randomised controlled trial of continuous positive airway pressure and standard oxygen therapy in...

7/25/2019 Randomised Controlled Trial of Continuous Positive Airway Pressure and Standard Oxygen Therapy in Acute Pulm

1/8

European Heart Journal(2002) 23, 13791386doi:10.1053/euhj.2001.3156, available online at http://www.idealibrary.com on

Randomised controlled trial of continuous positive

airway pressure and standard oxygen therapy in

acute pulmonary oedema

Effects on plasma brain natriuretic peptide concentrations

C. A. Kelly1, D. E. Newby2, T. A. McDonagh3, T. W. Mackay1, J. Barr1, N. A. Boon2,H. J. Dargie3 and N. J. Douglas1

1Department of Respiratory Medicine, Royal Infirmary, Edinburgh, Scotland, U.K.; 2Department of Cardiology,

Royal Infirmary, Edinburgh, Scotland, U.K.; 3Clinical Research Initiative in Heart Failure, Western Infirmary,Glasgow, Scotland, U.K.

Aims The study aim was to compare the effects of continu-

ous positive airway pressure (CPAP) on clinical outcomes

and plasma neurohormonal concentrations in patients withacute pulmonary oedema.

Methods and Results In addition to standard therapy,

58 consecutive patients were randomized to receive 60%

inhaled oxygen with or without CPAP at 75 cmH2O press-

ure. Clinical variables, symptoms and oxygenation were

monitored and plasma epinephrine, norepinephrine and

brain natriuretic peptide (BNP) concentrations estimated at

0, 1, 6 and 24 h. CPAP was associated with less breathless-

ness at 1 h (P


2/8

haemodynamic variables more rapidly than standardoxygen therapy or CPAP, the data on this form oftreatment in acute heart failure are limited[10,11].

Heart failure is characterized by activation of severalneurohormonal systems, particularly the sympatheticnervous system and the renin-angiotensin-aldosteronesystem[1214]. In addition, counteregulatory mechanismssuch as the natriuretic peptide system are activatedproducing increased plasma concentrations of bothatrial natriuretic peptide (ANP) and brain natriureticpeptide (BNP)[15,16]. Although BNP is a good indicatorof subsequent prognosis and morbidity in chronic heartfailure[17,18], there is little information on the significanceof BNP in acute heart failure.

The aim of this study was to compare the effectsof oxygen therapy and CPAP on clinical outcomesand plasma neurohormonal concentrations in patientspresenting with acute pulmonary oedema.

MethodsStudy population

Patients requiring hospitalization for acute pulmonaryoedema were recruited into the trial. Fifty-eight consecu-tive patients fulfilling the following entry criteria wererecruited: (1) acute onset of breathlessness (within 6 h),(2) respiratory rate >20/min, (3) bilateral basal crackleson chest auscultation, and (4) typical chest X-rayappearance of pulmonary oedema. Patients wereexcluded if they had chest X-ray features consistent withpneumonia or pneumothorax, or if they had receivedpre-hospital treatment with interventions other than

oxygen, diuretics or opiates. The trial was conducted inaccordance with the Declaration of Helsinki (1989) ofthe World Medical Association and with the approval ofthe local research ethics committee. All patients gavewritten informed consent although, in some cases, theclinical condition necessitated initial witnessed verbalconsent before subsequent written consent could beobtained.

Study protocol

Patients had baseline measurements of heart rate, blood

pressure, respiratory rate, oxygen saturation, creatinekinase and arterial blood gases. Standard therapy foracute pulmonary oedema was administered: 50100 mgintravenous frusemide, 5 mg buccal nitrate (if systolicblood pressure >90 mmHg) and 2510 mg intravenousmorphine sulphate. Patients were randomized, bybalanced blocks using sealed envelopes, to receive either60% oxygen delivered through a Venturi mask or CPAP(Vital Signs, NJ, USA) using a full face mask with apressure of 75 cm water and an inspired oxygen con-centration of 60% confirmed by an integral oxygenanalyser. Treatment was commenced in the emergency

department and continued in the high dependency unitfor a minimum period of 6 h, or until no longer clinicallyindicated. Further drug therapy was given at thediscretion of the treating physician.

