American Journal of Respiratory and Critical Care Medicine

Owing to difficulties in measuring ventilation symmetry, good evidence of different right/left respiratory movements has not yet been provided. We investigated VT differences between paretic and healthy sides during quiet breathing, voluntary hyperventilation, and hypercapnic stimulation in patients with hemiparesis. We studied eight patients with hemiparesis and nine normal sex- and age-matched subjects. Right- and left-sided VT was reconstructed using optoelectronic plethysmography. In control subjects, no asymmetry was found in the study conditions. VTs of paretic and healthy sides were similar during quiet breathing, but paretic VT was lower during voluntary hyperventilation in six patients and higher during hypercapnic stimulation in eight patients (p = 0.02). The ventilatory response to hypercapnic stimulation was higher on the paretic than on the healthy side (p = 0.012). In conclusion, hemiparetic stroke produces asymmetric ventilation with an increase in carbon dioxide sensitivity and a decrease in voluntary ventilation on the paretic side.

Breathing can be activated volitionally through corticospinal pathways or automatically via bulbospinal pathways (13). The cerebral cortex does not contribute to respiratory drive during quiet breathing (4). In contrast, cortical and subcortical activation elicited by chemical and mechanical stimuli may result in breathing alterations (2, 57) and in its breath-to-breath variability (8, 9). Direct stimulation of the cortex in animals and humans provides evidence of the cortical influence on ventilation. The major effect of cortical stimulation on ventilation is inhibitory (10). Voluntary hyperventilation (VH) is under the control of the corticospinal pathway both for excitation (premotor potentials) (4) and inhibition (cortical silent period) (11). Chemically induced activation of cells contained within the red nucleus will elicit significant respiratory inhibition (12). Furthermore, deep nuclei within the cerebellum exert well defined modulatory influences on the respiratory responses induced by increased activation of peripheral chemoreceptors (13).

Focal destructive hemispheric lesions result in contralateral dysfunction of ventilatory muscles (1416). Using a mechanical caliper to measure anteroposterior movements of a point on each side of the thorax, Fluck (16) observed a reduction in chest wall movements on the side of the paralysis when patients took a voluntary deep breath. Employing electromyography (14) and ultrasonography (15), more recent studies have clearly shown a reduction in EMG activity and movement on the paretic side during voluntary ventilation. However, in hemiplegia, results with volitional hyperventilation differed from those with chemical stimuli. In 56 patients with acute hemispheric or brain stem infarcts, Klassen and coworkers (17) showed, compared with control subjects, a significantly higher carbon dioxide (CO2) sensitivity, apparently due to the increases in both respiratory frequency and Vt. The authors hypothesized that the increase in CO2 sensitivity was related to a loss of normal inhibitory or “damping” influences on the brain stem–mediated ventilatory response to hypercapnia. However, owing to technical difficulties, no evidence of asymmetrical ventilation of the two sides of the chest wall has yet been provided with hypercapnia, so it is still unknown whether volitional ventilation and the ventilatory response to CO2 are asymmetrical. We reasoned that if the inhibitory control of the cortex is unilateral and each cerebral hemisphere inhibits the ventilatory response to CO2 of the opposite side of the chest wall, inhibition of the CO2 ventilatory response would be less on the paretic side in patients with a unilateral hemispheric stroke, thereby resulting in a greater ventilatory response than on the healthy side. On the other hand, during VH, the ventilatory response would be lower on the paretic than on the healthy side.

To test our hypothesis and to assess how cerebrovascular disease could modify the cortical control of ventilation, we evaluated the breathing pattern of the two sides of the chest wall during chemical stimulus and VH by employing optoelectronic plethysmography (18). This is a motion analysis system that dynamically measures the three-dimensional coordinates of 89 markers applied to the surface of the chest wall to evaluate differences in reconstructed Vt between the sides of the chest wall. A parallel processing computer reconstructs the whole chest wall and measures its volume and that of its compartments and how they change with breathing. We used this system to quantify the volume changes of the left and right sides of the chest wall in patients with hemiparetic stroke.


