Respiratory Medicine 148 (2019) 43–48

Contents lists available at ScienceDirect

Respiratory Medicine
journal homepage: www.elsevier.com/locate/rmed

Monitoring peak expiratory flow could predict COPD exacerbations: A T

prospective observational study
Jie Cena,b, Hongying Maa, Zhongbo Chena, Lei Wengb,∗∗,1, Zaichun Denga,∗,1
a
Department of Respiratory Medicine, The Affiliated Hospital of Medical School of Ningbo University, Ningbo, Zhejiang, China
b
Department of Respiratory Medicine, Ningbo No. 9 Hospital, Ningbo, Zhejiang, China

ARTICLE INFO ABSTRACT

Keywords: Background: Exacerbation of chronic obstructive pulmonary disease (ECOPD) is an important event during the
Chronic obstructive pulmonary disease course of the disease. It causes a more rapid decline in lung function, which is associated with hospitalization
PEF and the risk of death. Therefore, it is essential to discover approaches to early detection and prevention of
Predict ECOPD. Peak expiratory flow (PEF) can be safely used instead of spirometry which can assess the severity of
COPD as a standard tool. We hypothesized that monitoring PEF could possibly be used to predict the ECOPD.
Method: To verify this hypothesis, daily morning PEF was monitored for 6 months in 53 patients with moderate
to severe COPD (mean FEV1 31.53%predicted) who were enrolled in Ningbo, China.
Result: A total of 69 exacerbations of COPD (63 of gradual onset, six of sudden onset) were recorded in this
study. Thirty cases (43.5%) of gradual onset exacerbations needed to be hospitalized, and the mean PEF sig-
nificantly decreased (vs baseline) during the 5 days that preceded those exacerbations (from 161.9 ± 39.4 L/
min to 137.9 ± 36.1 L/min, P < 0.05, statistical power = 0.92). However, this was not the case with non-
hospitalized exacerbations (from 175.4 ± 42.5 L/min to 161.5 ± 39.3 L/min, P = 0.172, statistical
power = 0.63). The ROC analysis demonstrated that 24 h before hospitalized exacerbation, the optimal cutoff
value of ΔPEF for its prediction was 28 L/min (17% from baseline), with a sensitivity and specificity of 76.7%
and 72.7%, respectively (area under the curve [AUC] = 0.84, P < 0.05, statistical power = 0.78). While 48 h
before hospitalized exacerbation, the optimal cutoff value of ΔPEF for its prediction was 14 L/min (9% from
baseline), with a sensitivity and specificity of 86.7% and 66.7%, respectively (AUC = 0.863, P < 0.05, statis-
tical power = 0.87).
Conclusions: As a rapid, inexpensive method, PEF could be used for the prediction and early detection of hos-
pitalized exacerbation of COPD. This may provide opportunity for early intervention of ECOPD.

1. Introduction hospitalization results in reduced quality of life, and increased risk of

Chronic obstructive pulmonary disease (COPD) is currently the third
leading cause of death worldwide and has become an important public
health problem. It is characterized by persistent respiratory symptoms
and airflow limitation, and is both a treatable and preventable disease
[1]. Exacerbation of chronic obstructive pulmonary disease (ECOPD) is
an important event during the course of the disease that can cause a
more rapid decline in lung function [2,3], which is associated with
increased frequency of exacerbations and hospitalization [4], and
subsequent risk of death [5,6]. In addition, ECOPD leading to
readmission. Therefore, the prevention and treatment of ECOPD is an
important step to limit disease progression [7].
Spirometry is a standard tool used for the clinical assessment of
COPD. However, this tool is inconvenient for patients with COPD to
master, particularly for long-term tracking and dynamic observation.
Peak expiratory flow (PEF) can be safely used instead of spirometry as
an inexpensive and easier method [8–12]. Previously, PEF was often
used to monitor patients with asthma, and was able to detect a reduc-
tion in PEF values before the onset of asthma exacerbations [13]. We
hypothesized that these reductions could be detected before ECOPD,

