Original Article 2012 NRITLD, National Research Institute of Tuberculosis and Lung Disease, Iran ISSN: 1735-0344
Tanaffos 2012; 11(1): 49-54
TANAFFOS
Logistic Regression Model for Prediction of Airway Reversibility Using Peak Expiratory Flow Javad Shakeri, Omalbanin Paknejad, Keivan Gohari Moghadam, Maryam Taherzadeh Department of Pulmonary Medicine, Shariati Hospital, Tehran University of Medical Sciences, Tehran, Iran. Received: 25 October 2011 Accepted: 30 December 2011 Correspondence to: Paknejad O Address: Department of Pulmonary Medicine, Shariati Hospital, Nasr Bridge, Jalal Al Ahmad Highway, Tehran, Iran Email address:
[email protected] Background: Using peak expiratory flow (PEF) as an alternative to spirometry parameters (FEV1 and FVC), for detection of airway reversibility in diseases with airflow limitation is challenging. We developed logistic regression (LR) model to discriminate bronchodilator responsiveness (BDR) and then compared the results of models with a performance of >18%, >20%, and >22% increase in ΔPEF% (PEF change relative to baseline), as a predictor for bronchodilator responsiveness (BDR). Materials and Methods: PEF measurements of pre-bronchodilator, postbronchodilator and ΔPEF% of 90 patients with asthma (44) and chronic obstructive pulmonary disease (46) were used as inputs of model and the output was presence or absence of the BDR. Results: Although ΔPEF% was a poor discriminator, LR model could improve the accuracy of BDR. Sensitivity, specificity, positive predictive value, and negative predictive value of LR were 68.89%, 67.27%, 71.43%, and 78.72%, respectively. Conclusion: The LR is a reliable method that can be used clinically to predict BDR based on PEF measurements.
Key words: Peak expiratory flow; Spirometry; Airway reversibility; Logistic regression INTRODUCTION
capacity. However, spirometry is not widely available and
Asthma and chronic obstructive pulmonary disease
its use is limited in primary care clinics (4-9). Moreover, a
(COPD) are respiratory diseases that are increasing in rate
general practitioner, who may not have access to a
all over the world especially in developing countries (1).
spirometer, visits patients with asthma or COPD. In this
Assessing the presence and degree of airflow limitation
regard, both physician and patient need a cheaper, simpler
and reversibility in airways is critical, especially in the
and more accessible method with high levels of reliability.
elderly with asthma or COPD (2, 3). Spirometry is a
Peak expiratory flow (PEF) is maximal expiratory flow that
standard tool that is used for the diagnosis of airflow
is utilized clinically in asthma monitoring (10-15). PEF
limitation
bronchodilator
measures large airways’ (diameter>2mm) function, which
responsiveness (BDR) test is carried out by performing
is effort dependent whereas FEV1 mirrors large and
baseline spirometric evaluation, with repeated spirometry
intermediate airways (15). It is assumed that FEV1 is
after administration of a short-acting bronchodilator, and
moderately dependent on PEF and there is a high
assessing the change in forced vital capacity (FVC), forced
correlation between them within and between patients (16,
expiratory volume in the first second (FEV1), or vital
17). Thus, measurement of PEF is suggested as an
and
its
severity.
The
50 Predicting Airway Reversibility
alternative for spirometry and meets those criteria (3, 4, 1820).
Confirmed asthmatic and COPD patients who had used bronchodilators and steroids and were able to do
Since bronchodilator response in asthmatic patients
lung function maneuvers were selected and entered the
occurs first in large airways and to a lesser extent with
study consecutively. Patients’ demographic data and
delay in small airways, PEF may have some correlations
clinical
with FEV1 after bronchodilator administration. Although
examinations and interview and entered into standard
there is evidence regarding the correlation of PEF and
forms. FEV1, FVC and FEV1/FVC were measured using a
spirometry parameters (16, 17), there are controversies in
Jaeger SpiroPro hand-held spirometer (Viasys Healthcare,
previous findings and this correlation is questioned in
Hoechberg, Germany) and PEF was measured by mean of
airway limitation (9, 15). We need more accurate evidence
peak flow meter (PFM20; OMRON Healthcare Ltd; UK).
regarding validity of measuring peak flow to assess airflow
As a routine activity, spirometer was calibrated each day
limitation and reversibility in asthmatics or COPD patients.
using a 3-liter syringe.
conditions
were
obtained
by
means
of
The aim of this study was to evaluate the efficacy of
Three efforts of patients (18, >20, and >22 increase in ΔPEF% in subjects correctly identified with airway reversibility was assessed. All recordings were verified by the referrer pulmonologist.
have caffeine or alcohol 4 hours before the tests.
