Original Contribution

Natriuretic peptide testing in EDs for managing acute dyspnea: a meta-analysis^?

Ludovic Trinquart MSc ^a,b,1,2,3, Patrick Ray PhD, MD ^c,d,e,1,3, Bruno Riou PhD, MD ^c,d,e,1,3, Antonio Teixeira PhD, MD ^f,g,?,1,2,3,4

^aEvidence-based Medicine Center, Paris, France

^bINSERM CIE 4, Paris, France

^cUPMC Univ Paris 06, Paris, France

^dAssistance Publique-Hopitaux de Paris, Hopital La Pitie-Salpetriere, Service des Urgences, Paris, France

^eINSERM UMRS 956, Paris, France

^fUnite ER 10 Biologie et Physiologie des interactions neurorespiratoires et cardiopulmonaires,

UPMC Univ Paris 06, Paris, France

^gAssistance Publique-Hopitaux de Paris, Groupe Hospitalier Lariboisiere-Fernand Widal, Service de Geriatrie, Paris, France

Received 26 January 2010; revised 25 February 2010; accepted 25 February 2010

Abstract

Purposes: The aim of the study was to assess the usefulness of systematic natriuretic peptide testing in the management of patients presenting with acute dyspnea to emergency departments (EDs).

Methods: We performed a systematic review and meta-analysis of randomized controlled trials assessing the usefulness of B-type Natriuretic Peptide or its N-terminal fragment (NT-proBNP) in the management of patients presenting with dyspnea into ED. We searched Medline, Embase, and conference proceedings without restriction on neither language nor Publication year. Selection of studies, data collection, and assessment of risk of bias were performed by 2 reviewers independently and in duplicate. Outcomes included Hospital admission rate, Time to discharge, and length of hospital stay, mortality and rehospitalization rates, and total direct Medical costs. Combined risk ratios were estimated using fixed or random effects model. Duration and cost data were not combined.

Findings: Four randomized controlled trials, representing 2041 patients, were selected. In 4 trials, there was a tendency for hospital admission to be reduced in the intervention group (combined risk ratio, 0.95; 95% confidence interval, 0.89-1.01). Time to discharge was significantly reduced in 2 trials, whereas there was no significant reduction in hospital length of stay in 3 trials. There was no significant effect on in-hospital and 30-day mortality rates or rehospitalization rates (3 trials reporting each outcome). Two trials found significant reduction in direct costs.

^? Competing interests: All authors except Dr Patrick Ray (PR) declare that they have no competing interest. PR has received honoraria from bioMerieux.

* Corresponding author. Assistance Publique-Hopitaux de Paris, Groupe Hospitalier Lariboisiere-Fernand Widal, Service de Geriatrie, 200 rue du Faubourg Saint-Denis 75010 Paris, France. Tel.: +33 00331 40 05 42 90; fax: +33 00331 40 05 42 91.

E-mail address: [email protected] (A. Teixeira).

¹ Conceived and designed the study.

² Responsible for the electronic search, data collection, and data analysis.

³ Drafted the manuscript. All authors revised the manuscript for important intellectual content and approved the final version.

⁴ Is the guarantor.

0735-6757/$ - see front matter. Crown Copyright (C) 2011 Published by doi:10.1016/j.ajem.2010.02.026

Conclusions: The current evidence remains inconclusive on whether systematic natriuretic peptide testing is useful for the management of patients presenting to ED with acute dyspnea.

Crown Copyright (C) 2011 Published by

Introduction

Background

heart failure is a major Public health concern that affects more than 15 million people in North America and Europe. About 1.5 million new cases are diagnosed every year, most of them being admitted through the emergency departments (EDs) [1]. Acute dyspnea is a key symptom of HF and one of the most common causes of admission to ED. Other frequent causes of dyspnea include pneumonia, asthma, exacerbation of chronic obstructive pulmonary disease, and pulmonary embolism [2]. However, in obese or elderly patients or in patients with history of chronic obstructive pulmonary disease, clinical presentation of HF is often atypical, and etiologic diagnosis of acute dyspnea is challenging, especially when several etiologies are present.

For acute dyspnea, emergency routine investigations usually include chest radiography, electrocardiography, and arterial blood gases analysis. Doppler-echocardiography is considered to be the diagnostic reference standard for establishing HF. However, this rapid but costly evaluation is seldom feasible in Acute care settings. Unfortunately, delays in diagnosing and initial misdiagnosis of HF result in higher in-hospital Morbidity and mortality rates [3,4]. Given the increasing burden of HF in Western countries, an accurate diagnostic test that would allow diagnosing rapidly HF as the cause of dyspnea is needed.

The cardiac natriuretic hormones family has been shown to be useful in diagnosing HF in dyspneic patients. B-type natriuretic peptide (BNP) is a biologically active peptide of 32 amino acids that is cleaved from the 108 amino acid hormone pro-B-type natriuretic peptide released predomi- nantly from the left cardiac ventricles [5,6]. Its N-terminal fragment, NT-proBNP, is a biologically inactive peptide of 76 amino acids that circulates at higher concentrations than BNP [5,6]. Both BNP and NT-proBNP are released by cardiomyocytes into the bloodstream in response to increased ventricular wall stretch, Volume expansion, and overload.

