E&M Coding Changes 2026: How to Pick the Right Level and Avoid Audits

Two patients are sitting in the same waiting room. Both are 60 years old. Both have diabetes and hypertension.  Both take multiple medications and come in for a follow-up. One visit gets billed at a higher code level and paid without issue. The other gets denied months later, forcing the practice to appeal or absorb the loss.

The patients were nearly identical. The difference was the documentation.

In U.S. healthcare, every patient visit must be assigned an Evaluation and Management (E&M) code before a practice can bill for it. These codes, ranging from 99202 to 99215 for office visits, tell insurance payers how complex the visit was. Higher complexity means higher payment. 

But payers do not take a provider’s word for it. They look at the clinical note to see if the documentation actually supports the code that was billed. When it does not, claims get denied. When the mismatch is repeated across many claims, practices face audits and are sometimes required to return money already paid.

The E/M coding changes 2026 continue building on changes introduced in 2021, which moved the focus away from counting how many body systems a doctor examined and toward what the doctor actually thought about and decided during the visit.

That repositioning made documentation more meaningful clinically, AI SOAP note generator but it also made the rules more nuanced. For providers managing this every day, the challenge is real: how do you document visits in a way that is accurate, defensible, and not enormously time-consuming?

This blog answers that question with specific thresholds, real note examples, and a daily workflow you can put to use immediately.

Why Documentation Quality Determines Everything

E&M services account for 40 to 60 percent of primary care revenue. That makes AI Medical Billing solution accurate coding for E&M one of the most important financial functions a practice performs, even though most providers think of it as administrative work.

The core principle, reinforced by AAFP E&M guidance and the 2026 updates, is this: document the complexity you actually managed, not what you hope to bill. This sounds obvious, but in practice it is easy to write a note that reflects what happened clinically without capturing enough detail to justify the code level. A note that says “follow-up for diabetes, A1C 7.8, stable, continue medications” describes a visit. It does not document a visit in a way that supports billing above the lowest complexity level. There is no evidence of analysis, no record of what the provider was thinking, and no indication of what risks were weighed.

Both notes describe the same type of visit. The second one shows what the provider noticed, why it mattered, what risk it created, and what decision was made as a result. That is the standard 2026 E&M guidelines are held to.

Medical Decision Making: The Exact 2026 Thresholds

The most common method for determining a visit’s code level is called Medical Decision Making (MDM). Think of MDM as a structured way of capturing how hard a visit actually was to manage. Payers use it to verify that the code billed reflects real clinical complexity, not an optimistic estimate.

MDM is assessed across three columns. Two of the three must meet the threshold for a given level. Here is what each column measures.

Column 1: Problems addressed

This column captures how complex the patient’s health issues were during the visit, not overall, but specialty-specific EHR templates, specifically what was addressed that day.

Straightforward (99202/99212): One self-limited or minor problem, meaning something expected to resolve on its own. Example: Acute ear infection expected to clear without treatment.

Low (99203/99213): Two or more minor problems, or one chronic illness that is currently stable and well-managed. Example: Contact dermatitis and vitamin D deficiency, both stable.

Moderate (99204/99214): One or more chronic illnesses that are not well controlled, two stable chronic illnesses together, or one acute illness without serious systemic effects. Example: COPD flaring and requiring oral steroids, or a patient with both hypertension and stable diabetes, or a urinary tract infection without fever or other systemic symptoms.

High (99205/99215): A chronic illness in severe exacerbation, or an acute illness that threatens life or the functioning of a major body system. Example: A heart attack or pneumonia leading to sepsis.

Column 2: Data reviewed and analyzed

This column is not just about what information the provider looked at. It is about what the provider did with that information. A list of lab results sitting in a chart does not count. Interpreting those results in light of the patient’s condition does.

Minimal: No data reviewed.

Low: Test results from any number of sources reviewed during the encounter, or relevant history gathered from a family member or caregiver rather than the patient directly.

Moderate: Any two items from distinct data categories, or one test with the provider’s own interpretation recorded, or a discussion with another physician outside the practice.