Measurements

Arterial blood gas analysis was performed on admissionand repeated 1 h after commencing treatment. Heartrate, blood pressure, respiratory rate and oxygensaturation were recorded on admission and at 1, 6 and24 h. Blood samples were taken for the estimation ofplasma neurohormone concentrations on study entry.Epinephrine and norepinephrine concentrations wererepeated at 1 h and BNP concentrations at 1, 6 and 24 hafter starting treatment. Venous blood was withdrawnand 10 ml admixed with each of 1 ml of 1% disodiumEDTA and 1 ml of disodium EDTA/2% sodium meta-bisulphite. Blood samples were placed on ice and imme-diately centrifuged at 2000g for 15 min. Plasma was

separated and stored at 70 C before assay. PlasmaBNP was measured usinga standard radioimmunoassay(Peninsula, CA, U.S.A.)[19]. Plasma epinephrine andnorepinephrine concentrations were determined by anelectrochemical method after separation by reversephase high performance liquid chromatography[20].

Echocardiography

To assess left ventricular function, an echocardiogramwas performed within 24 h of admission by the sameinvestigator (D.N.). The left ventricular end systolic and

diastolic dimensions were measured from the M-moderecordings at the tips of the mitral valve. The dimensionswere measured in both the long and short axis para-sternal views and the mean of the two measurementstaken. The ejection fraction was estimated by theTeicholz method[21].

Visual analogue scale of dyspnoea

Patients were instructed in the use of a visual analoguescale and were asked to score the severity of theirbreathlessness at 1 and 6 h after commencing treatment.

Since many patients were acutely distressed and severelybreathless on admission, it was not feasible to instructpatients in the use of visual analogue scale at this time.

Follow-up

Treatment failure was defined as: (1) the need forintubation and ventilation for any reason, (2) deteriorat-ing hypoxaemia or hypercapnia, or (3) progressive res-piratory distress. Patients were followed up until thetime of discharge from hospital to determine length of

1380 C. A. Kellyet al.

Eur Heart J, Vol. 23, issue 17, September 2001


3/8

stay and in-hospital mortality. Myocardial infarctionwas defined as an elevated creatine kinase more thantwice the upper limit of the laboratory reference rangewith associated accepted ECG changes.

Statistical analysis

Data are expressed as meanstandard error of themean and were examined, where appropriate, by analy-sis of variance (ANOVA) with repeated measures,regression analysis and chi-square test using StatViewv5.0.1 (SAS Institute Inc., Cary, North Carolina,U.S.A.). Where ANOVA demonstrated significant dif-ferences in responses, post-hoc comparisons were madeusing the Fisher protected least significant difference test(StatView v5.0.1). Statistical significance was taken atthe 5% level.

Results

Fifty-eight patients were recruited into the trial and theirbaseline characteristics are shown inTable 1.The studypopulation was elderly with evidence of significant leftventricular systolic dysfunction. On admission, patientswere tachypnoeic, tachycardic and hypertensive.

In comparison to conventional therapy alone, patientsrandomized to CPAP therapy reported less breathless-ness at 1 h (Table 2). This was associated with a more

rapid reduction in respiratory rate, heart rate and aci-dosis at 1 h in patients treated with CPAP ( Fig. 1). Thesetreatment group differences were not present at 6 h,but patients treated with CPAP continued to feelsubjectively less breathless.

There were no treatment failures in the CPAP group.None of the patients reported adverse effects fromCPAP other than minor discomfort from the facial maskafter 6 h. There were two treatment failures in thenon-CPAP group. The first patient was withdrawn fromthe study after 35 min because of respiratory distress andfalling oxygen saturation. Since no intensive care unitbeds were immediately available, he was given a trial ofCPAP while awaiting review by an intensivist. Followinginitial rapid improvement in both clinical variables andarterial blood gases, the patient later deteriorated, devel-oped cardiogenic shock and renal failure, and died after48 h. The second patient was also withdrawn after

45 min because of deteriorating clinical features andincreasing respiratory distress. He was given a trial ofCPAP with rapid improvement in symptoms and arterialblood gases. He was taken off CPAP at 6 h andwas subsequently discharged home with no furtherproblems.

