The A–L grid, with the numbers filled in
Every nursing home citation carries a letter from A to L. Every published explanation of that grid shows you the same empty 12-cell matrix and stops there. So here it is with the actual frequencies behind it, counted across 418,479 federal inspection citations — including the two cells that turn out to be almost entirely theoretical.
First, a correction worth making
A lot of what you will read — including answers generated on top of it — says that D, E and F citations indicate actual harm. They do not. D, E and F sit in the “no actual harm” band: an inspector found something that could cause more than minimal harm. Actual harm begins at G.
This is not a technicality. D–F is 92.2% of every citation ever written. If those letters meant a resident was hurt, then essentially every nursing home in America would have hurt someone — which is exactly the wrong conclusion to draw when you are holding your parent's inspection report.
How often each cell is actually used
Rows are severity, columns are scope. The share is out of all 418,479 health-inspection citations in the current federal file.
| Letter | Severity | Scope | Citations | Share |
|---|---|---|---|---|
| A | No actual harm, minimal potential | Isolated | 0 | 0% |
| B | No actual harm, minimal potential | Pattern | 5,077 | 1.213% |
| C | No actual harm, minimal potential | Widespread | 4,463 | 1.066% |
| D | No actual harm, potential for more than minimal | Isolated | 262,910 | 62.825% |
| E | No actual harm, potential for more than minimal | Pattern | 94,108 | 22.488% |
| F | No actual harm, potential for more than minimal | Widespread | 28,834 | 6.89% |
| G | Actual harm | Isolated | 12,946 | 3.094% |
| H | Actual harm | Pattern | 460 | 0.11% |
| I | Actual harm | Widespread | 20 | <0.01% |
| J | Immediate jeopardy | Isolated | 6,844 | 1.635% |
| K | Immediate jeopardy | Pattern | 2,170 | 0.519% |
| L | Immediate jeopardy | Widespread | 647 | 0.155% |
Source: CMS health-citation file, 2026-08-09. Counted by Haven across every row, not a per-facility sample.
Level A has never been cited
Not once in 418,479 citations. An isolated finding with no harm and no real potential for harm is, in practice, not written up at all — if it were that minor, it did not become a citation. The grid is printed as 12 cells; it operates as 11.
Widespread actual harm is nearly a null category
Level I — actual harm affecting residents throughout a facility — appears just 20 times (0.005%). When harm is that broad, inspectors escalate to immediate jeopardy instead, which is why K and L together (2,817) vastly outnumber it.
Why the letter matters more than the count
Citation volume and citation seriousness are close to inverted. The most-cited tags in America are mostly procedural; the tags that actually hurt people are cited far less often. These are the tags with the highest share of harm-or-worse findings.
| Tag | What it covers | Citations | Involved harm (G–L) |
|---|---|---|---|
| F0678 | Provide basic life support, including CPR, prior to the arrival of emergency medical perso… | 771 | 49.3% |
| F0600 | Protect each resident from all types of abuse such as physical, mental, sexual abuse, phys… | 8,625 | 38.1% |
| F0689 | Ensure that a nursing home area is free from accident hazards and provides adequate superv… | 21,289 | 32.8% |
| F0835 | Administer the facility in a manner that enables it to use its resources effectively and e… | 1,217 | 29.1% |
| F0603 | Protect each resident from separation (from other residents, his/her room, or confinement … | 157 | 24.2% |
| F0686 | Provide appropriate pressure ulcer care and prevent new ulcers from developing. | 8,486 | 22.6% |
Compare that with the three most-cited tags overall — F0880 (23,906), F0689 (21,289), F0812 (19,968). Volume tells you what inspectors look at on every visit. The letter tells you what happened.
Common questions
What do the letters A through L mean on a nursing home inspection?
They encode two things at once: how serious the problem was, and how widespread. Severity runs in four bands — no harm with minimal potential (A–C), no harm but potential for more than minimal harm (D–F), actual harm (G–I), and immediate jeopardy (J–L). Scope runs across three columns — isolated, pattern, and widespread. So a D is an isolated problem that could have caused harm, and an L is a widespread situation where serious injury or death has occurred or is imminent.
Does a D, E, or F citation mean a resident was harmed?
No, and this is the most common error in published explanations of the grid. D, E and F all sit in the "no actual harm" band — they mean an inspector found something that COULD cause more than minimal harm, not that it did. Actual harm starts at G. This matters because D–F is the overwhelming majority of all citations: 385,852 of 418,479 (92.2%). If D meant harm, nearly every nursing home in America would have harmed someone.
What is the most common nursing home citation level?
D, by an enormous margin — 262,910 citations, 62.825% of everything in the federal file. An isolated, no-harm-but-possible finding is the default outcome of a nursing home inspection. Seeing D citations on a facility's record tells you almost nothing on its own; it tells you the facility was inspected.
How often are the most serious levels actually cited?
Rarely. Actual harm (G–I) accounts for 13,426 citations and immediate jeopardy (J–L) for 9,661 — together 5.5% of the file. Two cells are effectively unused: level A has never been cited once in 418,479 citations, and level I appears just 20 times. In practice the grid is 11 cells, and widespread actual harm is close to a null category — when a problem is that bad, inspectors reach for immediate jeopardy instead.
Is the letter or the number more important?
The pair. A tag tells you what went wrong, the letter tells you how bad it got, and neither is meaningful alone. F0880 (infection control) is the most-cited tag in the country but is overwhelmingly cited at no-harm levels. Other tags are cited far less often but involve real harm a large share of the time. A count of citations is not a quality signal; a count of G-and-above citations is.
Looking at a specific report?
Every commonly cited code has its own page with national frequency and severity behind it, so you can tell a routine finding from one that hurt someone.