Note Bloat: How to Write Notes Colleagues Actually Read
Note bloat makes charts longer, not clearer. What the evidence shows, what the E/M rules let you cut, and a before and after note rewrite.
Note bloat is text that makes a clinical note longer without making it more informative: copied narrative, auto-imported lab tables, templated exams nobody performed. The fix is not typing faster or buying another tool. It is deciding what belongs in the note and deleting the rest, which the 2021 and 2023 evaluation and management (E/M) changes now let you do without giving up a billing level [1][2].
Most of us learned to write notes in a system that paid for documented bullets. That system changed. Our notes mostly did not.
Key takeaways
Median outpatient progress note length grew 60.1% in a decade, from 401 words in 2009 to 642 in 2018, while text identical to earlier notes rose from 47.9% to 58.8% [3].
Across 104 million notes at one health system, 50.1% of all text was duplicated from earlier text about the same patient, up from 33.0% in 2015 to 54.2% in 2020 [4].
Since January 1, 2021 for office visits and January 1, 2023 for most other E/M families, history and exam no longer determine your visit level. Medical decision making or total time does [1].
One note writing session plus a leaner template cut medical student note length about 35% and raised measured note quality, so this is teachable in an hour [5].
What note bloat actually is, and what it costs you
The best data comes from Oregon Health and Science University, where researchers analyzed more than 2.7 million outpatient notes written between 2009 and 2018. Median length went from 401 words to 642, a 60.1% increase. Median redundancy, meaning text identical to something already in the chart, climbed to 58.8%. By 2018, only 29.4% of note text was typed by the clinician. The rest was templated (55.9%) or copied (14.7%) [3].
Penn Medicine looked at 104,456,653 notes covering nearly two million patients from 2015 through 2020. Half the record was echo: 50.1% of total text was duplicated from earlier text about that patient, rising from 33.0% to 54.2%. Slightly more than half came from the same author, and 45.9% came from someone else [4].
That last number should bother you clinically. When almost half the duplicated text was written by a different clinician, you cannot tell by reading whether a finding was assessed today or on admission by an intern who has since rotated off service.
American notes are also outliers. A 2018 Annals of Internal Medicine commentary using international Epic data reported that US physicians' notes ran roughly four times as long as notes written elsewhere in the same software [6][7]. Same vendor, same fields, different incentives.
The cost shows up in chart review. A systematic review by the Partnership for Health IT Patient Safety found 66% to 90% of clinicians routinely use copy and paste, and cited a diagnostic error analysis in which copy and paste contributed to 2.6% of errors where a missed diagnosis sent the patient back for unplanned care [8]. Not a large percentage. Also not zero, and each of those cases involved a clinician who read a note and believed something no longer true.
The Penn authors also found that duplication and information scatter trade off against each other, so banning copy and paste can push you from one hazard into the other [4]. The goal is not shorter notes. It is notes where the signal is findable.
See how your offer compares
Compare real pay data before you sign anything
Why EHR note bloat happens, and why it was rational at the time
Nobody set out to write a 1,400 word progress note about a patient whose only change today was a potassium of 3.4.
A 2025 review in JACEP Open names three drivers: documentation written to support billing, documentation written to manage malpractice risk, and EHR tooling that makes inserting text essentially free [9]. The fourth is habit. Once a template exists, deleting from it feels like an act of courage.
The copy paste charting problem
Copy forward is not forbidden. It is also not neutral. The Partnership's four safe practice recommendations are specific: make copied material easy to identify, make its provenance available, train clinicians on appropriate use, and monitor copy and paste practice like any other quality measure [8]. The 2013 AMDIS consensus statement reached a similar place, endorsing reasonable restrictions on copy and paste specifically to limit note bloat [10].
In practice the problem is rarely the problem list or the medication list. Those are supposed to carry forward. The problem is narrative: a hospital course paragraph that grows by one sentence a day for nine days and is never pruned.
And if you assume a scribe shortens the note, the data says otherwise. A study of more than 50,000 outpatient notes from 70 providers across 27 specialties found scribed notes were consistently longer, with most of the extra text coming from templates [11]. A scribe working inside your template will faithfully fill your template.
The billing excuse expired, and most of us did not notice
Under current CMS guidance, E/M visits require a medically appropriate history and examination, and the treating clinician decides what is medically appropriate. Those elements do not determine the visit level. You select the level based on medical decision making or total time on the date of the encounter. That applied to office and other outpatient visits from January 1, 2021, and to most remaining E/M families, including hospital inpatient and observation, from January 1, 2023 [1].
The AMA is equally blunt. Its guidance, updated July 25, 2025, states that the 2021 revisions "no longer include the documented history and physical exam in determining the appropriate level of service," and that when you select a level by medical decision making there is no requirement to document total time at all [2].