The length of stay in high dependency areas and thetotal hospitalization times were similar for both treat-ment groups. However, there was a trend for a reductionin the overall hospital mortality in patients treated withCPAP (2 vs 7; 010>P>005, chi-squared test). Alldeaths were related to cardiovascular disease with the

Table 1 Admission patient characteristics

Conventionaltherapy alone

Conventionaltherapy and

CPAP

Number 31 27Age 782 772 (years)

Male:Female 16:15 10:17Shortening fraction 162 162 (%)Ejection fraction 403 383 (%)Heart rate 1136 1153 (/min)Systolic blood pressure 1526 1557 (mmHg)Diastolic blood pressure 904 935 (mmHg)

Respiratory rate 271 311 (/min)Oxygen saturation 932 931 (%)

Hydrogen ion 392 482 (nmol . l1)Partial pressure of oxygen 10108 13515 (kPa)Plasma bicarbonate 231 231 (mmol . l1)Partial pressure of carbon dioxide 4802 6104 (kPa)

Serum urea 15245 12725 (mmol . l1)Serum creatinine 13411 16529 (mol . l1)

Serum creatine kinase 16969 11529 (IU . ml

1

)Frusemide 764 774 (mg)Nitrate 3104 2804 (mg)Morphine 1003 2505 (mg)

P=037; ANOVA, conventional therapy vs conventional therapy+CPAP.

Trial of CPAP in acute pulmonary oedema 1381



4/8

exception of one CPAP patient who died of metastaticadenocarcinoma.

Plasma BNP concentrations were markedly elevatedon admission and correlated inversely with left ven-tricular ejection (r =024, P


5/8

a more rapid response, falling at 1 h (Table 3). Therewere no statistically significant differences in the plasmaneurohormone concentrations between the two treat-ment groups (P=ns, 2-way ANOVA). Plasma catecho-lamine and BNP concentrations were significantlyhigher in those patients who died or sustained amyocardial infarct (Figs 2and 3).

Discussion

We have shown that CPAP is an effective treatment foracute pulmonary oedema with a more rapid improve-ment in symptoms, respiratory rate, heart rate andacidosis as well as a trend for a reduction in overallmortality. These findings are in keeping with previous

Table 3 Plasma neurohormone concentrations during first 24 h

Conventionaltherapy alone

Conventionaltherapy and

CPAP

Plasma norepinephrine concentration Normal range


6/8

similar studies using CPAP and NIPPV[79]. Moreover,for the first time, we have described the plasma profile ofBNP concentrations during the first 24 h of hospital-ization. Although plasma neurohumoral concentrationswere unaffected by CPAP treatment, plasma BNP andcatecholamine concentrations were elevated in patientswho subsequently died or sustained an acute myocardialinfarction.

Unlike previous studies, we had a very low treatmentfailure rate, but this is unlikely to be explained bydifferences in the study population since our baselineclinical variables, left ventricular ejection fraction andarterial blood gases were similar[710]. None of thepatients assigned to CPAP required intubation andventilation or failed to improve with treatment. Therewere, however, two treatment failures in the non-CPAPgroup. These patients would have met the criteria forintubation and ventilation in other published studies,but intubation was averted by the introduction ofCPAP to which both made a rapid initial response.Unfortunately one patient later died of cardiogenicshock but a decision was made when he subsequently

deteriorated, that he would not be ventilated. Wehad no reports of intolerance to CPAP or any ofthe associated complications, such as nasal skin necro-sis, conjunctivitis, aspiration of gastric contents orbarotrauma.

Consistent with previous studies, we failed to demon-strate a significant improvement in the length of hospitalstay or on overall mortality. The aim of the study was toassess the effect of CPAP on clinical outcomes andplasma neurohormonal concentrations, and it was notsufficiently powered to show a significant difference inhospital mortality. However, there was a trend for a

reduction in mortality and pooling our data with thoseof previous studies wouldsuggest that CPAP confers anoverall survival benefit[22].

The main clinical benefits of CPAP occurred withinthe first hour, suggesting that treatment should becommenced as soon as possible after admission. In mosthospitals treatment should ideally commence within theemergency department or the medical assessment unit.We did not encounter any difficulties commencingCPAP in the emergency department. Previous exper-ience has also shown that non-invasive ventilationhas been successfully used in the emergency departmentfor a variety of clinical conditions, including acutepulmonary oedema[2325].

