Today I review, link to, and Emergency Medicine Cases‘ Ep 197 Acute Heart Failure Risk Stratification and Disposition.*
*Helman, A. Episode 197 Acute Heart Failure Risk Stratification and Disposition. Emergency Medicine Cases. August, 2024. https://emergencymedicinecases.com/acute-heart-failure-risk-stratification-disposition. Accessed August 23, 2024.
We’d like to think that we’re getting better at identifying acute heart failure and making good disposition decisions. But the facts are, ED physicians are only about 80% accurate diagnosing acute heart failure, and 30-day mortality outcomes and readmission rates for acute heart failure have been about the same in North America for the last few decades – plus we often send home patients who should be admitted and admit patients who should be sent home. This is because there are many challenges for EM docs when it comes to acute heart failure. First, the diagnosis can be tough, and often is only made once the patient is admitted. There is no one clinical feature that is a slam dunk for the diagnosis, and even combinations of features are not great. There are many flavours of heart failure – diastolic, systolic, right-sided, left-sided, SCAPE, cardiogenic shock – just too complicate things further. Even if we’ve made an accurate diagnosis, the decision of whether to admit or send home can be challenging. We have risk stratification decision scales to help, but are they good enough? Are we using them appropriately? In this podcast, with the help of Dr. Doug Lee, cardiologist and researcher at Sunnybrook Hospital in Toronto and Dr. Clare Atzema, our go to EM Atrial fibrillation expert, an EM doc and researcher also from Sunnybrook Hospital, with a special appearance by Ian Chernoff on the role of PoCUS in the diagnosis and risk stratification of acute heart failure, we’ll dig into how to improve our diagnostic accuracy of acute heart failure in the ED and how to improve our disposition decision making so that just the right number of people are admitted and just the right number of people of sent home safely…
Accuracy of initial evaluation findings in the diagnosis of acute heart failure
Value of NT-pro-BNP in risk stratification of acute heart failure remains controversial
Based on our in depth review of the world’s literature in 2018 in this Journal Jam podcast, and as detailed on First10EM, BNP has limited, if any, value in risk stratification of acute heart failure in the ED. However, a subsequent study and guidelines suggest that NT-pro-BNP is highly accurate at the extremes (NT-pro-BMP <300 highly unlikely acute heart failure and NT-pro-BNP ≥5,000 highly likely acute heart failure). There remains an argument to be made that in patients who obviously do not have acute heart failure clinically or obviously do have heart failure clinically, a BNP is not going to shift your diagnostic certainty significantly, and for the intermediate cases where BNP would be most valuable, it is seldom discerning, and may be misleading.
Cardiology guidelines from Europe and Canada seem to have settled on NT-pro-BNP <300pg/ml as “rule-out” threshold, while implementing an age-adjusted “rule-in” threshold:
- Rule out – <300
- Rule in (“consider admission”)
- <50y – >450
- 50-75 – >900
- 75+ – >1800
- Rule in (“admit, close monitoring”)
- ≥5,000
Source: Practical algorithms for early diagnosis of heart failure and heart stress using NT-proBNP: A clinical consensus statement from the Heart Failure Association of the ESC. Eur J Heart Fail. 2023 Nov;25(11):1891-1898.
It is important to realize that BNP may be elevated by advanced age, renal dysfunction, ACS, ARDS, lung disease with right heart failure, pulmonary embolism, high output states and atrial fibrillation, and lowered by obesity, cardiac tamponade and pericardial constriction.
Importance of co-diagnoses and underlying causes and triggers of acute heart failure in risk stratification
The most common cause of acute heart failure decompensation is medication/dietary non-compliance on a background of known heart failure (up to 20% in North America). Despite our best efforts to identify a precipitant, about 30% may have no precipitants identified.
Important co-diagnoses and/or underlying triggers/causes to identify in the ED which help inform disposition decisions include:
- Cardiac
- ACS
- Dysrhthmias *atrial fibrillation with a rate that is relatively well controlled, specifically a HR < 110 can still trigger heart failure; BBB, LVH with strain and paced rhythm portend higher risk (see EHMERG score below)
- Acute myocarditis
- Endocarditis
- Mechanical cause (VSD, acute MR, cardiac tamponade)
- Pulmonary
- COPD exacerbation or asthma
- Pulmonary embolism
- Pneumonia
- Other
- Hypertensive emergency
- Thyroid dysfunction
- Anemia
- Cardio-renal syndrome
- Acute aortic syndromes
- NSAID use
- Active cancer (high risk feature included in EHMERG score – see below)
Importance of mildly elevated troponin in informing disposition decisions in acute heart failure
About 75% of acute heart failure patients will have a high sensitivy troponin (hsTnT) value at or above the 99th percentile reference limit. While a slightly elevated hscTnT has little prognostic value in the acute heart failure patient, a hscTnT of ≥35-45 ng/L portends a significant increase in 30 day mortality in patients with acute heart failure, and this should help guide disposition decisions.
