Caregos
Trust & transparency

Methodology & sources

How we source, compute, and maintain everything on Caregos. Transparency is the product: if you can’t see how a number was made, you shouldn’t trust it.

Data sources & cadence

All facility data originates from the Centers for Medicare & Medicaid Services (CMS). We do not generate our own ratings.

DatasetPowersRefresh
Care Compare — Provider Info5-star ratings, beds, ownershipMonthly
Health DeficienciesInspection findings, F-tags, scope/severityMonthly
PenaltiesFines & payment denialsMonthly
Payroll-Based Journal (PBJ)Nurse staffing hours, turnoverQuarterly
Special Focus Facility listSFF & candidate flagsMonthly

Every data block on the site shows the dataset’s “as of” date. When CMS revises historical data, we re-import in full rather than patch.

Derived signals

A few signals are computed by Caregos on top of raw CMS data. Each is labeled and linked to this section wherever it appears.

Local percentile

Ranks a facility’s overall star rating against all CMS-certified nursing homes in the same county.

percentile = share of same-county facilities with a lower or equal overall rating, ties resolved by health-inspection score.

18-month trend

A visual summary of the official overall rating across the last six published quarters. It is not a separate score and never overrides the current CMS rating.

Deficiency severity

Aggregates findings using CMS’s own scope-and-severity grid (A–L). Actual-harm (G+) and immediate-jeopardy (J–L) findings are weighted most heavily and surfaced explicitly.

Note: we never invent thresholds that imply a facility is “safe” or “dangerous.” Derived signals add context to CMS data — they don’t replace your judgment.

Enforcement framing

Enforcement records are reproduced as published by CMS, in neutral language. We state what was cited, when, and at what severity — and stop there.

  • No fear-based or conclusory language (“dangerous,” “worst”).
  • Every record carries its survey date and source.
  • A citation reflects a point in time and may have been corrected since.
  • A correction link appears on every enforcement record.

Editorial team

Real, accountable people maintain the data pipeline, review methodology, and handle corrections.

DK
Dana Koenig
Editor — data sourcing & methodology
RM
Dr. Ruth Mensah, RN
Clinical reviewer — staffing & quality measures
AT
Andre Trevino
Data engineer — CMS ingestion & QA

Corrections policy

If a record is wrong, tell us. We verify against the source dataset, correct within five business days, and log every change with a date and reason on our corrections page.

Report a correction

Contact

Questions about the data or method? Reach the editorial team at editor@nh-data.example. We are an independent project and do not sell facility placement or referral leads.