Importance

Several studies have assessed the diagnostic usefulness of both Natriuretic peptides in ED. Systematic reviews [7-12] showed that (1) BNP and NT-proBNP yielded moderate to good diagnostic accuracy for HF in the ED, with positive likelihood ratios ranging from 1.33 to 33.53 and 1.85 to 3.35, respectively, and negative ranging from 0.01 to 0.17 and

0.03 to 0.18, respectively, and (2) although variable

threshold values have been identified for each peptide according to the population studied, the existing evidence rather supports the use of 2 Diagnostic thresholds, with a high value supporting the diagnosis of HF and a low value to rule it out. However, diagnostic accuracy does not imply that outcomes of patients who undergo natriuretic peptide testing in the ED improve compared to that of patients who do not. Direct evidence requires randomized controlled trials [13]. The first trial, using BNP, showed a decrease of time to discharge and in direct costs [14]. However, the most recent multicenter randomized trial found no significant effect of early BNP testing neither on clinical outcomes nor on resource use [15].

Goals of this investigation

We performed a systematic review of all randomized controlled trials that assessed the usefulness of BNP or NT- proBNP in the management of patients presenting with dyspnea into ED regarding short-term clinical outcomes and resource use.

Materials and methods

The reporting of this meta-analysis was prepared in accordance with the Preferred Reporting Items for System- atic Reviews and Meta-Analyses statement [16].

Selection criteria and search strategy

Randomized controlled trials were eligible when patients presenting with acute dyspnea as the main symptom to ED were assigned to have natriuretic peptide testing or not, in addition to routine care. Results of the natriuretic peptide (BNP or NT-proBNP) measurement were communicated only to emergency physicians who managed the patients randomized to the intervention group.

The studies were searched into the Cochrane Central Register of Controlled Trials on the Cochrane Library (Issue 3, 2009), MEDLINE (up to September 2009), EMBASE (up to September 2009), and LILACS (up to September 2009). The search strategy used combinations of free text words and index terms (MeSH/EMTREE), corresponding to the following concepts: “dyspnea” and “natriuretic peptide” (Table 1). It was further restricted to identify randomized controlled trials only, using the Cochrane Highly Sensitive Search Strategy [17]. No language restrictions were applied.

We hand-searched the reference lists of primary selected studies and that of relevant reviews.

To identify recent studies not yet published as full articles, we also searched books of abstracts from recent conferences (American Heart Association congress 2007, 2008, and 2009; European Congress on Emergency Medicine 2008 and 2009; annual meeting of the Society for Academic Emergency Medicine 2007, 2008, and 2009). Clinical trials registration Web sites were also searched (www.clinicaltrials.gov; www.controlled-trials.com).

Selection of studies, data collection, and synthesis

Two authors independently reviewed titles, abstracts, and full-text articles identified by the search strategy and independently extracted data in selected trials, using standardized electronic forms, and also assessed the risk of bias within trials according to 6 criteria pertaining to sequence generation; allocation concealment; blinding of patients, physicians, and outcome assessors; incomplete outcome data; selective outcome reporting; and other potential threats to validity [18]. Data relevant to inten- tion-to-treat analyses were extracted. There was no disagreement between the 2 authors regarding study selection and data extraction, except for assessment of allocation concealment and blinding, which was resolved by consensus discussion.

The primary outcome in this review was the hospital admission rate. Other outcomes measurements included (1)

Table 1 Search strategies in Medline and Embase

Medline

(“Atrial Natriuretic Factor”[Mesh] OR “Natriuretic Peptide, Brain”[Mesh] OR “BNP”[tiab] OR “NT-proBNP”[tiab]) AND (“Dyspnea”[Mesh] OR “dyspnea”[tiab] OR “dyspneic”[tiab] OR “dyspnoea”[tiab] OR “dyspnoeic”[tiab] OR “shortness of breath”[tiab]) AND ((“Randomized Controlled Trial”[ptyp] OR “Controlled Clinical Trial”[ptyp] OR “randomised”[tiab] OR “placebo”[tiab] OR “Clinical Trials as Topic”[Mesh:noexp] OR “randomly”[tiab] OR “trial”[ti]) NOT (“Animals”[Mesh] NOT (“Humans”[Mesh] AND “Animals”[Mesh])))

Embase

(‘atrial natriuretic factor’/exp OR ‘brain natriuretic peptide’/exp OR ‘amino terminal pro brain natriuretic peptide’/exp OR ‘BNP’:ti,ab,de OR ‘NT-proBNP’:ti,ab,de) AND (‘dyspnea’/ exp OR ‘dyspnea’:ti,ab,de OR ‘dyspneic’:ti,ab,de OR ‘dyspnoea’:ti,ab,de OR ‘dyspnoeic’:ti,ab,de) AND (random$ OR factorial$ OR crossover$ OR cross over$ OR cross-over$ OR placebo OR doubl$ adj blind$ OR singl$ adj blind$ OR assign$ OR allocat$ OR volunteer$ OR ‘crossover- procedure’/exp OR ‘double-blind procedure’/exp OR ‘randomised controlled trial’/exp OR ‘single-blind procedure’/exp)

intensive care unit admission rate, (2) time to discharge and length of hospital stay (ICU and overall), (3) in-hospital and 30-day mortality rates, (4) 30-day rehospi- talization rates, and (5) total direct Medical costs. We also extracted the rates of reviewer-adjudicated diagnoses of HF (final HF diagnosis).

