help button home button
AJRCCM
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Online Data Supplement
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Habib, R. H.
Right arrow Articles by Courtney, S. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Habib, R. H.
Right arrow Articles by Courtney, S. E.
American Journal of Respiratory and Critical Care Medicine Vol 166. pp. 950-953, (2002)
© 2002 American Thoracic Society

Original Articles

Optimal High-Frequency Oscillatory Ventilation Settings by Nonlinear Lung Mechanics Analysis

Robert H. Habib, Kee H. Pyon and Sherry E. Courtney

Mercy Children's Hospital at St. Vincent Mercy Medical Center; Department of Pediatrics, Medical College of Ohio, Toledo, Ohio; and Department of Pediatrics, Cooper Hospital and Robert Wood Johnson Medical School, Camden, New Jersey

Correspondence and requests for reprints should be addressed to Robert H. Habib, Ph.D., Cardiopulmonary Research, St. Vincent Mercy Medical Center, 2213 Cherry Street, ACC Building, Suite 309, Toledo, OH 43608. E-mail: robert_habib{at}mhsnr.org


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Use of nontidal high-frequency oscillatory ventilation (HFOV) while the lungs are expanded by an imposed airway pressure (Paw) in neonates is increasingly based on evidence of decreased risk of lung injury. However, an objective method to optimize Paw is lacking. We measured lung volume changes ({Delta}VL[t]) via respiratory inductance plethysmography over a range of Paw settings in five piglets before and after lung lavage. These multiple {Delta}VL(t) were then simultaneously fit by an exponential rise to maximum model, {Delta}VL(t, Paw) = {Delta}VL,max · (1 - e–(t/{tau})), where {Delta}VL,max was a sigmoidal function of Paw and {tau} varied with lung volume. Postlavage, the effective compliance (CEFF = {Delta}VL,max/Paw) was generally decreased, whereas {tau} increased, indicating a slower paced volume recruitment. Model-derived CEFF–{Delta}VL,max relationships were altered substantially after lavage and were sigmoidal with a bell-shaped derivative function. The maximum of its derivative corresponded to a favorable (or optimal) {Delta}VL/Paw where the maximal increase in compliance is achieved. In conclusion, CEFF–{Delta}VL,max data available from respiratory inductance plethysmography provided important insight to changes in lung mechanics. These also provided a basis of an objective method (1) to optimize Paw during HFOV and (2) to assess the efficacy of treatments and progression/regression of underlying disease in neonates managed with HFOV.

Key Words: respiratory inductance plethysmography • respiratory distress syndrome • preterm infants • lung volume • overdistention


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
High-frequency oscillatory ventilation (HFOV) is preferred by many as the method for ventilatory support of very low birth weight infants in respiratory failure (1, 2). The primary rationale for using HFOV is that the nontidal ventilation used to affect gas exchange potentially reduces the incidence of chronic lung disease compared with conventional methods (2–8).

A critical feature of HFOV is the necessary concomitant lung volume recruitment ({Delta}VL) achieved by an induced positive mean airway pressure (Paw) about which the oscillations are applied (9–11). In simple terms, optimal Paw aims to maximize the alveolar volume available for gas exchange to levels well above functional residual capacity (Paw = 0) without overstretching the lung tissues. If such overdistention is not avoided, cardiopulmonary function may be adversely affected (12) and pneumothorax or airway leaks may result (10). However, the optimal Paw setting is dependent on the underlying lung mechanics, which vary (1) with the severity of the disease, (2) with the efficacy of treatments (e.g., intratracheal surfactant delivery), and (3) as the disease process progresses or regresses. Accordingly, a reliable and clinically applicable method to ascertain the optimal Paw might be a valuable tool to neonatologists as they manage preterm infants on HFOV.

