The Insulin Pump Claim Dies on a Lab Draw — Not the Diabetes (E0784)

The referral that looks like the easy one

An endocrinologist's office faxes over an insulin pump order. Type 1 diabetes since she was nineteen. She's 68 now, on Medicare, and she's been fighting overnight lows for a year. The endo has written a two-page letter explaining exactly why a pump is the right call, and every word of it is true.

Your coordinator sees a clean clinical story and moves it forward. Six weeks later the claim denies — not the accessories, not one line item, the whole thing. Pump, tubing, reservoirs, insulin. Denied as not reasonable and necessary.

What was missing wasn't clinical judgment. It was a fasting blood sugar drawn at the same time as the C-peptide.

Two gates, and most files only clear one

Under Medicare's External Infusion Pumps LCD, continuous subcutaneous insulin infusion is covered when criterion A or B is met, and criterion C or D is met. A and B are the lab gate. C and D are the history gate.

The LCD is unusually blunt about what happens otherwise: if criterion A or B is not met, the pump and related accessories, supplies and insulin are denied as not reasonable and necessary. Same sentence again for C or D. One denial reason, four codes' worth of revenue.

Criterion A: the lab you can't reconstruct later

Criterion A has three parts, and the third one is where files quietly fail.

First, the C-peptide level has to be at or below 110 percent of the lower limit of normal for the laboratory's own measurement method. Note whose number that is — the lab's, not a textbook's. A C-peptide result printed without its reference range is a result nobody can adjudicate.

Second, there's a renal accommodation. For beneficiaries with renal insufficiency and a creatinine clearance at or below 50 ml/minute — actual or calculated from age, weight and serum creatinine — the threshold moves to 200 percent of the lower limit of normal. That's a meaningful loosening for a population that overlaps heavily with long-standing diabetes, and it gets missed constantly because nobody thinks to pull the creatinine.

Third, and this is the one: a fasting blood sugar obtained at the same time as the C-peptide level must be at or below 225 mg/dl. Not a fasting glucose from the same week. Not the one in the office note. The same draw.

"A C-peptide without its paired fasting glucose isn't weak evidence. For this policy, it's no evidence at all."

Here's why this breaks so reliably. C-peptide gets ordered by endocrinology for clinical reasons, usually years before anyone thought about a pump, and often as a standalone test. The paired fasting glucose exists in a different lab panel on a different day. By the time your intake team is assembling the file, the patient is on a pump-eligible path and re-drawing means another fasting appointment, another lab order, another two weeks.

Teams that catch this early don't catch it because they're more careful. They catch it because someone made "C-peptide and fasting glucose, same draw, with the lab's reference range" a single line item on the intake screen instead of a lab result to be reviewed by a human at 4pm on a Friday.

Criterion B: the shortcut worth checking first

Criterion B is one sentence: the beta cell autoantibody test is positive. No paired glucose, no renal math, no threshold interpretation.

For a type 1 patient diagnosed in adulthood, there's a reasonable chance GAD-65 or islet cell antibody testing was run at diagnosis and is sitting in the referring practice's chart. It's worth asking for before you go chasing a fasting draw. Plenty of intake teams default straight to C-peptide because that's the criterion they've heard of, and never ask the second question.

Criterion C: eight months of history that has to already exist

The history gate is the one that can't be fixed after the fact, because it's retrospective by construction. Criterion C requires that before the pump starts, the beneficiary:

Completed a comprehensive diabetes education program. Was on multiple daily injections — at least three per day — with frequent self-adjustment of insulin dose, for at least six months prior to initiation of the pump. And has documented glucose self-testing averaging at least four times per day during the two months prior to initiation.

Then, on top of that, at least one of five glycemic findings while on the injection regimen: HbA1c above 7 percent, a history of recurring hypoglycemia, wide fluctuations in blood glucose before mealtime, dawn phenomenon with fasting sugars frequently exceeding 200 mg/dL, or a history of severe glycemic excursions.

Read that timeline again. Six months of injections. Two months of four-times-daily testing. Both measured backward from the day the pump starts. If the referral lands on your desk and the patient has been on a twice-daily regimen for four months, there is no packet you can assemble that fixes it — and the honest move is a conversation with the prescriber, not a submission.

Criterion D: the patient who already had a pump

Criterion D exists for a specific and very common situation: someone who was on an external insulin pump before enrolling in Medicare. For them, the requirement is documented glucose self-testing averaging at least four times per day during the month prior to Medicare enrollment.