Time to discharge was defined in both initially hospital- ized and nonhospitalized patients as the time elapsed from presentation at the ED to hospital discharge. Length of stay was defined in hospitalized patients only as the time elapsed from hospital admission at the end of ED triage to hospital discharge. Among patients initially hospitalized, the rehos- pitalization rate was defined as the number of second admissions to hospital divided by the number of patients initially discharged alive (ie, patients initially hospitalized minus in-hospital deaths). Among patients not initially hospitalized, the secondary hospitalization rate was defined as the number of new admissions to hospital divided by the number of patients not initially hospitalized. Denominators were corrected by losses to follow-up, when appropriate.

Two authors (PR and BR) contacted the 4 first authors of primary studies. We requested missing data because all outcome measurements were not reported exhaustively. We also requested readmission data: we were interested in the rates of 30-day secondary admission in patients who had not been admitted initially. All authors answered to us, but only one provided the requested data [19].

The effect of intervention was measured using a risk ratio for binary outcomes. Combined estimates were calculated using Mantel-Haenszel fixed effect or random effects methods. Inconsistency of findings across studies was assessed using Cochran’s Q statistic and the I² statistic with associated 95% confidence interval (CI), the latter being the percentage of variability that is due to between-study heterogeneity rather than sampling error (chance): heteroge- neity was classified as moderate (I² >= 30%), substantial (I² >= 50%), or considerable (I² >= 75%) [20]. In all cases, we considered the results from both fixed effect and random effects models. For the latter, it is known that the estimate of the heterogeneity parameter is likely to be unreliable when the meta-analysis is based on a small number of studies. Hence, when the results from the trials were consistent, we preferred fixed effect analysis [21]. The Cochrane Collab- oration RevMan computer program was used for statistical analyses [22]. All P values were 2-sided and P b .05 was deemed significant.

Time to discharge and length of stay are known from primary studies to have left-skewed distributions, and these outcomes were therefore appropriately summarized using the median (25% percentile; 75% percentile). However, current methods for meta-analysis of continuous data involve combining raw or standardized differences in means. Although methods have been proposed to combine log- transformed data [23], medians for untransformed data cannot be mixed in a meta-analysis. Therefore, we chose to report individual results only.

Results

The electronic search yielded 100 references (59 from Medline, 41 from Embase), among which 18 abstracts and then 13 full-text articles were retrieved (Fig. 1). Four randomized controlled trials met the selection criteria and were included in this review [14,15,19,24]. Among them, the BASEL (B-Type Natriuretic Peptide for Acute Shortness of Breath Evaluation) trial gave rise to six subgroup analyses which were not included in our review. They concerned patients with diabetes [25], with kidney disease [26], with previous pulmonary disease [27], women [28], elderly patients [29], and obese patients [30]. Moreover, 2 Cost-effectiveness analyses [31,32], reporting 180-day and 360-day cost data, and a subanalysis concerning multimarker strategy [33] also derived from the BASEL trial were not included in our review.

Table 2 summarizes the study characteristics. All studies included patients presenting to ED with acute dyspnea as primary symptom. However, each study used its own way to grade severity of shortness of breath. All studies used the same exclusion criteria (Table 2), except 2 studies in which patients who requested an early transfer to another hospital were excluded [14,15].

Two trials concerned BNP [14,15], and 2 trials concerned NT-proBNP [19,24]. In all studies, the routine diagnostic

Fig. 1 Flow diagram of study selection.

investigations included standard blood tests, electrocardiogra- phy, and chest x-ray. Doppler-echocardiography and pulmo- nary-function tests were not routinely performed but mostly performed during hospitalization or on an outpatient basis. The 2 studies concerning BNP used the same threshold values to support (500 pg/mL) or to rule out the diagnosis of HF (100 pg/ mL). Concerning NT-proBNP, the study from Moe et al [24] used age-dependent threshold values based on the PRIDE study (N-terminal PRo-BNP Investigation of Dyspnea In the emergency DEpartment) [34] to confirm HF, whereas Rutten et al [19] used their own threshold values. Three trials of 4 were sponsored by pharmaceutical companies, whereas financial relationships between authors and the pharmaceutical industry were disclosed in 2 studies [15,24].

The assessment of risk of bias within selected studies is reported in Table 2. All studies used computer-generated or random number-generated randomization sequences. Sealed envelopes were used in 2 trials to conceal the randomization sequence, but only one [19] had enough information (ie, sealed opaque envelopes) to consider allocation concealment to be adequate. In all trials, patients were blinded to the intervention. However, we considered that no study responded to blinding criteria as physicians who managed the patients knew the assignment by design and that time to discharge was likely to be influenced by lack of blinding. It is to be noted that in the trial conducted by Mueller, end points were assessed in a blinded fashion by physicians who were not involved in patient care. There were no missing data in 2 studies, whereas, in the 2 others, losses to follow-up were less than 5% and were balanced across intervention groups, with similar reasons for missing data. Details concerning study protocol were available for 2 studies on trial registration Web sites [15,19]. Only one study could be considered to be free of selective reporting [19]. In the other, registered primary outcomes included the use of early continuous positive airway pressure ventilation, nitrates, Loop diuretics, and angiotensin-converting enzyme inhibitors and also included the number of endotracheal intubations (clinicaltrials.gov) but were not reported. Finally, we considered that reporting the 60-day rather than 30-day outcomes could be a form of selective reporting in the study of Moe.

