Long-term outcome for obstructive sleep apnea syndrome patients. Mortality

DOI 10.1378/chest.94.6.1200 1988;94;1200-1204Chest

 M Partinen, A Jamieson and C Guilleminault syndrome patients. Mortality.Long-term outcome for obstructive sleep apnea

  http://chestjournal.chestpubs.org/content/94/6/1200

can be found online on the World Wide Web at: The online version of this article, along with updated information and services 

) ISSN:0012-3692http://chestjournal.chestpubs.org/site/misc/reprints.xhtml(without the prior written permission of the copyright holder.reserved. No part of this article or PDF may be reproduced or distributedChest Physicians, 3300 Dundee Road, Northbrook, IL 60062. All rights

ofbeen published monthly since 1935. Copyright1988by the American College is the official journal of the American College of Chest Physicians. It hasChest

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

1200 Long-term Outcome in OSA (Partinen, Jamieson, Guilleminau!t)

Long-term Outcome for Obstructive SleepApnea Syndrome Patients*MortalityMarkku Partinen, M.D.; AndrewJamieson, M.D.; and

Christian Guilleminault, M.D.

As the actual mortality and morbidity of obstructive sleepapnea syndrome (OSAS) have been unknown heretofore,we undertook a follow-up study of 198 OSAS patients seen

at the Stanford Sleep Disorders Clinic between 1972 and1980, for whom either tracheostomy (71 patients) or weight

loss (127 patients) had been recommended. At five-year

follow-up, all of the deaths (14) had occurred among those

conservatively treated with weight-loss (a mortality rate of

1 1 per 100 patients per five years). These patents also hada higher five-year crude vascular mortality rate: 6.3 per100 patients per five years, with an age-standardized

T he obstructive sleep apnea syndrome (OSAS) can

be associated with severe cardiac arrhythmia and

with hemodynamic changes during sleep. Anecdotally,

death during the night has been related to OSAS, but

the morbidity and mortality associated with OSAS

have been nebulous heretofore.24 Significant contro-

versy exists over the risks induced by mild to moderate

OSAS when severity is defined by the “apnea index”

(Al), defined as the number of apneas lasting longer

than 10 seconds per hour of sleep.

A follow-up study was undertaken to assess the

health risks associated with OSAS and the long-term

effects of two different treatments. Between 1972 and

1980, most OSAS patients seen at the Stanford Sleep

Disorders Clinic were examined by one physician and

were offered only two options: tracheostomy or a more

conservative medical approach consisting of weight

loss and better sleep hygiene. Our follow-up study

assessed the health risks associated with OSAS and

evaluated the long-term outcome for adult OSAS

patients based on type of treatment.

Criteria for Inclusion in the Study

Before searching our clinic records to form two

retrospective cohorts of surgically (tracheostomy) and

conservatively treated patients, we established the

*Fr�n� the Sleep Research Center, Stanford University School of

Medicine, Stanford.

A portion of this work was presented orally at the annual meetingofthe Sleep Research Society (Columbus, Ohio, USA, 1986).’

Manuscript received August 29; revision accepted September 21,for rapid publication.

vascular mortality rate of5.9 per 100 patients per five years(95 percent confidence interval [CI] 2.5-11.6) vs 0 per 100

for the surgically treated population; this despite a lowermean apnea index (43 versus 69) and a lower mean body

mass index (31 versus 34 kg/rn’) in the conservatively treated

group. With the fictional adjunction of one possible deathat five-year follow-up in the surgically treated group, theage-adjusted odds of vascular mortality at five years for the

conservatively treated group was 4.7. Our data thereforeencourage “aggressive” treatment for patients with OSAS.(Cheat 1988; 94:1200-04)

following requirements for subjects:1. Residence in a western state of the USA (California,

Nevada, Arizona, Oregon, or Washington), to facilitate

follow-up.

2. Previous polygraphic monitoring, with sufficient infor-

mation in the chart to calculate Al. (Hypopneas or partial

obstruction of the upper airway were not scored before

availability ofan accurate ear oximeter.)

3. Information on age, height, weight, cardiovascular dis-

eases, arterial hypertension, strokes, and cardiovascular

medication intake.

4. Age at least 16 years at entry, with a minimum of five

years since the initial polygraphic monitoring.

5, Signed general consent form, giving permission to use

data for research purposes and supplying the name of a

relative who might be contacted.

A total of 200 patients were potential candidates;

198 were identified and included in the study.

MATERIAL AND METHODS

DefinitionsArterial hypertension: high blood pressure diagnosed by a cardi-

ologist, and continuous administration of hypertensive medication

or blood pressure repeatedly recorded at 160/90 mm Hg or more.

Coronary artery disease: diagnosis by a cardiologist with pre-

scribed nitroglycerin, beta blockers, or calcium antagonists; or

bypass surgery.

Myocardial infarction, stroke: discharge diagnosis after hospital-

ization.

Excessive daytime sleepiness: difficulty in driving, working, or

interacting socially because of sleepiness.