We recently demonstrated that {Delta}VL during HFOV can be accurately and noninvasively quantified by direct current–coupled respiratory inductance plethysmography (RIP) when compared with simultaneous independent estimates via body plethysmography (11), and at least in theory, such measurements may facilitate the detection/avoidance of lung overdistention. Conceptually, this may be done via estimates of the effective respiratory system compliance (CEFF = {Delta}VL/Paw) derived from the corresponding nonlinear quasistatic Paw–{Delta}VL relationship. The latter may be constructed from repeated {Delta}VL measurements at multiple Paw settings.

The objectives of this study were (1) to measure the time-dependent rise in lung volume {Delta}VL(t) induced by multiple Paw settings during HFOV in newborn piglets before (control) and after lung lavage and (2) to model in each case the {Delta}VL(t) derived over a wide range of Paw in terms of a mathematical model incorporating nonlinear static pressure–volume relationship and allowing for volume-dependent time constants to model varying dynamics of alveolar recruitment at different Paw.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Animals
With Institutional Animal Care and Use Committee approval according to National Institutes of Health guidelines, we studied five newborn piglets (age, 6–15 days; weight, 2.2–4.2 kg) under control and postlavage (surfactant deficiency) conditions. A detailed description of the experimental protocol and basic data was recently described by us (11).

Briefly, anesthesia was initiated with ketamine hydrochloride intramuscularly (14 mg/kg) and was maintained with intravenous sodium pentobarbital. Animals were intubated using a 3.0-mm endotracheal tube inserted via a tracheostomy to a depth of 4 cm. The trachea was securely tied around the endotracheal tube to prevent airway leaks. Temporary conventional mechanical ventilation was initiated followed by pancuronium bromide (0.1 mg/kg) to induce paralysis with maintenance doses every 20 minutes. The carotid artery (blood–gas sampling, medications, pressure monitoring) and jugular vein (10 ml/hour infusion of 5% dextrose and lactated Ringer's) were canulated. Initial HFOV (3100 Oscillator; SensorMedics, Yorba Linda, CA) settings varied slightly among piglets (Paw, 5–8 cm H2O [control], 10–16 cm H2O [postlavage], frequency 8–10 Hz, amplitude 55–70 cm H2O, fraction of inspired oxygen 1.0, and inspiratory time at 33% of the total breathing cycle) based on blood gas values.

Serial lung lavages were performed using aliquots of 30 ml/kg of warmed normal saline. The chest was massaged to circulate the liquid, and the lung effluent was allowed to exit passively while the animal was placed in a gravity-dependent drainage position. Between lavages, piglets were ventilated using with 5 cm H2O of positive end-expiratory pressure and allowed to recover (stable heart rate and oxygen saturation of more than 90%). Lavages continued until PaO2 was less than 100 mm Hg on fraction of inspired oxygen of 1.0 after which postlavage HFOV measurements were started. At the conclusion of the experiment, animals were sacrificed (pentobarbital sodium overdose).

RIP Measurements/Analysis
Abdominal and ribcage RIP bands (RespiBands; SensorMedics) were placed just above the umbilicus and at the axillae, respectively. The effective band lengths were secured by tight clips to avoid loosening, and their positions were marked to ensure similar placement after lavage. Absolute lung volume changes were computed from the sum of the ribcage and abdominal data (sampled at 50 Hz; Somnostar PT; SensorMedics) against a linear calibration (11). Before analysis, RIP data were digitally low-pass filtered (fc = 2Hz, -92 dB) to remove the superimposed oscillatory ventilation effects on both signals (MP 100; BioPac Systems Inc., Santa Barbara, CA).

Control and postlavage Paw-induced {Delta}VL(t) were measured at the initial and at multiple higher Paw settings (8 to 10 measurements before and after lavage) to elicit larger maximal {Delta}VL. After data collection with each Paw, changes in HFOV frequency and amplitude were made in response to variations in arterial blood gas values. Derecruitment back to FRC, or Paw = 0, after each measurement was confirmed by RIP returning to baseline.