Tanaffos 2012; 11(1): 49-54
Shakeri J, et al. 51
Logistic regression model: LR is useful for situations
(∆FVC%) and PEF (∆PEF%) after BD were 0.941±0.114 L
where researcher wants to be able to predict the presence
(12.09±1.35), 0.254±0.045 L (11.18±3.16), and 39.74±27.1
or absence of a characteristic or outcome based on values
L/min (20.49±4.88), respectively. Although ∆PEF and
of a set of predictor variables. LR regresses a dichotomous
∆PEF%
dependent (target) variable on a set of independent
changes in FEV1 and FVC (p20 for
absence of the BDR.
estimating airway reversibility was better than the other two cut-off values (>18 and >22 increase in ∆PEF%), we used ∆PEF%>20 for data analysis.
Statistical analysis: Correlation
between
changes
in
PEF
and
a
corresponding change in FEV1 or FVC were assessed by Pearson’s correlation coefficient. Accuracy (the number of correct
predictions
divided
by
total
Table 1. Comparison of predictive performance of ∆PEF% and LR for predicting BDR
predictions),
∆PEF%>18
∆PEF%>20
∆PEF%>22
LR
Accuracy (%)
56.67
63.33
55.56
68.89
Sensitivity (%)
67.27
60
43.64
67.27
Specificity (%)
40
68.57
74.29
71.43
PPV (%)
63.79
75
72.73
78.72
LR were calculated. The discriminating power of these
NPV (%)
43.75
52.17
45.61
58.14
methods to predict BDR can be measured by receiver
LR+
1.12
1.91
1.70
2.35
LR-
0.82
0.58
0.76
0.46
sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio positive (LR+) and likelihood ratio negative (LR−) of ΔPEF% and
operating characteristic (ROC) curves.
PPV positive predictive value, NPV negative predictive value, LR+ likelihood ratio positive, LR− likelihood ratio negative
RESULTS Of the 133 participants in this Study, 43 subjects were excluded because they did not meet ATS acceptability and reproducibility
criteria.
Data
from
90
subjects
(44
asthmatics and 46 COPD patients) were analyzed. The mean age of patients was 47.3±12.4 years (range 20-69 yrs). Gender distribution was 63% males and 37% females, 23.4% of patients reported being active smokers, 23.3% reported being former smokers (quitted more than a year
Figure1 depicts the ROC curves for the %∆PEF>20 and LR model. Results of the comparison of the ROC curves are shown in Table 2. ROC analysis estimates a curve that describes the inherent trade-off between sensitivity and specificity of a prediction tool. Discriminatory power is measured by area under the curve (AUC), which is a particularly important metric for evaluating prediction
ago), and 53.3% mentioned that they never smoked. BDR
tools because it is the average sensitivity over all possible
was documented by standard spirometric criteria in 55
specificities. AUC may range from 0 to 1, with area of 1.0
(61.1%) of these patients.
representing perfect discrimination and an area of 0.5
The mean values of respiratory parameters: FEV1, FVC
representing what is expected by chance alone. According
and PEF were 2.103±0.286 (L), 2.593±0.296 (L), and 296±85
to Figure 1 and Table 2, the LR model significantly
(L/min), respectively; the changes in FEV1 (∆FEV1%), FVC
outperformed the %∆PEF>20 (p20 and LR for predicting BDR
volume increase of 200ml. Dekker et al. could detect BDR (>9% increase in FEV1) in patients with asthma and COPD based on PEF (60 l/min increase in PEF) with a sensitivity,
Table 2. Comparison of ROC (receiver operating characteristic) curves
specificity and PPV of 68%, 93%, and 87%, respectively ∆PEF%>20 LR
AUC
Standard error
95% confidence interval
(27). In another study on asthmatic patients, a >18%
0.643 0.694
0.060 0.058
0.526-0.760 0.581-0.806
increase in the PEF predicted FEV1 >15% with a sensitivity, specificity, PPV and NPV of 85%, 79%, 77%, and 86%, respectively (28). Since the patients and the definition of
AUC: area under the curve
BDR were different in previous studies, it is difficult to
DISCUSSION
compare their results with our findings. However, our
In this study, we developed LR model in order to
results confirmed previous investigations regarding the
predict BDR based on PEF values in patients with asthma
fact that PEF was a poor predictor of airway reversibility.
and COPD. Although ΔPEF% was a poor discriminator,
In this research, BDR (>12% increase in FEV1 or FVC) was
the LR model could improve the accuracy of BDR finding
detected using a >20% increase in ΔPEF% with a
using PEF values (preBD and postBD) and its changes as
sensitivity, specificity, PPV and NPV of 60%, 68.57%, 75%,
inputs.
and 52.17%, respectively.
