You have a range of options, from simple H&P to using every test in the book. Based on the following article, I hope you’ll consider a new diagnostic option for this type of patient: selectively utilize BNP for patients remaining in the intermediate probability range after history, physical exam, and chest X-ray are available.
The Evidence: “N-terminal Pro-B-Type Natriuretic Peptid Testing Improves the Management of Patients with Suspected Acute Heart Failure: Primary Results of the Canadian Prospective Randomized Multicenter IMPROVE-CHF Study” Circulation 2007 Nov 11; 115: 3103-3110
In North America 15 million people live with the diagnosis of CHF with 1.5 million new cases diagnosed each year. Failure to diagnose CHF increases morbidity and can delay hospital discharge. In the US the total costs of CHF are estimated at $27.9 billion. A recent Rational Clinical Exam detailed the poor diagnostic test characteristics of such findings as peripheral edema, rales, or wheezing. Another study questioned whether the chest X-ray was a reliable test for CHF. Accordingly, a 2007 ACEP clinical policy for the diagnosis of CHF offered Level B recommendations supporting the addition of a single BNP or NT-proBNP “to improve the diagnostic accuracy compared to standard clinical judgment” among patients presenting to the ED with acute dyspnea. They offered a BNP an industry-sponsored randomized controlled trial in Switzerland demonstrated several benefits of rapid BNP testing including a reduction in the initial hospitalization rates and total hospital length-of-stay (from 14 to 10 days) with a $2500 savings. Of course, shorter stays and lower costs could have occurred simply because more BNP patients died, but incremental analysis suggested that BNP guidance lowered mortality and lowered costs in 81%, while it increased mortality and lowered cost in just 19%.
The current study occurred in seven Canadian hospitals where BNP or NT-proBNP are not currently available and only excluded those with chronic renal insufficiency (creatinine > 2.8 mg/dL). Patients were randomly assigned to two groups: clinician aware of NT-proBNP result and clinician unaware. The majority of the subjects were elderly and Caucasian. The Gold standard for the diagnosis of CHF represents the Achilles’ heal of this study: two cardiologists used the medical records (discharge summery, echocardiography, labs minus the NT-pro BNP level) and a 60-day telephone follow-up interval to identify those with and without the diagnosis of CHF at the time of enrollment. Though some criticize this as an inadequate Gold standard, none have identified a better alternative so this remains the criterion standard for now.
The most impressive results occurred in the intermediate probability subgroup (219/500 – the majority of enrolled subjects) where knowledge of NT-proBNP reduced ED length of stay from 7.5 hours to 5.4 hours and ED visit expense from $2324 to $1759 ($2005 US). To a lesser extent, these results were obtained in the overall cohort as well. Clinical judgment alone had an overall accuracy of 83% which was augmented to 90% with the addition of NT-proBNP. Finally, at 180-days re-hospitalization rates were 7% lower in the NT-proBNP group.
As a caveat, this was an industry-sponsored trial just as was last year’s Swiss BNP publication. Furthermore, these results lack face validity for some of the reported outcomes. For example, how does knowledge of the BNP today impact 6-month outcomes? Is the cause-effect relationship tied to an enhanced clinician commitment to the diagnosis of CHF with appropriate application of guideline-recommended therapeutic and preventative interventions as a result? Although the results of the current well-conducted randomized controlled trial, taken in conjunction with the 2006 Swiss study, are promising beginnings for the addition of BNP and/or NT-proBNP to the diagnostic armamentarium for CHF these lingering questions and industry-bias concerns should be answered by further research before economical clinicians can be confident that their use is truly cost-effective in improving patient outcomes. Several recent meta-analyses, all pre-dating this RCT, have questioned the optimal cut-point for BNP and NT-proBNP and the current study does not address this question. Those with pre-existing CHF, renal disease, obesity, and the elderly may have different baseline levels of these markers than those without disease.
In conclusion, when used on patients in whom the diagnosis of CHF remains uncertain after standard testing, BNP and NT-proBNP appear to decrease ED and hospital length-of-stay, healthcare costs, and 6-month re-hospitalization rates.
Christopher R. Carpenter, MD, MSc
Chief Clinical Editor, Emergency Physicians Monthly
Faculty, Best Evidence in Emergency Medicine (BEEM)
Assistant Professor, Division of EM, Washington University in St. Louis
So, how does BNP perform in the gray-zone cases? Dyspneic patients who do not clearly have a pulmonary etiology or obviously decompensated congestive heart failure pose a real diagnostic dilemma for us. Unfortunately, BNP just doesn’t perform well in these cases. Although several studies illustrate this limitation, one meta-analysis, Schwam, E., Academic Emergency Medicine, June 2004, is particularly valuable, confirming that BNP was not very helpful unless it was very high or very low. Unfortunately, in intermediate prior probability cases, the ones we need help with, the BNP was not reliably high or low. Thus, gray zone cases usually have gray zone BNPs. In addition, Hohl, C.M., et al, in the Canadian Journal of Emergency Medicine, found the sensitivity and specificity to be 79% and 71%, respectively, in this subset of patients.