We studied eight male patients at an average of 26 days (range 14–55) after onset of the symptom of hemiplegia due to a cerebrovascular accident. Patients' ages and computed tomography scan findings are listed in Table 1

TABLE 1. Anthropometric and clinical data of the patients

Age (yr)
Height (m)
FEV1 (% pv)
FVC (% pv)
TLC (% pv)
FRC (% pv)
Computed Tomography Scan Findings
1631.6096959381Right nuclei pontis ischemia
2561.60123111124113Right subcortical and basal ganglia ischemia
3531.75738583112Left extracapsular ischemia
4621.8082878080Left capsular ischemia
5291.78898783125Left basal ganglia ischemia
6661.6575809085Right capsulolenticular hemorrhage
7641.5711310298109Right temporotalamic hemorrhage
8461.708174*8590Left extracapsular ischemia
Control  subjects

*Value outside the 95% confidence limits for normal values.

Definition of abbreviations: pv = predicted value; TLC = total lung capacity.

. The lesions were in the area of the middle cerebral arteries (right or left) and affected the basal ganglia and the substantia alba. Tomography was not able to define exactly the magnitude of the lesions, but the clinical evidence is generally proportional to the magnitude of the lesions. For this reason, the clinical score (19) was chosen to recruit patients classified as moderate. The major neurologic manifestation in all patients was hemiplegia (four right and four left); five patients presented a mild distal hypoesthesia of the lower limbs to the knee. Two had no sensory abnormalities. All patients were fully conscious and able to understand and carry out verbal orders. The neurocognitive and logopedic evaluation showed a mild heminegligence in three left-hemiplegic patients and amnesic aphasia in two right-hemiplegic patients. Two patients were current mild smokers (< 5 pack years). No patient was overweight, the body mass index being less than 28 kg/m2. None had scoliosis or any other abnormalities of the vertebral column.

Nine normal male subjects matched for age (range 32–70 years, mean age 46.7 years ± 12.1) were studied as a control group. We selected healthy, nonsmoking volunteers from our Foundation staff (see Table 1) with no history of respiratory, cardiac, or neurologic diseases.

The study was approved by the local ethics committee, and subjects gave their informed consent.

Compartmental Volume Measurements

Kinematic analysis of the chest wall was computed by using the optoelectronic plethysmography (OEP) system. Details of the technique are reported elsewhere (18). In brief, four cameras, two 4 m in front of the subject and two 4 m behind, tracked the three-dimensional movements of 89 small surface markers attached to the skin of the trunk with double-sided adhesive tape. The markers, 5-mm hemispheres coated with reflective paper, were positioned according to the method of Cala and coworkers (18) along seven horizontal and vertical lines both anteriorly and posteriorly to the chest wall and abdomen. Their movements were tracked by the cameras that lit them through infrared light-emitting diodes coaxial with the lenses. According to Cala and coworkers (18), the positioning of the markers allows not only computation of the entire volume of the chest and abdomen beneath the markers with great accuracy but also its partitioning into right and left compartments. We used this marker configuration to define anatomically the right and left chest wall compartments and the rib cage (rc) and abdomen (ab) compartments.

Functional Evaluation

Routine spirometry, obtained with the patients seated in a comfortable armchair, was measured as previously described (20, 21). Functional residual capacity was measured by the helium dilution technique. The normal values for lung volumes were those of the European Community for Coal and Steel (22).

Maximum static inspiratory (MIP) and expiratory (MEP) pressures at functional residual capacity, measured against an obstructed mouthpiece with a small leak to minimize oral pressure artifacts, were recorded using a differential pressure transducer (Valydine, Northridge, CA). The subjects were comfortably seated, wearing a nose clip, and performed maximal inspiratory efforts, maintaining maximal pressures for at least 1 second. The maneuvers were repeated until three measurements with less than 5% variability were recorded, and the highest value obtained was used for analysis.

After baseline routine testing, the ventilatory pattern was evaluated with subjects sitting comfortably during room-air breathing and during CO2 rebreathing. In the pneumotachographic apparatus we used, the inspiratory line was separated from the expiratory one by a one-way valve (Hans-Rudolph) connected to a Fleisch type 3 pneumotachograph. The flow signal was integrated into volume. Expired CO2 was sampled continuously at the mouth by an infrared CO2 meter (Datex Normocap 200, Helsinki, Finland). Lung volumes by both pneumotacograph and OEP system were contemporaneously recorded. From both volume signals we derived the total time of the respiratory cycle, Vt, respiratory frequency, and V̇e. The Vt derived by OEP was partitioned into its right- and left-sided compartments.