Table 3 summarizes participant characteristics. The 4 randomized controlled trials represented 2041 patients. Patients’ mean age was greater than 70 years in all studies except in one with a mean age less than 60 years old [19]. The rate of symptoms or signs at admission was similar in all studies except for dyspnea at rest that varied from 27% [14] to 52% [24], and orthopnea that varied from 20% to 45%.

Hospital and ICU admission

The 4 trials were included in the comparison of hospital admission rate. Combining the results of the 4 trials showed no statistically significant difference in hospital admission, although a trend was toward reduced

Table 2 Methods of selected studies

Setting

Mueller et al, 2004 (n = 452) Moe et al, 2007 (n = 500) Rutten et al, 2008 (n = 477) Schneider et al, 2009 (n = 612)

Country Switzerland Canada Netherlands Australia Single/multicenter Single-center 7 centers Single-center 2 centers Selection criteria

Grading of severity of shortness of breath

While walking up a slight incline While walking on level ground At rest

At rest or not New York Heart Association II-IV

Category 1-3, the latter requiring assessment by a physician immediately to within 30 min after arrival

Exclusion criteria Obvious Traumatic cause of dyspnea Severe renal disease

Cardiogenic shock

Patients who requested an early transfer to another hospital

B-type natriuretic peptides measurement

Obvious cause of dyspnea (including pneumothorax and chest wall trauma) Severe renal disease

Acute myocardial infarction Malignant disorders

Obvious traumatic cause of dyspnea

Severe renal disease Cardiogenic shock

Obvious traumatic cause of dyspnea Severe renal disease

Cardiogenic shock Age younger than 40 y

Patients who requested an

early transfer to another hospital

Peptide BNP NT-proBNP NT-proBNP BNP

Kit Biosite diagnostics Roche Diagnostics GmbH Elecsys Hoffman-La Roche Elecsys Abbott AxSYM

Dosage method Fluorescence immunoassay Electrochemiluminescence immunoassay

Electrochemiluminescence immunoassay

Microparticle enzyme immunoassay

Threshold to rule in HF 500 pg/mL 450 pg/mL for patients b50 y of age 900 pg/mL for patients >=50 y of age

1017 pg/mL 500 pg/mL

Threshold to rule out HF 100 pg/mL 300 pg/mL 144 pg/mL for women 100 pg/mL

93 pg/mL for men

Time from blood sampling to delivery of natriuretic peptide dosage results

15 min Not clear Not clear Within 60 min

Funding by the pharmaceutical industry

Yes (Biosite, San Diego, Calif) Yes (Roche Diagnostics Canada,

Laval, Quebec, Canada)

None declared Yes (Janssen-Cilag, North Ryde, New South Wales, Australia)

Financial relationship between authors and the pharmaceutical industry

Risk of Bias assessment

None declared Yes None declared Yes

Adequate sequence generation?	Yes	Yes	Yes	Yes
Allocation concealment?	Not clear	Not clear	Yes	Not clear
Blinding? a	No	No	No	No
Incomplete outcome data addressed?	Yes	Yes	Yes	Yes
Free of selective reporting?	Not clear	No b	Yes	No c
Free of other bias?	Not clear	No d	Not clear	Not clear

^a We considered that time to discharge likely to be influenced by lack of blinding.

Natriuretic peptide testing in ED for acute dyspnea

761

^b Sixty-day rather than 30-day mortality and readmission rates were reported.

^c Registered secondary end points included the use of continuous positive airway pressure ventilation, nitrates, loop diuretics, and angiotensin-converting enzyme inhibitors and also included the numbers of endotracheal intubations, which were not reported (clinicaltrials.gov).

^d Fixed block size of 4 was used so that investigators could easily deduce the next treatment allocations if they discovered the block size.

Mueller et al, 2004 (n = 452)		Moe et al, 2007 (n = 500)	Rutten et al, 2008 (n = 477)	Schneider et al, 2009 (n = 612)
Participants
Age, y	71	70	59	73
Male	58%	51%	54%	53%
Current smoking	24%	Not clear	36%	13%
Medical history
Heart failure	Not clear	34%	43%	36%
Coronary artery disease	50%	Not clear	21%	41%
Hypertension	50%	53%	25%	50%
Diabetes	23%	25%	16%	20%
Chronic kidney disease	25%	Not clear	8%	11%
Chronic obstructive pulmonary disease	31%	29%	26%	63% a
Asthma	6%	Not clear	13%
Symptoms/signs at admission
Shortness of breath at rest	27%	52%	32%	Not clear
Orthopnea	Not clear	45%	31%	20%
Coughing	49%	Not clear	57%	50%
Elevated jugular venous pressure	14%	26%	14%	28%
S3 gallop	2%	Not clear	2%	3%
Rales	46%	47%	37%	Not clear
Wheezing	22%	32%	24%	28%
Lower extremity edema	34%	45%	20%	15%
Heart rate, beat/min	97	86	97	96
SBP/DBP, mm Hg	145/85	135/78	Not clear	142/73
Final heart failure diagnosis	48%	46%	Not clear	45%
SBP indicates systolic blood pressure; DBP, diastolic blood pressure. a Chronic obstructive pulmonary disease or asthma.

hospital admissions in the intervention group (combined risk ratio, 0.95; 95% CI, 0.89-1.01; P = .08; Fig. 2A),

Table 3 Participant characteristics in selected studies

with moderate heterogeneity for the overall comparison (?2 test, P = .24; I² = 29%). Three trials [14,15,19] reported ICU admission rate, which was nonsignificantly lower in the intervention group (combined risk ratio, 0.74; 95% CI, 0.48-1.14; P = .17; Fig. 2B), with substantial heterogeneity (P = .14; I² = 50%). However, the decrease in hospital and ICU admission rates originated mainly from the BASEL trial.

