Author + information
- Received August 11, 2011
- Revision received November 26, 2011
- Accepted November 29, 2011
- Published online April 17, 2012.
- Justin A. Ezekowitz, MBBCh, MSc⁎,⁎ (, )
- Adrian F. Hernandez, MD‡,
- Christopher M. O'Connor, MD‡,
- Randall C. Starling, MD§,
- Guy Proulx, MD†,
- Mason H. Weiss, MD∥,
- Jeffrey A. Bakal, PhD⁎,
- Robert M. Califf, MD‡,
- John J.V. McMurray, MD¶ and
- Paul W. Armstrong, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Justin A. Ezekowitz, Cardiology, University of Alberta, 2c2, 8440-112 Street, Edmonton, Alberta T6G 2B7, Canada
Objectives This study hypothesized that peak expiratory flow rate (PEFR) would increase with acute heart failure (AHF) treatment over the first 24 h, related to a Dyspnea Index (DI) change and treatment effect.
Background Dyspnea is a key symptom and clinical trial endpoint in AHF, yet objective assessment is lacking.
Methods In a clinical trial substudy, 421 patients (37 sites) underwent PEFR testing at baseline, 1, 6, and 24 h after randomization to nesiritide or placebo. DI (by Likert scale) was collected at hours 6 and 24.
Results Patients were median age 70 years, and 34% were female; no significant differences between nesiritide or placebo patients existed. Median baseline PEFR was 225 l/min (interquartile range [IQR]: 160 to 300 l/min) and increased to 230 l/min (2.2% increase; IQR: 170 to 315 l/min) by hour 1, 250 l/min (11.1% increase; IQR: 180 to 340 l/min) by hour 6, and 273 l/min (21.3% increase; IQR: 200 to 360 l/min) by 24 h (all p < 0.001). The 24-h PEFR change related to moderate or marked dyspnea improvement by DI (adjusted odds ratio: 1.04 for each 10 l/min improvement [95% confidence interval (CI): 1.07 to 1.10]; p < 0.01). A model incorporating time and treatment over 24 h showed greater PEFR improvement after nesiritide compared with placebo (p = 0.048).
Conclusions PEFR increases over the first 24 h in AHF and could serve as an AHF endpoint. Nesiritide had a greater effect than placebo on PEFR, and this predicted patients with moderate/marked improvement in dyspnea, thereby providing an objective metric for assessing AHF. (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure [ASCEND-HF]; NCT00475852)
Shortness of breath is 1 of the principal presenting symptoms of patients with acute decompensated heart failure (AHF) (1), and relief of this dyspnea is a commonly employed endpoint for clinical care and randomized clinical trials of AHF therapy (2). Yet assessment of dyspnea is subjective, difficult to validate, and the relationship to objective measures is unknown (1,3,4). Despite this shortcoming, 3 major clinical trials in AHF (supported by the European Medicines Agency and U.S. Food and Drug Administration) have used dyspnea as a primary endpoint, given the importance assigned to this symptom (5–7). Whereas a recent study enrolling patients with AHF very early in the emergency department reported that patient-assessed visual analogue and Likert scales evaluating the change in dyspnea were highly correlated, these results differed when provocative orthopnea testing was introduced. Hence, the relationship between dyspnea and more objective measures remains unclear (1).
Although the mechanisms of dyspnea are multiple, bronchoconstriction is thought to be a key modulator (8). Peak expiratory flow rate (PEFR) is a measure that reflects airflow limitation and is primarily determined by large airway caliber, expiratory muscle strength, elastic recoil, and resistance of the smaller intrathoracic airways (9). Because PEFR is inexpensive, portable, and widely used for the assessment of asthma and chronic obstructive lung disease, it is frequently used as an endpoint in clinical trials of asthma control (10).
Three studies have evaluated the use of PEFR for the diagnosis of patients with AHF (11–13). All found that patients with AHF could be readily distinguished from those with chronic lung disease by PEFR; however, its value for the diagnosis of heart failure (HF) is limited. However, only baseline PEFR measures were used in these mixed cardiovascular and respiratory disease populations with the goal of demonstrating diagnostic utility, rather than potential as an objective clinical trial endpoint. To date, no studies to our knowledge have described change in PEFR from baseline over time in patients with AHF.
We evaluated the baseline and sequential change in PEFR over the first 24 h from admission in a prospective substudy of the ASCEND-HF (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure), a large trial of patients with AHF. Additionally, we examined the relationship of PEFR to clinical outcomes, and the effect of nesiritide or placebo on PEFR. Finally, we explored the relationship of a subjective Dyspnea Index (DI) to PEFR.