High: Any three items from distinct data categories, or one test with a separate detailed written interpretation, or a discussion with an external physician that includes written interpretation.

The most common mistake here is listing tests without any analysis. “Labs reviewed” earns zero credit. “Creatinine doubled from 1.2 to 2.4 mg/dL, concerning for acute kidney injury versus dehydration” demonstrates actual clinical thinking and earns moderate data credit.

Column 3: Risk of management

This column reflects the risk involved in the treatment decisions made during the visit, not the patient’s overall risk level. A patient can be very sick while the visit itself involves only low-risk management. The column asks: what did the provider actually decide to do, and how risky was that decision?

Low: Recommending over-the-counter medications, simple diagnostic tests, or rest.

Moderate: Starting or changing a prescription medication, making a decision about minor surgery where some risk exists, referring to social work, or ordering a test that carries moderate risk.

High: Managing drug therapy that requires close toxicity monitoring (such as warfarin or chemotherapy agents), deciding about elective major surgery, or managing an acute illness threatening normal body function.

This note does not just record what was prescribed. It shows that the provider weighed options, identified a contraindication, selected an alternative, and counseled the patient on a specific risk. That is moderate prescription drug management.

Once all three columns are assessed, the level follows logically. Moderate across problems, data, and risk supports 99214. 

But if MDM is not the strongest basis for a visit, there is another option.

Total Time Method: When And How to Use It

Some visits are complex not because of the diagnosis but because of how much time the provider spent thinking through the situation, counseling the patient or family, or coordinating care with other providers. 

For those visits, documenting total time can be more straightforward than trying to parse MDM.

Total time, in E&M coding, includes all time the provider personally spends on that patient’s care on the day of the visit. It is not limited to face-to-face time in the exam room.

What counts toward total time:

• Preparing the chart and reviewing external records before the visit
• Obtaining history from the patient, family, or through an interpreter
• Ordering medications, tests, or procedures
• Writing the note
• Counseling the patient or family
• Communicating with or referring to other providers

What does not count:

• Time spent by nurses, medical assistants, or other clinical staff
• Time spent performing a separately billable procedure
• Time reviewing records on a different day than the visit

That note would survive an audit because it is specific, it accounts for the time, and it makes clear the provider was personally involved throughout.

Vague time statements are one of the most common audit flags. The note must explain what the time was spent on.

Real-World Documentation Examples By Level

99213 (Low complexity)

 “Ms. Chen, 58, routine hypertension follow-up. BP 132/78 on lisinopril 20 mg daily. Patient compliant. Reviewed home BP log (average 135/80). Recent lipid panel stable compared to three months ago. Plan: continue current regimen, return in six months.” Problems: One stable chronic illness. Data: Minimal. Risk: Low.

99214 (Moderate complexity) 

“Mr. Rodriguez, 62, type 2 diabetes mellitus and hypertension. A1C increased to 8.4 percent from 7.6 percent last quarter. Fasting blood sugars averaging 210 mg/dL. Patient admits missing metformin doses for one month. Reviewed 90-day glucose log and renal panel (CrCl stable at 65 mL/min). Risk of renal progression and cardiovascular events elevated. Plan: add Jardiance 10 mg daily, increase metformin to 2000 mg daily, endocrinology referral, extensive counseling on medication adherence.” Problems: One unstable chronic illness. Data: Moderate (labs reviewed with interpretation and patient history). Risk: Moderate (two medications adjusted with documented clinical rationale).

99215 (High complexity)

 “Mrs. Thompson, 72, presents with progressive shortness of breath on exertion and known heart failure history. O2 saturation 91 percent on room air, BNP 1200 pg/mL (previously 400). Reviewed chest x-ray showing mild pulmonary edema, recent echocardiogram (EF 35 percent), called accepting cardiologist who recommends urgent diuresis. High risk of acute decompensation. Plan: Lasix 40 mg twice daily, daily weights, follow-up in 48 hours or emergency department if symptoms worsen.” Problems: Acute illness affecting major body system function. Data: High (imaging, labs, external specialist consultation). Risk: High (urgent medication escalation with close monitoring required).

