When an AI scribe documents the wrong type of heart failure, it doesn't just affect the note. It affects the diagnosis, the medications, and the billing code — all at once.
HFrEF and HFpEF are not interchangeable terms. They represent two distinct conditions with different pathophysiology, different evidence-based treatments, and different prognoses. AI documentation errors range from dangerous abbreviations to subtle AI scribe hallucinations — but HFrEF vs HFpEF confusion is in a different category because it affects diagnosis selection. Yet busy cardiologists, trusting the output, sign the note without catching it.
This is one of the most consequential documentation errors in cardiology today.
What HFrEF and HFpEF Actually Mean
HFrEF — Heart Failure with Reduced Ejection Fraction
EF below 40%
Previously called systolic heart failure. The left ventricle is not pumping effectively. Evidence-based treatments include ACE inhibitors, ARBs, beta blockers, mineralocorticoid receptor antagonists, SGLT2 inhibitors, and Entresto (sacubitril/valsartan).
HFpEF — Heart Failure with Preserved Ejection Fraction
EF 50% or above
Previously called diastolic heart failure. The ventricle pumps normally but fills poorly. The treatment approach is meaningfully different — focused on symptom management, comorbidity control, and SGLT2 inhibitors, which now have evidence in HFpEF from the EMPEROR-Preserved and DELIVER trials.
HFmrEF — Heart Failure with Mildly Reduced Ejection Fraction
EF 40–49%
A third category recognized in the 2022 AHA/ACC guidelines. AI scribes almost never document this correctly.
The ejection fraction value in the note tells you everything. If an AI scribe writes the wrong heart failure type relative to that number, the entire note is clinically inconsistent.
How AI Scribes Get This Wrong
There are three common patterns:
The scribe defaults to HFrEF
HFrEF is the more historically prominent term. AI scribes trained on large clinical text datasets encounter it more frequently and default to it — even when the physician's dictation mentioned an EF of 55%.
The scribe copies forward
A patient was diagnosed with HFrEF three years ago. Their EF has recovered to 52% — now HFpEF by definition. The AI scribe pulls the prior diagnosis forward and documents HFrEF again. The note contradicts the echo result in the same visit.
The scribe mishears or misprocesses the EF value
The physician says “EF sixty-five percent” and the scribe documents HFrEF. The EF value and the diagnosis are internally inconsistent within the same note.
In all three cases, the note looks complete on the surface. The error only becomes visible when you compare the documented EF value against the documented heart failure type — something easy to miss when you're between patients.
Why This Error Matters Clinically
Entresto is the clearest example.
Entresto (sacubitril/valsartan) is FDA-approved for HFrEF and has strong mortality benefit data from the PARADIGM-HF trial. Its evidence base in HFpEF is weaker. If a note documents HFrEF but the patient actually has HFpEF with an EF of 58%, a medication plan including Entresto is built on a wrong diagnosis.
The contraindication risk compounds it.
Entresto cannot be combined with ACE inhibitors — the combination causes dangerous hypotension. If the wrong heart failure type leads to the wrong medication selection, and that medication interacts with something already on the list, the note now contains a layered error.
Billing follows the documented diagnosis.
HFrEF and HFpEF have different ICD-10 codes. If the documented type doesn't match the clinical picture, that's a billing inconsistency — and a RAC audit target.
What to Check Before Signing
When reviewing any AI-generated cardiology note involving heart failure:
Find the EF value
It should be documented somewhere in the note — either from a recent echo or a prior result referenced in context.
Match it against the documented heart failure type
Check the medication plan against the diagnosis
If the note documents HFpEF but lists Entresto as a new initiation with no prior HFrEF history, that's a flag worth reviewing.
Verify the EF is actually documented
A heart failure note with no EF documented is incomplete regardless of which type is listed. For any HF visit, the EF should be present — either from the current encounter or explicitly referenced from a recent study.
How VerifyChart Catches This
VerifyChart runs an independent analysis of the note before you sign — built on the PDSQI-9 framework, a peer-reviewed clinical quality instrument (JAMIA 2025).
For heart failure documentation specifically, it checks:
- HFrEF/HFpEF contradiction — if the documented EF value and the documented heart failure type are internally inconsistent, it flags it as a critical finding
- Missing EF documentation — any heart failure note without an EF value documented triggers a critical flag
- AFib anticoagulation gaps — if AFib is documented without a rate/rhythm status and anticoagulation decision, it flags the omission
- Entresto + ACE inhibitor conflict — if both are listed in the medication plan without a hold noted, it flags the contraindication
The analysis takes under 60 seconds, works with any AI scribe, and requires no EHR integration.
The Bottom Line
HFrEF and HFpEF look similar in a note. They are not similar in clinical management. An AI scribe that documents the wrong type — or leaves the EF undocumented — creates a note that is internally inconsistent and potentially dangerous to sign.
The physician signing that note is legally responsible for every word in it.
A 60-second independent review before signing is the most efficient way to catch this class of error before it reaches the chart. Try VerifyChart free at verifychart.ai — no credit card, no EHR integration required.
Protect your license. Prove you verified.