Time to discharge and hospital length of stay

Data on time to discharge and hospital length of stay were reported by 2 [14,19] and 3 [15,19,24] trials, respectively. As durations were skewed and were summa- rized using medians, data were not pooled. However, time to discharge was significantly reduced in the intervention group of the 2 individual trials, whereas no individual study among the 3 showed any reduction in hospital length of stay (Table 4).

In-hospital and 30-day mortality

The effect of BNP and NT-proBNP testing on in- hospital mortality was examined in 3 trials [14,19,24].

There was no significant difference in rates of in-hospital mortality in the intervention group compared with the control group (combined risk ratio, 1.02; 95% CI, 0.61-

1.72; P = .93; Fig. 3A). The 3 trials reporting the 30- day mortality rates showed no significant effect of natriuretic peptide testing on 30-day mortality (combined risk ratio, 0.86; 95% CI, 0.61-1.21; P = .38; Fig. 3B),

with a similar finding in the trial [24] reporting the 60- day mortality rate (risk ratio, 1.22; 95% CI, 0.56-2.67;

P = .58).

30-day readmission

When pooling the data from the 2 trials [14,19] reporting 30-day secondary hospitalization rates, there was no difference between the intervention and control groups (combined risk ratio, 0.93; 95% CI, 0.59-1.46; Fig. 4; P =

.74), whereas the study [24] examining the 60-day rate showed a nonsignificant reduction of risk of readmission (risk ratio, 0.68; 95% CI, 0.46-1.02; P = .07).

The secondary hospitalization rate among patients not initially hospitalized could be extracted in one study [14] and was provided by the authors of another one [19]; we found a nonsignificant reduction of risk of secondary hospitalization in the intervention arm (risk ratio, 0.64; 95% CI, 0.22-1.84; P = .40).

Fig. 2 Hospital (upper panel) and ICU (lower panel) admission.

Final heart failure diagnosis

Two trials [14,15] reported the comparison of final HF diagnosis rates. In the study by Mueller [14,15], the final discharge diagnosis was less frequently HF in the interven- tion group than in the control group (absolute difference,

-6%; 95% CI, -15% to 3%; P = .19), whereas in the trial by

Schneider [14,15], 2 physicians who were blind to BNP results but not to group assignment made the final diagnosis and diagnosed HF more frequently in the intervention group than in the control group (absolute difference, 7%; 95% CI,

-1% to 15%; P = .07).

Costs

Two trials [14,24] examined the impact of natriuretic peptide testing on direct health care costs and found that the median total treatment cost was significantly reduced in the

intervention group (median costs [25%-75% percentile], US

$5410 [US$4516-US$6304] vs US$7264 [US$6301-US

$8227] in the control group, respectively; P = .006 [14], and US$5180 [US$3005-US$8416] vs US$6129 [US$3384-

US$9991] in the intervention and control group, respective- ly; P = .0232 [24]). In another trial [19], a post hoc subgroup analysis showed a nonsignificant decrease in total direct costs (mean reduction in intervention vs control group, US

$2627; 95% CI, -US$1506 to US$6753 in patients with cardiac dyspnea, and US$150; 95% CI, -US$1386 to US

$1626 in patients with noncardiac dyspnea).

Discussion

Several studies have demonstrated that BNP and NT- proBNP are accurate biomarkers of HF in ED, even in several subgroup populations (elderly, patients with kidney

Duration of ED visit, h
NT-proBNP
Moe et al, 2007	246	5.6 (4.0-8.0)	254	6.3 (4.3-8.6)	.031
Rutten et al, 2008	236	2.8 (2.0-3.7)	241	2.9 (2.2-3.9)	.12
Time to discharge, d
BNP
Mueller et al, 2004 b	212	8.0 (1.0-16.0)	206	11.0 (5.0-18.0)	.001
NT-proBNP
Rutten et al, 2008	236	1.9 (0.12-8.4)	241	3.9 (0.16-11.0)	.04
Length of stay, d
BNP
Schneider et al, 2009	262	4.4 (2.0-9.0)	265	5.0 (2.0-9.0)	.93
NT-proBNP
Moe et al, 2007	139	6.0 (4.0-11.0)	146	7.0 (4.0-13.0)	.30
Rutten et al, 2008	147	7.8 (4.8-13.9)	162	8.1 (4.4-15.6)	.48
a Nonparametric 2-sample test. b Patients who died in hospital were excluded.

or pulmonary diseases). However, the main question is does the routine use of this accurate biomarker in patients presenting to ED with acute dyspnea improve their outcomes or process of care?

