How many units of 95004 can you bill for a skin test panel?

How many units of 95004 can you bill on one date of service?

Eighty. The Medicare Medically Unlikely Edit for 95004 is 80 units per date of service, and the unit is one antigen, so 80 units means 80 antigens tested on that day. Bill 95 and the claim does not pay 80 and drop 15. It denies all 95, because the edit carries a Medically Unlikely Edit Adjudication Indicator of 3, which is adjudicated against the date of service rather than the claim line.

The MAI matters more than the ceiling. An MAI of 1 is applied at the line level and can be bypassed with modifiers such as 76, 77, 91, RT and LT. An MAI of 2 is absolute, and CMS has identified no circumstance in which a higher value is payable. An MAI of 3 sits in between: the value is unlikely on a correctly coded claim but can be payable in unusual circumstances with supporting documentation. That is the category 95004 lives in, which is why an over-the-ceiling denial is worth appealing rather than writing off. Across every payer and code, appeals we file are overturned 68% of the time, and 19% of denied claims were never reworked or appealed in the practices we review before we take over.

The neighbouring allergy codes have their own ceilings, and they are much lower: 95024 intradermal testing stops at 40 units, 95017 venom testing at 27, and 95018 drug and biological testing at 19.

What is the 95004 CPT code description, and what counts as one unit?

The 95004 CPT code description covers percutaneous tests, meaning scratch, puncture or prick tests, with allergenic extracts and an immediate type reaction. The descriptor includes test interpretation and report, and it ends with the instruction to specify the number of tests. That closing phrase is the whole billing rule: the code is reported once, with the antigen count in the units field, box 24G on the CMS-1500.

Medicare contractor guidance is blunt about it. The CMS article on allergy testing states that a physician performing 25 percutaneous tests must bill 95004 and specify 25 in the units field. Eighteen scratch tests is 18 units. Thirty-six is 36 units.

Three things that are not units. Scratches are not units, antigens are: two pricks of the same extract is still one unit. Positive and negative controls are performed and charted but are not counted in the reported test total under Medicare rules, a point the American College of Allergy, Asthma and Immunology makes directly. And dilutions of the same antigen are not additional 95004 units, though they do matter for the sequential and incremental codes. We find the control problem in 1 in 6 prick panels we audit, and it is the quietest overbilling risk in allergy coding because it inflates every single claim by two units.

What happens when a prick panel is larger than 80 tests?

You have two real options, and only one of them is safe by default. Splitting the panel across two dates of service is the clean answer when the clinical plan allows it. Billing the full count on one date and appealing is the answer when the patient cannot come back and the record supports the size of the panel. Guessing between them after the denial arrives is what costs money.

Here is the sequence we run when a scheduled panel is going to clear 80 antigens:

1. Check the panel size against the MUE before the visit, not after the claim. A 95-antigen order is a scheduling decision, not a billing one.
2. Check the payer. Medicare applies the MUE. A commercial plan may apply its own annual cap instead, and that cap can bind long before 80 units ever do.
3. Pull the patient year-to-date test count for 95004, 95017, 95018, 95024 and 95027. Several policies count those codes together against one annual total.
4. Split the order across two dates where the clinical plan allows it, with the second date documented as a planned continuation rather than a repeat.
5. If the panel must run on one date, document why the antigen count was individualised to this patient, and file the appeal with that documentation attached rather than resubmitting the claim.
6. Never resubmit a denied line as a duplicate. Duplicate claim denials made up 9% of denials in our claim audit, mostly from resubmitting instead of correcting.

Splitting oversized panels across two dates of service recovered a median $4,100 a year per allergy provider in our client data. That is not a large number on its own. It is a large number relative to the effort, which is one scheduling rule.

What is the 95004 CPT code cost per unit, and what does a panel pay?

95004 carries 0.11 total RVUs. At the 2026 Medicare conversion factor of $33.4009 for non-qualifying-APM clinicians, that is about $3.67 per test in the national non-facility setting. The 2025 national amount was $3.56, and the 2024 amount was $3.66, so this is a code that has been flat to slightly down for years. 95024 intradermal testing carries 0.23 RVUs, or roughly $7.68 per test at the same conversion factor, against $7.44 in 2025.

