Quadriceps isometric strength as a predictor of exercise capacity in coronary artery disease patients

Abstract

Background

Quadriceps strength is related to exercise capacity in normal subjects and different patient populations, but the relationship between maximal quadriceps isometric strength (QIS) and different exercise capacity levels in coronary artery disease (CAD) patients has not been systematically evaluated yet.

Method

We studied 621 patients (60.6 ± 9.9 years, 538 males) with recent coronary artery bypass grafting or myocardial infarction, who underwent treadmill exercise testing, maximal QIS measurement (hand-held dynamometry), and coronary arteriography. Maximal QIS was expressed as absolute value (kg), %bodyweight, and %predicted maximum. Logistic regression was used to assess the relationship of maximal QIS, age, sex, number of diseased coronary vessels, peak systolic blood pressure, peak heart rate, brain natriuretic peptide, and left ventricular ejection fraction with 5, 7, and 10 estimated metabolic equivalents (eMETs) exercise capacity levels.

Results

Maximal QIS %bodyweight was the strongest predictor of exercise capacity in each eMETs category. Receiver-operating characteristics curves identified maximal QIS of 46, 51, and 59 % bodyweight as the best predictive cut offs for 5, 7 and 10 eMETs, respectively, with positive predictive values of 0.72, 0.66, and 0.67, respectively.

Conclusions

Maximal QIS is related with eMETs levels reached at exercise testing in CAD patients, and identified maximal QIS cut-off values for eMETs prediction may be used to set strength training goals according to patients’ needs with regard to habitual physical activity level. Hand-held dynamometry may meet the need of easiness of use and low cost required for strength evaluation in large-scale clinical trials.

Keywords

Cardiac rehabilitation coronary artery disease exercise capacity muscle strength

Introduction

Exercise capacity, as assessed by peak VO₂ or estimated metabolic equivalents (eMETs), is known to be an important predictor of physical disability and mortality in healthy individuals and patients with cardiovascular disease.^1,2 For this reason, improvement in exercise capacity is one of the main goals of cardiac rehabilitation, resulting in both a reduction of mortality risk and an increased level of everyday habitual activities.^3,4 The reduction of mortality risk for each 1-eMET increase in exercise capacity ranges between 10% and 25% and is evident in young and elderly subjects of both sexes with and without cardiovascular disease with mortality risk decreasing with increasing exercise capacity and reaching an asymptote at around 10 eMETs.²

Besides aerobic exercise, many reports document that strength training also effectively improves exercise capacity in individuals with and without cardiovascular disease.^5–8 The bulk of evidence suggests that loss of muscle strength is an important predictor of physical disability⁹ and reduced exercise capacity in cardiac patients.^11–15 Moreover, analogously to and independently of exercise capacity, muscle strength has been shown to predict mortality in both young and elderly normal and in different patient populations.^16–19 Hence, resistance training is currently accepted as a main component of comprehensive cardiac rehabilitation, leading to improvements in quality of life and raising the performance level for occupational and recreational activities.^6,20

Several papers have assessed the relationship between quadriceps strength and exercise capacity in normal subjects and chronic heart failure patients,^11,12,21 but such relationship has been addressed to date only in small groups of patients with coronary artery disease (CAD) with contradictory results.^9,13–15,22 Our working hypothesis was that a positive correlation would exist between maximal quadriceps isometric strength (QIS) and exercise capacity in CAD patients and that, if this were true, such a relationship would provide useful information in the clinical setting for a strength training prescription tailored on daily habitual activities in this population. The purpose of the present study was thus to investigate the relationship between maximal QIS and exercise capacity and to identify the levels of maximal QIS corresponding to those of given habitual activities in a large cohort of patients with CAD.

Methods

Study population

This study considered 772 consecutive patients aged 30–80 years with a recent acute cardiac event, referred to our institution for a cardiac rehabilitation programme from 1 January 2005 to 30 December 2010. Patients’ diagnosis was determined as follows: a recent coronary artery bypass grafting (CABG) replaced other possible admitting diagnoses, such as unstable angina or myocardial infarction (MI); on the other hand, a recent MI without subsequent CABG was coded as MI irrespective of possible subsequent percutaneous coronary interventions. Exclusion criteria were: (1) treadmill exercise testing stopped for reasons other than muscular fatigue; (2) orthopaedic, neurological, and/or pulmonary comorbidities limiting functional capacity; and (3) clinical instabilization. According to such criteria, 621 patients were included in the study. Of the 151 excluded patients, 111 (73%), 27 (18%), and 15 (9%) met the exclusion criteria 1, 2, and 3, respectively. The clinical characteristics of the study population are shown in Table 1.