The values for dead space and resistance of the system up to a flow of 4 L were 201 ml and 0.94 cm H2O/L/second, respectively.

The outputs of the CO2 meter and volume signals were recorded on a PC hard disk using an eight-channel analogic/digital board at a sampling rate of 50 Hz. After a 10-minute adaptation period, evaluation began. Signals were recorded over a 10-minute period. Details of the procedures have been described elsewhere (20, 21).


All subjects were tested in the morning and were well acquainted with the laboratory and equipment before the experiment. Changes in volume and flow were recorded during quiet breathing, during VH, and during hypercapnic–hyperoxic stimulation.

The patients hyperventilated voluntarily by performing repetitive maximal respiratory efforts over a period of least 15 seconds according to American Thoracic Society guidelines (23). Then a hypercapnic–hyperoxic rebreathing (HCS) test was performed according to the procedure recommended by Read (24). Rebreathing was terminated when the CO2 reached 72 to 74 mm Hg. Changes in CO2, volume, and time components of breathing pattern were recorded continuously.

Statistical Analysis

Volume and time components of the respiratory cycle were averaged in each patient over 30 consecutive breaths during quiet breathing. During hypercapnic rebreathing, the Vt values were the mean of 4 to 5 breaths recorded at 40, 50, 60, and 70 mm Hg CO2.

Values are reported as mean ± SD. A nonparametric statistical procedure was used to test differences: the Wilcoxon test for paired samples and the Mann–Whitney test for unpaired samples. Regression analysis was performed using Pearson's correlation coefficient. The level of significance was set at p values less than 0.05. All statistical procedures were performed using the Statgraphics Plus 3.1 statistical package (Manugistics, Rockville, MD).

Patients' respiratory function and clinical data are shown in Table 1. The height of the control group men (mean 1.81 ± 0.05 m) was significantly higher (p < 0.0013) than that of the patients (mean 1.68 ± 0.09 m).

Both maximum inspiratory and expiratory pressure values were lower in patients (p < 0.0000) than in control subjects (maximum inspiratory pressure, 53.43 ± 21.4 and 99.4 ± 8.4 cm H2O, respectively; maximum expiratory pressure, 61.6 ± 16 and 121.8 ± 18.1 cm H2O, respectively).

In patients and control subjects, we compared Vt estimated by OEP and Vt measured by pneumotachograph (pn) during quiet breathing using the Bland and Altman test. In control subjects, the mean difference between Vtpn and VtOEP was 0.03 L and limits of agreements were 0.07 L and −0.01 L; in patients, the mean difference between Vtpn and VtOEP was −0.08 L and limits of agreement were 0.04 L and −0.2 L.

The right and the left Vt during quiet breathing, VH, and CO2 rebreathing (HCS) were not different in control subjects.

No significant differences were found between the Vt of the paretic (416 ± 105 ml) and the healthy sides (403 ± 84 ml) in patients during quiet breathing (Figure 1)

. In contrast, the paretic Vt was consistently higher during HCS (1,114 ± 203 ml vs. 805 ± 134 ml; p = 0.02) (Figure 1) compared with the healthy side and was lower during VH (582 ± 155 ml vs. 783 ± 317 ml; p = 0.04) than the healthy side. As shown in Table 2

TABLE 2. Compartmental distribution of chest wall volumes in patients under study conditions



Paretic side416 ± 1051114 ± 203*582 ± 155*
Healthy side403 ± 84805 ± 134783 ± 317
Paretic side153 ± 43416 ± 96*268 ± 79*
Healthy side148 ± 36272 ± 61371 ± 113
Paretic side263 ± 81698 ± 174*314 ± 95*
Healthy side
255 ± 52
533 ± 127
412 ± 87

*p < 0.05 between healthy and paretic side.

Definition of abbreviations: HCS = hypercapnic–hyperoxic rebreathing; QB = quiet breathing; Vab = volume of abdomen; Vcw = volume of chest wall (Vcw = Vrc + Vab); VH = voluntary hyperventilation; Vrc = volume of rib cage.

Values are mean ± SD.