The fact is that much of the literature supporting the use of BNP has serious methodological limitations. For instance, many of the studies are manufacturer sponsored. Although manufacturer sponsorship doesn’t automatically make an article and its conclusions worthless, it should raise some serious skepticism with respect to the motivation for performance of the study and, what data they chose to report and how they presented it. Another substantial limitation in much of the BNP literature is the “Gold standard” used. In many of these studies, the “Gold standard” was retrospective chart review by a cardiologist. So, instead of using pulmonary capillary wedge pressures or echocardiogram, they chose to have a cardiologist review the chart after the fact. Such retrospective reviews cannot be considered a standard of any kind. At best, they are educated, personal opinions.
Making things worse, those studies frequently were subject to incorporation bias. The cardiologists were not blinded to the BNP, allowing them to use the BNP to determine if the patient’s etiology for dyspnea was from congestive heart failure or not. It is truly bad methodology to use the test you are “testing” to see how well it functions diagnostically. Although there are many, Steg, P.G., et al., Chest 128 (1):21, July 2005, is an excellent example of this.
Although it probably isn’t practical to measure pulmonary capillary wedge pressures (PCWP) on patients in the ED or to obtain 2-D echos either, the lack of practical, better standards doesn’t excuse the use of weak ones. Furthermore, the data on BNP in ICU patients with respiratory failure shows a poor correlation with elevated PCWPs. Hemodynamic monitoring has more credibility with me than a cardiologist’s retrospective review. (J Am Coll Cardiol. 2005 May 17;45(10):1667-71.; Am Heart J. 2005 Dec;150(6):1213-9).
What is a positive BNP? There is no consistency in the reference ranges set by manufacturers for BNP. However, they tend to be set artificially low. One study by Januzzi, J.L., et al., in the European Heart Journal in 2006, noted that BNP levels should be age-stratified and be set much higher than most manufacturers have recommended. In this study, they utilized the pro-BNP with cutoffs of 450, 900 and 1800 for ages 75 years, respectively. Even with these generous cutoffs, the sensitivity and specificity was 90% and 84%.
Finally, is the NT-pro BNP any better than standard BNP? The answer is no. Why was the pro-BNP developed? Manufacturers sensed the inadequacies of BNP, and thus, began working on the second-generation test. Just as with the multiple versions and of d-dimer assays we’ve seen come and go, the primary reason to develop a new one is due to the inadequacies of the old one. Mueller, in Heart, 2005 and Januzzi, in American Journal of Cardiology, 2005, both found the pro-BNP to function identically to the standard BNP assay.
Thankfully, investigators are attempting to better answer this question, such as the study by IMPROVE-CHF published in Circulation in 2007. And yet, while a methodological improvement, this randomized controlled trial leaves us with more questions. The fact that they report cost-savings and small improvements over clinical judgment alone doesn’t surprise me. The previous studies all make similar claims. They make a compelling argument that the time to diagnosis is less in the BNP-guided group, particularly with gray zone cases, but is it time to the right diagnosis? We don’t really know. The investigators admit that many of their assertions lack statistical significance and they state, “NT-proBNP testing was not statistically superior to clinical judgment alone.” Finally, their gold standard was retrospective cardiology review (blinded to BNP results) and the study was sponsored by Roche Diagnostics.
In the final analysis, BNP may be able to tell you what you already know. However, it can’t reliably tell you what you need to know. The best predictor of CHF is your clinical judgment.
Kevin Klauer, DO
Editor-in-Chief, Emergency Physicians Monthly
Director, Quality and Clinical Education, Emergency Medicine Physicians
Faculty, Center for Emergency Medical Education (CEME)
I think this is an interesting discussion. I think, though, we have to rememeber that we live and work in a murky grey zone most of the time, and for multiple reasons, a little more data sometimes helps. In an institution where Internists admit dyspnea, and Cardiologists admit CHF, it can provide a disposition. It, potentially, becomes part of the lexicon we use when swe speak with the admiting service. I think, ultimately, does it make a huge difference? No, if it looks like a duck and quacks like a duck it probably is, and by the time the result returns, the patient has likely been on Bi-pap for an hour. But in cases of undifferentiated dyspnea, I use it to push me in a direction. If the CXR in normal as is the BNP I am reassured that CHF as well, to some very limited extent, PE is unlikely. If the cxr is negative but the BNP is positive, I sit back and make myself prove there isn’t inappropriate cardiac stretch i.e. PE, early CHF, etc. That is where I think it comes in handy.
I agree that BNPs are only one tool in diagnosing CHF. Clinical exam is, and should always be, our number one diagnostic tool. However, I also feel that sometimes when on the phone with an admitting physician, every piece of information given can help them better “see” what we are looking at in the ER. Also, the multi-diagnosis patient can be sorted out better with more information at our fingertips.