---
title: Atlas Commons research objects — the national distribution of \"Patients who got better at taking their drugs correctly\"
license: CC0-1.0
source: CMS Home Health Care Compare
---

# Atlas Commons research objects — the national distribution of "Patients who got better at taking their drugs correctly"

The national distribution of "Patients who got better at taking their drugs correctly" across 7,937 U.S. home-health agencies, with each figure's status labeled: each per-agency rate is a CMS-measured OASIS outcome; the cross-agency distribution statistics (median 86.3%, p10 55.4% to p90 97.97%) are an in-house aggregation over the public panel, which CMS does not publish; the boundary is context — the national picture, not a per-agency verdict or a personal prediction. Source: CMS Home Health Care Compare. CC0.

## Claims

### "Patients who got better at taking their drugs correctly" is a CMS-measured home-health outcome, reported by 7,937 agencies

**RECORDED.** Patients who got better at taking their drugs correctly is a measured outcome rate (from OASIS patient assessments at admission and discharge) that 7,937 U.S. home-health agencies report to CMS. It captures how often patients actually manage medications — an outcome (a change in the patient), not an input or a rating.

Source: CMS Home Health Care Compare — https://www.medicare.gov/care-compare/
Falsifier: A RECORD of measured/reported events (occurred=true) from CMS Home Health Care Compare; superseded when CMS Home Health Care Compare publishes its next release.

### Nationally, the median agency's "Patients who got better at taking their drugs correctly" rate is 86.3% (middle half 75.13–92.22%, p10 55.4% to p90 97.97%)

**DERIVED.** Across the 7,937 reporting agencies: median 86.3%, mean 80.99%, range from p10 55.4% to p90 97.97%. The per-agency rates are CMS-RECORDED; these DISTRIBUTION statistics are an in-house aggregation over the public panel — CMS does not publish the cross-agency distribution. They show where any given agency falls on the national curve.

Source: Care Atlas — national aggregation over the CMS Home Health Care Compare panel — https://www.medicare.gov/care-compare/
Falsifier: occurred=FALSE (a calculation, not an event) and adjudged=FALSE (a transparent in-house formula, not an authority's verdict — no external adjudicator). Reproduce it from its named inputs; falsified if the formula or inputs are wrong. Lower warrant than the adjudged figures above.

### What this distribution is — and is NOT (the epistemic boundary)

**INTERPRETATION.** This is the NATIONAL picture of one outcome across all agencies — useful for seeing where a provider falls on the curve, and for understanding what "good" looks like. It is NOT a rating of any single agency (see that agency's page), NOT a prediction of how YOU or your relative will do (your clinician knows your case), and NOT a substitute for CMS Care Compare. The spread is real: choosing well moves a patient across a wide distribution of measured recovery.

Source: Care Atlas framing (over the CMS outcome panel) — https://care-atlas.ryanjhunter.workers.dev/
Falsifier: occurred=FALSE (a framing, not an event) and adjudged=FALSE (the atlas's reading, not a CMS verdict) — the LOWEST-warrant claim here. The measure (recorded) and the distribution stats (derived) above are grounded; this is the only interpretation, and the four-status says so.

---

Every claim above carries a four-status warrant (RECORDED / MODELED / DERIVED / INTERPRETATION) and a named source you can re-fetch — this is a research object, not a flat fact. Structured: `https://care-atlas.ryanjhunter.workers.dev/care/outcomes/medication-self.jsonld`

## Data

### National distribution of "Patients who got better at taking their drugs correctly"

24-bin histogram (0%–100%): ▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▃▃▅▆█▆▆

| Statistic | Value | Warrant |
| --- | --- | --- |
| n (agencies) | 7,937 | RECORDED |
| median | 86.3% | DERIVED |
| mean | 80.99% | DERIVED |
| IQR (p25–p75) | 75.13%–92.22% | DERIVED |
| p10–p90 | 55.4%–97.97% | DERIVED |
| min / max | 0% / 100% | DERIVED |