Per unit the code looks trivial. Per encounter it is not, because the unit count does the work. A 40-test panel is about $147, a 60-test panel about $220, and an 80-test panel about $294. A 95-test panel billed on one date is about $349 of charges and $0 of payment.

A worked example: a three-provider allergy practice

Take a three-provider allergy practice collecting $95,000 a month. It runs 18 prick panels a week, averaging 60 antigens each. That is 1,080 units a week, or about $3,964 a week at $3.67 a unit, which works out to roughly $17,200 a month from 95004 alone, close to 18% of collections.

Now change one thing. Two panels a week run at 95 antigens on a single date. Those two panels carry about $698 in charges a week, and under an MAI 3 edit they pay nothing until they are appealed or rebilled. Over a year that is roughly $36,300 sitting in denials generated by a scheduling habit. Splitting each of those two panels into a 50-unit and a 45-unit visit costs the practice two extra appointment slots a week and converts the whole amount to paid claims. That is the entire trade.

A denial category that repeats weekly is never a billing problem. It is a front-desk rule that has not been written down yet.

95004 vs 95024: can you bill both on the same date?

Yes, when the tests are for different allergens or different dilutions. CMS contractor guidance is explicit that a single test code and a sequential and incremental test code may both be reported on the same date of service if the tests cover different allergens or different dilutions of the same allergen. What is not reportable is a single test and a sequential test for the same dilution of the same allergen.

The common clinical pattern is a prick panel first, then intradermal follow-up on the antigens that came back negative. That is 95004 for the prick count and 95024 for the intradermal count, two lines, each with its own unit total, each against its own ceiling of 80 and 40. Where practices get into trouble is running the intradermal on the same antigen at the same dilution the prick test already covered.

95004 CPT code modifier rules

Two modifiers come up. Modifier 25 goes on the evaluation and management code, not on 95004, and only when the visit was significant and separately identifiable from the testing. The testing code already includes interpretation and report, so an E/M billed for reading the test is not payable. Modifier 59 or the X series is sometimes requested by commercial plans to distinguish the 95004 and 95024 lines, though Medicare guidance does not require it when the allergens differ. Check the payer, not the habit: 95004 and 95024 billed on the same date were denied for a missing or wrong modifier on 12% of same-day 95004 and 95024 claims in our audit. Coding and modifier errors caused 21% of denials across every specialty we audit, so this is not an allergy-specific weakness.

What does the LCD for 95004 require in the record?

The CPT 95004 billing guidelines that contractors actually adjudicate against sit in the local coverage article on allergy testing, and they are shorter than most people expect. The antigen count must be individualised to the patient based on history and environmental exposure. Testing should not exceed two strengths per unique antigen. Allergy testing performed as part of a rapid desensitisation kit under 95180 is not separately reportable. Testing and immunotherapy on the same day falls outside standard practice.

None of that gives a number, which is exactly why the MUE catches so many people. The coverage article tells you how to count and what to document. The MUE table tells you where payment stops. They are different documents, and the one that ranks in search is not the one that denies the claim.

95004 interpretation and report

Interpretation and report is bundled into the 95004 descriptor. That means the record needs a documented reading of the results, with wheal and flare measurements and a clinical conclusion, and it means you cannot bill an E/M for producing that reading. The two rules run together: the interpretation is required for the unit to be payable, and it is not separately payable. Practices that skip the documented interpretation are exposed on audit even when the unit count was perfect, and 63% could not name their top three denial reasons in our practice manager survey, which is usually how that exposure goes unnoticed for a year.

What are the mistakes that cost the most on allergy test units?

Five patterns account for most of what we recover on allergy testing claims. Controls billed as units, which inflates every claim and is the hardest to defend on audit. Panels scheduled over the daily ceiling with no split rule. Annual payer caps never checked, which is where the 23% of allergy testing denials in our claim audit come from. Same-day 95004 and 95024 lines missing the modifier a specific plan wants. And an E/M billed alongside testing for what was really the test interpretation.

None of those are exotic. They sit inside the same three buckets that drive denials everywhere: eligibility and coverage errors caused 24% of denials, coding and modifier errors 21%, and missing or invalid prior authorisation 17% of denials in the Luxen claim audit.

