How the flags work

Every lead is generated from published federal records — currently the Centers for Medicare & Medicaid Services hospice enrollment and ownership files, refreshed from the government's own data feed. Addresses are normalised so that "123 Main Street, Suite 4" and "123 MAIN ST STE 4" group together, then each pattern below is applied to the whole file.

Shared street address

Several separately enrolled providers give the exact same street address. Severity rises with the count: two or three is worth a look, fifteen at one door is extraordinary. Shared medical office buildings are a legitimate explanation — the point is to check.

PO box or private mailbox

The enrollment address matches a post office box, PMB or mailbox-store pattern rather than a place where care is coordinated.

Residential unit

The address carries an apartment, unit or trailer designation. Small agencies legitimately run from homes, so this is a low-severity nudge, not a red alert.

Owner cluster

An owner or managing party listed on this provider also appears on many other hospice enrollments. Chains are lawful; rapid accumulation of licences by one party is a documented fraud structure.

New-entity cluster

Two or more providers incorporated in the last four years share a single address — the churn pattern seen in recent hospice fraud waves.

Suite stacking

One building hosts providers registered across three or more different suite or unit numbers.

What a flag is not

A flag is a statistical anomaly in a public file. It is not an allegation, a finding, or evidence that anyone has done anything wrong. Most flagged providers will have ordinary explanations. Your job as an investigator is to establish which ones do not — with documents, records and observation — before anything is reported.

Community notes

Anyone signed in can add a note to a flagged provider saying whether it looks like real fraud, has an innocent explanation, or cannot be judged yet — with their reasoning and links. Others rate those notes helpful or not helpful.

Ranking is not a popularity vote. We group raters by how they usually vote: people who tend to back fraud verdicts sit on one side, people who tend to back innocent explanations on the other. A note rises to the top only when both groups rate it helpful, so the leading note is one that convinces people who normally disagree. Until enough people from both sides have rated, notes are shown newest first and marked "Needs more ratings". Notes most readers rate unhelpful collapse behind a click, and anyone can report a note for an administrator to review.

How claiming works