In a clinical setting, FEV1, FVC and FEV1/FVC before
The AUC is a measure of a model’s discriminatory
and after BD are the main parameters in airway
power. According to the observation by Swets et al. (29), an
reversibility
an
AUC of ≥0.7 is diagnostically useful. In our study, LR
expensive apparatus and is inaccessible in most primary
model could discriminate BDR well (Table 2). The LR
care centers and physician offices. Alternatively, we
model
suggest peak flow meter for substitution which is a
(AUC=0.694 vs. 0.643, p20
Shakeri J, et al. 53
To
our
knowledge,
there
is
no
study
using
3.
Dekker FW, Schrier AC, Sterk PJ, Dijkman JH. Validity of peak
mathematical model to reveal BDR based on PEF values.
expiratory flow measurement in assessing reversibility of
Our findings show that the LR has a good ability to predict
airflow obstruction. Thorax 1992; 47 (3): 162-6.
airway reversibility by using different values of PEF as
4.
Pellegrino R, Viegi G, Brusasco V, Crapo RO, Burgos F, Casaburi R, et al. Interpretative strategies for lung function
input.
tests. Eur Respir J 2005 Nov; 26 (5): 948-68.
The real time use of the LR is not difficult. The number of hospitals that have an electronic medical record is
5.
Miller MR, Crapo R, Hankinson J, Brusasco V, Burgos F,
increasing rapidly. Once trained, the LR could reside in the
Casaburi R, et al. General considerations for lung function
background of the clinical information systems. The data
testing. Eur Respir J 2005; 26 (1): 153-61.
used by our LR is the standard information routinely
6.
Standardization of Spirometry, 1994 Update. American
collected from the patients. Once entered into the electronic
Thoracic Society. Am J Respir Crit Care Med 1995; 152 (3):
record, these data could then be used by the LR to generate
1107-36.
the probability of the predicted outcome. LR accuracy can
7.
Spirometry in primary care practice: the importance of quality
be continuously improved over time because it can
assurance and the impact of spirometry workshops. Chest
constantly be retrained as more patients are added to the
1999; 116 (2): 416-23.
system (23, 30). Some limitations to this study need to be addressed.
8.
physicians will respond if they are provided with LR predicted outcome of BDR. However, a reliable second opinion is often helpful in medical decision making with or without a detailed understanding of how it works. Secondly, this study was carried out at a single institution. These findings must be corroborated on patients from multiple locations.
indicated
that
ΔPEF%
is
a
poor
improve the accuracy of BDR finding using PEF values, and can be used clinically as an inexpensive and rapid method.
Aggarwal AN, Gupta D, Jindal SK. The relationship between FEV1 and peak expiratory flow in patients with airways obstruction is poor. Chest 2006; 130 (5): 1454-61.
10. Expert Panel Report 2: guidelines for the diagnosis and management of asthma. Bethesda, MD: U.S. Department of Health and Human Services, Public Health Service, National
11. Practice parameters for the diagnosis and treatment of asthma. Joint Task Force on Practice Parameters, representing the American Academy of Allergy Asthma and Immunology, the American College of Allergy, Asthma and Immunology, and the Joint Council of Allergy, Asthma and Immunology. J Allergy Clin Immunol. 1995 Nov; 96 (5 Pt 2): 707-870. 12. Boulet LP, Becker A, Bérubé D, Beveridge R, Ernst P.
REFERENCES
Canadian Asthma Consensus Report, 1999. Canadian Asthma
A.S. Fauci, E. Braunwald. Harrison's Principles of internal medicin, 17th, U.S.A, MacGraw Hill. 2008. 1596-1606, 1635-42.
2.
9.
1997.
discriminator of BDR. However, the LR model could
1.
practice. Aust Fam Physician 2005; 34 (3): 201-3.
Institutes of Health, National Heart, Lung, and Blood Institute,
CONCLUSION results
Walters JA, Hansen E, Mudge P, Johns DP, Walters EH, Wood-Baker R. Barriers to the use of spirometry in general
First, the LR was not tested in real time. It is not clear how
Our
Eaton T, Withy S, Garrett JE, Mercer J, Whitlock RM, Rea HH.