Finally, the family of RIP recruitment curves (obtained at the multiple Paw settings obtained before or after lung lavage) was analyzed by fitting a two-dimensional mathematical model (modeling analysis details are posted in an online data supplement), which allowed for (1) an exponential rise in lung volume toward {Delta}VL,max in response to a given imposed Paw and defined by a time constant related to the underlying mechanical properties and (2) a nonlinear (sigmoidal) relationship governing the effective lung volume change in terms of its distending pressure or {Delta}VL,max–Paw.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
{Delta}VL(t) measurements in control piglets were done for Paw ranging between 5 and 15 cm H2O, whereas higher values (12–24 cm H2O) were generally needed after lung lavage. Expectedly, {Delta}VL,max was systematically greater as Paw increased for both control and postlavage (Figure 1) . Also, irrespective of Paw setting, the time-dependent volume recruitment or {Delta}VL(t) rise toward {Delta}VL,max was generally more rapid in healthy lungs compared with surfactant-deficient lungs (Figure 1).



View larger version (21K):
[in this window]
[in a new window]
  		Figure 1. {Delta}VL(t) data (solid line) measured in an example piglets before (control, left panel) and after lung lavage (postlavage, right panel) shown for multiple Paw spanning the range of settings used. Other {Delta}VL(t) data corresponding to Paw within the shown range were intentionally excluded for clarity of illustration. Dashed lines illustrate the simultaneous model fit to the {Delta}VL(t) at all Paw settings (Equations 1, 2, and 3 in Appendix in online data supplement).

 
The nonlinear model detailed in the METHODS section almost always fit the data closely with all seven parameters generally (9 of 10 fits) needed (see example fits in Figure 1). These model parameters were then used to derive {Delta}VL,max–Paw curves in individual piglets for healthy and diseased conditions as shown in Figure 2 .



View larger version (18K):
[in this window]
[in a new window]
  		Figure 2. (Left panel) Model-derived {Delta}VL,max–Paw curves in individual piglets for control (open symbols) and postlavage (closed symbols) conditions. Note that higher Paw settings were used in piglets after lavage to compensate for the volume derecruitment after repeated warm saline lavages and to facilitate adequate gas exchange in surfactant-deficient piglets. Higher Paw settings in control animals were avoided as these were unnecessary for gas exchange and will necessarily lead to lung tissue overdistention and possibly to air leaks caused by pneumothorax. (Right panel) Mean ± SD of the CEFF derived from all piglets at each Paw setting before (open symbols) and after lung lavage (closed symbols). CEFF = {Delta}VL,max/Paw. Both the {Delta}VL,max–Paw and CEFF–Paw curves were generally nonlinear over the range of Paw. Lines through the control (solid lines) and postlavage (dashed line) CEFF data represent sigmoidal fits in terms of Paw.

 
The effective compliance (CEFF = {Delta}VL,max/Paw ) for each {Delta}VL,max and Paw combination was derived in each piglet before and after lavage and was averaged (Figure 2). The resulting comparison indicated that lung compliance varied substantially and nonlinearly in terms of Paw under both healthy and diseased conditions. Moreover, CEFF (ml cm H2O-1) at similar Paw values was substantially lower after lavage, reflecting both the effects of surfactant deficiency and the corresponding lung volume derecruitment after lung lavage. For example, at Paw = 14 cm H2O, postlavage CEFF (1.93 ± 0.06) was decreased by approximately 35% relative to control (2.96 ± 0.23) values (Figure 2).

Because the range of Paw settings used for control and postlavage conditions was different, we plotted CEFF and {tau} in terms of the {Delta}VL,max corresponding to each of the Paw settings. This allowed comparing the underlying lung mechanics at similar changes in lung volume. These results showed that for the same {Delta}VL,max, CEFF was consistently and substantially lower after lung lavage. Also, lung volume recruitment was consequently slower, as is illustrated by the contrasting {tau} values before and after lung lavage (Figure 3) . Note, {tau} was not a constant for a given fit but was rather assumed to be a dynamic or varying property during each recruitment that varied as a function of the underlying change in lung volume (or compliance); {tau} generally decreased with increasing Paw (or equivalently {Delta}VL,max) except at the highest Paw, where this trend appeared to reverse probably because of lung tissue overdistention.