Abbreviations: ROC, receiver operating characteristic; Δ PEF, change in PEF relative to baseline
∗
Corresponding author. Department of Respiratory Medicine, the Affiliated Hospital of Medical School of Ningbo University, 247 Renmin Road, Ningbo, 315020,
China.
∗∗
Corresponding author. Department of Respiratory Medicine, Ningbo No. 9 Hospital, 68 Xiangbei Road, Ningbo, 315020, China.
E-mail addresses: wengjensen@sina.com (L. Weng), dzcrespiratory123@163.com (Z. Deng).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.rmed.2019.01.010
Received 12 October 2018; Received in revised form 11 January 2019; Accepted 12 January 2019
Available online 24 January 2019
0954-6111/ © 2019 Elsevier Ltd. All rights reserved.
J. Cen et al. Respiratory Medicine 148 (2019) 43–48

Fig. 1. Profile of the patients and their exacerbations recorded in the study.

and this may provide a predictive indicator for early detection and
prevention before the onset of ECOPD. The present study was con-
ducted to monitor daily morning PEF of a cohort of patients with
moderate to severe COPD over 6 months, to determine whether it could
be useful for the prediction of ECOPD.
2. Methods
2.1. Study population
increase in respiratory symptoms, both major (dyspnea, sputum puru-
lence, or sputum volume) and minor (wheeze, sore throat, cold, or
cough) symptoms were also recorded at the same time.
Exacerbation was defined as an increase in respiratory symptoms for
two consecutive days, including at least one major symptom and an-
other major or minor symptom. The first of the two consecutive days
was defined as the day of exacerbation onset. The above definition is
consistent with that used in a previous study [14].

2.3. Statistical analysis

This observational, prospective study was conducted over 6 months
between November 1, 2017 and May 1, 2018 in China. Patients were
Data were analyzed using the SPSS (version 23) software and the
recruited from the Affiliated Hospital of Medical School of Ningbo
statistical power was calculated by the Power Analysis and Sample Size
University and Ningbo No. 9 Hospital from September 1 to November 1,
software (PASS) (version 16). Based on the normality of data distribu-
2017. Inclusion criteria were: (1) a clinical diagnosis of COPD ac-
tion, all quantitative variables were presented as mean and standard
cording to the Global Initiative for Chronic Obstructive Lung Disease
deviation (SD), or medians and interquartile range (IQR); whereas
(GOLD) guidelines [1]; (2) age > 50 years; (3) in a stable stage of the
qualitative data were presented as percentages. The analyses were
disease, with no history of exacerbations over the previous 8 weeks.
performed by Mann–Whitney U test or Student's t-test, and Pearson's
Patients with any other significant respiratory diseases (such as asthma,
chi-squared test or Fisher's test. A P value of < 0.05 was regarded as
bronchiectasis, or pulmonary fibrosis), and those unable to complete
statistically significant. The discriminating power of the change in PEF
the study were excluded. At recruitment, the following baseline data
to predict hospitalized exacerbation was analyzed by receiver operating
were collected from all patients: age, sex, body mass index (BMI),
characteristic (ROC) curves. Data were analyzed from two periods: the
spirometric data, history of sputum production, blood gas analysis,
baseline period (5 consecutive days within stable stage of COPD), and
smoking history, and comorbidities. The study was approved by the
another period of 5 days preceding ECOPD. Changes in PEF during the
hospital's Medical Ethics Committee. All patients gave written informed
days that preceded ECOPD were compared with the baseline period. To
consent after understanding the nature and purpose of the study.
describe day-to-day PEF variation at baseline, a coefficient of variation
(CV) of morning PEF (%) was calculated as follows: (SD of morning PEF
2.2. Monitoring and definition of exacerbations values)/(mean of morning PEF values) × 100 for each patient, aver-
aged over at least 5 days. [15].
All patients were instructed to use the Mini-Wright peak flow meter
(Keka, Shanghai) correctly. Their daily 8am PEFs were monitored tel-
ephonically, prior to the use of any respiratory medication.
The highest of three values was recorded for further analysis. Any

