E/M Undercoding: How to Tell If Your Practice Is Billing Too Low

Why undercoding is so easy to miss

Most revenue cycle problems leave a trail.

A denied claim comes back with a reason code. A rejected claim returns from the clearinghouse. An underpayment can be identified by comparing the payment to the contracted rate. Each of these problems creates something for your billing team to investigate.

Undercoding is different.

When a visit is coded lower than the documentation supports, the payer typically processes the claim exactly as submitted. The claim is clean. It pays quickly. Your clean claim rate looks great. Your denial rate stays low. Your days in A/R may even look impressive.

Meanwhile, the practice may be leaving revenue behind on encounter after encounter.

The evidence is usually sitting quietly in the medical record: the provider documented more work, complexity, or time than the code submitted reflects. If no one is routinely comparing the documentation to the code, that difference can go unnoticed for years.

How office visit levels are actually selected

Since the 2021 changes to office and outpatient E/M coding, visits in the 99202–99215 range are generally selected using one of two methods: medical decision making (MDM) or total time on the date of the encounter.

History and physical examination remain important parts of patient care and should be documented when medically appropriate, but they no longer determine the E/M level for these office and outpatient services.

Most practices adjusted their templates when these rules changed. What we continue to see, however, is that many practices did not fully adjust the way they select their codes. That is where undercoding can begin.

Route one: medical decision making

Medical decision making consists of three elements. To qualify for a particular MDM level, two of the three elements must meet or exceed that level.

1. Problems addressed

This looks at the number and complexity of the conditions actually evaluated and managed during the encounter. The key phrase is addressed during the encounter.

A diagnosis appearing on the patient's problem list does not automatically count. The documentation should show what the provider evaluated, monitored, treated, or considered as part of the visit.

2. Data reviewed and analyzed

Depending on the circumstances, this may include:

• Unique tests or documents reviewed or ordered
• Review of external notes
• Use of an independent historian when applicable
• Independent interpretation of a test performed and separately reported by another physician or qualified healthcare professional
• Discussion of management or test interpretation with an external physician or other appropriate source

This is an area where we frequently see work performed but not clearly captured in the documentation.

3. Risk

Risk considers the potential complications and/or morbidity associated with the patient's management.

Prescription drug management is one common example of moderate risk, but risk can also be reflected in other treatment decisions, including decisions regarding procedures, hospitalization, treatment limitations, or other management choices. Importantly, the decision not to pursue a particular treatment may also be clinically significant when appropriately documented.

Route two: total time

Time is another opportunity that is frequently overlooked.

For office and outpatient E/M services, total time is not simply the number of minutes the provider spends face-to-face with the patient. When applicable under current CPT guidelines, the physician or qualified healthcare professional's total time on the date of the encounter may include activities such as:

• Preparing to see the patient
• Reviewing records and test results
• Obtaining or reviewing history
• Performing the medically appropriate examination or evaluation
• Counseling and educating the patient or family
• Ordering medications, tests, or procedures
• Communicating with other healthcare professionals when not separately reported
• Documenting clinical information in the medical record
• Care coordination performed on the date of service

Clinical staff time and time spent performing separately reported services are not included.

The important point is that practices do not have to force every visit through the same pathway. A lengthy counseling visit involving relatively straightforward medical decision making may appropriately be coded based on time. A relatively short encounter involving significant medical complexity may be better represented by MDM.

If your providers only think about one pathway, legitimate work captured by the other can easily be missed.

A simple example

Consider an established patient with type 2 diabetes and hypertension. Both conditions have worsened since the previous visit. The provider reviews the patient's recent A1c and metabolic panel, increases metformin, adds an ACE inhibitor, discusses why the patient could not tolerate a previous medication change, and orders repeat laboratory testing in six weeks.

Now look at the encounter through the MDM framework.

• Two chronic illnesses with exacerbation or progression can support a moderate level of problem complexity when the documentation supports it.
• Prescription drug management can support moderate risk.

That gives us two of the three MDM elements at the moderate level, which may support 99214 based on MDM alone. Yet visits like this are often submitted as 99213.

Why?

Usually, no one consciously decided that the visit should be undercoded. Instead, 99213 gradually became the practice's “comfortable” code. That comfort can become expensive. Even a relatively small reimbursement difference multiplied across one or two encounters per provider, per day, across an entire year can turn into a meaningful amount of lost revenue.

Five common reasons practices undercode

1. Fear of audits

Sometimes a practice has experienced an audit. Sometimes a colleague has. Sometimes providers have simply been told that billing a higher-level visit is “risky.” The result is defensive coding. A provider may think: I'd rather bill a 99213 and be safe.

But intentionally choosing a lower code is not the goal. Accurate coding is the goal. If the documentation supports 99214, consistently billing 99213 does not make the practice more compliant. It simply means the code does not accurately reflect the documented service.

