Case Study 10.1 — Medical Debt and the Credit Report: What Changed, What Didn't, and Which Layer It Moved

DataField.Dev

Case Study 10.1 — Medical Debt and the Credit Report: What Changed, What Didn't, and Which Layer It Moved

Type: real, public — regulatory and industry developments Tier: Tier 1 for the statutes, agencies, and the bureaus' publicly announced changes; Tier 2 for program-guideline specifics and for anything with a current effective date. No statistics or enforcement figures are asserted in this case study. Where a figure would ordinarily appear, you are sent to the source instead.

Background: why medical debt is not like other debt

Every collection on a credit report tells a story about a borrower's willingness or ability to repay. That is the premise the whole scoring apparatus rests on, and for most collections it is a reasonable premise. Someone stopped paying a credit card. That is information.

Medical debt breaks the premise in several specific ways, and the ways are worth naming precisely because they are the entire basis for everything that followed.

It is rarely a borrowing decision. Nobody comparison-shops an ambulance. A consumer who chooses to finance a sofa has made a decision about their own balance sheet; a consumer who arrives at an emergency department has not.

The amount is frequently disputed between two institutions, neither of which is the patient. The provider bills, the insurer pays part, the provider bills again, a coordination-of-benefits question arises, and months pass. The patient is a spectator to a negotiation about their own liability.

The consumer often does not know the bill exists. Medical billing generates statements to addresses that may be stale, from entities the patient never chose — an out-of-network radiologist, a pathology group, an anesthesiologist — for services rendered while the patient was unconscious. It is entirely ordinary for the first notice of a medical bill to be a collection notice.

And it is common. The Consumer Financial Protection Bureau has published research on medical debt in consumer credit reporting, including work examining how medical collections compare with other collections and a broader study of medical debt burden in the United States. Those reports are public and free; read them rather than any summary, including this one, and do not carry a number from secondhand sources into a conversation with a borrower.

The issue: a signal that may not be a signal

The question the industry and its regulators spent a decade working on was narrow and technical: does a medical collection predict future credit performance the way a non-medical collection does?

If it does, it belongs on the report and in the model. If it predicts less well — if a consumer with a medical collection and no other derogatory information performs materially better than a consumer with, say, an unpaid credit card in collection — then treating the two identically produces a score that is wrong in a specific direction, and wrong most often for households that experienced an illness.

The CFPB's published research examined precisely that question. Two families of responses followed, and the distinction between them is the reason this case study is in a credit chapter rather than a policy chapter.

Response one: change the model

Scoring model developers can weight medical collections differently. They did.

FICO Score 9, introduced in 2014, differentiated medical collections from non-medical collections and also disregarded collections that had been paid.
VantageScore 3.0 and 4.0 likewise treat medical collections differently from other collections.

This is a model-layer change. It reaches only the files that are scored with those models.

Response two: change the data

The three nationwide consumer reporting agencies can change what appears on the report at all. They did that too, jointly and publicly.

In March 2022, Equifax, Experian, and TransUnion announced a set of changes to medical collection reporting. As announced:

Effective July 1, 2022: paid medical collections were removed from consumer credit reports, and the period before an unpaid medical collection could appear was extended from six months to one year — giving consumers and insurers longer to sort out a disputed bill before it becomes a credit event.
In the first half of 2023: medical collections under \$500 were removed.

This is a data-layer change. It reaches every model, because every model reads the same underlying report.

Response three: change the rule

In January 2025, the CFPB finalized a rule amending Regulation V that would have removed most medical bills from consumer credit reports and restricted creditors' consideration of medical information. The rule was challenged in federal court and its status has been the subject of litigation.

Do not rely on this paragraph. Verify the current state of that rule, of any successor rule, and of your own state's law before you tell a borrower what will or will not appear on their report. This is exactly the category of fact that §7.1 of this book's own standards calls perishable, and it is perishable right now.

What it shows: the layer distinction, and why it matters to you specifically

Here is the analytical point, and it is the reason a loan officer should be able to explain this sequence at a whiteboard.