One month, not eight. But it's a specific month, and it's usually the month when the patient was transitioning insurance, moving between practices, and least likely to have been handing meter downloads to anyone. Suppliers who serve a lot of aging-in patients should be asking for those downloads at first contact, not at claim assembly.

What each gate needs in the file

Continued coverage: the every-three-months rule

Getting month one paid is not the same as getting the rental series paid. Continued coverage of an external insulin pump and supplies requires that the beneficiary be seen and evaluated by the treating practitioner at least every three months.

There's a second requirement in the same paragraph that almost nobody tracks: the pump must be ordered — and follow-up care rendered — by a practitioner who manages multiple beneficiaries on continuous subcutaneous insulin infusion therapy, working closely with a team that includes nurses, diabetic educators and dieticians knowledgeable in that therapy. If a patient's endocrinologist retires and care shifts to a primary care office, that condition is worth a second look before the next resupply ships.

The coding traps that deny clean files

Even a file that clears both gates can lose money on mechanics. Four things worth checking against your current billing setup:

E0784 is the only pump

Subcutaneous insulin is administered using ambulatory infusion pump E0784. Claims for insulin delivered by any other infusion pump code are denied as not reasonable and necessary. There's no interpretation available here.

Supply codes are pump-specific in both directions

For dates of service on or after January 1, 2017, billing A4221, A4222 or K0552 with E0784 is denied as incorrect coding. The mirror image is also true: A4224 or A4225 billed with a pump other than E0784 is denied as incorrect coding. Two different denial patterns from the same misunderstanding, and neither one reads as a coverage problem on the remittance.

The refill rule has three numbers

Contact with the beneficiary or their designee must happen no sooner than 30 calendar days before the expected end of the current supply. Delivery must happen no sooner than 10 calendar days before that end date. And regardless of utilization, a supplier must not dispense more than a three-month quantity at a time. Shipments made without a valid documented refill request are denied as not reasonable and necessary — a rule that costs suppliers real money on autoship programs built for convenience rather than for audit.

Pumps with CGM functionality carry two policies

E0784 billed with E2103 — a pump with dose rate adjustment using non-adjunctive continuous glucose sensing — has to meet the insulin pump criteria andevery CGM criterion in the Glucose Monitors LCD, L33822. Same for E2102 on dates of service from April 1, 2022. If your CGM documentation and your pump documentation live in different workflows, that's the file where they have to meet.

KX is an assertion, not a formality

For dates of service on or after January 1, 2023, the KX modifier goes on the pump, drug and supply lines only if all of the coverage criteria in the LCD have been met — and the evidence supporting it has to be retained in the supplier's files and produced for the DME MAC on request. For E0784 and J1817, that requirement continues to apply for any date of service.

When the criteria aren't met, GA or GZ belongs on the line instead. The distinction matters: GA says you have a properly executed ABN on file and the beneficiary can be billed. GZ says you don't. Appending KX to a file where the C-peptide never had a paired glucose isn't a shortcut — it's a written statement that the criteria were met, made in a claims system that keeps records.

Why this is an intake problem, not a billing problem

Look at where the failures actually sit. A C-peptide drawn without its glucose. A creatinine nobody pulled. Meter downloads that existed on a manufacturer portal for ninety days and then rolled off. A quarterly visit that slipped to five months because the patient felt fine.

None of that is a billing error. Every one of them is a fact that existed somewhere, at some point, and didn't make it into the file — or stopped being true after the file was built. By the time a remittance explains the problem, the patient has been wearing the pump for two months and the conversation with the prescriber is much harder than it would have been on day one.

Insulin pumps also punish partial success in a way most DME categories don't. The denial doesn't take a line item. It takes the pump, the accessories, the supplies and the insulin, in one motion, on one reason code.

Three things to do this week

1. Pull every C-peptide result in your last twenty insulin pump files and look for the paired glucose

Not the nearest fasting glucose. The one from the same draw. Count how many you actually have. That number tends to surprise people, and it's the fastest read available on whether your lab gate is real or assumed.

2. Add "beta cell autoantibody" to your first records request

It costs one line on a form. When it comes back positive, criterion A becomes irrelevant and a two-week lab delay disappears from the file.

3. Put the three-month evaluation on the same calendar as the resupply

Resupply has a schedule because it has a rule attached. The practitioner evaluation has a rule too, and almost never has a schedule. Attaching one to the other costs nothing and closes the most common continued-coverage gap.

The clinical case for a pump is usually the easiest part of these files. The hard part is that Medicare asks for two independent proofs, one of which is a lab pairing nobody orders by accident and the other of which is eight months of history that has to already be true. Both are knowable on day one. Neither is fixable on day sixty.