Follow-up Information Search

Investigators rarely saw any of the patients at the time of the

follow-up study. Contacts were mainly by phone. The following

sources were contacted in order, until the necessary information

was obtained:

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

CHEST I 94 I 6 I DECEMBER, 1988 1201

1. Phone number on chart (minimum of three calls at short

intervals during one month).

2. Current phone number.

3. Letter to address of record, asking that we be contacted.

4. Work phone number on chart if authorized to call there.

5. Relative, if listed.

6. City directory: neighbors contacted.

7. Department of Motor Vehicles, if authorized by law.

8. As a last resort, the National Death Index. Ifa record was

available, we contacted the family about the cause of

death and obtained a death certificate from the appropri-

ate vital statistics office. (At the end of the study, these

certificates were disposed of per state regulations.)

Evaluation at Entry

The clinical interview, physical evaluation, and objective poly-

graphic recording were reviewed. Polygraphic recording was always

one all-night monitoring, It included electroencephalogram (EEG)

(C3/A2-C4/A1 of the 10-20 international electrode placement sys-

tem), electro-oculogram (EOG), chin electromyogram (EMG), elec-

trocardiogram (ECG-modified V2 lead), monitoring of respiration

by abdominal and thoracic strain gauges, and airflow by nasal and

mouth thermistors.’

Questionnaire

At follow-up, a standardized questionnaire ensured that the three

investigators asked identical questions in the same order. The test

took 15 to 30 minutes to administer. Ifa point was unclear, we asked

permission to contact the patient’s current physician.

( Following the initial evaluation and test, a recommendation for

treatment was sent to the referring physician and this recommen-

dation was available for review at the time of the follow-up study.

Frequently, however, the Stanford Sleep Disorders Clinic had done

no follow-up prior to this retrospective study, except for patients

who had been subjects of other research protocols.)

Outcome Measures

Mortality: Expected vs observed death rates and age-standardized

death rates were tabulated. The direct method with minimum

variance weights was used for the age�standardization.6 Time of

death was subdivided into three segments (2400-0800 h, 0800-1600

h, 1600-2400h) and cause ofdeath was analyzed, with identification

of cardiovascular and stroke-related deaths. Mortality was also

considered in comparison with body mass index (BMI)7 and with

BMI and age.

Statistical Analysis: Descriptive statistics with means, median

and standard deviation were used to analyze total population

and subgroups. Analysis of variance, Student’s t test, and non-

parametric statistics (depending on the distribution) were per-

formed using the BMDP Statistical package.’ Similarly, the BMDP

statistical package 1 L (life table method) was used to construct the

five-year survival curve. In comparing survival between the tra-

cheostomized and weight-loss-treated subgroups, the Wilcoxon-

Breslow and Mantel-Cox statistics were used.8 Crude mortality

rates (MR) and age-standardized mortality rates (SMR) were calcu-

The 95 percent CI for mortality rates and odds ratios were

determined as indicated by Schoenberg� and by Kahn.#{176}We also

calculated the CI for survival time between entry and five-year

follow-up with death as subsequent event, as described by Machin

and ‘#{176}The 1979 vital statistics for the US population were

used as comparison with the plotted five-year survival curve

(National Center for Health Statistics, 1984).”

RESULTS

The mortality rate was evaluated at five-year follow-

up for each patient.

Total Population

The initially seen population included 198 patients

with a median age of 52 years (mean 51.3, SD 11.3,

range 18-78). There were 190 men, median age 52

years (mean 51 . 1 , SD 1 1 .4, range 18-78) and eight

women, median age 57 years (mean 55.8, SD 9.1,

range 41-67) (t 1 . 16, p 0.24). At entry, 112 patients

(56.6 percent) presented with arterial hypertension.

Coronary artery disease had been diagnosed in 33

(16.7 percent) patients (31 men, two women). Previous

Table 1-Total Deaths in Population of198 OSAS Patients

Cause

Case Sex Age Al BMI HTA CHD CVD Tr Fol of Death

1 M 36.0 110.4 48.0 + - - C 4.90 Cardiac

2 M 66.6 68.0 28.4 + + + C 3.00 Cardiac

3 M 46.9 19.6 25.8 + - - C 0.94 Cardiac

4 M 56.4 46.5 31.9 + - - C 0.75 Cardiac5 M 49.9 20.0 31.0 - - - C 3.71 Cardiac

6 M 67.6 32.0 28.2 + + + C 2.49 Cardiac

7 M 70.3 32.0 25.4 - + - C 4.59 Cardiac

8 M 69.3 58.0 37.2 - - - C 2.24 Stroke9 M 69.3 12.0 24.3 + + - C 2.32 Cancer

10 M 56.9 64.7 28.4 - - - C 4.73 Cancer

11 M 54.7 66.0 30.9 + - - C 1.00 Cancer

12 M 55.7 19.0 39.8 + - + C 2.80 Cirrhosis*

13 M 56.9 59.0 28.4 - - - C 1.43 Shy-Dragert

14 M 56.7 77.3 37.5 - + - C 3.18 Infection

Mean 58.1 48.9 31.8 2.72

±SD 9.9 28.0 6.6 1.40

Age: age at entry; Al: apnea index; BMI: body mass index; CHD: coronary heart disease at entry; CVD: cerebrovascular disease at entry; Fol:years fbllowed; HTA: arterial hypertension at entry; Tr: treatment, conservative (C) or surgical.