The study design and results of the ASCEND-HF study have been previously published (2,5). Briefly, the ASCEND-HF study evaluated nesiritide versus placebo in 7,007 patients with AHF enrolled within 24 h of first intravenous HF–related treatment. Participants were required to have each of the following at time of randomization: dyspnea at rest or with minimal activity; ≥1 accompanying sign (respiratory rate ≥20 breaths/min or pulmonary congestion/edema with rales ≥1/3 base); and ≥1 objective measure of HF (evidence of congestion/edema on chest x-ray; B-type natriuretic peptide [BNP] ≥400 pg/ml or N-terminal pro-BNP ≥1,000 pg/ml; pulmonary capillary wedge pressure >20 mm Hg; or left ventricular ejection fraction <40% in the previous 12 months). Among others, 1 exclusion criteria germane to this substudy was severe pulmonary disease (full exclusion criteria are available elsewhere) (5).
Respiratory substudy design
Patients in the prospectively planned respiratory substudy were enrolled between May 2007 and August 2010 at 37 sites in Canada and the United States. Consecutive patients were approached for enrollment at the time of entry into the main trial. The institutional review board at each participating hospital approved the protocol, and patients were required to provide written informed consent. The study team at each site completed standardized training including how to perform PEFR, and the same type of PEFR meter was used across all sites along with standardized patient and study team instructions.
PEFR, that is, the maximal flow generated on expiration after full inspiration, was performed at baseline (after randomization, but before drug infusion) and at 1 h, 6 h, and 24 h thereafter. Three measurements were completed at each time point, with the highest PEFR recorded in l/min (11–13). Other data captured simultaneously included respiratory rate in breaths/min, oxygen saturation (%), and liters of oxygen delivered (in l/min).
Sample sizes were estimated from 3 prior studies of dyspneic patients with cardiovascular disease that showed baseline PEFRs of 224 l/min (SD: 82 l/min) (12), 253 l/min (SD: 93 l/min) (13), and 267 l/min (SD: 97 l/min) (11), and 1 study of patients with an acute exacerbation of chronic obstructive pulmonary disease (COPD), which had a mean PEFR of 113 l/min (SD: 51 l/min) (14). An improvement in the PEFR between 10 to 25 l/min, or approximately 5% to 12%, was considered clinically significant based on asthma studies (15,16). We assumed that with a baseline mean PEFR of 200 l/min (SD: 80 l/min), 200 patients per treatment arm would provide 80% power to detect a clinically meaningful 20% change (between nesiritide and placebo groups) in PEFR with a p value <0.05. A total of 400 patients would also result in adequate measurements to ascertain the time course of change, detect any differences between groups, and allow for drop-outs and missing data of <10% of PEFR measurements.
Data for continuous variables are presented as medians with 25th and 75th percentiles, and categorical variables are presented as frequencies and percentages. Wilcoxon rank sum tests were used to measure differences for continuous variables, and the chi-square test was used for categorical variables. Differences in continuous variables over time followed a Gaussian distribution and are reported as means with confidence intervals (CI) using t tests for comparison. Adjustment for baseline differences was done using covariate adjustment. Repeated-measures analysis of covariance was used to model the effects of treatment group and to control for baseline PEFR. The groups were defined by the intention-to-treat groups as per the main trial protocol for the principal analysis. An additional “on-treatment” group was used to evaluate effects of patients receiving study drug or placebo as a sensitivity analysis. Outliers were examined for PEFR using the methods of Shiffler (17). Log transformation was used to normalize the PEFR data for evaluation. A correlation analysis was used to evaluate the association of the DI score and the change in PEFR. Additionally, a series of parsimonious logistic regression models were developed for predicting both significant improvement in dyspnea score, and for the composite outcome of HF rehospitalization or death at 30 days. Candidate variables included age, sex, PEFR, and treatment assignment. A p value of <0.05 was used as the level of statistical significance for all tests. A statistical analysis plan was developed before data analysis; all analyses were pre-specified unless otherwise stated. All analyses were performed using R Software (version 2.12.2, R Foundation for Statistical Computing, Vienna, Austria).
There were 430 patients enrolled in the respiratory substudy. PEFR data was available for 421, 406, 405, and 410 patients at baseline, hour 1, hour 6, and hour 24, respectively. The 9 patients without baseline PEFR data were excluded from further analyses. Of the 421 patients, 208 (49.4%) were in the nesiritide arm, and 213 (50.6%) in the placebo arm. Comparisons with the 7,007 patients in the overall trial as well as all the 2,728 ASCEND patients enrolled in Canada or the United States are shown in Table 1. As compared with patients from Canada or the United States not in the respiratory substudy, those in the respiratory substudy were older, weighed less, and had higher respiratory rates; however, measurements of BNP, N-terminal pro–BNP, creatinine, and ejection fraction were similar. Differences between patients in the nesiritide or placebo arms are shown in Table 1. Small differences between groups in the percent of patients on beta-blockers or creatinine clearance were not associated with PEFR (p = 0.7 and p = 0.4, respectively).