Table 1.

Patients’ characteristics

Characteristic	Study population (n = 621)
Sex (male)	538 (87)
Age (years)	60.6 ± 9.9
Weight (kg)	63.6 ± 11.8
Body mass index (kg/m²)	23.6 ± 3.4
Diagnosis
Myocardial infarction	425 (68)
Coronary artery bypass grafting	196 (32)
Time since cardiac event (days)	56 ± 21
Brain natriuretic peptide (pg/ml)	169 ± 281
Left ventricular ejection fraction (%)	48.1 ± 10.6
Medications
Angiotensin-converting-enzyme inhibitors	254 (41)
Angiotensin II receptor blockers	310 (50)
β-Blockers	477 (77)
Maximal QIS (kg)	34.6 ± 11.4
Maximal QIS (%bodyweight)	54.4 ± 15.4
Maximal QIS (%predicted maximum)	93.8 ± 25.2
Exercise capacity (eMETs)	7.8 ± 1.5
Peak heart rate (beats/min)	138 ± 19
Peak heart rate (%predicted maximum)	87 ± 12
Peak systolic blood pressure (mmHg)	169 ± 28
Peak rate-pressure product (beats × mmHg)	23,332 ± 5372
Peak Borg scale	17.5 ± 1.1

Values are mean ± SD or n (%).

eMETs, estimated metabolic equivalents; QIS, quadriceps isometric strength.

All patients underwent an echocardiography and a selective coronary arteriography as a part of their routine clinical/instrumental evaluation. Coronary lesions were considered haemodynamically significant when they caused a 75% reduction of the coronary luminal diameter (50% reduction for the left main coronary artery). The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, and was approved by the Ethics Committee of Kitasato University Hospital. All patients gave their informed, written consent to participation in the study.

Treadmill exercise testing

All patients underwent a symptom-limited treadmill exercise testing according to the Bruce protocol. ECG was continuously recorded with an ergometric stress test system (ML-6500; Fukuda Denshi, Tokyo, Japan), and blood pressure was measured every minute by the cuff method with an automatic manometer (FB-300; Fukuda Denshi). Rating of perceived exertion (6–20 Borg scale) was assessed every 3 minutes during the exercise phase and just before its interruption. Exercise capacity was expressed in eMETs, calculated according to the following Bruce protocol-specific equation accounting for handrail use:²³ eMETs = (2.587 × TT + 6.004)/3.5, where TT is exercise time in minutes.

Maximal quadriceps isometric strength measurement

A few days after exercise testing, maximal QIS was measured by a hand-held dynamometer (µ Tas MT-1, ANIMA, Tokyo, Japan). Patients sat on a bench, and the dynamometer was fixed to a rigid bar.²⁴ Three 5-second maximal isometric voluntary contractions of the quadriceps were collected for both legs, with the knee joint angle fixed at 90° of flexion and hip joint angle set at approximately 90° of flexion, respectively. Right and left quadriceps were tested consecutively. A rest period of 30 seconds was provided between sets of bilateral contractions. Electrocardiographic data were continuously monitored via telemetry and blood pressure was evaluated manually before and immediately after strength measurements. Patients were told not to hold breathing during contractions to avoid the Valsalva manoeuver. The highest right and left side strength values were averaged and expressed as absolute value (kg), relative to body weight (%bodyweight), and relative to age- and sex-predicted values (%predicted maximum) based on available normative data.²⁵

