A reviewer's-eye read on why dementia recerts get denied more often than the disease severity would suggest.
Adam Gupilan, LVN — Utilization Management
Dementia is the hardest hospice diagnosis to recertify, and the reason is structural, not clinical. Every other terminal diagnosis produces a measurable slope inside a 60-day window. Dementia does not. A patient can be profoundly, unmistakably end-stage and look completely flat across two consecutive periods.
So the reviewer's question shifts. On a CHF recert I am looking for progression. On a dementia recert I am looking for two different things:
Almost every dementia denial I see is a failure of one of those two, not a failure of severity.
Skipped: "FAST 7." "FAST 7, advanced dementia." "Patient at FAST stage 7 and declining."
Carries: The substage, the observation that establishes it, and the date it was reached.
"FAST 7b as of 8/12. Verbal output over the 40-minute visit was a single repeated word ('no'); no phrases, no intelligible response to name. At the 6/4 visit she produced approximately five different intelligible words across the visit (7a)."
Why it matters mechanically: FAST is an ordinal scale that assumes ordered progression. The substages are:
| 7a | Speech reduced to roughly six or fewer intelligible words in a day |
| 7b | Speech reduced to a single intelligible word |
| 7c | Ambulation lost — cannot walk unassisted |
| 7d | Cannot sit up unassisted |
| 7e | Loss of ability to smile |
| 7f | Loss of ability to hold head up |
The hospice LCD threshold is 7a or beyond. But "or beyond" means reached in sequence. A patient who lost ambulation because of a hip fracture, a CVA, or a fall — while still speaking in sentences — is not FAST 7c. The scale was never validated for out-of-order loss, and a reviewer who sees "FAST 7c, nonambulatory" next to a nursing note describing conversational speech reads the score as unsupported and moves to the rest of the record with a lowered prior on everything in it.
The single most common fixable error: claiming a substage from the deepest deficit rather than from the sequence. If the patient's losses are out of order, say so and argue eligibility on the secondary conditions instead. That is an honest record and it reviews better than a score that doesn't hold.
Skipped: "Nonverbal." "Unable to communicate." "No longer speaks."
Carries: What was actually observed, and over what window.
"Two intelligible words ('water,' 'no') elicited across a 45-minute visit; no spontaneous speech. Same finding at the prior two visits."
Why it matters: nonverbal is a conclusion, and it collapses three different clinical states — a patient with expressive aphasia from a stroke, a patient who is oversedated, and a patient at FAST 7b — into one word. Only one of those three supports the score. FAST 7a is explicitly a per-day count, not a per-visit count, so a visit-level observation should be framed as consistent with the daily pattern, ideally corroborated by facility or caregiver report.
Reaching FAST 7a is the entry criterion. The LCD also asks for at least one specified condition in the past 12 months:
Skipped: "History of UTIs." "Frequent infections." "Poor PO intake."
Carries: The event, dated, with what was done about it.
"Aspiration pneumonia 7/19 — treated with oral levofloxacin at facility, hospitalization declined per POLST. Second episode of fever to 101.4 on 8/26, twelve days after antibiotic completion, no source identified."
Why it matters: these are discrete, checkable events. A reviewer can act on a dated event and cannot act on a pattern description. Note the specificity traps:
Skipped: "Weight 118 lbs, down from 126." with no other context.
Carries: Weight with the method, plus a non-weight corroborator.
"118 lbs on chair scale 8/28; 126 lbs on 5/30, also chair scale. MAC 21.8 → 20.4 cm over the same interval. Diet advanced from mechanical soft to puree 7/2; nectar-thick liquids initiated 8/4 after two witnessed coughing episodes during meals."
Why it matters: in advanced dementia, serial weights are frequently taken by different methods — standing scale, chair scale, bed scale, or "per facility record" — and switching methods produces swings of several pounds that have nothing to do with the patient. Reviewers who see an implausible weight trajectory discount the whole series. Naming the method costs one word and preserves the evidence.
The intake criterion in the LCD is commonly operationalized as roughly 10% body weight loss over six months, or albumin below 2.5 g/dL. Two cautions on that:
Refusal versus inability. These are charted interchangeably and mean opposite things to a reviewer. A patient who declines food may be exercising preference. A patient who pockets, coughs, or cannot initiate a swallow is losing a brainstem-level function. Write which one it is.
At the third benefit period and every one after, the face-to-face encounter must occur within the 30 days preceding the period, and the physician or NP who performed it must compose a narrative explaining how the clinical findings support a prognosis of six months or less.
Skipped: A narrative that restates the diagnosis. "Patient has end-stage dementia and continues to be appropriate for hospice."
Carries: A narrative that names findings from the encounter and connects them.
"On examination 8/29 she was nonambulatory, produced one intelligible word, and required full feeding assistance with visible coughing on puree. Since the June encounter she has had one aspiration pneumonia and an 8 lb weight loss. Given the pace of these events I do not expect survival beyond six months."
Why it matters: the narrative is a physician attestation, not a form field. A narrative that could have been written without seeing the patient is the single most common reason an otherwise well-documented recert gets sent back — and it is the one element the nursing team cannot fix after the fact.
A dementia recert that contains these will clear review in most cases:
In most diagnoses, the recert argument is the slope. In dementia there often isn't one — so the argument becomes the events, and the FAST score is only there to prove you have the right patient.
A FAST stage asserted without its sequence isn't a weak argument. It's an unusable one.
Written from the utilization management side. Nothing attached to this — if it's useful, take it.
This series is written one diagnosis at a time, and the useful ones to write next are the ones people are actually losing. If there's a recertification or authorization that keeps coming back on your charts and it isn't covered above, name it and I'll write it up.
No cost, no pitch attached, and nothing patient-specific — the diagnosis and the setting is all I need. If it's a common one, it goes into the series so everyone gets it.
Name a diagnosis →I do utilization review on the payer side, so this is written from the chair the note actually lands in. I also build software for this, at getclinicalflow.com — but nothing above depends on it and there's no obligation attached to asking.
Adam Gupilan, LVN — utilization management. I review the prior authorizations and recertifications that providers submit, which is where everything above comes from.
Nothing here is attached to a product. If it's useful in your charts, use it. More at the full series, or getclinicalflow.com.
Written from the reviewer's side and offered as general documentation guidance. It is not clinical, legal, or billing advice, and it does not reproduce any proprietary criteria text. Eligibility determinations are made by the certifying physician and the reviewing plan.