The role of bilevel ventilation in the management ofacute pulmonary oedema is more controversial. Twostudies have shown that bilevel ventilation may beassociated with an increased frequency of myocardialinfarction. The study by Mehta et al .[10] which washalted after an interim analysis indicated that, despite amore rapid improvement in ventilation, patients treatedwith bilevel ventilation had twice the rate of myocardial

infarction as those treated with CPAP. However, arecent study comparing NIPPV with conventional oxy-gen therapy demonstrated that NIPPV was associatedwith a more rapid improvement in oxygenation andresolution of respiratory symptoms without a significantincrease in the rate of myocardial infarction[11]. I n afurther study comparing NIPPV with high doseintravenous isosorbide dinitrate[26], a higher rate ofintubation, death and myocardial infarction and aslower resolution of clinical variables were reported inthe NIPPV group. However, data from this latter studyis difficult to compare with other published series since

024

1800

Time from admission (h)

Plasmabrainnatriur

eticpeptide

concentration(p

g.m

l1)

1600

1400

1200

1000

800

600

400

200

610

*

Figure 3 Plasma brain natriuretic peptide (BNP) concentrations during the first 24 h ofadmission in patients with (closed circles) or without (open circles) death or myocardialinfarction. Two-way ANOVA, P


7/8

the investigators used much lower mean inspiratory andexpiratory pressures, and treatment was delivered pre-hospital by mobile intensive care units rather than onarrival at hospital.

It has been suggested that bilevel ventilation mayreduce stroke volume and cardiac output to a greaterextent than CPAP, leading to impaired myocardialperfusion that may contribute to the higher rate ofmyocardial infarction. Takeda and colleagues showedno difference in hospital mortality rates for patients withacute myocardial infarction and pulmonary oedematreated with nasal CPAP[27]. Like Masip et al.[11], wedemonstrated a poorer outlook for patients with myo-cardial infarction which was independent of treatmentused. Rusterholtz et al .[28] also found that 80% ofpatients who failed to respond to NIPPV and subse-quently died had suffered myocardial infarction. Thiswould suggest that patients with myocardial infarctionhave a poorer prognosis generally but further research isrequired to determine whether bilevel ventilation, inparticular, has detrimental effects on these patients.

Neurohormonal activation has been extensivelystudied in chronic heart failure, but little is known in thecontext of acute heart failure. After initiation of treat-ment, plasma norepinephrine concentrations fell rapidlywithin the first hour, with a similar trend for epine-phrine. In contrast, BNP concentrations continued torise, peaking at 6 h before falling below baseline valuesby 24 h. This latter finding is consistent with an animalmodel of acute heart failure where rapid atrial pacingcaused a marked rise in plasma ANP concentrationswithin 1 hbut increases in BNP concentrations took 8 hto develop[29]. This is likely to represent the delay in theup-regulated synthesis and release of BNP since it israpidly cleared from the plasma through metabolismby neutral endopeptidase 24.11 and via binding toclearance receptors[30].

In patients with chronic heart failure or a recentmyocardial infarction, BNP is known to be elevated inproportion to the severity of left ventricular dysfunctionand has been shown to bea usefulprognostic marker ofmorbidity and mortality[1518,31,32]. It is also secretedbyinfarcted ventricular myocardium[33], peaking at 16 h[34].All patients in our study, including those with myocar-dial infarction, had a peak plasma BNP concentration at6 h. While there did not appear to be a treatment effectof CPAP, plasma epinephrine, norepinephrine and BNPconcentrations were higher in those patients who died or

sustained an acute myocardial infarction. Althoughthe association of raised catecholamine concentrationsand risk of mortality has been demonstrated pre-viously[14,35,36], this is the first description of an associ-ation between in-hospital outcome and elevated plasmaBNP concentrations in patients with acute heart failure.

Study limitations

Although fully randomized, there appeared to besome differences in baseline variables between the two

treatment groups. Patients assigned to CPAP therapyhad more severe disease with a slightly greater acidosisand hypercapnia on admission. However, despite this,patients felt subjectively much better on CPAP therapywith a two-fold reduction in the dyspnoea score at 1 and6 h. The absence of an admission dyspnoea score waspredicted and resulted from the inability of many very illpatients to comprehend and complete the required scaleswhen pilot studies were done.

In conclusion, we have shown that CPAP is a safe,well-tolerated treatment for acute heart failure. It causesa more rapid improvement in symptoms and clinicalvariables without any adverse haemodynamic effects.Plasma neurohormonal concentrations and the fre-quency of myocardial infarction were unaffected byCPAP treatment. Maximum clinical benefits appear tooccur within the first hour of treatment with CPAP,making this a useful adjunctive treatment for the man-agement of patients with acute heart failure in theemergency setting.