Statistical analysis

Spearman’s rank correlation was calculated to evaluate the correlation between maximal QIS strength (absolute value, %bodyweight, and %predicted maximum) and eMETs. The maximal QIS parameter most correlated to eMETs was adopted in following analyses. Stepwise logistic regression was used to evaluate the relationship between exercise capacity categories of 5, 7, and 10 eMETs and maximal QIS, age, sex, number of diseased coronary vessels, brain natriuretic peptide level, left ventricular ejection fraction, peak systolic blood pressure, and heart rate. The above-mentioned eMETs categories were chosen as energetic costs equal to 5, 7, and 10 METs and include, respectively, most light-to-moderate, moderate-to-high, and high-to-vigorous intensity household, occupational, and sports activities.²⁶ In further analyses, we calculated the area under the curve (AUC) acquired from receiver-operating characteristics (ROC) curves to evaluate the discriminating power of maximal QIS in prediction of the three eMETs categories. The relationship between sensitivity and specificity of maximal QIS for prediction of 5, 7, and 10 eMETs attainment was examined by constructing sensitivity and specificity curves.²⁶ An optimal cut-off point maximizing sensitivity and specificity for each eMETs category was determined from the intersection of the sensitivity and specificity curves.²⁷

All analyses were carried out with the use of SPSS for Windows 20.0 statistical package (SPSS Japan, Tokyo, Japan). The level of statistical significance was set at a 2-tailed p-value of < 0.05.

Results

Clinical characteristics

The study population was composed mainly by males (Table 1). About one-half of the patients had 2/3-vessel CAD, with a left ventricular systolic function on average at lower normal limits and a quite well-preserved haemodynamic balance, as witnessed by only slightly elevated mean brain natriuretic peptide levels; most patients were on angiotensin-converting enzyme inhibitors/sartanics and/or beta-blockers (Table 1).

Maximal QIS corresponded on average to around one-half of body weight (even if with a large standard deviation) and was very close to predicted maximum (Table 1). Mean exercise capacity was >5 eMETs, attesting a good average exercise tolerance of the study population, but also in this case with a quite wide dispersion of individual values around the mean (Table 1). Mean peak heart rate expressed as a percentage of predicted maximum and mean peak rate-pressure product were consistent with significant exercise-induced whole body and myocardial stress, respectively (Table 1).

Exercise capacity predictors

Maximal QIS was significantly related to eMETs (r = 0.42 for absolute value, r = 0.43 for %bodyweight value, and r = 0.29 for %predicted maximum value, all p < 0.001). By design, maximal QIS expressed as %bodyweight was used in following analyses. At stepwise logistic regression, maximal QIS and peak heart rate were the only independent predictors of a 5 eMETs exercise capacity attainment (Table 2). On the other hand, the predictive models of 7 and 10 eMETs attainment included several other independent predictors of exercise capacity in addition to maximal QIS and peak heart rate (Table 2). Of note, maximal QIS was one of the strongest independent predictor in all eMETs predictive models (Table 2).

Table 2.

Logistic regression models for different eMETs levels

eMET level	B ± SE	p	Odds ratio (95% CI)
5 eMETs
Maximal QIS (%bodyweight)	0.082 ± 0.017	<0.001	1.086 (1.050–1.122)
Peak heart rate	0.060 ± 0.010	<0.001	1.062 (1.041–1.084)
7 eMETs
Maximal QIS (%bodyweight)	0.046 ± 0.009	<0.001	1.047 (1.029–1.066)
Peak heart rate	0.044 ± 0.006	<0.001	1.045 (1.032–1.059)
Peak systolic blood pressure	0.014 ± 0.004	0.001	1.014 (1.006–1.023)
Male sex	0.908 ± 0.300	0.002	2.480 (1.378–4.463)
Age	−0.035 ± 0.013	0.009	0.966 (0.941–0.991)
LVEF	0.025 ± 0.011	0.019	1.026 (1.004–1.048)
10 eMETs
Maximal QIS (%bodyweight)	0.030 ± 0.008	<0.001	1.030 (1.013–1.048)
Peak heart rate	0.025 ± 0.008	0.002	1.025 (1.009–1.041)
Peak systolic blood pressure	0.011 ± 0.004	0.015	1.011 (1.002–1.020)
Age	−0.029 ± 0.013	0.028	0.971 (0.947–0.997)
Male sex	0.945 ± 0.548	0.084	2.573 (0.880–7.529)

eMETs, estimated metabolic equivalents; LVEF, left ventricular ejection fraction; QIS, quadriceps isometric strength. SE, standard error; CI, confidence interval.

