TL;DR: 99203 is the new-patient office visit for low medical decision making or at least 30 minutes of total time. Payers cut it two ways: down to 99202 when the MDM or time does not hold, or over to 99213 when their claim history shows a same-specialty visit in your group within three years. The first is an appeal, and a registration fix handles the second.
Your remit says you billed 99203 and got paid for 99202. Or it says 99213, for a patient your front desk swore was new. Nothing in the chart looks wrong. The provider is annoyed, and next month the same payer makes the same adjustment. That repeat is the tell: a level-three cut usually traces to a rule the payer applies the same way every time. Correcting one claim does not touch the rule. Practices running medical billing services in-house often spot these adjustments only after they age, because nobody owns the level-of-care line on the remittance.
The code describes a level-three new-patient office or outpatient visit. It requires a medically appropriate history or exam, plus either low medical decision making or 30 minutes of total practitioner time on the date of the encounter. Under the current CPT office visit rules, that time is a floor to meet or exceed, not a range. The AAFP's table of time and MDM levels lays out the full ladder.
|
New patient code |
Medical decision making |
Total time on the date (met or exceeded) |
|---|---|---|
|
99202 |
Straightforward |
15 minutes |
|
99203 |
Low |
30 minutes |
|
99204 |
Moderate |
45 minutes |
|
99205 |
High |
60 minutes |
Level three sits in the middle of that ladder. A level-four claim that a payer edit reduces by one step lands there. Going the other way, a level-three claim that a reviewer finds short drops to 99202. Pressure arrives from both directions, so this is one of the codes worth watching closest on a remit.
A note holds the level when it proves either the MDM or the time, and it only has to prove one. Providers who document both give the claim two ways to survive review instead of one.
Medical decision making has three elements: the problems addressed, the data reviewed or ordered, and the risk of the management plan. Low MDM requires two of the three to reach the low level. In practice that usually means one stable chronic illness or an acute uncomplicated injury, paired with either limited data or a management plan carrying low risk. Reviewers look for the reasoning, not the diagnosis list. A problem the provider names but never assesses does not count toward MDM.
Time-based leveling counts everything the billing practitioner did for that patient on the date of service. Reviewing outside records before the visit, counseling, ordering tests, and documenting after the patient leaves all count. Clinical staff time does not. Write the total minutes plainly, because a payer reviewer will not infer 30 minutes from a long note.
Four things show up in nearly every level-three note that survives review:
That last item is the one documentation cannot fix.
A claim that comes back paid at another code has gone one of two directions. Telling them apart is the first decision. One is a documentation dispute, while the other is a registration problem.
Here the payer accepted that the patient was new but decided the visit supported only straightforward MDM or under 30 minutes. Payers that use it report this as a level-of-care adjustment such as CO-186, and the difference lands in the contractual adjustment column. When the note supports the billed level, this cut is appealable.
In this case the payer decided the patient was established. Under the three-year rule in the CMS Evaluation and Management Services guide, a patient is established if they received a professional service from the same physician, or from another physician of the same specialty and subspecialty in the same group, within the past three years. Nurse practitioners and physician assistants count as the same specialty as the physician they work with. Some payers deny the line. Others reprocess it as an established-patient code such as 99213. Documentation does not move this one. Claim history does.
Provider education is the obvious fix: better notes, clearer MDM, stated time. It works for the claims a person reviews. Against the claims an algorithm cuts, it does nothing, because the algorithm never opened the note.
Automated E/M leveling runs on claim data. The edit compares the billed level with the diagnosis codes on the claim and, at some payers, with the provider's billing pattern over time. Cigna's policy for dates of service on or after October 1, 2025, for example, allows a one-level reduction on level-four and level-five E/M claims when the submitted diagnoses do not appear to support the complexity billed. A new-patient level-four claim cut under that logic lands on 99203. Better charting changes what sits in the chart. The edit reads only what sits on the claim. When the next claim carries the same diagnosis pattern, the same cut fires, and it keeps firing after every round of education.
The new-patient reclass recurs for a parallel reason. Payers store claim history by billing group and specialty, not by your scheduling system. Your front desk asks whether the patient has been to this office. The payer asks whether anyone of that specialty under your group billed a professional service in three years. A departed physician counts. So does a colleague at another location, an acquired practice now billing under your tax ID, or an NP in the same specialty. Every new hire or acquisition widens that history overnight. Meanwhile registration staff keep following the same script they always have, and the reclass rate climbs anyway.
|
Fix the practice tries |
What the payer edit reads |
Why the downcode returns |
|---|---|---|
|
Provider documentation training |
Billed level against claim diagnoses |
The note improves, but the claim fields that trigger the edit stay the same |
|
Asking patients "have you been here before" |
Group and specialty claim history across three years |
The question tests the location, while the payer tests the whole group |
|
Appealing each cut one claim at a time |
The same rule on every new claim |
A won appeal pays that claim and leaves the trigger in place |
Stopping the repeat means changing what the claim carries, or changing what the front end checks before anyone codes the visit. Neither shows up in a chart review.