, the compartmental (rc and ab) volume distribution was maintained in the patients during all the study conditions (quiet breathing, VH, and HCS). There were uniform differences between sides at different levels of CO2. The paretic Vt was always greater than the healthy Vt, the difference being statistically significant at 60 and 70 mm Hg of CO2 (Table 3)

TABLE 3. Tidal volume of the paretic and healthy side of the thorax during progressive hypercapnic stimulation

Healthy Side

Paretic Side

p Value
CO2 40 mm Hg, QB403 ± 84416 ± 105NS
CO2 50 mm Hg495 ± 268649 ± 344NS
CO2 60 mm Hg610 ± 305856 ± 4280.01
CO2 70 mm Hg
805 ± 134
1114 ± 203

Definition of abbreviations: CO2 = carbon dioxide; NS = not significant; QB = quiet breathing.

Values are means ± SD.


In Figure 2

we selected breaths of equal Vt on the healthy side during both HCS and VH, and then we found the corresponding Vt on the paretic side. The Vt on the healthy side (circles) was the same with HCS as it was during VH, whereas the paretic side Vt (triangles) comparatively increased in seven subjects with HCS and decreased in six with VH; in the remaining two subjects the paretic side Vt was either similar or greater with VH.

The ventilatory response to CO2 was similar on the right and left sides in control subjects, whereas it was significantly higher on the paretic side than on the healthy side (p = 0.012; F = 6.35) in patients. Individual data points are close to the line of identity in control subjects (triangles) but are above the identity line in four of the patients (circles) (Figure 3)


Optoelectronic plethysmography allowed us to document the asymmetry of respiratory movements of the chest wall during hemiplegia. In particular, the paretic side showed reduced expansion during VH (when the drive is under cortical control) and increased expansion during chemical stimulation (when the drive is under brain stem control).

Because of differences in anthropometric characteristics between the two groups (see Table 1), the ventilatory response to CO2 or VH or Vt in liters cannot be compared between groups and the analysis has to be restricted to left- versus right-side ventilation in a given subject or patient.

One concern regarding the present research is the small number of patients recruited and the heterogeneous lesions of the patients. However, we need to underline the extreme difficulty of selecting patients with all of following characteristics: (1) presence of hemiplegia without impairment of comprehension, (2) ability to maintain the sitting position and to use the mouthpiece correctly, (3) absence of respiratory disease, and (4) ability to provide reproducible maneuvers. As we were very strict about the selection criteria, we were only able to find eight patients with heterogeneous lesions over a period of 16 months. However, the results in patients with hemorrhage did not differ from those of the other patients with ischemia. This is in line with studies showing no correlation between respiratory pattern and site of lesion or clinical assessment or presence of blood in the cerebrospinal fluid (25, 26).

To our knowledge, this is the first study that has successfully reconstructed ventilation of the two sides of the thorax. Our findings do not show agreement with the data of Fluck (16) who found asymmetric movements during quiet breathing and during VH but not when breathing was driven by CO2. However, the method used by Fluck (16) (a caliper to measure linear displacement in the parasagittal plane) allows evaluation of only the anteroposterior movement at a single point on each side of the chest wall, which moves with more than two degrees of freedom, especially during stimulated breathing (27). This technique, which measures a distance to estimate Vt, ignores small but systematic distortions of the rib cage. In addition, the validity of the calibration coefficients obtained experimentally to convert one or two dimensions to volume is limited to the estimation of Vt under conditions matching those during which the calibration was performed. Optoelectronic plethysmography that measures the three-dimensional coordinates of several markers applied to the chest wall overcomes this technical inaccuracy and does not depend on any assumptions of the number of degrees of freedom of the chest wall.

De Troyer and coworkers (14) observed a striking reduction in electromyographic activity of the intercostal muscles on the side of the paresis in all patients and of the diaphragm in the large majority of cases during progressive voluntary increases in Vt. Cohen and coworkers (15) confirmed these results by ultrasound: in four of eight patients with hemiplegia, a reduced diaphragmatic movement was present on the paralyzed side during volitional breathing compared with automatic breathing. In line with the above studies, we found a reduction in respiratory movements on the paretic side during VH. The reduction in Vt on the paretic side during VH confirms that the involvement of voluntary control of the diaphragm in ischemic hemiplegia is not different from that of other skeletal muscles (28).