Validity of maximal QIS as an exercise capacity predictor

The AUC acquired from ROC curves for maximal QIS as a predictor of a 5, 7, and 10 eMETs exercise capacity attainment was 0.784 (95% CI 0.715–0.854), 0.719 (95% CI 0.672–0.767), and 0.720 (95% CI 0.634–0.806), respectively (all p < 0.001). The sensitivity and specificity curves for maximal QIS as a predictor of each eMETs level attainment are shown in Figure 1. Maximal QIS of 46, 51, and 59 %bodyweight were identified as optimal cut-off points maximizing sensitivity and specificity for prediction of 5, 7, and 10 eMETs exercise capacity attainment, respectively, with positive predictive values of 0.72, 0.66, and 0.67, respectively.

Figure 1.

Sensitivity and specificity curves for maximal quadriceps isometric strength (%bodyweight) as a predictor of eMET attainment.

Discussion

The main finding of this study was that maximal QIS is significantly related to exercise capacity in patients with CAD; as a consequence, exercise capacity, as expressed in eMETs categories, can be predicted according to maximal QIS measurement in this population. Overall, these results can help clinicians to set the goal of strength training programmes in CAD patients according to patients’ needs about household, occupational, and sports activity level and to ascertain patients’ exercise capacity when exercise testing is not available or feasible.

Maximal QIS and exercise capacity in CAD patients

It is well known that exercise capacity is related to age, physical activity level, and cardiovascular clinical status. Additionally, many reports refer to muscle mass or strength as a factor linked to exercise capacity both in healthy subjects⁵ and patients with chronic heart failure.^8,10,11,21 However, the relationship between muscle strength and exercise capacity in CAD patients has been evaluated only in small patient groups, with contradictory results.^9,13–15,22 The present study is the first o systematically examine the association between maximal QIS and exercise capacity in a large cohort of CAD patients. Our results show that maximal QIS was among the strongest independent predictors of each exercise capacity category after adjusting for several confounding factors. Of note, the logistic regression model for the 5 eMETs category included only maximal QIS and heart rate, whereas the 7 and 10 eMETs exercise capacity levels were independently predicted also by several other parameters. These findings lend support to the hypothesis that, in more unfit people (i.e. older and/or diseased), exercise capacity may be limited not only by central cardiovascular factors, but also by leg strength itself.⁵

Clinical implications of maximal QIS cut offs in CAD patients

Strength training is widely accepted as a primary component of comprehensive exercise programmes of cardiac rehabilitation. Hence, the identification of strength values corresponding to specific exercise capacity levels and leisure-time and occupational activities may be useful to properly define the goals of strength training protocols. The results of the present study show that a maximal QIS of around 45, 50, and 60 %bodyweight is needed to achieve exercise capacities of 5, 7, and 10 eMETs, respectively. As given eMETs levels can be linked to a broad range of specific household, occupational, and sports activities,²⁶ these findings underlie the possible correspondence between maximal QIS and habitual activity levels in CAD patients (Table 3). This holds true especially for the 5 eMETs category, in which the highest positive predictive value was observed. For example, a maximal QIS of 45 %bodyweight would be necessary for patients to perform most light-to-moderate intensity activities; a maximal QIS of 50 %bodyweight to perform most moderate-to-high intensity activities; and a maximal QIS of 60 %bodyweight to perform most high-to-vigorous intensity tasks (Table 3). In the light of the above, the maximal QIS cut-off values identified in this study may be used in CAD patients as targets of strength training programmes, according to individual needs about household, occupational, and sports activity levels. Moreover, the possibility to indirectly evaluate by maximal QIS measurement the eMETs level attainable at exercise testing could be useful to evaluate the exercise capacity of patients who cannot undergo formal exercise testing because of clinical conditions and to those settings that do not have exercise testing equipment available. It must be acknowledged, however, that isometric strength evaluation does not account for safety aspects that would be identified if a symptom-limited exercise test with ECG and blood pressure monitoring was conducted.

Table 3.

Association of quadriceps isometric strength with levels of exercise capacity and habitual activities

QIS (%bodyweight)	eMETs	Habitual activity intensity	Habitual activity type
45	5	Light to moderate	Cleaning, kitchen activity, laundry, machine tooling, electrical work
50	7	Moderate to high	Digging, carrying/chopping wood, mowing lawn, lifting items continuously (4.5–9 kg), walking on job (5.5 km/h) carrying objects ≤11 kg
60	10	High to vigorous	Tennis single, golf without cart, bicycling (19–22 km/h), using heavy tools, walking carrying objects (23–34 kg)

Habitual activity types reported for each quadriceps isometric strength category are not intended to be exhaustive. Ainsworth et al.²⁶ gives a complete list of activities associated with the different QIS/eMETs categories.

eMETs, estimated metabolic equivalents.