View larger version (24K):
[in this window]
[in a new window]
  		Figure 3. (Top panel) Mean ± SD of the derived time constant ({tau}) corresponding to each {Delta}VL,max–Paw combination. Note that {tau} values were derived by plugging each Paw setting and the corresponding {Delta}VL,max (as derived from model fit Equation 3—and in online data supplement) then averaged for specific Paw settings across piglets. (Middle panel) Mean ± SD of the CEFF derived from all piglets at each Paw setting plotted as a function of {Delta}VL,max before and after lung lavage. This presentation of CEFF illustrates the volume dependence of CEFF and contrasts the control and postlavage CEFF corresponding to similar {Delta}VL,max above the resting lung volume or when Paw = 0. Note that resting lung volume is reduced after lavage due to the derecruitment induced by surfactant deficiency. Lines through control (solid lines) and postlavage (dashed lines) data depict four-parameter sigmoidal fits to these curves. (Bottom panel) Derivative functions of the later sigmoid functions exhibiting significant and distinct changes postlavage (see online data supplement for details).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
In a related study (11), we showed that direct current–coupled RIP allows reliable and accurate estimation of the lung volume recruitment induced by positive pressure at the airway opening during HFOV. Furthermore, it was proposed that the availability of such measurements provided the basis for a practical method to help in the optimization of HFOV settings. This study describes a potential method, or paradigm, that combines RIP measurements with HFOV to achieve this important task.

Attractive features of RIP include its noninvasive nature and the fact that it is measured distally at the chest wall. The latter is important as it means RIP (1) will not interfere with the delivery of ventilation, (2) will not interfere with access to the patient (as opposed to body plethysmography), and (3) is insensitive to airway leaks at the airway/tracheal tube interface—a fact particularly relevant in infants intubated with uncuffed endotracheal tubes. Alternatively, RIP currently remains more of a research tool that is commercially available in alternating current–coupled form for measuring tidal ventilation only. Fortunately, direct current coupling of RIP devices may be easily achieved to facilitate measurement of both static and tidal volume changes. Also, the potential for skin irritation in the preterm infants from the RIP bands may need to be addressed.

The proposed method for the optimization of HFOV settings is based on a series of short RIP measurements of the changes in lung volume over a clinically relevant range of Paw settings. Once available, the family of recruitment curves {Delta}VL(t, Paw) may then be analyzed simultaneously in terms of a mathematical model describing lung recruitment as an exponential rise to maximum occurring over time. Importantly, this model has to incorporate the nonlinearity or volume dependence of lung mechanics (11, 13, 14). In addition, we allowed for a varying time constant ({tau}) to define the exponential rise to maximum at the multiple Paw settings, as {tau} was assumed to be a dynamic function of the underlying lung mechanical properties that themselves will vary as the lung inflates during the recruitment maneuver.

The devised seven-parameter model fit the control and postlavage {Delta}VL(t,Paw) data well in all piglets, and the derived model parameters allowed the construction of {Delta}VL,max–Paw and CEFF–Paw relationships (Figure 2) and consequently CEFF–{Delta}VL,max and {tau}–{Delta}VL,max curves (Figure 3). These relationships provided considerable insight to the underlying lung mechanical properties that may then be used to ascertain the optimal operating range of Paw settings. One expects that attaining optimal Paw settings would allow for (1) maximizing lung volume to facilitate gas exchange, (2) maximizing lung compliance so that amplitude of the imposed pressure oscillations needed to ventilate may be minimized, and (3) avoiding lung overdistention that can lead to tissue injury including airway leaks. A useful qualitative/quantitative approach to interpret the CEFF–{Delta}VL,max relationship is through its derivative function (see Appendix in online data supplement), which may provide a straightforward means to ascertain changes in the underlying lung mechanics, such as expected changes with surfactant treatments or with progression/regression of disease.

