Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

BNP for AF - update

Systematic review
Results
Comment

Systematic review and meta-analysis of good studies of diagnostic tests are rare. One on BNP tests for congestive heart failure is important as diagnosis is not easy in primary care because symptoms are unspecific. Diagnosis usually requires expensive echocardiography or radionuclide scan. BNP ruling out heart failure would save money and time.


Systematic review

The review [1] had a thorough search of general and specialist diagnostic test databases. Studies compared any type of BNP assay in asymptomatic patients or those with suspected congestive heart failure with gold standard of echocardiography or radionuclide scan, with or without additional diagnostic criteria, and have information on true and false positive and negative. Cut off was that used by studies.


Results

Nineteen studies on 22 populations with 9,093 patients were available. Studies were generally of good quality, with prospective design, consecutive cohort, and blind test interpretation of test results common. Mean ages ranged from 51 to 79 years, and percentage of men from 35% to 95%. Some studies examined secondary care patients (acute dyspnoea, after myocardial infarction, with an existing diagnosis of heart failure), others primary care referrals, and some were screening studies of patients with risk factors.

The results for ELISA and RIA methods are shown in Table 1 as the proportion of all positive tests that were true positives and the proportion of all negative tests that were true negatives. The ideal test would score 100% for both. The proportion of patients who actually had heart failure by gold standard method varied from 50-60% in the secondary care setting, to about 20% in patients referred from general practice, and was 5% in screening studies. Where prevalence was low, both types of tests reliably ruled out heart failure, so that a negative test meant patients did not have heart failure.



Table 1: Results of BNP tests by ELISA or RIA in different populations with varying proportions of patients with true heart failure defined by gold standard diagnostic methods



ELISA
RIA
Population
Number of subjects
True HF
(%)
True positive/
all positive
(%)
True negative/
all negative
(%)
Number of subjects
True HF
(%)
True positive/
all positive
(%)
True negative/
all negative
(%)
All studies
2963
40
72
92
6130
11
28
98
Secondary care
1839
50
81
89
310
60
82
71
Primary care
1124
22
51
97
1391
17
32
96
Screening
4429
5
17
99



Comment

These tests are not cheap, but the gold standard diagnostic test is much more expensive. In a primary care population of 1,000 people in whom the GP has a clinical suspicion of heart failure, 200 will actually have heart failure and 800 will not. Figure 1 shows how the test will work out if we assume that in this population it picks up 50% of true positives and excludes 96% of true negatives.



Figure 1: Results in a hypothetical primary care population of 1,000 people where GP suspects heart failure





Of the 1,000 patients, the test would mean that 868 would not be sent for confirmatory testing, while 132 would be sent for confirmatory testing. The ratio of about 7 patients not sent for confirmatory testing because of the result of the BNP testing for every one sent for confirmation would imply cost saving if the confirmatory test cost about £100 or more, though it would also have major implications for waiting times.

The concern might be that 100 patients who truly had heart failure would not have confirmatory testing. Presumably these would be less severe cases that might return later to the GP, and would have other tests. More detailed thinking is needed to fully appreciate the possible cost-effectiveness on BNP testing in primary care.

Reference:

  1. M Battaglia et al. Accuracy of B-type natriuretic peptide tests to exclude congestive heart failure. Archives of Internal Medicine 2006 166: 1073-1080.

previous story