Read that again if you are still pasting a 12 system review of systems into every progress note to protect a level 4.
The uncomfortable follow-up: researchers studied what physicians did after the 2021 change and found billing shifted within months, with level 5 visits up 22.6% in relative terms, while note length and EHR time showed no meaningful change [12]. We updated our codes in weeks. We did not update our notes.
Coding requirements vary by payer, state, and organization, so confirm specifics with your coding and compliance team before changing a template others use.
Before and after: turning a bloated note into a useful progress note
An illustrative composite, not a real patient. Day 3 of an admission for community acquired pneumonia in a 68 year old with COPD.
Before (mostly recycled text):
Subjective: Patient seen and examined. No acute events overnight per nursing. Reports feeling "about the same." Denies chest pain, palpitations, abdominal pain, dysuria, headache, visual changes, focal weakness. ROS otherwise negative in 10 systems.
PMH: COPD, HTN, HLD, OA, remote tobacco (quit 2011), cholecystectomy 2009, NKDA. Family history: mother with CAD, father with CVA. Social: lives with spouse, independent at baseline.
Vitals and labs: [auto-imported table: 24 hours of vitals, CBC, CMP, magnesium, phosphorus, two prior chest radiograph impressions]
Exam: General: no acute distress. HEENT: normocephalic atraumatic. Neck supple. CV: regular rate and rhythm, no murmurs rubs or gallops. Pulm: crackles at right base. Abd: soft, nontender, nondistended. Ext: no edema. Neuro: alert and oriented times three, CN II through XII intact, strength 5/5 throughout. Skin warm and dry.
Assessment/Plan: 1. CAP: continue ceftriaxone and azithromycin, day 3. 2. COPD: continue bronchodilators. 3. HTN: continue home meds. 4. HLD: continue statin. 5. DVT ppx: enoxaparin. 6. Dispo: pending.
After (115 words, written for whoever covers tonight):
Day 3, community acquired pneumonia with COPD. Better: temp max 37.6 overnight (38.9 on admission), oxygen down from 4 L to 2 L nasal cannula, eating, walked to the door with physical therapy. Still has right basilar crackles and a productive cough.
Assessment and plan
1. CAP, improving. Day 3 of 5 of ceftriaxone and azithromycin. Blood cultures negative at 48 hours. Convert to oral amoxicillin-clavulanate and azithromycin tomorrow if afebrile and on 2 L or less. No repeat imaging unless he worsens.
2. COPD, no exacerbation features. Scheduled bronchodilators. No steroids.
3. Potassium 3.4 this morning, replaced orally, recheck in the morning.
4. Chronic meds unchanged (lisinopril, atorvastatin).
5. Discharge likely in 48 hours pending oxygen wean. Home with spouse.
Same patient, same reasoning. The second one answers the three questions a cross-covering physician has at 2 a.m.: what is wrong, is it getting better, and what do I do if it is not.
Shorter is not the point. The after version is more specific, not less: it names the antibiotic day count, the culture result, the oxygen trend, and the condition for stepping down. Bloat and vagueness usually travel together.
Generate cleaner notes with Physicians' Copilot.
Seven habits that produce concise clinical notes
Write the assessment and plan first. Put it at the top if your institution allows the APSO format, which the AMDIS consensus endorsed for readability [10]. Writing the reasoning first forces you to pull up only the data you need.
Turn off automatic lab and imaging imports. One pediatric hospital medicine group rebuilt its templates to discourage laboratory SmartLinks, dropping their use in progress notes from 92.6% to 17.8% and cutting character count 18.7%, with no change in billing audit performance [13].
Quote a number only if it changed your plan. A potassium of 3.4 in the plan beats a full metabolic panel in a table above it.
Copy forward structured data, not narrative. Problem list, medications, allergies: yes. Hospital course and subjective: rewrite daily.
Attribute and date anything you reuse. If you bring forward a consultant's impression, say whose it was and when [8].
Delete your own template monthly. Open it and remove anything you have not personally used in the last month. Templates accrete.
Reread the note as the night covering physician. If you cannot answer "what do I do if this patient spikes a fever" from your own note in fifteen seconds, the note failed.
Two of these are tooling and five are judgment, which is roughly the right ratio. For workflow level tactics, see our guide to the best ways to speed up clinical documentation, and for the effect on your evenings, finishing notes before you go home.
Teaching this to residents without making them feel unsafe
Residents write long notes because length is the only visible proof of thoroughness. Nobody has ever been pimped for failing to document a negative psychiatric ROS, but plenty of trainees have been told their note was "thin."