An important aspect of the method proposed in this study is that it (1) does not necessarily require the absolute calibration of RIP, (2) does not entail an invasive placement of an esophageal balloon/catheter to estimate intrapleural pressure, and (3) does not rely on an airway opening flow measurement in a setting where air leaks at the airway–ventilator interface are the norm rather than the exception. Uncalibrated RIP precludes having a sensible physiologic value for compliance, yet knowing the change in CEFF should be sufficient to quantify effects of treatments or to determine optimal settings within a given subject. Importantly, for RIP to provide accurate {Delta}VL, it is critical that one be then meticulous and careful about the constancy of the positioning and the effective length of the RIP bands.

Other factors not addressed in this study may affect the optimal HFOV settings. Overdistention stemming from the superimposed oscillatory ventilation is plausible, but this risk is small given the amplitude of these oscillations and may be accounted for by maintaining a margin of safety when selecting Paw. Another—the implicit assumption that the lungs are expanded as a homogenous mechanical unit—is not always accurate, particularly in the surfactant deficient lung during treatment. Although respiratory distress syndrome affects the infant fairly homogeneously, delivery of exogenous surfactant in the premature infant lung is almost invariably nonuniform (15) and will lead to nonhomogeneous lung mechanical properties. Arguably, overdistention of healthier lung units (or those with the highest regional compliance) may go undetected if a cumulative or single compartment measure such as CEFF is used. Related to this is that, even if inhomogeneity is not present, overdistention may occur at lower Paw as the lungs become healthier during treatment. Conversely, if the disease worsens, applied HFOV settings may become suboptimal. We believe that these scenarios are best avoided, or minimized, by periodic (and perhaps frequent) application of the proposed method for determining appropriate support levels. Indeed, this emphasizes the need for the eventual automation of this method.

Conclusion
We have shown that {Delta}VL,max measured with RIP at multiple Paw can provide important insight to changes in lung mechanics during HFOV via model derived CEFF versus Paw and CEFF versus {Delta}VL,max relationships. The latter formed the basis of a potentially clinically applicable model analysis approach for determining optimal HFOV settings. Although promising, important hurdles remain for future research to overcome before routine clinical application is possible. First, to test clinical feasibility, similar future animal models and infant studies should be done but where the clinically undesirable step of derecruitment back to Paw = 0 after each setting change is avoided. In such measurements, the CEFF estimate is replaced by a relative change in compliance but should convey similar information. Alternatively, however, the model parameters defining the time constant and their interpretation may differ substantially. Second, and given the underlying nonlinearity of lung mechanics, it is important to explore in similar fashion the changes in compliance as Paw is systematically (or in random fashion) decreased back from high toward lower values following the lung inflation. Third, future research should extend these measurements to infants with respiratory distress syndrome before and after surfactant treatments, as such studies will more fully elucidate their potential in the neonatal setting that is in need of reliable methods to assess changes in respiratory mechanics with treatments (16). Finally, if such a method is to become routine in a clinical setting, it is essential for the RIP and HFOV systems be integrated seamlessly, including automation of the proposed procedure to facilitate real-time assessments. The latter would include (1) sweeping through Paw settings, (2) simultaneously fitting the {Delta}VL(t), (3) computing plotting of the CEFF–{Delta}VL,max relationships and (4) modeling of CEFF–{Delta}VL,max.


   Acknowledgments
 
The authors thank Kaye Weber, M.S., R.R.T., Paresh B. Pandit, M.D., and Gordon Y. Chang, M.D., for their assistance during data acquisition.


   FOOTNOTES
 
Supported by grants from the Whitaker Foundation and the Foundation of the University of Medicine and Dentistry of New Jersey.

This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournals.org.