Table 4 Time to discharge and length of stay outcomes in selected studies

Intervention group Control group

P a

n Median (25%-75% percentiles) n Median (25%-75% percentiles)

Based on the examination of 4 randomized controlled trials representing 2041 patients, the present meta-analysis showed that natriuretic peptides testing had limited impact on the outcome of patients presenting to ED with acute dyspnea. Two trials showed significant reduction for time to discharge, and 2 trials found significant reduction in direct costs. Yet, individual trials found no effect on hospital length of stay. Moreover, our meta-analysis did not find any reduction in the rates of hospital and ICU admission, and there was no impact on readmission rates and mortality.

Any group of patients that would have the most to gain from natriuretic peptide assay has yet to be identified. Several arguments highlight that the early use of natriuretic peptides could be of particular interest in older patients. First, HF is the leading cause of admission in patients 65 years and older in the Western world [35] and is associated with an in- hospital Death rate higher than 15% [4]. Second, the determination of the cause of dyspnea is highly challenging in elderly patients, and the positive likelihood ratio of BNP is this population is particularly high [36-38]. Lastly, the BASEL trial showed a significant impact in the subgroup of older patients, even on mortality, although this analysis was not prespecified.

Another subgroup of patients that could benefit from systematic natriuretic peptide testing could be those with a intermediate clinical suspicion for HF, that is, a pretest physician-estimated probability between 20% and 80%, who represent about 30% of emergency patients, and is associated to a specific higher mortality [24,39]. Unfortunately, none of

the study evaluated both natriuretic peptides based upon prespecified classification according to pretest probability.

Lastly, the diagnostic accuracy of BNP and NT-proBNP is not satisfactory for patients within the “gray zone” level, that is, level between the rule in value and the rule out value. These patients usually represent less than 20% of emergency patients. In the 4 studies, authors do not specify the rate of patients in this challenging situation.

Limitations

The eligible trials had limitations. Firstly, sample size calculation was based on time to discharge in 2 trials and on ED Visit duration in another. They might have been underpowered for hard clinical outcomes such as read- missions or mortality. Secondly, as the unit of randomization was the patient, there may have been a possibility of contamination in the trials, physicians being able to order natriuretic peptide testing on patients in the control arms. Thirdly, all studies but Mueller’s failed to distinguish true readmissions from secondary admissions in patients who had not been admitted initially. Yet, to assess the absence of increase risk of secondary admissions is crucial, given the reduced risk of initial hospital admission in the intervention group. In the 2 trials in which these data were obtained, we did not find any increase in the secondary hospitalization rate among patients not initially hospitalized. Lastly, there was an imbalance in the proportion of final diagnoses of HF with the trial conducted by Schneider [15] showing more HF diagnoses in the intervention than in the control group and that in the trial conducted by Mueller showing the other way around. Although nonsignificant, these differences might

Fig. 3 In-hospital (upper panel) and 30-day (lower panel) mortality.

Fig. 4 30-day readmission.

question the quality of randomization or allocation conceal- ment. When considering the distribution of baseline characteristics, the study by Schneider [15] showed signif- icantly more patients with orthopnea at admission in the BNP group than in the control group (absolute difference, 9.2%; 95% CI, 2.8-15.4; P = .005). In the trial conducted by Mueller, orthopnea at admission was not reported, but there was no imbalance in other baseline characteristics. More- over, we cannot exclude a possible lack of allocation concealment considering the use of envelopes in 2 trials and the unclear information in the other two.

The chief limitation of this analysis is the limited number of available randomized trials that precluded formal investigation of sources of heterogeneity. First, evidence of impact of BNP testing originated mainly from the BASEL trial, which showed decreased hospital and ICU admission rates, decreased time to discharge, and costs. However, this may reflect a specific caring of patients. In fact, European emergency care systems differ from Anglo-American health system, the former showing longer length of hospital stay than the latter. The effect of diagnostic testing upon clinical and process outcomes is likely to be dependent upon the setting. Factors such as the expertise of physicians, the availability of facilities (such as observation wards, inpatient beds, coronary care beds) and other diagnostic tests (especially echocardiography) and the presence of financial incentives to admit or discharge will influence the intervention effect. Moreover, 2 trials were single-center studies, which may be prone to overestimation of intervention effect and may suffer from lack of external validity. Beyond differences in caring of patients, diversity of study populations could explain discrepancies between studies results. For instance, Rutten et al [19] included a much younger population, with less comorbidity. Second, the severity of shortness of breath varied between studies. In particular, that reported by Schneider et al [15] included explicitly patients with severe dyspnea. However, it is likely that natriuretic peptide testing in the ED would yield greater benefit in the most severe patients, that is, in patients with Multiple comorbidities in which the diagnosis of HF is challenging. Unfortunately, Schneider et al [15] did not show any benefit in the group of patients with an early measurement of BNP. Third, variability in the method and intervention could be of importance. Only one study

[14] specifically used the result of BNP associated to recommendations (ie, if BNP is N500 pg/mL, consider HF and give vasodilators, diuretics, and ask for a cardiologic consultation). However, other interventional studies with biomarkers demonstrated that decision making based upon the level of the biomarker needs to be associated with guidelines [40]. Along this line of thinking, providing physicians with further guidelines could impact process of care measures in HF patients, such as initiation of appropriate therapies. Only Schneider et al [15] reported on the treatment initiated in the ED. They reported that the use of appropriate heart failure medication was not

increased in patients with heart failure in the intervention group vs the control group.

