Skip to content
View commentaries on this research

Please note that this summary was posted more than 5 years ago. More recent research findings may have been published.

This is a plain English summary of an original research article. The views expressed are those of the author(s) and reviewer(s) at the time of publication.

The cut-off level for the blood test NTproBNP appears to provide the best balance of detecting true cases while excluding false positives when lowered to 125 pg/ml.

The trial supported by the NIHR included a sample of people presenting to their GP with suspected heart failure. It aimed to see which method was best for identifying those who needed referral: the blood test alone; clinical decision rules based on clinical symptoms; or the combination of both.

The blood test alone at the 125 pg/ml cut-off correctly identified 94% of people with heart failure but led to 50% of people who did not have heart failure being referred for further investigation. This threshold is lower than the level currently recommended by NICE (≥400 pg/ml), which could miss up to one in five people.

Clinical decision rules used alone or in combination with the blood test did not improve detection rates.

However, the lower cut-off would also mean that one in two people without heart failure would be referred for unnecessary expensive tests. A cost-effectiveness analysis is being prepared, which will enable decision-makers to balance the effectiveness of the different approaches.

Why was this study needed?

Heart failure is when the heart is not pumping blood around the body as well as it should. Around half a million people in the UK are living with heart failure. It is a serious and long-term condition, but can be managed to help people to control their symptoms.

Timely and accurate diagnose is important. People often present to their GP with tiredness, breathlessness or swollen legs. However, symptoms can develop gradually and can be confused with other conditions.

Clinical decision rules are a way of assessing how likely the diagnosis is based on the presence of characteristic features, like previous heart attack. They aim to help GPs make consistent and evidence-based decisions.

This study looked at the best way of identifying heart failure: using only the rule, combining the rule with a blood test, or doing the blood test only.

What did this study do?

The REFER study (Primary care REFerral for EchocaRdiogram) included 304 adults aged over 55 years who were recruited to the study when they presented with suspected heart failure at 28 UK GP practices. The final number of participants was lower than planned, but is still larger than previous, similar studies.

Clinical assessment, blood tests, electrocardiograph (ECG) and echocardiograms were carried out within seven days. Three heart specialists independently assessed each case. They were given patient data in three instalments: clinical assessment, ECG and echocardiogram results; addition of clinical decision rule data; and finally blood levels of NTproBNP (N-Terminal pro-B type natriuretic peptide).

The accuracy of heart failure diagnosis using the decision rule, NTproBNP level (at different cut-offs), or combination of the two was compared against a reference standard. In this case, the standard was a diagnosis confirmed by consensus of three experts using agreed standard definitions.

What did it find?

  • 104 out of 304 participants (34.2%) were confirmed as having heart failure.
  • The most accurate overall diagnosis came from using NTproBNP level alone at a low cut-off of ≥125 pg/ml. This had sensitivity of 94.2% (95% confidence interval [CI] 87.9 to 97.9), meaning that the majority of those with heart failure would be picked up by this test. Specificity was much lower at 49.0% (95% CI 41.9 to 56.1) meaning that about half of those without heart failure would be referred for an echocardiogram.
  • Combining the clinical decision rule with the 125 pg/ml NTproBNP cut-off did not improve accuracy. It missed a few more people with heart failure (sensitivity 90.4%, 95% CI 83.0 to 95.3) while not improving the false positive rate, (specificity 45.5%, 95% CI 38.5 to 52.7).
  • Using NTproBNP alone at a higher cut-off ≥400 pg/ml led to fewer false positives (specificity 91.5%, 95% CI 86.7 to 95.0) meaning less people would be referred for heart failure unnecessarily, but sensitivity was low at 76.9% (95% CI 67.6 to 84.6) meaning that the blood test could not be used alone.
  • Combining the clinical decision rule with the higher NTproBNP cut-off of ≥400 pg/ml had slightly better sensitivity (78.8%, 95% CI 69.7 to 86.2) but poorer specificity (63.5%, 95% CI 56.4 to 70.2).
  • Using the clinical decision rule alone gave very poor sensitivity of 44.2% (95% CI 34.5 to 54.3) and specificity of 64.0% (56.9 to 70.6).

What does current guidance say on this issue?

NICE guidelines (2010) recommend that people with suspected heart failure who’ve had a previous heart attack are referred by their GP for echocardiogram and specialist assessment within two weeks.

The advice is that those with suspected heart failure who have not had a heart attack should have a blood test to measure BNP or NTproBNP. The criteria for urgent echocardiogram within two weeks is NTproBNP >2000 pg/ml. An NTproBNP 400-2000 pg/ml is the threshold for referral within six weeks.

What are the implications?

This study suggests that the NTproBNP test at the cut-off of ≥125 pg/ml may be an appropriate threshold for identifying those in general practice who should be referred for an echocardiogram.  However this comes with some down-sides too. More people will be referred for testing unnecessarily and this can cause extra worry for patients.

Lowering the cut-off would mean that more people without heart failure are referred for further tests, impacting on service use and costs too.

Using the higher threshold as NICE recommends (NTproBNP cut-off of ≥400 pg/ml) could mean that up to one in five people with heart failure face a delay in diagnosis.

The researchers say that results of this study will be used in a cost effectiveness analysis, which will help decision-makers to inform how and whether to implement changes to practice.


Citation and Funding

Taylor CJ, Roalfe AK, Iles R, et al; REFER investigators. Primary care REFerral for EchocaRdiogram (REFER) in heart failure: a diagnostic accuracy  study. Br J Gen Pract. 2017;67(655):e94-102.

This project was funded by the Efficacy and Mechanism Evaluation (EME) Programme (09/160/13), a Medical Research Council and National Institute for Health Research (NIHR) partnership.



BHF.  Heart failure. London: British Heart Foundation.

NICE. Chronic heart failure in adults: management. CG108. London: National Institute for Health and Care Excellence; 2010.

NHS Choices.  Heart failure. London: Department of Health; 2016.

Produced by the University of Southampton and Bazian on behalf of NIHR through the NIHR Dissemination Centre


NIHR Evidence is covered by the creative commons, CC-BY licence. Written content may be freely reproduced provided that suitable acknowledgement is made. Note, this license excludes comments made by third parties, audiovisual content, and linked content on other websites.

  • Share via:
  • Print article


The clinical decision rule used in this study was developed by a previous systematic review of general practice studies. The “MICE” clinical decision rules indicated immediate referral if the person had any one of the below:
  • Past history of heart attack.
  • Crackles at the base of the lungs when listening with a stethoscope.
  • A male with ankle swelling.
If the person did not meet these criteria they had an NTproBNP blood test, with referral as follows:
  • Level 190-520 pg/ml in a female with ankle swelling.
  • Level 390-660 pg/ml in a male without ankle swelling.
  • Level 620-1060 pg/ml in a female without ankle swelling.
Back to top