Received in original form May 6, 2002; accepted in final form June 28, 2002


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Clark RH, Gerstmann DR, Null DM, Yoder BA, Cornish JD, Glasier CM, Ackerman NB, Bell RE, Delemos RA. Prospective randomized comparison of high-frequency oscillatory and conventional ventilation in respiratory distress syndrome. Pediatrics 1992;89:5–12.[Abstract/Free Full Text]
  2. HiFO Study Group. Randomized study of high-frequency oscillatory ventilation in infants with severe respiratory distress syndrome. J Pediatr 1993;122:609–619.[Medline]
  3. Clark RH. High-frequency ventilation. J Pediatr 1994;24:661–670.
  4. Gerstmann DR, Minton SD, Stoddard RA, Meredith KS, Monaco F, Bertrand JM, Battisti O, Langhendries JP, Francois A, Clark RH. The Provo multicenter early high-frequency oscillatory ventilation trial: improved pulmonary and clinical outcome in respiratory distress syndrome. Pediatrics 1996;98:1044–1057.[Abstract/Free Full Text]
  5. Plavka R, Kopecky P, Sebron V, Svihovec P, Zlatohlavkova B, Janus V. A prospective randomized comparison of conventional mechanical ventilation and very early high-frequency oscillatory ventilation in extremely premature newborns with respiratory distress syndrome. Intensive Care Med 1999;25:68–75.[CrossRef][Medline]
  6. Rettwitz-Volk W, Veldman A, Roth B, Vierzig A, Kachel W, Varnholt V, Schlosser R, von Loewenich V. A prospective, randomized, multicenter trial of high-frequency oscillatory ventilation compared with conventional ventilation in preterm infants with respiratory distress syndrome receiving surfactant. J Pediatr 1998;132:249–254.[CrossRef][Medline]
  7. Sasse S, Kribs A, Vierzig A, Roth B. A staged protocol for the treatment of persistent pulmonary hypertension of the newborn. Klin Padiatr 1997;209:301–307.[Medline]
  8. Jackson JC, Truog WE, Standaert TA, Murphy JH, Juul SE, Chi EY, Hildebrandt J, Hodson WA. Reduction in lung injury after combined surfactant and high-frequency ventilation. Am J Respir Crit Care Med 1994;150:534–539.[Abstract]
  9. Saari AF, Rossing TH, Solway J, Drazer JM. Lung inflation during high-frequency ventilation (note). Am Rev Respir Dis 1984;129:333–336.[Medline]
  10. Venegas JG, Fredberg JJ. Understanding the pressure cost of ventilation: why does high frequency ventilation work? Crit Care Med 1994;22:S48–S57.
  11. Weber K, Courtney SE, Pyon KH, Chang CY, Pandit PB, Habib RH. Detecting lung overdistention in newborns treated with high-frequency oscillatory ventilation. J Appl Physiol 2000;89:364–372.[Abstract/Free Full Text]
  12. Cournand A, Motley HL, Werko L, Richards DW. Physiologic studies of the effects of intermittent positive pressure breathing on cardiac output in man. Am J Physiol 1948;167–174.
  13. Venegas JG, Harris SR, Simon BA. A comprehensive equation for the pulmonary pressure-volume curve. J Appl Physiol 1998;84:389–395.[Abstract/Free Full Text]
  14. Narusawa U. General characteristics of the sigmoidal model equation representing quasi-static pulmonary P-V curves. J Appl Physiol 2001;91:201–210.[Abstract/Free Full Text]
  15. Balaraman V, Sood SL, Finn KC, Hashiro G, Uyehara CF, Easa D. Physiologic response and lung distribution of lavage versus bolus Exosurf in piglets with acute lung injury. Am J Respir Crit Care Med 1996;153:1838–1843.[Abstract]
  16. Joint committee of the ATS Assembly on Pediatrics and the ERS Paediatrics Assembly. Respiratory mechanics in infants: physiologic evaluation in health and disease. Am Rev Respir Dis 1993;147:474–496.[Medline]



This article has been cited by other articles:


Home page
 Am. J. Respir. Crit. Care Med.Home page
M. J. Tobin
Critical Care Medicine in AJRCCM 2002
Am. J. Respir. Crit. Care Med., February 1, 2003; 167(3): 294 - 305.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Online Data Supplement
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Habib, R. H.
Right arrow Articles by Courtney, S. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Habib, R. H.
Right arrow Articles by Courtney, S. E.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Proc. Am. Thorac. Soc. Am. J. Respir. Cell Mol. Biol.
Copyright © 2002 American Thoracic Society