Peak expiratory flow rate
As shown in Table 2, Overall, median PEFR increased from 225 l/min (interquartile range [IQR]: 160 to 300 l/min) at baseline, to 230 l/min (IQR: 170 to 315 l/min) at hour 1, 250 l/min (IQR: 180 to 340 l/min) at hour 6, and 273 l/min (21% increase, IQR: 200 to 360 l/min) at hour 24 (Table 2). Compared with baseline, PEFR increased by hour 1 (per patient average: 8.39 l/min; 95% CI: 2.89 to 13.89), hour 6 (26.32 l/min; 95% CI: 19.09 to 33.61), and at 24 h (43.92 l/min; 95% CI: 36.39 to 51.47).
Respiratory rate and oxygen saturation
Median respiratory rate was 23.0 breaths/min at baseline and similar between the nesiritide and placebo groups (22 [IQR: 20 to 24] vs. 23 [IQR: 21 to 24]; p = 0.30). Respiratory rate declined per patient by 2.9 breaths/min (95% CI: 2.56 to 3.22) over the first 24 h in the overall group from a median of 23.0 (IQR: 21 to 24) to 20 (IQR: 18 to 22; p < 0.0001).
Median oxygen saturation was 96% (IQR: 95% to 98%) at baseline and remained similar over the next 24 h to a median of 97% (IQR: 95% to 98%); the mean per patient difference was 0.08% (95% CI: −0.19 to 0.34). Initially, patients were on a median of 2 l/min of oxygen (IQR: 0 to 2.8 l/min), which declined to a median of 0 l/min by 24 h; the mean per patient decrease was 0.36 l/min (95% CI: 0.20 to 0.53). The difference was similar between the nesiritide and placebo groups.
Nesiritide versus placebo
The change in PEFR from baseline to 24 h between the nesiritide and placebo arms was statistically significant, with greater improvement in the nesiritide arm compared with placebo (unadjusted mean difference: +52.4 l/min vs. +35.9 l/min; 95% CI of difference: 1.37 to 31.54; p < 0.05). Individual pairwise comparisons at each time point did not show any significant difference between nesiritide and placebo. However, a repeated measures model incorporating time and treatment group showed a significant result favoring an improvement in nesiritide versus placebo (p = 0.05) (Fig. 1) for the improvement in PEFR.
Respiratory rate tended to decline more following nesiritide treatment as compared with placebo, that is, nesiritide patients had an additional 0.67 breaths/min reduction at 24 h (p = 0.07) not evident earlier at baseline, 1 h, or 6 h. There were no significant differences between the nesiritide and placebo groups in either oxygen saturation or delivered oxygen over the 24 h.
DI and PEFR
Baseline PEFR was unrelated to the 6-h or 24-h DI result (rho = −0.017, p = 0.73; rho = −0.08, p = 0.11, respectively). Additionally, the change in PEFR from baseline to 6 h did not correlate with the simultaneous DI measurement at 6 h (rho = 0.03, p = 0.5). At 24 h, the DI showed a small but significant correlation with change in PEFR from baseline to 24 h (rho = 0.15, p = 0.002).
The relationship between PEFR and DI at 24 h was further explored to evaluate whether the improvement in PEFR predicted patients with a moderate or markedly improved response on the DI. The regression model at 24 h shows a significant relationship: an increased PEFR predicted improved DI at 24 h (C-index: 0.60; odds ratio: 1.04 for each 10 l/min improvement [95% CI: 1.01 to 1.07]; p < 0.01). Figure 2 shows the relationship divided into 2 groups: patients with an improvement in DI (n = 272) and those without improvement (n = 127) demonstrating an improving PEFR over time by group.
No relationship was seen between the change in DI and change in respiratory rate (i.e., the change in respiratory rate by 24 h was unrelated to the DI response at 24 h).
Similar to the overall study, the 30-day mortality and HF re-hospitalization rate in the substudy was 2.8% (n = 12 events) and 10.0% (n = 43 events), respectively, with the composite rate of 12.1% (n = 52 events) (5). Using a logistic regression model incorporating 4 variables (age, sex, treatment assignment, and PEFR), the effect of baseline PEFR (and change from baseline to 24 h in PEFR) on the 30-day composite outcome was evaluated. Each 10 l/min increase in baseline PEFR was associated with an adjusted odds ratio of 0.96 (95% CI: 0.92 to 0.98, p < 0.01), and the change from baseline to 24 h in PEFR was similar but nonsignificant (odds ratio: 0.95 for 10 l/min change over 24 h, 95% CI: 0.92 to 1.01; p = 0.1).
