The $84 CPAP Resupply Order Never Gets Audited Before It Ships. In 2026 Checking Every One Costs Less Than the Shipping Label.

Eighty-four dollars. That is roughly what a routine CPAP resupply order bills out at when the cushion, the filters and the tubing all fall due in the same ninety-day cycle. Now look at your denial report for last quarter and find how many of those $84 orders came back — plan termed, compliance documentation not on file, quantity over the allowed cadence, patient in a covered Part A stay. You shipped the product, paid the freight, and got nothing.

You have never audited those orders before they ship. Nobody does. Auditing a resupply order costs a person four to six minutes, and four to six minutes against $84 of billed revenue with a thin margin on the product is not an argument anybody wins. So the resupply program runs on volume and eats the denials as a cost of doing business, while the pre-billing review that your reimbursement manager built gets applied to the power mobility files and the ventilators, where the dollars justify it.

The check you skipped, and why skipping it was correct

Be clear about this: skipping the check was the right call under the old arithmetic. A resupply coordinator making $22 an hour loaded costs about $1.83 for five minutes. On an $84 order at a 7% denial rate, the expected loss per order is about $5.88, and you would only catch part of it. Once you add the fact that most of the checks come back clean and the coordinator is doing nothing productive on those, the case falls apart. Anyone who told you in 2024 to inspect every resupply order was telling you to spend a dollar to find eighty cents.

So you did what the industry does. Quarterly eligibility sweeps instead of per-order checks. A compliance report pulled from ResMed AirView or Philips Care Orchestrator once a month instead of at order time. A resupply cadence enforced by the calendar rule in Brightree or Bonafide rather than against the specific plan's actual limits, which differ between straight Medicare, a Medicare Advantage plan and your state Medicaid.

What actually changed in the arithmetic

Running frontier AI got roughly ten times cheaper from 2025 into 2026. Capable models now cost around two dollars for a million words' worth of work, and running smaller models on your own hardware costs roughly ninety percent less than cloud for high-volume repetitive jobs. That is the entire development. There is nothing clever in it.

But it moves the decision. A per-order documentation and eligibility check that cost you a resupply coordinator's five minutes in 2025 now costs a fraction of a cent in 2026, which means the checks you reserved for the ventilator files can run on every single $84 cushion order. The check did not get smarter. It got cheap enough to run everywhere, which for a resupply program shipping thousands of orders a month is a bigger change than any accuracy improvement would have been.

flowchart TD
  A["Order builds in the 90-day resupply queue"] --> B["Coverage still active on this plan today?"]
  A --> C["PAP compliance download on file and current?"]
  A --> D["Quantity within this payer's allowed cadence?"]
  B --> E{"All three clear?"}
  C --> E
  D --> E
  E -->|Yes| F["Releases to the warehouse pick list"]
  E -->|No| G["Held with the exact missing item named"]
  G --> H["Resupply coordinator works a 40-line list, not 6,000 orders"]

What runs on every order now

The useful checks in a resupply program are not exotic. They are the ones your best billing specialist would run if she had unlimited time, and there are about six of them.

• Is coverage active today, not as of the last quarterly sweep — including whether the patient moved onto a Medicare Advantage plan on 1 January, which is the single most common January surprise in this business.
• Is the patient in a covered inpatient or skilled nursing stay right now, which makes the supply the facility's responsibility and your claim a refund waiting to happen.
• Does the last documented usage download support continued need, and is the initial adherence documentation — four hours a night on seventy percent of nights in a consecutive thirty-day period within the first ninety days — actually in the file rather than assumed?
• Does the quantity on this order exceed what this specific payer allows for these codes in this window, rather than what the default cadence rule says?
• Is there a current Standard Written Order on file covering the items being shipped, and is the ordering physician still active and not on a payer exclusion list?
• Does the shipping address match the address on file, since a returned box on an $84 order eats the margin twice.

The output is not a report. It is a hold. Clean orders release straight to the pick list; the ones that fail land on a short exception list that names the specific missing item, so your resupply coordinator spends her morning on forty real problems instead of spot-checking six thousand orders she cannot possibly review.

The arithmetic on 6,000 orders a month

Illustrative numbers for a mid-sized resupply program. Substitute your own volume and your own denial rate off last quarter's report.

The line that matters is the twenty dollars. Not because twenty dollars is impressive, but because it is small enough that the decision is no longer about the running cost at all — it is about whether the checks are wired correctly into your system and whether somebody works the exception list every morning. That is a management problem, which is a much better problem to have than an economics problem.

One caution on the savings figure. Preventing a denial is not the same as collecting the money; some of those 210 orders should genuinely not ship, and that revenue simply does not exist. The honest way to state the benefit is that you stop paying product and freight on orders that were never going to pay, and you stop spending billing time on appeals that were never winnable. Both are real. Neither is $17,640 landing in the bank.

Where the cheap check will happily lead you wrong

A check that runs everywhere also fails everywhere, quietly, if it is wrong. Three specific hazards in this trade.

First, eligibility responses lie by omission. A real-time response can say active coverage and still not tell you the patient has a secondary that is now primary, or that a Medicare Advantage plan requires prior authorization through Evolent for exactly these codes. Treat a clear eligibility check as one input, not as a decision.

Second, do not let an automatic hold quietly stop a patient's oxygen or their ventilator circuit. There is a difference between a $60 mask cushion and a supply a patient depends on to breathe. Set the rule so that respiratory and life-supporting items never auto-hold; they flag to a respiratory therapist and ship while the documentation gets chased.

Third, the compliance question is clinical territory. Whether a patient's usage pattern supports continued need — and what to do about a patient whose usage is dropping — belongs to your RT and the ordering physician, not to a rule that reads a number off a download. Use the check to surface the patients whose usage fell off. Do not use it to make the coverage argument for you.

Frequently asked questions

Is this really worth doing if my denial rate is already low?

Maybe not on resupply. But the same near-zero cost per check applies to things you also skip for cost reasons: verifying the delivery address on every order, screening every ordering physician against exclusion lists, or checking every capped rental approaching month eleven. Pick whichever of those you currently sample rather than check, and start there.

What does the January deductible reset do to this?

January is exactly when the per-order check pays for itself. Plan changes land on the first, patients move between Medicare and Medicare Advantage without telling you, and a quarterly eligibility sweep run in December is already stale by 3 January. A check that runs on the order itself catches those on the day they matter instead of in April when the denials arrive.

Do I need my own hardware to get the cheap version?

Not at 6,000 orders a month. Running on your own machines makes sense when volume is very high and the job is repetitive, and it is a real option now that it costs roughly ninety percent less than cloud for that kind of work. But at your volume the cloud cost is already a rounding error, and the setup effort is better spent on wiring the checks into Brightree properly.

Who owns the exception list?

Give it to one named person — usually the resupply coordinator, with the reimbursement manager reviewing the categories weekly. An exception list nobody owns turns into a queue of 4,000 held orders inside a month, at which point somebody switches the whole thing off and you are back where you started.

The one thing to do this week

Pull last quarter's resupply denial report and sort it by reason code. Count how many of the top three reasons could have been caught before the box left the warehouse. That number, multiplied by your average billed amount, is the honest size of this opportunity for your company — and it takes an afternoon to produce.

Then look at what the exception list does to your phone. Every held order eventually becomes a patient call asking where their supplies are, and those calls arrive in a lump. CallSphere builds AI voice and chat agents that answer those calls and your web chat around the clock, tell the patient the actual status of their order and take the details you need, so cleaning up the back end does not simply move the pressure onto the front desk.