. Research using Epic Signal data across 371 health systems found that U.S. physicians receive roughly three times more inbox messages than physicians elsewhere in the world, and the AMA notes the commonly cited 13 minutes per day figure likely understates true inbox time by a factor of two to three, since the measurement stops counting after five seconds of inactivity [7]. The study also predates the pandemic surge in portal messaging.

Note bloat is a quality problem before it is a time problem

The strongest argument for shorter notes is not efficiency. It is that long notes fail at their job.

Downing, Bates, and Longhurst published a comparison in Annals of Internal Medicine in May 2018 showing that U.S. physicians' clinical notes ran roughly four times as long as those written by physicians in other countries using the same EHR platform. The authors attributed the gap to American compliance and reimbursement documentation rather than to clinical content [8].

Our notes are not four times more informative. They are four times longer, which is a different thing entirely, and the extra length is where errors hide. A copied assessment from three days ago, still stating that the patient is on levofloxacin after it was stopped, is not a harmless artifact. It is a wrong fact in the medical record that the next clinician has no easy way to spot inside 5,000 characters of surrounding text.

If you want the deeper treatment of why notes inflate and how to deflate them without losing clinical content, that is the subject of our companion piece on note bloat (/clinical/note-bloat).

Ask yourself what a covering hospitalist at 2 a.m. needs from your note. She needs to know what you think is wrong, what you are worried about, what you already ruled out, and what she should do if the patient decompensates. That is four sentences. Everything else is available elsewhere in the chart.

Documentation best practices that survive contact with a real clinic

Here are the documentation best practices that hold up across specialties. They are ordered by how much time they return relative to how hard they are to adopt.

1. Write the assessment and plan first. Lead with your reasoning. If you are on an inpatient service, consider APSO ordering (assessment and plan on top, subjective and objective below) so that readers hit the thinking first. It also forces you to commit to a diagnosis before you have written 800 words around it.

2. Document in the room, out loud. Say the plan to the patient while you type it. The patient hears the plan twice, which improves retention, and you finish the meaningful part of the note before you leave the room. This is the oldest advice in the literature and it still outperforms most software [9].

3. Delete the ceremonial sections from your template. Open your most used note template today and remove every element that the 2021 rules no longer require and that you do not actually use clinically [4]. If your template auto populates a full ten system ROS, it is generating false documentation on your signature.

4. Stop importing data that already lives in the chart. Reference results, do not paste them. "Potassium corrected to 3.9 this morning" beats a pasted basic metabolic panel table that will be stale by afternoon rounds.

5. Build five to ten precise dot phrases, not fifty vague ones. The highest yield ones are for your most repeated counseling, your standard follow up instructions, and your three most common diagnoses. Vague templates create bloat. Specific ones eliminate typing.

6. Set a note timer for two weeks. Colleen Fogarty, writing in Family Practice Management, recommends timing individual notes and setting a personal target, suggesting five minutes or less for residents [9]. You will discover your slow notes are concentrated in two or three visit types, and that is where your fix goes.

7. Close notes before you leave the building, even imperfect ones. A note signed at 5:15 with one unresolved lab is worth more than a perfect note signed at 10:40 p.m., because the second one costs you an evening and produces no additional clinical value.

For a specialty by specialty treatment of the fastest available techniques, see our guide on how to speed up clinical documentation (/clinical/speed-up-documentation).

See how Physicians' Copilot speeds your notes.

Charting efficiency at the system level, where the real leverage sits

Individual technique has a ceiling. If you are in a clinic with a badly built template library, no documentation support, and a 15 minute visit slot for a new diabetic on six medications, no dot phrase is going to save you.

The AMA's guidance on cutting documentation load is structural rather than personal, and worth pressing your leadership on. Their six steps: assemble a change team, decide which team members will help with documentation, choose a model (advanced team based care where clinical staff document during visits, or a clerical documentation assistant model), define the workflow explicitly, start with a pilot of one or two willing physicians, then assess and optimize. They note a realistic implementation timeline of three to six months [10].

Two points from that list are worth emphasizing because physicians consistently skip them. Defining the workflow explicitly matters more than choosing the model. And piloting with willing volunteers prevents the classic failure where a system rolls out a documentation tool to 400 skeptical clinicians at once and concludes it does not work.

Training is the other neglected lever. The KLAS Arch Collaborative, synthesizing feedback from tens of thousands of physicians, consistently finds documentation burden among physicians' top requests and finds that EHR optimization work meaningfully improves satisfaction, with the cost of a single burned out physician in turnover and lost productivity estimated at roughly $87,000 [11]. Two hours of personalized EHR training is one of the cheapest interventions available to a department, and almost nobody schedules it.

