Copycat Billers: Providers With Nearly Identical Patterns
When two independent providers have >95% cosine similarity in their billing patterns, it raises questions. Are they the same organization billing under multiple NPIs? Part of a coordinated billing scheme? Or just similar practices? We found 246 pairs among our 150 most-flagged providers.
246
Pairs found
100%
Highest similarity
67
Perfect matches
100%
At 99%+ similarity
How Cosine Similarity Works
Each provider's billing is represented as a vector across all HCPCS procedure codes — how much they billed for each code. Cosine similarity measures the angle between two providers' vectors: 1.0 means identical billing distributions (same codes in the same proportions), while 0.0 means completely different billing patterns. We computed pairwise similarity among the 150 most-flagged providers in our watchlist.
<90%
Normal range. Different providers naturally have different billing patterns.
95–99%
Highly similar. Same specialty in same region could explain this, but unusual.
99%+
Near-identical. Suggests shared operations, same management, or coordinated billing.
Billing Clusters: Groups of Identical Providers
When multiple providers all share 100% billing similarity with each other, they form a cluster. These groups are especially notable — 5 clusters of 3 or more providers bill in exactly the same pattern.
Of the 100 provider pairs analyzed, 67 pairs share a perfect 100% cosine similarity — meaning they bill for exactly the same procedure codes in exactly the same proportions. Another 33 pairs are above 99% similarity, making them nearly indistinguishable.
What Perfect Similarity Means
A 100% cosine similarity score means two providers allocate their billing across procedure codes in identical proportions. If Provider A bills 40% code X, 35% code Y, and 25% code Z, then Provider B does exactly the same. This is extremely unlikely to occur by chance between truly independent practices, especially among providers already flagged for other billing anomalies.
The cluster analysis reveals groups of providers that all share identical billing patterns with each other. The largest cluster contains 10 providers — all billing in exactly the same pattern. This could indicate a single organization operating under multiple NPIs, a franchise model with standardized billing, or a coordinated billing operation.
Important Caveat
Same specialty in the same region naturally creates some similarity. A group of home health agencies all billing primarily for personal care services (T1019) will have high similarity simply because they provide the same service. This analysis is most meaningful when high similarity appears between providers that are already flagged for other anomalies, or when the providers are in different geographic regions yet bill identically.
Top 30 Most Similar Provider Pairs
Key Takeaways
- ▸246 pairs of flagged providers share >95% cosine similarity in their billing patterns.
- ▸67 pairs have perfect 100% similarity — identical billing distributions across all procedure codes.
- ▸5 clusters of 3+ providers all share identical billing patterns with each other, suggesting shared operations.
- ▸Same specialty naturally produces some similarity. This analysis is most meaningful for providers already flagged for other anomalies.
Source: HHS T-MSIS Other Services File (2018–2024) · 227M records