QIS measurement by hand-held dynamometer

We used a hand-held dynamometer to measure maximal QIS. Such dynamometer is widely used in clinical practice or community medicine because of its simplicity, objectivity, and responsiveness in measuring quadriceps strength, as compared to manual muscle testing or isokinetic dynamometry.^25,28,29 Isokinetic dynamometers provide accurate assessments of dynamic and static muscle strength and are usually the preferred option for clinical studies. However, their use in large-scale epidemiological studies is limited, because the equipment is costly and not portable, and, as a consequence, muscle strength measurement is often omitted.²⁹ As the normative values for hand-held dynamometry and its test–retest reliability in measurement of the isometric muscle strength have already been established in healthy persons,²⁵ the expected values of muscle strength obtained in the present study can be universally applied in many kinds of hospital and home rehabilitation settings.

Study limitations

The use of an indirect descriptor of aerobic power (eMETs) for exercise capacity assessment may be questionable, since aerobic power can be overestimated when indirectly assessed using the Bruce protocol, especially in patient populations.³⁰ However, the equation for peak METs estimation used in the current paper has been shown to provide the lowest discrepancy between estimated and actual METs, with a mean value of 0.51 METs in patients with signs and symptoms suggestive for myocardial ischemia.²³ Such a value equals a mean difference of around 1.8 ml/kg/min between the estimated mean peak VO₂ of 27.3 ml/kg/min in the study population (calculated as 7.8 eMETs × 3.5 ml/kg/min) and the actual peak VO₂, i.e. a 6% discrepancy, which seems indeed acceptable. Also the exclusion from the study of patients who stopped exercise testing for reasons other than muscular fatigue may be questioned, as this group does include patients in whom a peak exercise capacity evaluation with tools different by exercise testing would be most indicated. However, the inclusion in the study of patients likely not attaining ‘true’ maximal effort, would have significantly confounded the interpretation of the maximal QIS vs. eMETs relationship.

Conclusions

The results of this study demonstrate that maximal QIS is related to exercise capacity in patients with CAD and that maximal QIS values of 45, 50, and 60 %bodyweight can be used to predict exercise capacity levels of 5, 7, and 10 eMETs, respectively. These findings can be used in cardiac rehabilitation for the definition of strength training goals according to patients’ needs about daily activity level, and to indirectly evaluate exercise capacity when exercise testing is not available. Finally, the results of this study lend support to the use of hand-held dynamometry for maximal QIS measurement in large-scale clinical trials in CAD patients.

Footnotes

Funding

This study was supported in part by the Ministry of Education, Science and Culture in Japan (no. 15500383).

Conflict of interest

The authors declare that there is no conflict of interest.

References

Myers

Prakash

Froelicher

. Exercise capacity and mortality among men referred for exercise testing. N Engl J Med 2002; 346: 793–801.

Kokkinos

Myers

. Exercise and physical activity: clinical outcomes and applications. Circulation 2010; 122: 1637–1648.

Vanhees

Fagard

Thijs

. Prognostic value of training-induced change in peak exercise capacity in patients with myocardial infarcts and patients with coronary bypass surgery. Am J Cardiol 1995; 76: 1014–1019.

Dorn

Naughton

Imamura

. Results of a multicenter randomized clinical trial of exercise and long-term survival in myocardial infarction patients: the National Exercise and Heart Disease Project (NEHDP). Circulation 1999; 100: 1764–1769.

Vincent

Braith

Feldman

. Improved cardiorespiratory endurance following 6 months of resistance exercise in elderly men and women. Arch Intern Med 2002; 162: 673–678.

Liu

Latham

. Progressive resistance strength training for improving physical function in older adults. Cochrane Database Syst Rev 2009; 3): CD002759–CD002759.

McCartney

McKelvie

Haslam

. Usefulness of weightlifting training in improving strength and maximal power output in coronary artery disease. Am J Cardiol 1991; 67: 939–945.

