There is a large amount of evidence showing the usefulness of natriuretic peptide (NP) testing - BNP and NT-proBNP - in the exclusion of heart failure as part of a heart failure diagnostic pathway. This was emphasized in the updated NICE guidelines for chronic heart failure (CG108)1.
At the time of publishing approximately one third of primary care organisations commissioned the test. This has now increased significantly but research demonstrates that implementation remains sub-optimal2 and it is important to try to understand the reasons for this and how these can be addressed.
Possible reasons include misconceptions about cost as the test is a relatively inexpensive test compared to the cost of an echocardiogram. When not used as an in-surgery device, the assay needs to be commissioned as an add-on to the existing pathology contract. Although it is not traditionally part of the pathology contract once implemented in line with the updated clinical guidance, NICE estimates an overall saving of £3.8 million from the whole pathway following the introduction of this test3.
There is still some concern around cut-off levels with some clinicians concerned the test may miss cases of heart failure and others worried that low cut off levels may open the flood gates to requests for echocardiography and specialist opinions, putting additional pressure on already stretched services.
However, we have addressed a lot of these issues in East Sussex and wider across Sussex through the Heart Network and wider again across the South East Coast via the Enhancing Quality Programme. In East Sussex we have used NT-proBNP since 2004 and have therefore been able to derive some valuable insights into its benefits and best implementation. Some of the following comments refer to our use of that assay locally.
Studies using NT-proBNP to exclude heart failure have shown high levels of sensitivity (88–96%) but relatively low levels of specificity (50–80%)4 with figures at the lower end of these ranges in primary care. This is why the test must only be used as a ‘rule out’ test and not to attempt to diagnose heart failure. The test should only be used if, prior to the availability of NP, the clinician would have referred the patient for an echocardiogram and specialist opinion.
Using the test like a biochemical searchlight in the dark of an ill-defined condition will lead to a significant increase in demand on cardiology services the majority of which will be inappropriate due to the many non-CHF causes of raised NP, most frequently impaired renal function in the mostly elderly population tested. Misunderstanding of this may explain some under use of the test. NP testing should not be used in patients with a history of myocardial infarction as this delays time to scan and advice in a population at particularly high risk of systolic heart failure.
Having decided to use the test, the application needs to be robust for it to be safe and effective. Potential pitfalls include excessive waiting times for phlebotomy in primary care. Also excessive turnaround times from phlebotomy to result. For example ‘batching’ of samples where demand is low and frequent testing has excess costs for biochemistry departments. Then the tests must be acted on quickly to ensure abnormal results have echocardiography and opinion in the times set out in NICE CG1081, particularly with ‘very high’ NP ( >400pg/ml for BNP and >2000pg/ml for NT-proBNP).
To avoid further strain on services we are advising clinicians not to use the test in patients with fast atrial fibrillation which in very many cases results in ‘very high’ NP levels and patients who may need seeing within 2 weeks but who are usually unsuitable for echocardiography until their heart rates are controlled.
The cut off levels raise another issue. In the draft guideline, the NICE Guideline Development Group wanted cut-off levels agreed locally but stakeholder involvement persuaded the group to advise the use of the cut-off levels from the European Society of Cardiology (ESC) guideline of 2009. However the ESC guideline stipulated that the cut-off levels were for ‘untreated’ patients.
In an audit of referrals to the Sussex Downs and Weald Open Access Echo Service in 2009, before NP testing was available, 179 out of 261 patients referred (69%) were on a drug that could lower NP levels at the time of referral. Consequently, only to use the ESC cut-off level could result in testing only 31% of the suspected heart failure patients. Therefore we have adopted lower cut off levels to include treated patients in line with the NICE Quality Standards for Heart failure 20115.
Some clinicians remain sceptical that even if requested appropriately, the tests still miss significant cardiac conditions. We have audited this in East Sussex, by looking at ‘normal’ results, where patients still end up having an echocardiogram at some point. The results remain very encouraging. About 8% have a significant valve abnormality or heart failure. However all those with valve disease had murmurs at presentation and the one patient with left ventricular systolic dysfunction, already known to have cardiomyopathy, was on optimised treatment and should not have been tested in the first place.
In summary I would urge all emerging CCGs to implement the NICE guideline and Quality Standards recommendations with regard to NP testing with the following advice:
- Don’t test post-myocardial infarction (in accordance with the NICE guideline and Quality Standards), in patients with fast atrial fibrillation or with a murmur.
- Do test if the path from request to actioning results is smooth and timely with access to echocardiography and specialist opinion.
- Do work closely with cardiology services and biochemistry to provide clear guidance to users of the test.
2. SNP Diagnostics Test Report. Ipsos MORI on behalf of Roche Diagnostics. 2012
3. National Institute for Health and Clinical Excellence. Costing report - Implementing NICE guidance. 2010
4. Mant J, Doust JA, Roalfe AK et al. Systematic Review and Individual Patient Data
meta-Analysis of Diagnosis of Heart Failure, with Modelling of Implications of
Different Diagnostic strategies in Primary Care. 2009.