This responds to a single intervention. At Johns Hopkins, pediatric clerkship students who got one note writing session plus a streamlined template with minimal auto-populated data wrote notes about 35% shorter than the prior cohort (median 68.5 lines versus 105), signed them earlier, and scored higher on a validated quality instrument for accuracy, organization, and comprehensibility [5]. Shorter and better, from an hour of teaching.
A 2025 AMA Journal of Ethics piece goes further: take the template away and ask the trainee to write the shortest note that still tells the patient's story [14]. Painful the first time, clarifying after that.
If you supervise, say the quiet part out loud. You will not be judged for a short note. You will be asked about a note that claims something you did not examine.
Will AI scribes fix note bloat?
Partly, inconsistently, and only if you tell them what you want.
A Stanford emergency department study of 8,740 encounters published in February 2026 found that when attendings used an ambient AI scribe, median on shift documentation time fell from 3 minutes 50 seconds to 2 minutes 45 seconds (28%), total EHR time fell 16%, and ambient notes were shorter [15]. A 2025 analysis of Epic Signal data at UC Irvine found the opposite on length: time spent on notes fell while note length rose, most sharply in the first month [16].
Zoom out and the trend is not encouraging. Epic Research analyzed 1.7 billion notes from 166,318 outpatient clinicians across 384 organizations and found that between May 2020 and April 2023, average note length rose 8.1% (4,628 to 5,002 characters) while time per note fell 11.1%. Tooling made writing faster, not shorter. Roughly 40% of clinicians did shorten their notes anyway, across primary care, surgical specialties, dermatology, cardiology, and psychiatry, which tells you it is achievable and personal rather than structural [17].
Generative tools inherit your intent. Ask for a complete note and you get a long one. Ask for a four line assessment and plan written for a cross-covering physician, with only the data that changed management, and you get that. The prompt is the template now.
One more reason to care what the note says: patients read it. Under the 21st Century Cures Act information blocking provisions, patients have a right to electronic access to their clinical information, including notes, without unnecessary delay [18].
No AI tool replaces clinical judgment. Verify all outputs against primary sources and your own assessment.
The one minute note audit
Run this on your next note, before you sign it.
□ Can someone tell from the first three lines what is wrong and whether it is improving?
□ Is every lab value in the note one that changed my plan?
□ Did I personally perform every exam maneuver documented?
□ Is there a sentence here I copied and cannot vouch for today?
□ Does each active problem have a next step with a condition attached, not just "continue"?
□ If I deleted the longest paragraph, would any reader lose something they need?
□ Would I be comfortable if the patient read this tonight?
If the last question makes you want to edit, edit. That instinct is usually right.
For the broader framework this sits inside, see our pillar guide to faster, better clinical documentation
REFERENCES
Centers for Medicare and Medicaid Services. "Evaluation and Management Services" (MLN006764). CMS Medicare Learning Network. May 2026. https://www.cms.gov/files/document/mln006764-evaluation-management-services.pdf. Accessed October 1, 2026.
American Medical Association. "Are physicians required to document the time spent on each specific task associated with an outpatient visit?" AMA Debunking Regulatory Myths series. Updated July 25, 2025. https://www.ama-assn.org/health-care-advocacy/administrative-burdens/are-physicians-required-document-time-spent-each. Accessed October 1, 2026.
Rule A, Bedrick S, Chiang MF, Hribar MR. "Length and Redundancy of Outpatient Progress Notes Across a Decade at an Academic Medical Center." JAMA Network Open. July 19, 2021;4(7):e2115334. doi:10.1001/jamanetworkopen.2021.15334. https://doi.org/10.1001/jamanetworkopen.2021.15334. Accessed October 1, 2026.
Steinkamp J, Kantrowitz JJ, Airan-Javia S. "Prevalence and Sources of Duplicate Information in the Electronic Medical Record." JAMA Network Open. September 1, 2022;5(9):e2233348. doi:10.1001/jamanetworkopen.2022.33348. https://doi.org/10.1001/jamanetworkopen.2022.33348. Accessed October 1, 2026.
McRae AE, Rowe JT, Friedes BD, et al. "Assessing the Impact of a Note-Writing Session and Standardized Note Template on Medical Student Note Length and Quality." Academic Pediatrics. March 11, 2023;23(7):1454-1458. doi:10.1016/j.acap.2023.02.014. https://doi.org/10.1016/j.acap.2023.02.014. Accessed October 1, 2026.
Downing NL, Bates DW, Longhurst CA. "Physician Burnout in the Electronic Health Record Era: Are We Ignoring the Real Cause?" Annals of Internal Medicine. May 8, 2018;169(1):50-51. doi:10.7326/M18-0139. https://doi.org/10.7326/M18-0139. Accessed October 1, 2026.