2. Time is not fully captured

Providers often think only about face-to-face time. They may spend 20 minutes with the patient but another 10 or 15 minutes reviewing records, ordering medications, coordinating care, and completing documentation on the same date.

When time is the appropriate method for selecting the service level, failing to document the total qualifying time can leave legitimate work unrecognized.

3. Data is being undercounted

This is one of the biggest missed opportunities we see.

A provider may independently review information, interpret a study, speak with another clinician about the patient's management, or perform other qualifying data-related work. But if that work is buried in free text — or never documented at all — the coder may have no way of knowing it happened.

If the work is not visible in the documentation, it cannot reliably support the code.

4. The EHR calculator is doing too much of the thinking

EHR coding tools can be helpful, but they should not replace professional judgment. An automated calculator can only evaluate the information it has been designed to recognize.

Important clinical work may be documented in narrative text, occur outside the examination room, or simply not map neatly into the structured fields the calculator uses. Think of the EHR's recommended level as a starting point — not the final authority.

5. No one owns the question

This is often the real issue.

The provider documents the visit. The coder assigns or validates the code. The biller submits the claim. But who is responsible for periodically asking: does the code we billed actually represent the work documented in this note?

If the answer is “no one,” undercoding can become part of the practice's normal workflow without anyone realizing it.

How to check your own practice

You do not need expensive software to perform an initial undercoding review. Start with your own data.

1. Pull 12 months of E/M utilization. At minimum, include CPT code, rendering provider, date of service, and new versus established patient.
2. Calculate each provider's E/M distribution. Determine what percentage of established patient visits are billed as 99212, 99213, 99214, and 99215. Do the same separately for new patient visits. Do not combine new and established patient codes — doing so can hide meaningful patterns.
3. Compare providers within your own practice. This can be one of the most useful comparisons available.
4. Compare against external benchmarks. CMS publishes Medicare Part B utilization data that can provide a useful reference point for understanding coding patterns by specialty.
5. Review the charts.Once you identify an unusual pattern, select approximately 20–30 representative encounters and compare the documentation to the code submitted.

If several physicians in the same specialty are treating a similar patient population but one provider bills almost exclusively 99213 while colleagues routinely have a mix of 99213 and 99214, that deserves a closer look. It does not automatically mean the provider is undercoding. It means you have found something worth investigating.

Benchmarks should never be treated as coding targets. Your patients, providers, subspecialty, and clinical complexity may legitimately produce a different distribution. Use benchmarks to identify questions — not to determine codes.

What does a “normal” distribution look like?

There isn't one. And that is important.

A primary care physician managing a large population of patients with multiple chronic conditions should not necessarily have the same E/M distribution as an urgent care provider treating mostly minor acute illnesses. Even providers within the same specialty can have legitimate differences based on subspecialization and patient complexity.

The goal is not to make every provider's bell curve look the same. The goal is to identify patterns that do not make clinical sense.

For example, a provider whose coding distribution is dramatically lower than colleagues treating a comparable patient population may warrant review. Likewise, a practice managing a medically complex population but reporting almost no higher-level E/M services should ask why.

Sometimes the answer is case mix. Sometimes the answer is documentation. And sometimes the answer is undercoding.

What should you fix first?

Fix the documentation before trying to fix the coding distribution. This is one of the most important parts of the process.

Simply telling providers to “bill more 99214s” is not a compliance strategy. If the documentation does not support the service, increasing the code creates a different — and potentially much more serious — problem.

Instead, focus on helping the documentation accurately reflect the work already being performed. Start with a few practical changes:

• Document total qualifying time with a clear, explicit statement when time is being used for code selection.
• Clearly document qualifying independent interpretations and discussions with external clinicians when they occur.
• Describe the status of the problems actually addressed: stable, worsening, exacerbated, progressing, improving, uncontrolled, and so on, when clinically accurate.
• Make management decisions visible in the note rather than assuming the diagnosis or medication list tells the story.
• Recheck provider E/M distributions after several months to determine whether documentation and coding patterns have changed.

The goal should be a natural, defensible change driven by better documentation and more accurate coding — not an artificial increase in higher-level services.

Related revenue leaks

Undercoding rarely exists in isolation. If a provider's E/M levels appear unusually low, it may also be worth reviewing whether modifier 25 is being missed when a significant, separately identifiable E/M service is performed on the same day as another procedure or service. It is also worth asking whether the documentation supports the level being reported in the first place.

These problems often travel together because they have the same underlying cause: no one is routinely comparing what happened in the exam room, what was documented in the medical record, and what ultimately went out on the claim.

The bottom line

Undercoding is dangerous precisely because it does not look like a revenue problem. The claims are clean. The payments arrive. The denial rate looks good. And the practice can still be losing thousands — or potentially much more — every year.

Do not start by asking, “How do we get more 99214s?” Start by asking: does the code on the claim accurately reflect the work documented in the medical record?

That question protects compliance, improves documentation, and helps ensure your practice is appropriately reimbursed for the care it is already providing.