44
J. Cen et al. Respiratory Medicine 148 (2019) 43–48

Table 1 Change in PEF before hospitalized and non-hospitalized ex-

Clinical characteristics of patients in the baseline. acerbation. Fig. 3 shows the time course of mean PEF preceding gra-
Parameter Result dual onset exacerbation. The mean PEF decreased significantly (vs.
baseline) during the 5 days before hospitalized exacerbation (from
Age, years(SD) 71.51 ± 8.47 161.9 ± 39.4 L/min to 137.9 ± 36.1 L/min, P < 0.05, statistical
Gender(males), n% 48(90.6)
power = 0.92). However, no significant decline in PEF was observed
BMI, kg/m2 (SD) 21.27 ± 4.15
smoking pack-years (IQR) 30.0(17.5–40)
preceding non-hospitalized exacerbation (from 175.4 ± 42.5 L/min to
current smokers, n% 21(39.6) 161.5 ± 39.3 L/min, P = 0.172, statistical power = 0.63).
History of chronic sputum production, n% 33(62.3)
History of cardiovascular disease, n% 22(41.5) 3.4. The predictive value of change in PEF for hospitalized exacerbation
FEV1, %predicted (SD) 31.53 ± 10.37
FVC, %predicted (IQR) 45.0(38.55–59.45)
FEV1/FVC (SD) 50.34 ± 9.68 Fig. 4 and Table 2 show the change in PEF (ΔPEF) at 24-h and 48-h
Baseline PEF, L/min (SD) 185.9 ± 51.86 predictions before hospitalized exacerbation (in gradual onset exacer-
PaO2, kPa (IQR) 10.45(8.62–11.70) bation). The mean ΔPEF of hospitalized exacerbation and that of non-
PaCO2, kPa (IQR) 5.87(5.31–6.71)
hospitalized exacerbation were compared by ROC analysis. At 24 h
Day-to-day PEF variation in baseline,% (SD) 3.3 ± 1.4
GOLD stage, n%
before hospitalized exacerbation, the optimal cutoff value of ΔPEF for
I 0(0) its prediction was 28 L/min (17% from baseline), with a sensitivity and
II 4(7.5) specificity of 76.7% and 72.7%, respectively. The ROC area under the
III 26(49.1) curve (AUC) was 0.84 (95% CI 0.744–0.935, P < 0.05, statistical
IV 23(43.4)
power = 0.78). At 48 h before hospitalized exacerbation, the optimal
GOLD category, n%
A 8(15.1) cutoff value of ΔPEF for its prediction was 14 L/min (9% from base-
B 2(3.8) line), with a sensitivity and specificity of 86.7% and 66.7%, respec-
C 27(50.9) tively. The ROC AUC was 0.863 (95% CI 0.776–0.950, P < 0.05, sta-
D 16(30.2) tistical power = 0.87).
BMI: body mass index; FEV1: forced expiratory volume in 1 s (liters); FVC:
forced vital capacity (liters); PEF: peak expiratory flow(L/min); PaO2: oxygen 4. Discussion
partial pressure measured by arterial blood gas analysis; PaCO2: carbon dioxide
partial pressure measured by arterial blood gas analysis; GOLD: Global In the present study, we monitored the daily PEF from baseline to
Initiative for Chronic Obstructive Lung Disease. exacerbation and prospectively evaluated whether this monitoring
could be used to predict ECOPD. We detected a significantly reduction
3. Results in PEF over 5 days before hospitalized exacerbation (in gradual onset
exacerbation). The ROC curves showed the predictive value of PEF for
3.1. Patient characteristics and baseline PEF variability hospitalized exacerbation at 24 h and 48 h before it occurred. Thus, this
study has demonstrated that monitoring PEF facilitates the early iden-
Sixty-one patients were enrolled initially and during the follow-up, tification of ECOPD leading to hospitalization (in gradual onset ex-
eight patients withdrew voluntarily. Thus, 53 patients were included in acerbation). This finding is significant because it presents the possibility
the final analysis (Fig. 1). The median adherence rate was 90%. Base- of early intervention in ECOPD.
line characteristics of the included subjects are shown in Table 1. Of the A previous study has shown that early detection and early therapy