Nonetheless, how can we explain the observation that in two of the eight patients the Vt on the paretic side was either similar or slightly greater than that of the healthy side during VH (see Figure 2)? In these two patients, the cortical representation of the diaphragm could be bilateral and symmetric and thus not damaged by hemiplegic stroke. This possibility is consistent with the interindividual variability in the cortical representation of the diaphragm (14, 15) or by the bilateral representation in regions other than the primary motor area (28). In humans, premotor areas seem to be linked to the diaphragm or at least to the act of inspiration (29).

Ours is the first demonstration that the ventilatory response to CO2 is increased on the paretic side, in keeping with the hypothesis of a loss of the cortical inhibition (10) on that side. The statement of an increased ventilatory drive requires explanation. Lanini and coworkers (21) and MacMahon and Heyman (30) have demonstrated an increase in dynamic elastance during CO2 rebreathing in patients with hemispheric stroke. In a unilateral muscle weakness disease, it is conceivable that the mechanical properties of the lung are not the same on the paretic side as on the normal side (15). Thus, considering the increase in ventilatory output we found on the paretic side (a greater increase in Vt), we can speculate that the increased respiratory drive also pertains to the paretic side. As to the hypothesis of a loss of cortical inhibition, cortical and subcortical lesions may result acutely in a transient decrease of cerebral inhibition of brain stem–mediated automatic bulbospinal pathway responses to a chemical stimulus (10, 17). This alteration, in our study, was present about 1 month after onset of the symptoms. Recent observations by Lefauncher and Lofaso (11), that the diaphragm can be subjected to intracortical inhibitory control, lend support to our hypothesis.

This descriptive study provides evidence, using OEP, of asymmetric ventilation in patients with hemiplegia. Transcranial magnetic stimulation (TMS) has gained widespread acceptance for study of pyramidal tract conduction and cortical excitability of somatic and respiratory muscles (28, 31). According to Lefauncher and Lofaso (11), transcranial magnetic stimulation, employed to study the cortical silent period, could also be used to assess some intracortical regulatory mechanisms influencing the activity of the diaphragm. In particular, in conditions of impairment of central diaphragmatic control, as in stroke, the study of the cortical silent period derived both from the paretic and the contralateral hemidiaphragm would enable us to confirm the hypothesis of a loss of inhibitory cortical control.

In conclusion, our study detects the alterations in the control of breathing in patients with stroke. The present data, showing that cerebrovascular disease can produce an asymmetric ventilatory involvement of the respiratory system with an increase in CO2 sensitivity and a decrease in volitional ventilation on the paretic side, lend support to the starting hypothesis of the existence of a unilateral crossed inhibitory cortical control of the ventilatory response to CO2.