Conclusions

This meta-analysis of 4 randomized controlled trials showed that natriuretic peptide testing in all emergency patients with shortness of breath had no apparent effects on patient outcomes (length of hospital stay, hospital admission, secondary admission, or mortality), except a significant reduction in time to discharge, whereas costs were reduced. The current evidence remains inconclusive on whether routine use of BNP or NT-proBNP measurement should be used in all dyspneic patients in the ED.

Large multicenter randomized trials may bring a more solid basis for generalization of results and may be particularly useful in patients with an intermediate pretest probability of HF, and/or including elderly patients, and using the same recommendations in both groups based upon the suspected emergency’s diagnosis.

Acknowledgments

The authors would like to thank Dr DJ Baker (Department of Anesthesiology, CHU Necker-Enfants Malades, Paris, France) for editing the manuscript.

References

1. Redfield MM. Heart failure-an epidemic of uncertain proportions. N Engl J Med 2002;347(18):1442-4.
2. Mulrow CD, Lucey CR, Farnett LE. Discriminating causes of dyspnea through clinical examination. J Gen Intern Med 1993;8(7):383-92.
3. Green SM, Martinez-Rumayor A, Gregory SA, Baggish AL, O’Donoghue ML, Green JA, et al. Clinical uncertainty, diagnostic accuracy, and outcomes in emergency department patients presenting with dyspnea. Arch Intern Med 2008;168(7):741-8.
4. Ray P, Birolleau S, Lefort Y, Becquemin MH, Beigelman C, Isnard R, et al. Acute respiratory failure in the elderly: etiology, emergency diagnosis and prognosis. Crit Care 2006;10(3):R82.
5. Maeda K, Tsutamoto T, Wada A, Hisanaga T, Kinoshita M. plasma brain natriuretic peptide as a biochemical marker of high left ventricular end-diastolic pressure in patients with symptomatic left ventricular dysfunction. Am Heart J 1998;135(5 Pt 1):825-32.
6. Nagagawa O, Ogawa Y, Itoh H, Suga S, Komatsu Y, Kishimoto I. Rapid transcriptional activation and early mRNA turnover of brain natriuretic peptide in cardiocyte hypertrophy: evidence for brain natriuretic peptide as an “emergency” cardiac hormone against ventricular overload. J Clin Invest 1995;96:1280-7.
7. Battaglia M, Pewsner D, Juni P, Egger M, Bucher HC, Bachmann LM. Accuracy of B-type natriuretic peptide tests to exclude congestive heart failure: systematic review of test accuracy studies. Arch Intern Med 2006;166(10):1073-80.
8. Clerico A, Fontana M, Zyw L, Passino C, Emdin M. Comparison of the diagnostic accuracy of Brain natriuretic peptide and the N-

terminal part of the propeptide of BNP immunoassays in chronic and acute heart failure: a systematic review. Clin Chem 2007;53(5): 813-22.

1. Doust JA, Glasziou PP, Pietrzak E, Dobson AJ. A systematic review of the diagnostic accuracy of natriuretic peptides for heart failure. Arch Intern Med 2004;164(18):1978-84.
2. Korenstein D, Wisnivesky JP, Wyer P, Adler R, Ponieman D, McGinn

T. The utility of B-type natriuretic peptide in the diagnosis of heart failure in the emergency department: a systematic review. BMC Emerg Med 2007;7:6.

