What to scrub before you submit a Medicare claim at a specialty clinic
The cheapest place to work a Medicare denial is before submission. The four-layer pre-submission scrub: NCCI edits, PECOS enrollment, OIG and SAM exclusion screening, and place-of-service, taxonomy, and NPI consistency.
The cheapest place to work a Medicare denial is before the claim is transmitted. NCCI edits get most of the attention, but a Medicare or Medicare Advantage claim can also deny because the rendering provider is not enrolled in PECOS, because an ordering or referring provider is on an exclusion list, or because the place-of-service code does not match how the service was delivered. Every one of those is checkable against a CMS or OIG source before you hit submit. Most clinics check one or two and find the rest at the denial stage.
This is the pre-submission scrub a specialty clinic should run on every Medicare and MA claim, what each layer catches, and why the hard part is keeping the underlying data current, not knowing the rules.
- 01NCCI PTP and MUE editsProcedure-to-procedure pairs and per-line unit ceilings, checked against the current quarter's tables.
- 02PECOS enrollmentThe rendering NPI is actively enrolled in Medicare. New hires often start billing before enrollment completes.
- 03OIG LEIE and SAM screeningNo rendering, ordering, or referring party is on a federal exclusion list, checked on the claims cadence, not the credentialing calendar.
- 04POS, taxonomy, and NPIHeader fields that match the enrollment record, including telehealth place of service and type-1 versus type-2 NPI.
NCCI: PTP pairs and MUE units
The National Correct Coding Initiative is two sets of edits. Procedure-to-procedure (PTP) edits define code pairs that should not be billed together for the same patient on the same date. When a pair fires, the Column One code is payable and the Column Two code is denied unless an appropriate modifier is attached and the record supports it. Each pair carries a modifier indicator.
| Indicator | Meaning | Operator note |
|---|---|---|
| 0 | No modifier can override the edit | The two codes will not be paid together |
| 1 | An NCCI-associated modifier may bypass the edit | Use 59 or the X{EPSU} modifiers only when the record supports it |
| 9 | Not specified | Used for same-day-deletion pairs; effectively not applicable |
Medically Unlikely Edits (MUE) are the unit ceiling, the maximum units of a code Medicare expects on a single line for one patient on one date. Bill four units against an MUE of two and you are paid for two and denied for two. An MUE denial is a coding denial, not a medical-necessity denial, so an ABN does not shift liability to the patient for the cut units.
Two operator mistakes turn NCCI into a denial stream. The first is reflexive modifier 59. CMS treats 59 as the modifier of last resort, and high 59 volume is a documented audit trigger. The X modifiers (XE separate encounter, XS separate structure, XP separate practitioner, XU unusual non-overlapping service) are more specific and harder to challenge. The second is stale edit tables. CMS updates the PTP and MUE files quarterly, with occasional mid-quarter replacements, so a scrubber running last quarter's pairs passes claims the current file would flag. That failure does not announce itself: the claims simply start denying a few weeks after a quarter boundary.
PECOS: the provider behind the claim
A correctly coded claim still denies if the rendering provider is not actively enrolled in Medicare. PECOS is the enrollment system of record, and Medicare expects the rendering, and where applicable the ordering or referring, NPI to map to an active enrollment. A provider mid-revalidation, newly hired and not yet enrolled, or recently moved between groups can code clean and still bounce.
This is checkable before submission, and the clinics that get burned are usually the ones adding providers fast: a new hire starts seeing patients, billing starts on day one, and enrollment is still in the queue. The claims accumulate in AR until someone notices they share one NPI. A pre-submission enrollment check raises an advisory warning so the batch is caught at once, not after the denials.
Exclusion screening: OIG LEIE and SAM
Federal payment is not available for items or services furnished by an excluded individual or entity. The OIG maintains the List of Excluded Individuals and Entities (LEIE), updated monthly, and the System for Award Management (SAM) carries a separate federal exclusion record. The obligation extends past the rendering provider to ordering and referring providers.