Esposito

Reese

Shabetai

. Isolated quadriceps training increases maximal exercise capacity in chronic heart failure: the role of skeletal muscle convective and diffusive oxygen transport. J Am Coll Cardiol 2011; 58: 1353–1362.

Ades

Savage

Tischler

. Determinants of disability in older coronary patients. Am Heart J 2002; 143: 151–156.

10.

Harrington

Anker

Chua

. Skeletal muscle function and its relation to exercise tolerance in chronic heart failure. J Am Coll Cardiol 1997; 30: 1758–1764.

11.

Clark

Rafferty

Arbuthnott

. Relationship between isokinetic muscle strength and exercise capacity in chronic heart failure. Int J Cardiol 1997; 59: 145–148.

12.

Cicoira

Zanolla

Franceschini

. Skeletal muscle mass independently predicts peak oxygen consumption and ventilatory response during exercise in noncachectic patients with chronic heart failure. J Am Coll Cardiol 2001; 37: 2080–2085.

13.

Gayda

Merzouk

Choquet

. Aerobic capacity and peripheral skeletal muscle function in coronary artery disease male patients. Int J Sports Med 2003; 24: 258–263.

14.

Ghroubi

Chaari

Elleuch

. The isokinetic assessment of peripheral muscle function in patients with coronary artery disease: correlations with cardiorespiratory capacity. Ann Readapt Med Phys 2007; 50: 295–301.

15.

Kida

Osada

Akashi

. The exercise training effects of skeletal muscle strength and muscle volume to improve functional capacity in patients with myocardial infarction. Int J Cardiol 2008; 129: 180–186.

16.

Ruiz

Sui

Lobelo

. Association between muscular strength and mortality in men: prospective cohort study. BMJ 2008; 337: 92–95.

17.

Hülsmann

Quittan

Berger

. Muscle strength as a predictor of long-term survival in severe congestive heart failure. Eur J Heart Fail 2004; 6: 101–107.

18.

Izawa

Watanabe

Osada

. Handgrip strength as a predictor of prognosis in Japanese patients with congestive heart failure. Eur J Cardiovasc Prev Rehabil 2009; 16: 21–27.

19.

Artero

Lee

Ruiz

. A prospective study of muscular strength and all-cause mortality in men with hypertension. J Am Coll Cardiol 2011; 57: 1831–1837.

20.

Beniamini

Rubenstein

Zaichkowsky

. Effects of high-intensity strength training on quality-of-life parameters in cardiac rehabilitation patients. Am J Cardiol 1997; 80: 841–846.

21.

Izawa

Watanabe

Oka

. Upper and lower extremity muscle strength levels associated with an exercise capacity of 5 metabolic equivalents in male patients with heart failure. J Cardiopulm Rehabil Prev 2012; 32: 85–91.

22.

Kida

Osada

Akashi

. Functional capacity, skeletal muscle strength, and skeletal muscle volume in patients with myocardial infarction. Int Heart J 2006; 47: 727–738.

23.

Arena

Pinkstaff

Peberdy

. Differences in equations used to estimate aerobic capacity in patients being assessed for suspected myocardial ischemia. Am J Cardiol 2011; 108: 1198–1199.

24.

Kamiya

Mezzani

Masuda

. Effects of electrical muscle stimulation in a left ventricular assist device patient. Int J Cardiol 2012; 160: e44–e45.

25.

Bohannon

. Manual muscle test scores and dynamometer test scores of knee extension strength. Arch Phys Med Rehabil 1986; 67: 390–392.

26.

Ainsworth

Haskell

Herrmann

. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc 2011; 43: 1575–1581.

27.

Hosmer

Lemeshow

. Applied logistic regression, New York: John Wiley and Sons, 2000.

28.

Noreau

Vachon

. Comparison of three methods to assess muscular strength in individuals with spinal cord injury. Spinal Cord 1998; 36: 716–723.

29.

Martin

Yule

Syddall

. Is hand-held dynamometry useful for the measurement of quadriceps strength in older people? A comparison with the gold standard Bodex dynamometry. Gerontology 2006; 52: 154–159.

30.

Pinkstaff

Peberdy

Kontos

. Overestimation of aerobic capacity with the bruce treadmill protocol in patients being assessed for suspected myocardial ischemia. J Cardiopulm Rehabil Prev 2011; 31: 254–260.