There was no effect of bolus (vs. no bolus) nesiritide use, baseline beta-blocker or other medication use, or other baseline characteristics (including chronic respiratory disease) on PEFR. Additionally, there was no effect on the overall results when time from presentation to the emergency department to enrollment in the trial was incorporated into the multivariable model.
This prospective substudy in AHF provides 3 novel observations: 1) PEFR shows a similar, clinically important pattern of improvement within 24 h of admission, analogous to that seen in acute respiratory diseases such as asthma or exacerbations in COPD; 2) changes in PEFR over 24 h predicted those AHF patients with a significant (moderately or markedly) improvement in their subjective DI; and 3) changes in PEFR demonstrated a small difference in respiratory function between nesiritide and placebo.
The minimal clinically important difference was postulated by Jaeschke et al. (18) to help place statistical differences in clinical context. In patients with asthma, the minimal clinically important difference for a PEFR has been proposed to be an absolute value of 18 l/min (16) or a 12% increase over baseline value (15), and for COPD trials, between 10 to 32 l/min (19–21). Hence, our observation of a mean increase of 44 l/min over 24 h falls well within this range and thus appears to reliably demonstrate an objective difference. Further studies will be required to establish whether the minimal clinically important difference in AHF should be expected to be greater, given the placebo group, on top of background medical therapy that included diuretics, also had an improvement in both PEFR and dyspnea. Additionally, PEFR should be examined over a longer duration of time (e.g., 5 days after admission) to see whether these findings correlate with other clinical improvements.
There are a number of potential mechanisms by which dyspnea may improve in patients with AHF with nesiritide, and objective measurements such as PEFR improvement may reflect 1 or more possibilities. For example, infusion of nesiritide to stable outpatients with asthma has been shown to increase forced expiratory volume in 1 s (FEV1) and forced vital capacity by 36% and 24%, respectively, indicating a significant direct bronchodilator effect of nesiritide (22). Intravenous and inhaled atrial natriuretic peptide have both been shown to improve FEV1 in patients with asthma (23,24). The significant 18 l/min PEFR improvement with nesiritide in our study is consistent with the impact of nesiritide in these studies, albeit in a different population, and should be put in context of the overall ASCEND-HF trial result. The overall mechanism of how nesiritide and other current therapies affect peak flow limitation and dyspnea improvement is poorly understood.
The measurement of the symptom of dyspnea is clinically important in AHF as it constitutes a treatment goal that is subjectively assessed by both clinician and patient on a day-to-day basis. Dyspnea is complex and has been defined as “a subjective experience of breathing discomfort that consists of qualitatively distinct sensations that vary in intensity…derived from interactions among multiple physiological, psychological, social, and environmental factors” (8). Within clinical trials, dyspnea has been challenging to measure as evident from the use of multiple different dyspnea instruments seeking to internally “validate” a specific scale without external validation of the mechanisms, psychometric properties, or objective measurement of dyspnea. Given the absence of a gold standard, clinical trials have frequently used simple 5- or 7-point Likert scales without further validation (25). Our findings, albeit only when predicting improvement (as few subjects worsened), demonstrate an objective correlate for the subjective measure of the 7-point DI.
Assessment of “soft” endpoints such as symptomatic changes used in clinical trials of AHF therapy are difficult to evaluate and deserve careful scrutiny. Until recently, the DI (as assessed by visual analogue scale or Likert scale) had yet to be linked to improved clinical endpoints such as re-hospitalization, mortality, or other established surrogates such as BNP within clinical trials (26). However, in a recent large randomized clinical trial enrolling patients with AHF, those with early dyspnea relief had a greater weight loss and a lower 30-day mortality than patients without dyspnea relief as measured by a 7-point Likert scale (27). When other surrogates such as pulmonary capillary wedge pressure are measured, they are perceived to be of clinical value since they can be re-assessed, standardized between patients, sites, and trials, and provide objective evidence of improvement. Given the disconnect between the measurement of dyspnea as a symptom and clinical outcomes, we contend that PEFR provides an improved approach for evaluating respiratory function (25). Prior studies evaluating respiratory flow limitation have demonstrated a relationship between posture and airflow limitation consistent with our overall results and may help explain the relationship between dyspnea, posture, pulmonary edema, and PEFR (28). Further work should refine the role of the provocative dyspnea test as a clinical trial endpoint (3) in the context of objective measures of respiratory function including PEFR and FEV1. In order for PEFR to be effectively used as a surrogate in a large clinical trial, further efforts to define the pathophysiology of peak flow limitation, measurement characteristics of the instrument used (e.g., visual analogue scale), and susceptibility of symptom measures to small variations in positioning are required (29).