Appeal the 99202 cut. The appeal puts the existing note in front of the payer and shows that it already supports low MDM or 30 minutes. It does not recode the visit. Our CO-186 downcoding appeal walkthrough covers the remittance codes, the payer dispute forms, and the deadlines. The same process applies whether you billed 99214 and the payer paid 99213, or you billed a level-three new patient and got level two.
A 99213 reclass is appealable only when the payer's history is wrong. That means the earlier visit came from a different specialty, a different group, more than three years back, or a service that was not a professional visit at all. When the payer's history is right, the fix belongs at registration. Appealing it anyway burns hours you could spend on cuts you can win.
The harder question is which of your own downcodes deserve the time. That depends on your payer mix, which payers run automated leveling, and whether your cuts cluster by provider, by diagnosis pattern, or by location. No published benchmark answers it. Your own remits can, once someone sorts them by payer and adjustment type and works them inside a denial management routine.
Catching a downcode is a visibility problem before it is a coding problem. What matters is who sees the level-of-care adjustment, how fast, and whether that person can trace it back to its trigger.
|
In-house billing team |
Offshore billing vendor |
HRG |
|
|---|---|---|---|
|
Who sees the adjustment |
Whoever posts payments, often as a contractual write-off |
A posting team working from exported data |
The billers on your account, working inside your EHR and payer portals |
|
How fast someone works it |
When a staff member has time between other tasks |
On the vendor's queue and time zone |
Within 24 to 48 hours |
|
Pattern tracking |
Depends on one staff member noticing |
Usually a summary report after the fact |
Weekly and monthly A/R reviews by the billers doing the work |
|
New-patient reclass check |
Rarely compared against group history |
Rarely visible to the vendor |
Our billers send it back to registration with the payer's history in hand |
|
Coding role |
Varies by staff |
Varies by vendor |
Coding consultation as needed, never coding itself |
Each setup can work for some practices. The difference is whether the adjustment reaches someone who can act on it while the appeal window is still open.
We work inside your EHR, PM system, and payer portals, so a level-of-care adjustment shows up where your team already works rather than in a separate report weeks later. Our 100 percent U.S.-based billers post the remit, see the 99202 or 99213, and decide which direction the cut went. When the note supports the billed level, the claim goes into appeal with the chart attached. For a reclass, our team traces the payer's history to the prior visit and sends your front desk the provider and date that caused it.
After over 26 years of working payer edits, few leveling patterns surprise us anymore. Most of them repeat by payer, by provider, or by diagnosis, and the repeat only shows up when the same billers look at the same accounts every week. That review, not the single appeal, is what stops the next downcode.
HRG clients see denials drop 15 to 30 percent once appeals and A/R follow-up run on schedule.
On scope, HRG provides coding consultation as needed. HRG does not perform coding. Your coders or coding vendor assign the level, and we work the claim from submission through payment, appeal included.
If your remits keep showing level-three visits paid as 99202, or new patients coming back as established, the pattern is usually sitting in your own claims data. Want to see which payers are driving it and which cuts are worth appealing? Schedule a strategy call with Andy Garcia. We will walk through what your numbers look like. No pitch, no pressure.
Yes, when the visit is significant and separately identifiable from the procedure and the claim carries modifier 25. Payers scrutinize that pairing closely. The note needs E/M work beyond the procedure's own pre-service evaluation. HRG tracks modifier 25 denials on new-patient visits during denial follow-up and routes coding questions to your coders.
It can. The three-year rule counts professional services, and a visit your physician billed in another setting counts the same way an office visit does. HRG traces each reclassified claim to the specific prior service so your registration team knows which history the payer used.
No. HRG provides coding consultation as needed and does not perform coding. Your coders keep the coding decision, and HRG manages the claim, the appeal, and the payer follow-up around it.
No. HRG works inside your existing EHR, PM system, and payer portals, so the adjustment, the note, and the appeal all stay in systems your team already uses. You keep real-time visibility without a separate dashboard or a PDF report.
One claim rarely justifies the time on its own. A downcode that repeats across a payer, a provider, or a diagnosis pattern does, because the same trigger will keep firing. HRG sorts adjustments by payer and type during A/R reviews, and our billers work the highest-recovery appeals first.
New-patient visits feed every specialty's growth. In practices built on them, like primary care, a steady stream of 99203 cuts compounds faster than any single denial. Catch the pattern and the cuts stop coming back.