Eligibility and coverage problems are the largest single slice at 24%, coding and modifier errors follow at 21%, and missing prior authorisation accounts for 17%. That leaves 38% spread across everything else. The first three are all preventable before the claim goes out, which is the point: for a per-unit code like 95004, the work that protects revenue happens at scheduling and eligibility verification, not in denial and AR follow-up afterwards. Correct unit counting sits with certified medical coding and is the cheapest control in the whole chain.

Should allergy testing billing stay in house or go outside?

Both work. The question is whether the volume of per-unit codes justifies a specialist. A practice running a handful of prick panels a month would do better fixing scheduling rules than changing billers.

The cost comparison is the honest way in. Fully loaded in-house billing cost 7.9% of collections for practices under $2M in our billing reviews, across 96 practices that shared payroll data. Outsourced full revenue cycle work, ours included, runs 3% to 6% of collections. That gap is real, but it is not the whole argument: in-house staff sit next to the schedule, which is where the unit problem starts. Open biller roles took a median 67 days to fill in our practice manager survey, and 34% of practice managers replaced a biller in the past two years, so the in-house option carries a continuity cost that rarely shows in the spreadsheet. Whichever way you go, compare on denial reporting rather than rate. For the selection side, see how billing companies are evaluated and the revenue cycle end to end. We handle the whole cycle as full-service medical billing inside your existing system, with no migration, month to month, with 30 days notice. A free billing review will tell you whether units are actually where your money is going.

How do payer limits on 95004 compare?

The Medicare ceiling is a daily one. Most commercial policies use an annual one instead, and the spread between them is wide enough that the same panel is routine at one plan and a denial at another. These are the published positions.

Two numbers get confused constantly. The 70 in the practice parameter is what allergists argue is clinically appropriate. The 80 in the MUE table is what Medicare pays. They are not the same document and neither one overrides a commercial contract.

Allergy and immunology

This is the home specialty for 95004, and the unit count is the revenue. A high-volume practice running 60-antigen panels is billing more than a thousand units a week, so a two-unit control error repeated across every claim compounds fast. Annual caps matter more here than anywhere else, because these are the practices that retest and run intradermal follow-up in the same year. Our allergy and immunology billing page covers the immunotherapy side, including how 95165 doses are counted, which is a separate unit problem with its own ceiling.

Dermatology

Dermatology bills patch testing under 95044 far more often than prick testing under 95004, and 95044 has its own MUE of 90 units, counted per patch applied rather than per antigen. The bigger exposure in dermatology is the same-day visit: same-day procedures with an E/M visit were denied for modifier 25 on 14% of claims in our audit. When a patch test read happens at a visit that also treated something unrelated, the 25 has to be defensible in the note. See dermatology billing for the wider pattern.

Primary care and pediatrics

Primary care runs prick panels at low volume, which is exactly why the unit rule gets missed: there is no one in the practice who bills it weekly. The same modifier 25 problem shows up here in a different shape: problem-oriented visits billed alongside an annual wellness visit lacked modifier 25 on 12% of claims. In pediatric billing, add the Medicaid layer, where state fee schedules rather than the national MUE often set the limit and the caps are usually lower.

Physical and occupational therapy

Therapy has no allergy codes, but it is the purest example of the same failure mode, because the 8-minute rule makes almost every therapy line a unit calculation. Unit errors appeared on 9% of therapy claims in our audit of 4,300 claims, and the KX modifier was missing on 21% of Medicare therapy claims past the threshold. The lesson transfers exactly: when the code is per unit, the count is the claim, and the count is set before anyone opens the billing software.

Behavioral health

Behavioral health codes are time-based rather than count-based, which creates the same exposure through documentation instead of arithmetic. In our audit, 18% of 90837 claims had documented session time under 53 minutes, meaning the code billed did not match what the note supported. It is the mirror image of billing 95 units of 95004: the service may well have happened, but the record does not carry the number the code requires.

Ambulance and EMS

Mileage is the per-unit code in EMS, and mileage units were wrong on 4% of Medicaid transport claims in our audit. The documentation parallel is stronger still: Physician Certification Statements were missing or unsigned on 18% of non-emergency transports, the same structural problem as an unreported test interpretation. The paperwork that makes the units payable is generated by clinical staff, not billing staff, and that is where it goes missing.