This study was supported by a grant from the San DiegoFoundation.

References

[1] Fagon JY, Chastre J, Hance AJ, Montravers P, Novara A,Gibert C. Nosocomial pneumonia in ventilated patients: acohort study evaluating attributable mortality and hospitalstay. Am J Med 1993; 94: 2818.

[2] Stauffer JL, Olsen DE, Petty TL. Complications and conse-quences of endotracheal intubation and tracheostomy: a pro-spective study of 150 critically ill patients Am J Med 1980; 70:6576.

[3] Katz JA, Marks JD. Inspiratory work with and without

continuous positive airway pressure in patients with acuterespiratory failure Anesthesiology 1985; 63: 598607.

[4] Naughton MT, Rahman A, Hara K, Floras JS, Bradley TD.Effect of continuous positive airway pressure on intrathoracicand left ventricular transmural pressure in patients withcongestive heart failure Circulation 1995; 91: 172531.

[5] Lenique F, Habis M, Lofaso F, Dubois-Rande J-L, Harf A,Brochard L. Ventilatory and haemodynamic effects of CPAPin left heart failure Am J Respir Crit Care Med 1997; 155:5005.

[6] Baratz DM, Westbrook PR, Shah PK, Mohsenifar Z. Effectof nasal CPAP on cardiac output and oxygen delivery inpatients with congestive heart failure Chest 1992; 103: 13971401.

[7] Bernsten AD, Holt AW, Vedig AE, Skowronski GA,Baggoley CJ. Treatment of severe cardiogenic pulmonary

edema with continuous positive airway pressure delivered byface mask N Engl J Med 1991; 325: 182530.

[8] Rasanen J, Heikkila J, Downs J, Nikki P, Vaisanen I, ViitanenA. Continuous positive airway pressure by face mask in acutecardiogenic pulmonary edema Am J Cardiol 1985; 55: 296300.

[9] Lin M, Yang YF, Chiang HT, Chang M-S, Chiang BN,Cheitlin MD. Reappraisal of continuous positive airwaypressure therapy in acute cardiogenic pulmonary edema: shortterm results and long term follow up. Chest 1995; 107:137986.

[10] Mehta S, Jay GD, Woolard RH et al. Randomized, prospec-tive trial of bi-level versus continuous positive airway pressurein acute pulmonary oedema Crit Care Med 1997; 25: 6208.

Trial of CPAP in acute pulmonary oedema 1385



8/8

[11] Masip J, Betbese AJ, Paez J et al. Non-invasive pressuresupport ventilation versus conventional oxygen therapy inacute cardiogenic pulmonary oedema: a randomised trialLancet 2001; 356: 212632.

[12] Levine TB, Francis GS, Goldsmith SR, Simon AB, Cohn JN.Activity of the sympathetic nervous system and renin-angiotensin system assessed by plasma hormone levels andtheir relation to haemodynamic abnormalities in congestiveheart failure Am J Cardiol 1982; 49: 165966.

[13] Francis GS, Benedict C, Johnstone DEet al. Comparison ofneuroendocrine activation in patients with left ventriculardysfunction with and without congestive heart failureCirculation 1990; 82: 17249.

[14] Swedberg K, Eneroth P, Kjekshus J, Wilhelmsen L.Hormones regulating cardiovascular function in patients withsevere congestive heart failure and their relation to mortalityCirculation 1990; 82: 17306.

[15] Yasue H, Yoshimura M, Sumida H et al. Localization andmechanism of secretion of B-type natriuretic peptide in com-parison with those of A-type natriuretic peptide in normalsubjects and patients with heart failure Circulation 1994; 90:195203.

[16] Wei C-M, Heublein DM, Perella MAet al. Natriuretic peptidesystem in human heart failure Circulation 1993; 88: 10049.

[17] Yu CM, Sanderson JE. Plasma brain natriuretic peptide anindependent predictor of cardiovascular mortality in acute

heart failure Eur J Heart Failure 1999; 1: 5965.[18] Tsutamoto T, Wada A, Maeda K et al. Attenuation of

compensation of endogenous cardiac natriuretic peptide sys-tem in chronic heart failure: prognostic role of plasma brainnatriuretic peptide concentration in patients with chronicsymptomatic left ventricular dysfunction Circulation 1997; 96:50916.