53 patients, 48 (90.6%) were male, and five (9.4%) were female. Their can improve the outcome of COPD exacerbation and reduce the risk of
mean age was 71.5 ± 8.47 years, mean FEV1% predicted was hospitalization [16]. However, patients often do not report to health
31.53 ± 10.37%, mean baseline PEF was 185.91 ± 51.86 L/min, and care professionals for treatment, owing to a poor understanding of their
day-to-day PEF variation in baseline was 3.3%. The BMI, smoking condition [17]. This has a negative impact on the prognosis of ECOPD.
status, GOLD stage, GOLD category, and history of chronic sputum Thus, an objective self-monitoring tool, such as that employed in the
production of all subjects are presented in Table 1. present study, which can predict the onset of ECOPD will likely fulfill
an important clinical need.
To date, few studies have attempted to predict ECOPD, and there is
3.2. Exacerbations
still a lack of effective assessment tools to achieve quantitative pre-
diction. To our knowledge, this is the first study to effectively accom-
Fig. 1 shows that 53 patients completed the study and were further
plish both the prediction and early detection of ECOPD by monitoring
analyzed. Among them, 35 patients had one or more exacerbations. A
daily PEF. The PEF is one of the most commonly used measures to check
total of 69 exacerbations, which were identified from the symptom
lung function. Studies have shown that PEF has a good correlation with
data, were recorded. Among them, 63 exacerbations (91%) were of
FEV1, and can be used as an indicator of airflow obstruction when FEV1
gradual onset and six exacerbations (9%) were of sudden onset. Among
cannot be measured [18]. It is an inexpensive, reliable method by
the gradual-onset exacerbations, 30 (48%) needed to be hospitalized,
which patients can assess lung function at home on a daily basis.
and another 33 (52%) did not lead to hospitalization. All sudden-onset
In this study, we found two distinct patterns of exacerbations:
exacerbations led to hospitalization.
sudden and gradual onset. This is consistent with the outcome of a
previous study, in which 212 patients with COPD were monitored for
3.3. Performance of PEF before exacerbation 2.8 years by observing daily symptom scores [19]. The main difference
between the two studies stems from our detection of these patterns by
Change in PEF before different patterns of COPD exacerbation observation of the change in PEF. Accordingly, our study complements
onset. The time course of the daily median percentage change in PEF existing researches.
from −10 d to onset of exacerbation is depicted in Fig. 2. Daily median Seemungal et al. [14] performed a study to monitor 101 COPD
PEF% began to decrease gradually 5 days (Mean ± SD = 3.22 ± 1.45 patients for 2.5 years by observing daily PEF and symptoms. The results
days) before gradual onset exacerbation. In cases of sudden onset ex- showed increased incidence of respiratory symptoms (dyspnea, sore
acerbation, the daily median PEF% decreased suddenly at day 0, but no throat, cough, and cold), without any significant decline in PEF over 7
significant changes were observed before that time. days before the onset of exacerbation. At exacerbation, PEF showed an

45
J. Cen et al. Respiratory Medicine 148 (2019) 43–48

Fig. 2. Median peak flow expressed as a percentage of baseline peak flow from −10 d to day 0. Daily median PEF％ decreased gradually before gradual onset
exacerbation and fell from baseline to day 0 at sudden onset exacerbation. (Day 0 marks the day of the onset of an exacerbation).