Outcomes and acceptable event rates of discharged patients diagnosed with acute heart failure
In the U.S., among ~1 million annual ED visits for acute heart failure, 80%–90% are hospitalized. In Canada, 40-60%. About 10% of our discharged patients will return and be admitted within 2 weeks of discharge. The 30-day mortality rate in Canada is ~4% and 23% at 1 year.
Our experts suggest that a reasonable acceptable event rate of discharged patients is a predicted risk of 0.5% or less within 7 days or 1% 30-day mortality. For repeat ED visit rates, bounceback rates from the ED for cardiovascular causes should be 15 to 20 or less, however there is little data to back these numbers.
A recent study proposed acceptable event rates in discharged ED AHF patients:
Acute heart failure validated risk stratification tools
While risk scores may help support decision making (and in certain cases are shown to be more accurate than physician judgement of mortality – e.g. in OHFRS study), they should not be used in isolation. This is the Level B recommendation of the American College of Emergency Physicians (ACEP) – that physicians should not rely on current risk stratification tools in isolation to determine disposition for these patients. There are 3 validated ED-specific acute heart failure risk stratification tools, the Ottawa Heart Failure Risk Score, the Emergency Heart Failure Mortality Risk Grade score and the MEESSI score. The MEESSI score was only briefly included in our discussion as it is exceedingly complicated to calculate (as it includes a separate calculation of the Barthel Index) and includes vague factors such as ‘low output symptoms’ and thus impractical. Of note, however, it is the best predictor of 30 day mortality of the 3 scores.
Ottawa Heart Failure Risk Score (OHFRS)
*Not intended to be used to determine disposition until after ED intervention (in the study, OHFRS was assessed 2-8 hours after initial ED treatment)
Exclusion criteria
- Resting O₂ sat <85% on room air on normal home O₂ for >20 minutes
- Heart rate >120 on arrival
- Systolic BP <85 mmHg on arrival
- Confusion, disorientation, or dementia
- Ischemic chest pain requiring nitrates on arrival
- ST segment elevation on EKG
- Death expected within weeks from chronic illness
- Nursing home or chronic care facility resident
- On chronic hemodialysis
fig5 when server ready
*Can be used without NT-proBNP without sacrificing accuracy significantly – including BNP improves sensitivity while worsening specificity
Outcome
Serious Adverse Events = 30-day all cause mortality or within 14 days any of the following: hospital admission, NIPPV/intubation, MI, Major procedure (CABG, PCI, cardiac surgery, hemodialysis)
Scoring
Cut-off of >1 or >2 to recommend admission for monitoring/further treatment (>3 was associated with sensitivities for serious adverse events markedly worse than current practice)
- In validation study, “current practice” was as follows:
- With BNP – Sensitivity 69.8%, Specificity 41.1% – Admission rate 60.8%
- Without BNP – Sensitivity 71.8%, Specificity 45.5% – Admission rate 57.2%
- >1 – increases sensitivity for serious adverse events but increases admission rates
- With BNP – Sensitivity 95.8%, Specificity 13.6% – Admission rate 88%
- Without BNP – Sensitivity 91.8%, Specificity 24.9% – Admission rate 77.6%
- >2 – similar sensitivity to current practice with reduction in admission rates
- With BNP – Sensitivity 79.8%, Specificity 40.5% – Admission rate 63%
- Without BNP – Sensitivity 71.2%, Specificity 55.9% – Admission rate 48.3%
*Green – >5% better than current practice, Red – >5% worse than current practice
Criticisms of Ottawa Heart Failure Score
- May have some selection bias due to convenience sampling
- Explicitly excludes patients who are “too ill” via its many exclusion criteria
- Due to method of scoring, a patient with a history of TIA will be given the same weight as a patient with SpO2 <90%
- Similar to EHMRG, may not be as applicable to other settings due to Canada’s relatively low rate of AHF admissions
- The patients studied using the Ottawa Heart Failure Risk score were healthier than heart failure patients than we see in community practice, since there was approximately 3.7% mortality at 30 days in the validation study for this score, compared to 7% in the COACH trial
- Urea is not routinely obtained for heart failure patients in many EDs
- There is a question of whether the lowest risk patients are truly low risk. Using the Ottawa HF risk score, the lowest total score is 0, but this confers a 2.8% risk of adverse events at 14 days. Is this sufficiently low risk to warrant sending a patient home directly from the ED?
Emergency Heart Failure Mortality Risk Grade score (EHMRG)
EHMRG estimates 7-day mortality of acute heart failure patients in the ED (another version of same model estimates 30-day mortality) to aid in disposition decisions.
Study design: multicenter, prospective validation study of patients with acute heart failure at 9 hospitals of previously derived EHMERG score
Definition of acute heart failure used in the study: Framingham Criteria and pragmatic independent final discharge diagnosis
Exclusion criteria
- Dialysis-dependent patients
- DNR patients
- Palliative patients
- BNP <100 or NTproBNP <300
fig6 when server ready.