Where ambient AI fits, and where the evidence currently stops

Ambient documentation tools listen to the visit and draft the note. The enthusiasm is enormous. The evidence is real, mixed, and more interesting than either the vendors or the skeptics want it to be.

The strongest well being signal comes from a multicenter quality improvement study published in JAMA Network Open in October 2025. Among 263 ambulatory clinicians across six health systems using an ambient AI scribe for 30 days, self reported burnout fell from 51.9 percent to 38.8 percent (odds ratio 0.26, 95% CI 0.13 to 0.54). Note related cognitive task load dropped 2.64 points on a 10 point scale, ability to give patients undivided attention improved 2.05 points, and after hours documentation time fell by 0.90 hours per week [12].

The time savings picture is more sober. In a randomized trial published in NEJM AI in November 2025, 238 outpatient physicians across 14 specialties at UCLA were randomized to one of two ambient scribe products or usual care. One product produced a 9.5 percent reduction in time in note versus control (95% CI, 17.2 percent to 1.8 percent reduction; P=0.02). The other produced no significant change (1.7 percent reduction; P=0.66). Both arms showed improvement in burnout and task load measures. Notably, the tools were only used in 33.5 percent and 29.5 percent of eligible visits respectively, and physicians reported clinically significant inaccuracies "occasionally" on a five point scale [13].

A larger real world analysis of roughly 1,800 clinicians across five academic medical centers, reported by STAT in April 2026, found about 16 minutes of documentation time saved per eight hour patient care shift and no significant effect on after hours EHR time, with primary care physicians benefiting most and use patterns inconsistent across adopters [14].

Put those together honestly. Ambient AI appears to reduce the felt weight of documentation more reliably than it reduces the clock time, the effect size varies substantially by product and by specialty, and the benefit depends heavily on whether clinicians actually turn it on. A tool used in a third of visits cannot deliver a third of anything.

That last point is the one I would press hardest with any vendor. Ask them for utilization data from real deployments, not pilot data from enthusiasts.

This article is educational and is not legal advice. Contract terms vary by state and employer. Have any agreement reviewed by a licensed attorney before signing.

No AI tool replaces clinical judgment. Verify all outputs against primary sources and your own assessment.

The verification burden is not theoretical. An ambient draft that invents a normal neurologic exam you did not perform is your signature on a false statement. Read what you sign, every time, and build the read into your workflow rather than treating it as optional diligence.

A 30 day plan for better physician notes

If you want better physician notes without a system wide project, run this sequence. It takes about 90 minutes of setup spread over a month.

Week 1, measure. Time ten notes. Record the visit type next to each. Pull your own EHR usage report if your system offers one (Epic Signal, Cerner Advance, or equivalent) and note your after hours minutes.

Week 2, prune. Open your three most used templates. Delete every element the 2021 E/M rules no longer require and that you do not use clinically [4]. Remove every auto populated data pull that duplicates a chart tab.

Week 3, shift the timing. Pick one clinic session and commit to writing the assessment and plan in the room. Accept that the first two sessions will feel awkward. Measure again at the end of the week.

Week 4, add help. Choose one of: a documentation role on your team, a personalized EHR optimization session, or an ambient tool with a defined verification step. One, not three. Then remeasure at day 30 against your week 1 baseline.

New attendings should pay particular attention to weeks 2 and 4, because the habits you set in your first year are the ones you will still be running in year seven. We cover that transition in what new attendings wish they had known about documentation (/clinical/new-attending-documentation).

Key takeaways

• Physicians averaged 57.8 hours per week in 2024, with only 27.2 spent in direct patient care, and 22.5 percent reported eight or more weekly hours in the EHR outside 5:30 p.m. to 7 a.m. [1]

• Since January 2021, history and exam no longer determine office visit code selection, yet average outpatient note length rose 8.1 percent between 2020 and 2023 [4][5].

• Clerical work accounts for about 44 percent of EHR time, and roughly 86 minutes per day happens after clinic ends [6].

• Ambient AI scribes show consistent improvement in burnout and cognitive load, but measured time savings range from none to about 9.5 percent depending on product, and real world utilization is often under one third of eligible visits [12][13][14].

• The highest yield individual changes are pruning your templates, writing the assessment and plan in the room, and referencing chart data instead of importing it.

Documentation will not fix itself, and waiting for your institution to solve it has been a losing strategy for fifteen years. Start with your templates this week. It is the only part of this entirely within your control.

See how Physicians' Copilot speeds your notes.