Cost effectiveness of BNP testing
Bandolier
149 examined a systematic review of the accuracy of BNP testing. It showed that
in secondary care situations where about 50% of patients had heart failure
according to gold standard diagnostic procedures, the test had reasonable
accuracy, though that depended somewhat on the cut-off values set in individual
studies. That leaves open the question of whether it is not only useful
clinically, but whether also cost effective. A new randomised trial suggests
that it is [1].
Trial
This
was conducted on 452 patients presenting with acute dyspnoea to the emergency
department of a Swiss university hospital. Trauma patients, and those with
severe renal disease or cardiogenic shock were excluded. Randomisation was to a
diagnostic strategy that included measurement of BNP, or a conventional
diagnostic strategy.
Two
BNP cut-offs were used to separate heart failure from other causes of acute
dyspnoea, less than 100 ng/L to rule out heart failure, and >500 ng/L to
rule in heart failure. In this case rapid therapy with diuretics,
nitroglycerin, and ACE inhibitors was recommended. Patients in the control
group were examined and treated according to most recent guidelines.
Various
outcomes were recorded over six months, includeing mortality, time in hospital,
and costs. These were analysed for the initial hospital visit, and at 90 and
180 days.
Results
Patients
had an average age of 71 years, about 40% were women, and they were comparable
in terms of baseline characteristics. Follow up was complete in all but one
patient.
Acute
heart failure was the final discharge diagnosis in 45% in the BNP group and 51%
in the control group, with exacerbated COPD or asthma in 23% and 11%
respectively. In the BNP group, BNP levels were less than 100 ng/L in 36%,
between 100 and 500 ng/L in 28%, and above 500 ng/L in 36%.
In
patients in whom BNP had been used diagnostically, compared with the control
group using standard diagnostic procedures, there was significantly less use of
intensive care beds, 15% of patients compared with 24%, 10 vs 18 hours average
use per patient, and 6% rather than 12% of patients required any ventilatory
support. There was insignificant but consistently less use of echocardiography
(40% vs 49%) and coronary angiography (5% vs 9%).
Patients
in the BNP group required fewer days in hospital, 8 vs 10 following the initial
visit, with a greater difference of 10 vs 14 days over 180 days. Initially and
over 180 days, mortality was lower, by about 2 or 3 per 100 patients.
Total
treatment costs were lower when BNP tests were used for the initial diagnosis,
at all stages during the first 180 days (Figure 1). The average saving per
patient initially was $1,900, rising to $2,600 over 180 days. Medication costs
were the same, and most of the saving came from reduced use of hospital beds.
Figure 1: Total treatment costs with emergency room BNP testing or conventional diagnostic strategy in patients admitted with acute dyspnoea
Comment
Isn't
it terrific when it all comes together? If good systematic reviews of good
studies of diagnostic tests are rare as hens' teeth (as in Bandolier
149), then good randomised trials of effectiveness with good cost information
are even rarer. This formidable study from Switzerland even comes with
modelling around how this looks in a cost-effectiveness plane looking at costs
and mortality. Most of 5,000 simulations were in the dominant quadrant of lower
cost and lower mortality with use of BNP tests in the emergency department.
A
word of caution, though. This is a particular, and probably most important, use
of BNP in secondary care. How the model would work in a primary care situation
is something we do not yet know with the same degree of certainty. More
thinking still needs to be done for that setting.
Reference:
- C Mueller et al. Cost-effectiveness of B-type natriuretic peptide testing in patients with acute dyspnoea. Archives of Internal Medicine 2006 166: 1081-1087.
previous story