Our study has both limitations and strengths. First, PEFR was used rather than FEV1. FEV1 has been shown to be an independent predictor of mortality (30) and have a modest improvement in response to diuresis (31). PEFR was selected because we sought to test a simple, portable, reproducible, and inexpensive measurement that could be applied in the acute setting across multiple sites with minimal additional training. FEV1 and forced vital capacity are also now available in portable units (albeit at a greater expense), and both metrics are used clinically in acute respiratory disease, with PEFR favored for asthma and FEV1 for COPD. Although FEV1 and PEFR are correlated, PEFR underestimates the degree of respiratory impairment (32), and further studies should test both PEFR and FEV1 in the same cohort to determine the relative merits of each measure. The baseline PEFR of 230 l/min in our study is lower than previously reported in a stable outpatient HF population of 420 l/min, likely reflecting the acute presentation of these patients with symptoms of dyspnea (33)—further PEFR data from a broad AHF cohort with repeated measures would aid the generalizability of our study. Second, our substudy was of modest size, but similar to the size of many phase II trials, and the largest study of PEFR in patients with AHF. This further highlights the modest sample size required for demonstrating change in a clinical study with PEFR. Further exploration of the dosing and timing of diuretics will need to be explored in the overall trial as well as in this cohort to examine what effect, if any, is present. Third, the DI used in ASCEND-HF had few patients who worsened, and significant interpatient variability existed, requiring the creation of a model highlighting moderate or marked improvement prediction rather than using the entire scale. Further work will be required to understand the patient population who deteriorated, and what, if any, subjective or objective measurement should be used to predict this population.
PEFR is a useful objective measure that improves over the first 24 h of treatment for AHF. In this substudy of the ASCEND-HF trial, the improvement in PEFR was greater with nesiritide compared with placebo, and PEFR was linked to the primary clinical outcome at 30 days. However, the change in the DI was only modestly correlated with the change in PEFR. Additional validation in cohorts using multiple patient-reported symptom instruments and serial measures of PEFR are required for the demonstration of efficacy before use as an endpoint in clinical trials.
Funding was provided by Janssen Inc. to support this substudy, and by Johnson and Johnson for the overall ASCEND-HF trial. Drs. Ezekowitz, McMurray, and Armstrong have received research grants from Scios Inc., Ortho-Biotech, Johnson & Johnson, and Jansen Ortho Inc. in conjunction with Duke Clinical Research Institute. Dr. Hernandez has received research support from Johnson & Johnson, Proventys, and Amylin, as well as honoraria from Amgen and Corthera. Dr. O'Connor has received research funding from JNJ. Dr. Califf has received research grants from Johnson & Johnson; and consulting fees from Johnson & Johnson. All other authors have stated that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute decompensated heart failure
- B-type natriuretic peptide
- confidence interval
- chronic obstructive pulmonary disease
- Dyspnea Index
- forced expiratory volume in 1 s
- heart failure
- peak expiratory flow rate
- Received August 11, 2011.
- Revision received November 26, 2011.
- Accepted November 29, 2011.
- American College of Cardiology Foundation
- Mebazaa A.,
- Pang P.S.,
- Tavares M.,
- et al.
- Pang P.S.,
- Cleland J.G.,
- Teerlink J.R.,
- et al.
- Gheorghiade M.,
- Follath F.,
- Ponikowski P.,
- et al.
- Quanjer P.H.,
- Lebowitz M.D.,
- Gregg I.,
- Miller M.R.,
- Pedersen O.F.
- Santanello N.C.,
- Zhang J.,
- Seidenberg B.,
- Reiss T.F.,
- Barber B.L.
- Szafranski W.,
- Cukier A.,
- Ramirez A.,
- et al.
- Vestbo J.,
- Pauwels R.,
- Anderson J.A.,
- Jones P.,
- Calverley P.
- Allen L.A.,
- Hernandez A.F.,
- O'Connor C.M.,
- Felker G.M.
- Metra M.,
- O'Connor C.M.,
- Davison B.A.,
- et al.
- Duguet A.,
- Tantucci C.,
- Lozinguez O.,
- et al.
- Allen L.A.,
- Metra M.,
- Milo-Cotter O.,
- et al.