Fig. 3. Box-plot of mean PEF before hospitalized exacerbation (n = 30) and non-hospitalized exacerbation (n = 33) during follow up. (HE: hospitalized exacerbation,
NHE: non-hospitalized exacerbation).

Fig. 4. Receiver operating characteristic (ROC) curves of ΔPEF for the prediction of hospitalized exacerbation 24 h and 48 h before its occurrence (in gradual onset
exacerbation). Areas under the ROC curve at 24 and 48 h were 0.84 (95% CI 0.744–0.935, P < 0.05, statistical power = 0.78) and 0.863(95% CI 0.776–0.950,
P < 0.05, statistical power = 0.87), respectively.

evident decline from baseline, which was associated with increased
incidence of symptoms (dyspnea, cold, and wheeze). Our study re-
ported the novel finding that PEF began to decrease 5 days before
gradual onset exacerbation, and showed significant reduction during
the 5 days before hospitalized exacerbation.
Van den Berge et al. [20] reported increased symptoms and lower
PEF preceding ECOPD, associated with a high risk of severe exacerba-
tion. Yet, their ROC curves showed predictive values that were too low
to be suitable for clinical application. Those results are partially in line
with our study, because we have demonstrated the significant pre-
dictive ability of PEF for hospitalized exacerbation, which was eval-
uated by the AUC. Generally, an AUC of ≥0.7 is considered useful for
assessment.
In the present study, the day-to-day PEF variation at baseline was
3.3%. According to previously published literature, between-day var-
iation in PEF above 8% is considered abnormal. Among 20–70 year-old
subjects with respiratory symptoms, a between-day PEF variation of
2.57–3.16% has been reported [21]. Higgins BG [15] reported the