[19] Heublin DM, Heser DW, Edwards BSet al. N-terminal (1-25)atrial natriuretic factor in human congestive heart failure: amarker for endogenous secretion J Am Coll Cardiol 1989; 13:9A.

[20] Goldstein DS, Feuerstein G, Kopin IJ, Keiser HR. Validityand reliability of liquid chromatography with electromechani-cal detection for measuring plasma levels of norepinephrineand epinephrine in man. Life Sci 1981; 28: 46775.

[21] Bellenger NG, Burgess MI, Ray SGet al. Comparison of left

ventricular ejection fraction and volumes in heart failure byechocardiography, radionuclide ventriculography and cardio-vascular magnetic resonance; are they interchangeable? EurHeart J 2000; 21: 138796.

[22] Kelly C, Newby DE, Boon NA, Douglas NJ. Supportventilation versus conventional oxygen Lancet 2001; 357:11257.

[23] Sacchetti AD, Harris RH, Paston C, Hernandez Z. Bi-levelpositive airway pressure support system use in acute conges-tive heart failure: preliminary case series Acad Emerg Med1995; 2: 71418.

[24] Pollack CV, Torres MT, Alexander L. Feasibility study of theuse of bilevel positive airway pressure for respiratory supportin the emergency department Ann Emerg Med 1996; 27:18992.

[25] Kelly AM, Bau S, Georgakos C, Rosengarten P. Experiencewith the use of continuous positive airway pressure (CPAP)therapy in the emergency management of acute severecardiogenic pulmonary oedema Aust NZ J Med 1997; 27:31922.

[26] Sharon A, Shpirer I, Kaluski E et al. High-dose intravenousisosorbide dinitrate is safer than Bi-PAP ventilation combinedwith conventional treatment for severe pulmonary edema JAm Coll Cardiol 2000; 36: 8327.

[27] Takeda S, Takano T, Ogawa R. The effect of nasal continuouspositive airway pressure on plasma endothelin-1 concen-trations in patients with severe cardiogenic pulmonary edemaAnesth Analg 1997; 84: 10916.

[28] Rusterholz T, Kempf J, Berton Cet al. Noninvasive pressuresupport ventilation (NIPSV) with face mask in patients withacute cardiogenic pulmonary edema Intensive Care Med 1999;25: 218.

[29] Qi W, Kjekshus J, Hall C. Differential responses of plasmaatrial and brain natriuretic peptides to acute alteration inatrial pressure in pigs Scand J Clin Lab Invest 2000; 60:5563.

[30] Lazzeri C, Franchi F, Porciani Cet al. Systemic hemodynam-ics and renal function during brain natriuretic peptide infusionin patients with essential hypertension Am J Hypertens 1995;8: 799807.

[31] Motwani JG, McAlpine H, Kennedy N, Struthers AD.Plasma brain natriuretic peptide as an indicator forangiotensin-converting-enzyme inhibition after myocardialinfarction. Lancet 1993; 341: 110913.

[32] Omland T, Aakvaag A, Bonarjee VVS et al. Plasma brainnatriuretic peptide as an indicator of left ventricular systolicfunction and long term survival after acute myocardialinfarction Circulation 1996; 93: 19639.

[33] Sumida H, Yasue H, Yoshimura M et al. Comparison ofsecretion pattern between A-type and B-type natriuretic pep-tides in patients with old myocardial infarction. J Am CollCardiol 1995; 25: 110510.

[34] Morita E, Yasue H, Yoshimura M et al. Increased plasmalevels of brain natriuretic peptide in patients with acutemyocardial infarction Circulation 1993; 88: 8291.

[35] Cohn JN, Levine TB, Olivari MT et al. Plasma norepine-phrine as a guide to prognosis in patients with chroniccongestive heart failure N Engl J Med 1984; 311: 81923.

[36] Missouris CG, Grouzmann E, Buckley MG, Barron J,MacGregor GA, Singer DR. How does treatment influenceendocrine mechanisms in acute severe heart failure? Effects oncardiac natriuretic peptides, the renin system, neuropeptide Yand catecholamines Clin Sci 1998; 94: 5919.

1386 C. A. Kellyet al.


randomised controlled trial of continuous positive airway pressure and standard oxygen therapy in...

Documents