Stanford Medicine. "U.S. doctors saddled with four times the amount of note taking as foreign counterparts." Stanford Medicine news, Scope. May 7, 2018. https://med.stanford.edu/news/insights/2018/05/u-s-doctors-saddled-with-four-times-the-amount-of-note-taking-as-foreign-counterparts. Accessed October 1, 2026.
Tsou AY, Lehmann CU, Michel J, Solomon R, Possanza L, Gandhi T. "Safe Practices for Copy and Paste in the EHR: Systematic Review, Recommendations, and Novel Model for Health IT Collaboration." Applied Clinical Informatics. January 11, 2017;8(1):12-34. doi:10.4338/ACI-2016-09-R-0150. https://doi.org/10.4338/ACI-2016-09-R-0150. Accessed October 1, 2026.
Genes N, Sills J, Heaton HA, Shy BD, Scofi J. "Addressing Note Bloat: Solutions for Effective Clinical Documentation." Journal of the American College of Emergency Physicians Open. January 13, 2025;6(1):100031. doi:10.1016/j.acepjo.2024.100031. https://doi.org/10.1016/j.acepjo.2024.100031. Accessed October 1, 2026.
Shoolin J, Ozeran L, Hamann C, Bria W. "Association of Medical Directors of Information Systems Consensus on Inpatient Electronic Health Record Documentation." Applied Clinical Informatics. June 26, 2013;4(2):293-303. doi:10.4338/ACI-2013-02-R-0012. https://doi.org/10.4338/ACI-2013-02-R-0012. Accessed October 1, 2026.
Rule A, Florig ST, Bedrick S, Mohan V, Gold JA, Hribar MR. "Comparing Scribed and Non-scribed Outpatient Progress Notes." AMIA Annual Symposium Proceedings. February 21, 2022;2021:1059-1068. https://pubmed.ncbi.nlm.nih.gov/35309010/. Accessed October 1, 2026.
Apathy NC, Hare AJ, Fendrich S, Cross DA. "Early Changes in Billing and Notes After Evaluation and Management Guideline Change." Annals of Internal Medicine. Published online February 22, 2022; April 2022;175(4):499-504. doi:10.7326/M21-4402. https://doi.org/10.7326/M21-4402. Accessed October 1, 2026.
Chong AZ, Lee B, Hollenbach K, Kuelbs CL. "Disappearing Help Text: Implementing a Note-Based Tool for In-Line Clinical Decision Support and Note Bloat Reduction." Applied Clinical Informatics. August 31, 2022;13(5):1033-1039. doi:10.1055/a-1934-8323. https://doi.org/10.1055/a-1934-8323. Accessed October 1, 2026.
Frayha N. "How to Teach Good EHR Documentation and Deflate Bloated Chart Notes." AMA Journal of Ethics. November 1, 2025;27(11):E808-814. doi:10.1001/amajethics.2025.808. https://journalofethics.ama-assn.org/article/how-teach-good-ehr-documentation-and-deflate-bloated-chart-notes/2025-11. Accessed October 1, 2026.
Preiksaitis C, Alvarez A, Winkel M, et al. "Ambient Artificial Intelligence Scribe Adoption and Documentation Time in the Emergency Department." Annals of Emergency Medicine. February 10, 2026;87(5):569-574. doi:10.1016/j.annemergmed.2025.12.017. https://doi.org/10.1016/j.annemergmed.2025.12.017. Accessed October 1, 2026.
Guo Y, Hu D, Wang J, Zheng K, Perret D, Pandita D, Tam S. "Ambient Listening in Clinical Practice: Evaluating EPIC Signal Data Before and After Implementation and Its Impact on Physician Workload." Studies in Health Technology and Informatics. August 7, 2025;329:653-657. doi:10.3233/SHTI250921. https://doi.org/10.3233/SHTI250921. Accessed October 1, 2026.
Epic Research. "Two Years After Coding Changes Sought to Decrease Documentation, Notes Remain 'Bloated'." Epic Research. July 6, 2023. https://www.epicresearch.org/articles/two-years-after-coding-changes-sought-to-decrease-documentation-notes-remain-bloated/. Accessed October 1, 2026.
Assistant Secretary for Technology Policy / Office of the National Coordinator for Health Information Technology. "Information Blocking." HealthIT.gov. Updated April 8, 2026. https://www.healthit.gov/topic/information-blocking. Accessed October 1, 2026.
FAQs
Note bloat is text in a clinical note that adds length without adding information, usually from copy forward, templates, and auto-imported data. Researchers measure it as note length and as redundancy, the fraction of text identical to earlier text in the same chart. At one academic center, median redundancy reached 58.8% by 2018 [3].