The Five Most Common Audit Triggers And How to Fix Them

Payers use automated systems to scan claims for patterns that suggest miscoding. These patterns do not prove fraud, but they do flag claims for closer review and can lead to audits, denials, or requests to return payment. Understanding them helps practices avoid getting caught in those systems in the first place.

Pattern 1

Over 80 percent of visits billed at 99214 or 99215 Every specialty has a typical distribution of code levels based on the complexity of the patient population it serves. A dermatology practice and a hospital medicine group will look very different. When a provider bills high-level codes at a rate far above their specialty’s norm, it triggers automatic review. Running monthly reports that compare each provider’s coding distribution to specialty benchmarks, and flagging anyone more than two standard deviations above the norm, creates an early warning system.

Pattern 2

Identical note language across patients. Electronic health record templates make it easy to insert the same phrase into every note, for example, “patient denies fever, chills, nausea, vomiting, chest pain, shortness of breath.” When payers see this across dozens of charts, it signals that the note was not written based on what actually happened that day. Smart phrases are useful, but they need to be customized with patient-specific findings. “Patient denies fever but reports a three-pound water weight gain and mild ankle swelling” is the same template with details that make it real.

Pattern 3

Vague time documentation As explained above, time-based claims need specifics. Phrases like “extended discussion with family” or “spent significant time coordinating care” will not hold up in an audit. Exact minutes and a breakdown of how time was spent are required.

Pattern 4

A mismatch between the code billed and what the note actually says Billing 99215 when the note reads “stable, continue current medications” is one of the most straightforward audit findings a reviewer can make. A pre-billing review step for all 99214 and above claims, using a simple checklist that matches E/M documentation requirements, denial management solutions catches these mismatches before they become denied claims.

Pattern 5

A sharp coding spike from a new provider When a provider joins a practice and immediately bills 99215 at a rate three or four times higher than the previous provider for the same patient population, it raises questions. Setting up 90-day coding pattern monitoring for new providers and providing early feedback prevents small habits from becoming large audit risks.

A 3-Minute Daily Validation Workflow

Building a brief check into the end of each note takes about three minutes and significantly reduces the chance of a denial or audit finding. The steps are simple.

For teams, pulling five random charts each week, scoring them as a group, and discussing borderline cases builds consistent habits across providers and surfaces problems before they become patterns.

Leadership action plan: practice-wide protection

In individual documentation discipline matters, but it is not enough on its own. Practices that consistently avoid audit problems treat AI medical coder accuracy as an organizational function, not just a provider responsibility.

• A monthly analytics dashboard that tracks 99214 and 99215 rates by provider and specialty, flagging anyone more than two standard deviations above the norm, gives leadership early visibility into risk.

• Quarterly mock audits, in which an external medical coding reviewer examines 20 randomly selected high-level charts, simulate what a real payer audit would find and give providers a chance to correct habits before they become liabilities.

• An annual review of every claim denial, organized by code, payer, and reason, reveals the patterns that keep appearing and points education efforts at the issues that actually matter for that practice.

EHR workflows can be configured to reinforce good habits: required fields for problems, data, and risk on 99214 and higher, and a prompt for time documentation whenever time-based coding is selected.

Monthly provider scorecards comparing coding patterns to peers, using anonymized comparisons, help providers calibrate their own habits without feeling singled out.

Three Principles That Hold Across Every Visit

Document what actually happened, including the clinical reasoning, not template filler. The note should reflect what the provider was thinking, not just what was observed.

Let documentation determine the code, never the other way around. When a provider decides on a code first and then writes a note to match it, the result is a note that looks constructed rather than clinical, and payers have learned to recognize that.

Review one chart per day against these guidelines and fix patterns immediately. Habits that go uncorrected for months become the basis of an audit finding. The same habits caught and corrected early are simply part of getting better.

The 2026 E&M guideline rewards clarity and specificity. Practices that document well get paid appropriately, defend their claims successfully, and give patients the benefit of more complete and thoughtful clinical records. The documentation and the reimbursement end up pointing in the same direction.