46
J. Cen et al. Respiratory Medicine 148 (2019) 43–48

Table 2 Author contributions

Predictive performance of PEF before hospitalized exacerbation.
Cutoff value (L/min) Sensitivity% Specificity% PPV% NPV% Jie Cen contributed to the study design and drafting of the manu-
script; Jie Cen, Hongying Ma, Zhongbo Chen contributed to data col-
24 h△ PEF≥ 18.5 100.00% 36.40% 58.80% 100.00% lection, analysis and interpretation; Lei Weng contributed to conception
24 h△ PEF≥ 22.5 80.00% 63.60% 66.60% 77.80%
and design of the study; Zaichun Deng contributed to the study concept
24 h△ PEF≥ 28 76.70% 72.70% 71.80% 77.50%
24 h△ PEF≥ 35 60.00% 90.90% 85.70% 71.40%
and design, editing of the manuscript and approval of the final version.
24 h△ PEF≥ 45 43.30% 97.00% 92.90% 65.30%
48 h△ PEF≥ 11 93.30% 60.60% 68.30% 90.90% Conflicts of interest
48 h△ PEF≥ 14 86.70% 66.70% 70.30% 84.70%
48 h△ PEF≥ 18 76.70% 72.70% 71.80% 77.50%
We declare that we have no financial and personal relationships
48 h△ PEF≥ 23.5 53.30% 90.90% 84.20% 68.20%
48 h△ PEF≥ 31.5 40.00% 97.00% 92.40% 64.00% with other people or organizations that can inappropriately influence
our work, there is no professional or other personal interest of any
ΔPEF = change in PEF relative to baseline; PPV: positive predictive value; NPV: nature or kind in any product, service and/or company that could be
negative predictive value. construed as influencing the position presented in, or the review of, the
manuscript entitled.
median CV of PEF in random subjects, and in subjects with wheeze as
4% and 5.9%, respectively. Among 22–58 year-old workers, in which Acknowledgements
PEF was measured two to five times a day, a mean day-to-day variation
of 5.1–5.4% has been reported [22]. The level of PEF variability ob- We appreciate all participants who took part in our study.
served in the present study is relatively lower than that of previous
studies. This may be explained by the fact that PEF was only measured References
in the morning during our study; whereas most other reports of PEF
variability were based on measurements taken twice daily, in the [1] C.F. Vogelmeier, G.J. Criner, F.J. Martinez, et al., Global strategy for the diagnosis,
morning and in the evening at bedtime. In our study, day-to-day PEF management, and prevention of chronic obstructive lung disease 2017 report.
GOLD executive summary, Am. J. Respir. Crit. Care Med. 195 (5) (2017 Mar 1)
variation at baseline was relatively stable, and a reduction of about
557–582.
9%–17% from baseline PEF change should cause high concern. [2] G.C. Donaldson, T.A. Seemungal, A. Bhowmik, et al., Relationship between ex-
However, this prediction does not apply to all the exacerbations, acerbation frequency and lung function decline in chronic obstructive pulmonary
including sudden onset exacerbation and non-hospitalized gradual disease, Thorax 57 (10) (2002 Oct) 847–852.
[3] R.E. Kanner, N.R. Anthonisen, J.E. Connett, et al., Lower respiratory illnesses pro-
onset exacerbation. In sudden onset exacerbation, PEF declined sharply mote FEV(1) decline in current smokers but not ex- smokers with mild chronic
at exacerbation, without any prior significant changes. Understandably, obstructive pulmonary disease: results from the lung health study, Am. J. Respir.
there may be insufficient time to predict and effectively intervene. Crit. Care Med. 164 (3) (2001 Aug 1) 358–364.
[4] H. Mullerova, D.J. Maselli, N. Locantore, et al., Hospitalized exacerbations of
Similarly, no significant decline in PEF was observed preceding non- COPD: risk factors and outcomes in the ECLIPSE cohort, Chest 147 (4) (2015 Apr)
hospitalized exacerbation. Therefore, in present study, 56.5% of the 999–1007.
exacerbations were not predicted by PEF decline. Among the gradual [5] A. Agusti, P.M. Calverley, B. Celli, et al., Characterisation of COPD heterogeneity in
the ECLIPSE cohort, Respir. Res. 11 (2010 Sep 10) 122.
onset exacerbations leading to hospitalization, 43.5% could have been [6] J.B. Soriano, B. Lamprecht, A.S. Ramirez, et al., Mortality prediction in chronic
predicted by PEF monitoring. In this case, our approach can provide obstructive pulmonary disease comparing the GOLD 2007 and 2011 staging sys-
sufficient time and quantitative indicators for early intervention. tems: a pooled analysis of individual patient data, Lancet Respir Med 3 (6) (2015
Jun) 443–450.
We acknowledge that our study has some limitations. First, our
[7] M.Y. Dai, J.P. Qiao, Y.H. Xu, et al., Respiratory infectious phenotypes in acute