1. Wang CS, FitzGerald JM, Schulzer M, Mak E, Ayas NT. Does this dyspneic patient in the emergency department have congestive heart failure? JAMA 2005;294(15):1944-56.
2. Worster A, Balion CM, Hill SA, Santaguida P, Ismaila A, McKelvie R, et al. Diagnostic accuracy of BNP and NT-proBNP in patients presenting to acute care settings with dyspnea: a systematic review. Clin Biochem 2008;41(4-5):250-9.
3. Busse JW, Bassler D, Guyatt GH. Evaluating the evidence for assessing BNP and Nt-proBNP levels. Clin Biochem 2008;41(4-5): 227-30.
4. Mueller C, Scholer A, Laule-Kilian K, Martina B, Schindler C, Buser P, et al. Use of B-type natriuretic peptide in the evaluation and management of acute dyspnea. N Engl J Med 2004;350(7):647-54.
5. Schneider HG, Lam L, Lokuge A, Krum H, Naughton MT, De Villiers Smit P, et al. B-type natriuretic peptide testing, clinical outcomes, and health services use in emergency department patients with dyspnea: a randomized trial. Ann Intern Med 2009;150(6):365-71.
6. Moher D, Liberati A, Tetzlaff J, Altman D, Group TP. Preferred reporting items for systematic reviews and meta-analysis. The PRISMA statement. Ann Intern Med 2009;151(4):164-9.
7. Lefebvre C, Manheimer E, Glanville J. Chapter 6. Searching of studies. In: Higgins J, Altman D, editors. Cochrane handbook for systematic reviews of interventions version 500 (updated February 2008); 2008.
8. Higgins J, Altman D. Chapter 8. Assessing risk of bias in included studies. In: Higgins J, Altman D, editors. Cochrane handbook for systematic reviews of interventions version 500 (updated February 2008); 2008.
9. Rutten JH, Steyerberg EW, Boomsma F, van Saase JL, Deckers JW, Hoogsteden HC, et al. N-terminal pro-brain natriuretic peptide testing in the emergency department: beneficial effects on hospitalization, costs, and outcome. Am Heart J 2008;156(1):71-7.
10. Deeks J, Higgins J, Altman D. Chapter 9. Analysing data and undertaking meta-analyses. In: Higgins J, Altman D, editors. Cochrane handbook for systematic reviews of interventions version 500 (updated February 2008); 2008.
11. Whitehead A. In: Senn S, editor. Meta-analysis of controlled clinical trials. Chichester, England: Wiley; 2002.
12. Review Manager (RevMan). 5.0 ed. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2008.
13. Higgins JP, White IR, Anzures-Cabrera J. Meta-analysis of skewed data: combining results reported on log-transformed or raw scales. Stat Med 2008;27(29):6072-92.
14. Moe GW, Howlett J, Januzzi JL, Zowall H. N-terminal pro-B-type natriuretic peptide testing improves the management of patients with suspected acute heart failure: primary results of the Canadian prospective randomized multicenter IMPROVE-CHF study. Circula- tion 2007;115(24):3103-10.
15. Mueller C, Laule-Kilian K, Christ A, Perruchoud AP. The use of B- type natriuretic peptide in the management of patients with diabetes and acute dyspnoea. Diabetologia 2006;49(4):629-36.
16. Mueller C, Laule-Kilian K, Scholer A, Nusbaumer C, Zeller T, Staub D, et al. B-type natriuretic peptide for acute dyspnea in patients with kidney disease: insights from a randomized comparison. Kidney Int 2005;67(1):278-84.
17. Mueller C, Laule-Kilian K, Frana B, Rodriguez D, Scholer A, Schindler C, et al. Use of B-type natriuretic peptide in the management of acute dyspnea in patients with pulmonary disease. Am Heart J 2006;151(2):471-7.
18. Mueller C, Laule-Kilian K, Scholer A, Frana B, Rodriguez D, Schindler C, et al. Use of B-type natriuretic peptide for the management of women with dyspnea. Am J Cardiol 2004;94(12):1510-4.
19. Mueller C, Laule-Kilian K, Frana B, Rodriguez D, Rudez J, Scholer A, et al. The use of B-type natriuretic peptide in the management of elderly patients with acute dyspnoea. J Intern Med 2005;258(1):77-85.
20. Noveanu M, Breidthardt T, Cayir S, Potocki M, Laule K, Mueller C. B-type natriuretic peptide-guided management and outcome in patients with obesity and dyspnea-Results from the BASEL study. Am Heart J 2009;158:488-95.
21. Breidthardt T, Laule K, Strohmeyer AH, Schindler C, Meier S, Fischer M, et al. Medical and economic Long-term effects of B-type natriuretic peptide testing in patients with acute dyspnea. Clin Chem 2007;53(8): 1415-22.
22. Mueller C, Laule-Kilian K, Schindler C, Klima T, Frana B, Rodriguez D, et al. Cost-effectiveness of B-type natriuretic peptide testing in patients with acute dyspnea. Arch Intern Med 2006;166(10):1081-7.
23. Christ M, Laule K, Klima T, Hochholzer W, Breidthardt T, Perruchoud AP, et al. Multimarker strategy for risk prediction in patients presenting with acute dyspnea to the emergency department. Int J Cardiol 2008;126(1):73-8.
24. Januzzi Jr JL, Camargo CA, Anwaruddin S, Baggish AL, Chen AA, Krauser DG, et al. The N-terminal Pro-BNP investigation of dyspnea in the emergency department (PRIDE) study. Am J Cardiol 2005;95 (8):948-54.
25. Stewart S, MacIntyre K, MacLeod MM, Bailey AE, Capewell S, McMurray JJ. Trends in hospitalization for heart failure in Scotland, 1990-1996. An epidemic that has reached its peak? Eur Heart J 2001;22(3):209-17.
26. Ray P, Arthaud M, Lefort Y, Birolleau S, Beigelman C, Riou B. Usefulness of B-type natriuretic peptide in elderly patients with acute dyspnea. Intensive Care Med 2004;30(12):2230-6.
27. Cotter G, Metzkor E, Kaluski E, Faigenberg Z, Miller R, Simovitz A, et al. Randomised trial of high-dose isosorbide dinitrate plus low-dose furosemide versus high-dose furosemide plus low-dose isosorbide dinitrate in severe pulmonary oedema. Lancet 1998;351(9100): 389-93.
28. Dickstein K, Cohen-Solal A, Filippatos G, McMurray JJ, Ponikowski P, Poole-Wilson PA, et al. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2008: the Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2008 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association of the ESC (HFA) and endorsed by the European Society of Intensive Care Medicine (ESICM). Eur Heart J 2008;29(19):2388-442.
29. McCullough PA, Nowak RM, McCord J, Hollander JE, Herrmann HC, Steg PG, et al. B-type natriuretic peptide and clinical judgment in emergency diagnosis of heart failure: analysis from Breathing Not Properly (BNP) Multinational Study. Circulation 2002;106(4):416-22.
30. Stolz D, Christ-Crain M, Bingisser R, Leuppi J, Miedinger D, Muller C, et al. Antibiotic treatment of Exacerbations of COPD: a randomized, controlled trial comparing procalcitonin-guidance with Standard therapy. Chest 2007;131(1):9-19.