1. Aminoff MJ, Sears TA. Spinal integration of segmental cortical and breathing inputs to thoracic respiratory motoneurons. J Physiol 1971;215:557–575.
2. Manning HL, Leiter JC. Respiratory control and respiratory sensation in a patients with a ganglioglioma within the dorsocaudal brain stem. Am J Respir Crit Care Med 2000;161:2100–2106.
3. Guz A. Brain, breathing and breathlessness. Respir Physiol 1997;109:197–204.
4. Macefield G, Gandevia SC. The cortical drive to human respiratory muscle in the awake state assessed by premotor cerebral potentials. J Physiol 1991;439:545–558.
5. Roger PS. Breathing and the nervous system. In: Aminoff MJ, editor. Neurology and general medicine: the neurological aspects of medical disorders. London, UK: Churchill Livingstone; 1989. p. 1–22.
6. Gozal D, Simakajornboon N. Passive motion of the extremities modifies alveolar ventilation during sleep in patients with congenital central hypoventilation syndrome. Am J Respir Crit Care Med 2000;162:1747–1751.
7. Kijima M, Isono S, Nishino T. Modulation of swallowing reflex by lung volume changes. Am J Respir Crit Care Med 2000;162:1855–1858.
8. Preas HL, Jubran A, Vandivier RW, Reda D, Godin PJ, Banks SM, Tobin MJ, Suffredini AF. Effect of endotoxin on ventilation and breath variability: role of cyclooxygenase pathway. Am J Respir Crit Care Med 2001;164:620–626.
9. Jubran A, Tobin MJ. Effect of isocapnic hypoxia on variational activity of breathing. Am J Respir Crit Care Med 2000;162:1202–1209.
10. Plum F. Neurological integration of behavioral and metabolic control of breathing. In: Porter R, editor. Breathing: Hering-Breuer centenary symposium. London, UK: Churchill Livingstone; 1970. p. 159–174.
11. Lefaucheur JP, Lofaso F. Diaphragmatic silent period to transcranial magnetic cortical stimulation for assessing cortical motor control of the diaphragm. Exp Brain Res 2002;146:404–409.
12. Gozal D, Gautier C. Evolving concepts of the maturation of central pathway underlying the hypoxic ventilatory response. Am J Respir Crit Care Med 2001;164:325–329.
13. Xu F, Frazier DT. Modulation of respiratory motor output by cerebellar deep nuclei in the rat. J Appl Physiol 2000;89:996–1004.
14. De Troyer A, Zegerdes De Beyl D, Thirion M. Function of respiratory muscles in acute hemiplegia. Am Rev Respir Dis 1981;123:631–632.
15. Cohen E, Mier A, Heywood P, Murphy K, Boultbee J, Guz A. Diaphragmatic movement in hemiplegic patients measured by ultrasonography. Thorax 1994;49:890–895.
16. Fluck DC. Chest movements in hemiplegia. Clin Sci 1966;31:383–388.
17. Klassen AC, Heaney LM, Lee MC, Kronenberg RS. Altered cerebral inhibition of respiratory and cardiac responses to hypercapnia in acute stroke. Neurology 1980;30:951–955.
18. Cala SJ, Kenyon CM, Ferrigno G, Carnevali A, Aliverti A, Pedotti A, Macklem PT, Rochester DF. Chest wall estimation by optical reflectance motion analysis. J Appl Physiol 1996;81:2680–2689.
19. Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS. Intracerebral hemorrhage versus infarction: a comparison of stroke severity, risk factors and prognosis: the Copenhagen stroke study. Ann Neurol 1995;38:45–50.
20. Gorini M, Misuri G, Corrado A, Duranti R, Iandelli I, De Paola E, Scano G. Breathing pattern and carbon dioxide retention in severe chronic obstructive pulmonary disease. Thorax 1996;51:677–683.
21. Lanini B, Gigliotti F, Coli C, Bianchi R, Pizzi A, Romagnoli I, Grazzini M, Stendardi L, Scano G. Dissociation between respiratory effort and dyspnea in a subset of patients with stroke. Clin Sci 2002;103:467–473.
22. European Community for Coal and Steel. Standardization of lung function tests. Eur Respir J 1993;6:1–100.
23. American Thoracic Society. Standardization of spirometry: 1994 update. Am J Respir Crit Care Med 1995;152:1107–1136.
24. Read DJC. A clinical method for assessing the ventilatory response to carbon dioxide. Australas Ann Med 1967;16:20–32.
25. Vingerhoets F, Bugousslavsky J. Respiratory dysfunction in stroke. Clin Chest Med 1994;15:729–737.
26. Rout MW, Lane DJ, Wollner L. Prognosis in acute cerebrovascular accidents in relation to respiratory pattern and blood gas tensions. BMJ 1971;3:7–9.
27. Ward ME, Ward JW, Macklem PT. Analysis of human chest wall motion using a two-compartment rib cage model. J Appl Physiol 1992;72:1338–1347.
28. Similowski T, Catala M, Rancurel G, Derenne JP. Impairment of central motor conduction to the diaphragm in stroke. Am J Respir Crit Care Med 1996;154:436–441.
29. Colebatch LG, Adams L, Murphy A, Martin AJ, Lammersma AA, Tochon-Danguy HJ, Clark JC, Friston KJ, Guz A. Regional cerebral blood flow during volitional breathing in man. J Physiol (Lond) 1991;443:91–103.
30. McMahon SM, Heyman A. The mechanics of breathing and stabilization of ventilation in patients with unilateral cerebral infarction. Stroke 1974;5:518–527.
31. Gea J, Espadaler LM, Guiu R, Aran X, Seoane L, Broquetas LM. Diaphragmatic activity induced by cortical stimulation: surface versus esophageal electrodes. J Appl Physiol 1993;74:655–658.
Correspondence and requests for reprints should be addressed to Giorgio Scano, M.D., Fondazione Don C. Gnocchi (IRCCS), Via Imprunetana, 124 50020 Pozzolatico, Florence, Italy. E-mail:


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record