Most clinics treat exclusion screening as a credentialing task: check at hire, maybe again at re-credentialing. The gap is the months in between. An exclusion that posts in March is not caught by an annual July check, and the claims submitted in between are exposed. Screening on the claims cadence, against the current LEIE and SAM files, closes that window.
Place of service, taxonomy, and NPI
The last layer is the set of header fields that look clerical and deny anyway. A telehealth visit billed with an in-office place of service, or the reverse, draws an edit or a denial, and specialty telehealth practices are exposed because the same provider may deliver the same service in two settings on one day. A taxonomy that does not match the enrollment record misroutes or denies. A type-1 NPI where a type-2 belongs produces denials that read as obscure payer errors but trace to one wrong field. None of these need payer correspondence to catch.
The real problem is data freshness
Read the four layers and the rules are not the hard part. NCCI pairs are published. PECOS enrollment is published. LEIE and SAM are published. The hard part is that every source changes on its own schedule and none of them tell you when.
| Source | Update cadence | What goes stale |
|---|---|---|
| NCCI PTP / MUE | Quarterly, plus mid-quarter replacements | Edit pairs and unit ceilings |
| OIG LEIE | Monthly | Newly excluded parties |
| SAM | Continuous | Federal exclusion records |
| PECOS | The day a revalidation lapses or completes | Active-enrollment status |
| POS / taxonomy | Per-payer, often invisible | Header-to-enrollment matching |
A scrub layer is only as good as the freshness of the tables behind it. The failure mode is always the same: the rules were right last quarter, the data went stale, and the denials arrived before anyone connected them to the source. That makes a pre-submission scrub an infrastructure problem more than a coding one. Knowing the rule is a one-time cost. Keeping every source current, on its own cadence, and running every claim against the current copy is ongoing.
What Foresight handles here
Foresight runs the four checks as one pre-submission scrub on Medicare and MA claims and keeps the reference data current automatically rather than on a person's calendar.
- NCCI PTP and MUE edits run against CMS-sourced tables that refresh on a schedule, so the scrub uses the current quarter's pairs and unit ceilings and surfaces the reason a line was flagged, distinguishing indicator 0 from indicator 1 pairs.
- PECOS enrollment is checked for the rendering NPI and raises an advisory warning, scoped to Medicare and Medicare Advantage, before the claim is transmitted.
- OIG LEIE and SAM screening runs on a scheduled sync with a staleness check, so a newly posted exclusion is caught on the next claim, not at the next credentialing review.
- Place-of-service, taxonomy, and NPI consistency are checked against the enrollment record, with correction rules that can be configured to apply automatically or to flag for review. High-risk modifiers such as 25, 59, and the X-series are never applied automatically.
Working these as four separate fire drills after the fact is the expensive version. Catching them in one scrub before submission is the cheap one, and it keeps the preventable denials out of your ERA reconciliation later.
01How often do NCCI edit tables change?
Quarterly for the regular PTP and MUE files, with occasional mid-quarter replacement files from CMS. A scrubber not refreshed on that cadence drifts out of date within a single quarter.
02Is an MUE denial appealable?
The units denied above the MUE can be appealed with documentation supporting medical necessity and the unit count, submitted to your MAC. But an MUE denial is a coding denial, so an ABN does not move liability to the patient for the cut units.
03Do we screen ordering and referring providers for exclusions, or just our own?
Federal payment is unavailable for services connected to an excluded party anywhere in the chain, so ordering and referring providers are in scope, not only the rendering provider.
04We add providers frequently. What is the most common preventable denial?
Enrollment timing. A new provider who starts billing before PECOS enrollment is active generates clean-coded claims that deny on enrollment. Checking the rendering NPI against current enrollment before submission catches the whole batch at once.
05Can the clearinghouse catch these?
Clearinghouse edits catch a useful subset, mostly format and some NCCI pairs, but enrollment status, exclusion screening, and taxonomy-to-enrollment matching are usually outside what a clearinghouse checks.