TWO LAYERS, TWO REACHES

  MODEL LAYER          FICO 9, VantageScore 3.0/4.0 treat medical
                       collections differently
                              │
                              ↓
                       reaches ONLY files scored with those models
                              │
                              ↓
                       mortgage lending has long used older
                       "classic" FICO versions  ──►  LARGELY DOES NOT REACH YOUR FILE

  ─────────────────────────────────────────────────────────────────────

  DATA LAYER           the bureaus remove paid medical collections,
                       delay unpaid ones, and remove those under $500
                              │
                              ↓
                       changes the REPORT ITSELF
                              │
                              ↓
                       every model reads the changed report  ──►  REACHES YOUR FILE

A loan officer who read only the headlines in 2014 concluded that medical collections had stopped mattering to mortgage borrowers. That was wrong, and it stayed wrong for years, because the mortgage models in use were older than the change. A loan officer who read only the headlines in 2022 concluded the same thing and was substantially closer to right — because that change was made to the data rather than to a model.

The transferable rule: when you hear that something "no longer counts," ask which layer moved. A model change reaches the files scored with that model. A data change reaches everything. A guideline change reaches only the program it was written for.

There is a fourth layer, and it is the one you actually work in.

The underwriting layer. How a lender treats a medical collection is a guideline question, entirely separate from how a model scores it. Broadly and as of this writing: conventional automated underwriting frequently does not require collections on a one-unit primary residence to be paid off, and FHA's handbook currently treats aggregate collection balances at or above a threshold specially while excluding medical collections from that calculation. Both of those are Tier 2 — they are current practice as understood at the time of writing, they have been revised before, and you must verify them in the Fannie Mae Selling Guide, the Freddie Mac Seller/Servicer Guide, or HUD Handbook 4000.1 before you tell a borrower what their program requires.

Outcome

Three things are true at once, and holding all three is the professional competence this case study is testing.

One: the reporting environment genuinely improved. Paid medical collections and small unpaid ones are largely off consumer reports as a result of the bureaus' 2022–2023 changes, and unpaid ones surface later than they used to. Households that had an illness are, in the aggregate, less penalized in credit reporting than they were.

Two: medical debt did not stop existing. The bill is still owed. The provider can still collect. Larger unpaid medical collections can still appear after the extended waiting period. A borrower whose report is clean of medical items may still have medical debt, and a borrower's ability to make a mortgage payment is affected by an obligation whether or not a bureau prints it.

Three: the rulemaking layer is unsettled. The CFPB's 2025 rule and the litigation around it mean the answer to "will this appear on my report?" is not stable. Say so.

The lesson

For the file: read the collection section of every report and read the original creditor field. Then find out what the program requires before you tell a borrower to spend a dollar. Paying a medical collection to chase a score is frequently pointless in the older mortgage models; paying one because the program requires it is not optional. Those are different reasons and they lead to different advice.

For the conversation: this is the material §10.5's phone callout exists for. A borrower who has a medical collection has usually already decided it is a moral failure and is bracing for you to confirm it. The accurate, useful, and kind response is the same response: this is the most common item I see, the rules treat it differently for a specific reason, and here is what I need from you.

For your own reliability: everything in this case study has a date attached to it, and half of it will be superseded. The durable skill is not the 2022 announcement or the 2025 rule. It is knowing which layer a change happened at, and knowing that the answer to "has this changed?" is always a lookup, never a memory.

Discussion questions

A borrower has a \$340 medical collection reported eighteen months ago and unpaid. Walk through, in order, the four questions you ask before advising them to do anything — and identify which of the four is answered by the credit report itself.
The chapter distinguishes a model-layer change from a data-layer change. Apply the same analysis to a different development: the announced transition toward newer scoring models in mortgage lending. Which layer is that, whom does it reach, and what would you have to verify before quoting a borrower under it?
Two borrowers have identical files except that one has a \$900 unpaid medical collection and the other has a \$900 unpaid credit card collection. Describe how each would be treated by (a) an older mortgage FICO model, (b) a newer consumer model, and (c) an FHA underwriter. Where you are uncertain, say what you would look up and where.
A loan officer tells a borrower, "Medical collections don't count anymore." Identify every way that sentence can be wrong, and rewrite it as something both accurate and reassuring in under thirty words.
This case study deliberately contains no statistics, though the underlying CFPB research contains many. Argue both sides: what does a reader lose by that choice, and what does the book protect by making it? Then state the rule you would apply in your own borrower communications.
Suppose the rulemaking described above is ultimately vacated, upheld, or replaced — three different futures. Describe how your day-one credit conversation with a borrower would differ under each, and identify what would stay identical in all three. The part that stays identical is the part worth memorizing.