Consensus Group. CMAJ 1999; 161(11): S1-61. 13. British Thoracic Society. The British guidelines on asthma
Banerjee DK, Lee GS, Malik SK, Daly S. Underdiagnosis of
management: 1995 review and position statement. Thorax
asthma in the elderly. Br J Dis Chest 1987; 81 (1): 23-9.
1997; 52:S1–21.
Tanaffos 2012; 11(1): 49-54
54 Predicting Airway Reversibility
14. Gibson PG. Monitoring the patient with asthma: an evidence-
23. Raoufy MR, Vahdani P, Alavian SM, Fekri S, Eftekhari P,
based approach. J Allergy Clin Immunol 2000; 106 (1 Pt 1): 17-
Gharibzadeh S. A novel method for diagnosing cirrhosis in
26.
patients with chronic hepatitis B: artificial neural network
15. Hegewald MJ, Lefor MJ, Jensen RL, Crapo RO, Kritchevsky SB, Haggerty CL, et al. Peak expiratory flow is not a quality indicator for spirometry: peak expiratory flow variability and FEV1 are poorly correlated in an elderly population. Chest 2007; 131 (5): 1494-9. 16. Rosenblatt G, Alkalay I, McCann PD, Stein M. The correlation of peak flow rate with maximal expiratory flow rate, onesecond forced expiratory volume, and maximal breathing
approach. J Med Syst 2011; 35 (1): 121-6. 24. Matsunaga K, Kanda M, Hayata A, Yanagisawa S, Ichikawa T, Akamatsu K, et al. Peak expiratory flow variability adjusted by forced expiratory volume in one second is a good index for airway responsiveness in asthmatics. Intern Med 2008; 47 (12): 1107-12. 25. Aggarwal AN, Agarwal R, Gupta D, Jindal SK. Use of peak expiratory flow for assessing bronchodilator responsiveness.
capacity. Am Rev Respir Dis 1963; 87: 589-91. 17. Kelly CA, Gibson GJ. Relation between FEV1 and peak expiratory flow in patients with chronic airflow obstruction.
Prim Care Respir J 2009; 18 (1): 50-2. 26. Slieker MG, van der Ent CK. The diagnostic and screening capacities of peak expiratory flow measurements in the
Thorax 1988; 43 (4): 335-6. 18. Hansen EF, Vestbo J, Phanareth K, Kok-Jensen A, Dirksen A.
assessment of airway obstruction and bronchodilator response
Peak flow as predictor of overall mortality in asthma and
in children with asthma. Monaldi Arch Chest Dis 2003 ; 59 (2):
chronic obstructive pulmonary disease. Am J Respir Crit Care
155-9.
Med 2001; 163 (3 Pt 1): 690-3.
27. Thiadens HA, De Bock GH, Van Houwelingen JC, Dekker FW,
19. Jackson H, Hubbard R. Detecting chronic obstructive
De Waal MW, Springer MP, et al. Can peak expiratory flow
pulmonary disease using peak flow rate: cross sectional
measurements reliably identify the presence of airway
survey. BMJ 2003; 327 (7416): 653-4.
obstruction and bronchodilator response as assessed by
20. Llewellin P, Sawyer G, Lewis S, Cheng S, Weatherall M, Fitzharris P, et al. The relationship between FEV1 and PEF in the assessment of the severity of airways obstruction. Respirology 2002 ; 7 (4): 333-7. 21. Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, et al. Standardisation of spirometry. Eur Respir J
FEV(1) in primary care patients presenting with a persistent cough? Thorax 1999; 54 (12): 1055-60. 28. Pino JM, García-Río F, Prados C, Alvarez-Sala R, Díaz S, Villasante C, et al. Value of the peak expiratory flow in bronchodynamic tests. Allergol Immunopathol (Madr) 1996; 24 (2): 54-7.
2005; 26 (2): 319-38. 22. Raoufy MR, Eftekhari P, Gharibzadeh S, Masjedi MR. Predicting arterial blood gas values from venous samples in
29. Swets JA. Measuring the accuracy of diagnostic systems. Science. 1988 ; 240 (4857): 1285-93.
obstructive
30. Azarkhish I, Raoufy MR, Gharibzadeh S. Artificial Intelligence
pulmonary disease using artificial neural network. J Med Syst
Models for Predicting Iron Deficiency Anemia and Iron Serum
2011; 35 (4): 483-8.
Level Based on Accessible Laboratory Data. J Med Syst 2011.
patients
with
acute
exacerbation
chronic
Tanaffos 2012; 11(1): 49-54