sample size was relatively small, based only on two hospitals. exacerbation of COPD: an aid to length of stay and COPD Assessment Test, Int. J.
Therefore, larger studies are needed to replicate and validate our Chronic Obstr. Pulm. Dis. 10 (2015 Oct 20) 2257–2263.
[8] J.L. Kelly, O. Bamsey, C. Smith, et al., Health status assessment in routine clinical
findings. In the future, we plan to perform larger multi-centric studies.
practice: the chronic obstructive pulmonary disease assessment test score in out-
Second, the patients selected for this study were those with moderate to patients, Respiration 84 (3) (2012) 193–199.
very severe stage of the disease, to the absence of those with mild-stage [9] N. Kokturk, H. Turktas, P. Kara, et al., A randomized clinical trial of magnesium
disease who might have yielded a lower exacerbation count. Third, sulphate as a vehicle for nebulized salbutamol in the treatment of moderate to
severe asthma attacks, Pulm. Pharmacol. Therapeut. 18 (6) (2005) 416–421.
most of the patients included in the study were male patients, who [10] D. Agarwal, P.P. Gupta, A comparison of peak expiratory flow measured from
account for a major proportion of COPD patients owing to their forced vital capacity and peak flow meter manoeuvres in healthy volunteers, Ann.
smoking habits. This may affect generalizability of our results to female Thorac. Med. 2 (3) (2007 Jul) 103–106.
[11] R.R. Kodgule, V. Singh, R. Dhar, et al., Reference values for peak expiratory flow in
patients. In future research, we aim to expand the population char- Indian adult population using a European Union scale peak flow meter, J. Postgrad.
acteristics and extend the observation time. Med. 60 (2) (2014 Apr-Jun) 123–129.
[12] A. Lahdensuo, T. Haahtela, J. Herrala, et al., Rando- mised comparison of guided
self-management and traditional treatment of asthma over one year, BMJ 312
(7033) (1996 Mar 23) 748–752.
5. Conclusion [13] A.E. Tattersfield, D.S. Postma, P.J. Barnes, et al., Exacerbations of asthma: a de-
scriptive study of 425 severe exacerbations. The FACET International Study Group,
Am. J. Respir. Crit. Care Med. 160 (2) (1999 Aug) 594–599.
This was an observational, prospective study, which indicated that
[14] T.A. Seemungal, G.C. Donaldson, A. Bhowmik, et al., Time course and recovery of
PEF monitoring could be useful for the prediction of ECOPD leading to exacerbations in patients with chronic obstructive pulmonary disease, Am. J.
hospitalization. Moreover, PEF declines by about 9% to 17% at least Respir. Crit. Care Med. 161 (5) (2000 May) 1608–1613.
[15] B.G. Higgins, J.R. Britton, S. Chinn, et al., The distribution of peak expiratory flow
48 h before hospitalization, which should cause our concern. As it is a
in a population sample, Am. Rev.Respir. 140 (5) (1989 Nov) 1368–1372.
convenient, rapid method, PEF could find wide application in clinical [16] T.M. Wilkinson, G.C. Donaldson, J.R. Hurst, et al., Early therapy improves outcomes
practice. Further larger multi-centric studies are required to validate of exacerbations of chronic obstructive pulmonary disease, Am. J. Respir. Crit. Care
these results. Med. 169 (12) (2004 Jun 15) 1298–1303.
[17] T.A. Seemungal, G.C. Donaldson, E.A. Paul, et al., Effect of exacerbation on quality
of life in patients with chronic obstructive pulmonary disease, Am. J. Respir. Crit.
Care Med. 157 (5 Pt 1) (1998 May) 1418–1422.
Financial/nonfinancial disclosures [18] C.L. Emerman, R.K. Cydulka, Use of peak expiratory flow rate in emergency de-
partment evaluation of acute exacerbation of chronic obstructive pulmonary dis-
ease, Ann. Emerg. Med. 27 (2) (1996 Feb) 159–163.
There is no potential conflict of interest. [19] S.D. Aaron, G.C. Donaldson, G.A. Whitmore, et al., Time course and pattern of

47
J. Cen et al.
COPD exacerbation onset, Thorax 67 (3) (2012 Mar) 238–243.
[20] M. Van den Berge, W.C. Hop, T. van der Molen, et al., Prediction and course of
symptoms and lung function around an exacerbation in chronic obstructive pul-
monary disease, Respir. Res. 13 (2012 Jun 6) 44.
[21] H.M. Boezen, J.P. Schouten, D.S. Postma, et al., Relation between respiratory
48
Respiratory Medicine 148 (2019) 43–48
symptoms, pulmonary function and peak flow variability in adults, Thorax 50 (2)
(1995 Feb) 121–126.
[22] M. Zureik, R. Liard, C. Segala, et al., Peak expiratory flow rate variability in po-
pulation surveys. Does the number of assessments matter? Chest 107 (2) (1995 Feb)
418–423.