Per-source quality-model sub-scores
NPPES (NPI registry)
94.2%
Completeness 99.1 · Consistency n/a
Match rate n/a · Timeliness 89.3
Model observation Jul 13
OIG LEIE exclusions
66.0%
Completeness 99.8 · Consistency n/a
Match rate 98.1 · Timeliness 0.0
Model observation May 26
CMS PECOS PPEF
88.0%
Completeness 97.4 · Consistency n/a
Match rate n/a · Timeliness 78.6
Model observation Jun 28
CMS Open Payments
83.8%
Completeness 96.2 · Consistency n/a
Match rate n/a · Timeliness 71.4
Model observation Apr 15
CMS Care Compare
92.2%
Completeness 98.7 · Consistency n/a
Match rate n/a · Timeliness 85.7
Model observation May 25
Methodology
How we score
The composite is a weighted mean of four sub-metrics — completeness, consistency, match rate, and timeliness. Inputs can be stored metric rows or documented fallbacks; observation dates do not prove source freshness.
Per-source completeness
For each source, the median proportion of public-displayable required fields that are non-null and non-empty in the documented measurement. The date below is a stored row-count observation or fallback-baseline date, not necessarily an upstream release or a fresh completeness scan. See the methodology for sampling and fallback behavior.
| Source family | Measurement / observation date | Row count | Median field completeness |
|---|---|---|---|
| NPPES (NPI registry) | 2026-07-13 | 9.0M | 99.1% |
| OIG LEIE exclusions | 2026-05-26 | 68K | 99.8% |
| CMS PECOS PPEF | 2026-06-28 | 3.0M | 97.4% |
| CMS Open Payments | 2026-04-15 | 13.4M | 96.2% |
| SAM.gov entities | 2026-05-20 | 1.2M | 95.3% |
| HRSA HPSA shortage areas | 2026-07-01 | 79K | 99.5% |
| CMS Care Compare | 2026-05-25 | 42K | 98.7% |
| BLS OEWS wages | 2026-05-01 | 4K | 99.8% |
| BEA Regional | 2026-04-30 | 3K | 99.9% |
Inter-source consistency
Where two independent federal feeds describe the same NPI, do they agree? Each rate is the share of NPIs whose normalized value matches across both sources. The denominator is the set of NPIs present in both snapshots with a non-null value on each side.
No cross-source consistency rate is published at this snapshot. The originally-specified NPPES ↔ PECOS checks join enrollment-specialty and enrollment-status fields that are not present in the provider schema, so no reproducible figure exists — and we do not publish a number we cannot recompute. The composite below is renormalized over the families that are published. See the methodology for the full correction note.
Temporal update lag
The model reports lag values from source-release and ingestion timestamps when at least two valid rows are available in the trailing 90 days. Otherwise it displays the documented fallback baseline. These values are model inputs, not a current SLA or proof that every source has a recent ingest.
| Source family | Median lag | p95 lag | Window |
|---|---|---|---|
| NPPES (NPI registry) | 18 h | 64 h | 90 d |
| OIG LEIE exclusions | 306 h | 582 h | 90 d |
| CMS PECOS PPEF | 36 h | 96 h | 90 d |
| CMS Open Payments | 48 h | 120 h | 90 d |
| SAM.gov entities | 24 h | 72 h | 90 d |
| HRSA HPSA shortage areas | 24 h | 72 h | 90 d |
| CMS Care Compare | 24 h | 96 h | 90 d |
| BLS OEWS wages | 72 h | 168 h | 90 d |
| BEA Regional | 72 h | 168 h | 90 d |
OIG LEIE byte-level match rate
The quality-model block reports matching stored source and upstream SHA-256 values in 51 of the last 52 weekly snapshots — a 98.1% byte-level agreement. When live comparison rows are unavailable, this block uses its documented fallback baseline; it is not evidence of a current-file comparison.
98.1%
51 / 52 weekly snapshots · Mondays
Methodology
The composite is a weighted average — 35% completeness, 30% consistency, 20% timeliness, 15% OIG LEIE byte match — pinned at methodology version quality/v1. A family with no reproducible value (currently consistency) is dropped and the remaining weights are renormalized, so the headline never absorbs a missing family as a zero. The methodology identifies formulas, denominators, and fallback inputs; readers should not treat the page-revalidation time as a source date.
Machine-readable twins: /quality.json (schema.org Dataset) · /api/v1/quality · Methodology pointer: https://fonteum.com/quality/methodology
Add federal-data citations to your agent in 60 seconds.
Connect the hosted MCP server to your agent, or call the API with an issued bearer credential. See the developer docs for the current public contract.