*}�fient with a liver cirrhosis died with a sepsis.

tPatient died enroute to hospital; respiratory arrest had followed repetitive obstructive apneas of long duration.

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

1202 Long-term Outcome in OSA (Partinen, Jamieson, Guilleminault)

medical history noted myocardial infarction in 16 (8.1

percent) and stroke in 14 (7. 1 percent).

The polygraphic recordings gave . the following in-

formation: the median Al was 54.8, (mean 52.4, SD

30.6, range 5.3-130.0). The median Al for men was

53.5 (mean 52.0, SD 20.8), for women, 58.4 (mean

56.0, SD 28.9) t=0.7, p=O.46). The median BMI was

31.0 kg/in2, mean 32.8, SD 8.0, range 18.7-61.8. The

median BMI for men was 30.7 (mean 31.9, SD 8.0),

for women, 38.4 (mean 35. 1 , SD 7.6) t = 1 .25, p = 0.21.

Description ofthe Two Subgroups

The conservatively treated (weight-loss recommen-

dation) group A included 127 patients with, at entry,

a mean age of53 ± 11 years, a mean BMI of3l ± 8 and

a mean Al of 43 ± 30.5. The surgically (tracheostomy)

treated group B (71 patients) had a mean age of

48.8 ± 1 1 years (p<O.O2), a mean BMI of 34 ± 7.7

(p<0.Ol8) and a mean A! of69±23 (p<O.000l).

No statistically significant difference existed at entry

for frequency of hypertension, coronary artery dis-

ease, cerebrovascular disease, and chronic obstructive

pulmonary disease between the two groups. However,

the number of patients who had myocardial infarction

prior to being monitored for OSAS was significantly

different (p<O.O2), with six patients in the conserva-

tively treated group vs ten patients in the surgically

treated group. As expected, the more severe disease

at entry in group B had led to acceptance of the

reconirnended surgical treatment.

Cause of Death

There were 14 deaths, all in men, during the firstfive-year follow-up period. All ofthe deceased patients

had been treated conservatively without surgery.

Eight of the deaths were considered “vascular,�’ (Table

1) and at entry most ofthe 14 had presented with one

or more of the following conditions: coronary artery

disease, cerebrovascular disease, or arterial hyperten-

sion (Table 1). The mortality rate for conservatively

treated patients was 11.0 per 100 patients per five

years (95 percent CI 6.0-18.5).

Vascular Death

As vascular death is considered a potential risk of

OSAS due to the marked hemodynamic and heart

conduction changes monitored in association with

obstructive apneas, a more specific analysis was per-

formed taking vascular death (myocardial infarction

and cerebrovascular accident) as the end-point. The

BMI of the eight men thus deceased was 32 ± 7.5

kg/m2 and their Al was 48.3 ± 30.5. Age at entry was

the only significant difference noted between the

vascular death group and the total group of patients

conservatively treated: mean age for the vascular death

group was 57.9 ± 12.7 vs 53 ± 11 years (p<O.O3).

For the conservatively treated group, the crude

vascular mortality rate was 6. 3 deaths per 100 patients

per five years (95 percent CI 2. 7-12.4). The age-

adjusted five-year vascular mortality rate was 5.9 per

100 patients (95 percent CI 2.5-11.6). As no death was

monitored in the tracheostomy group, the difference

in survival was highly significant using Mantel-Cox

statistics (M = 4.8, p<O.O3). Using the Kaplan-Meier

method,8 the estimated percentage of vascular mor-

tality to five years was 8 percent (with CI 1. 1-15.0

percent). Figure 1 presents the five-year survival

curve, taking into consideration only vascular death

risk and comparing it with the tracheostomy patient

group and the US age-adjusted death rate.4

Independent of the approach that we considered,

(total number of deaths, or the more limited number

of vascular deaths, or calculation of age-adjusted vas-

cular mortality rate or estimation of mortality to five

years with evaluation of survival time), conservatively

treated patients had a much higher mortality risk at

five years and higher vascular mortality risk at five

years. Finally, with a fictional adjunction of one possi-

ble death at five-year follow-up in the surgically treated

group to calculate the age-adjusted odds of vascular

mortality at five years, the conservatively treated group

had a five-year age-adjusted odds of 4.7 for vascular

death.

DISCUSsIoN

Our report presents a prospective follow-up study

based on two retrospectively selected cohorts of pa-

tients, seen as early as 1972 when tracheostomy was

the only surgical treatment4 and weight-loss the only

0.975

0.95

OSAS patients with an Al > 55-year survival curve

Vascular deaths (8 vs 0)

Proportion surviving

0.925Estimated mortality to 5 years - 8% (95% Cl: 1.1% 15.0%)

0.9 Wilcoxon-Bresiow #{149}4.719. Mantel-Cox #{149}4.724. p - 0.03

0.875 �-�-�- -----��� L�. � � A

0 1 2 3 4 5

�-,-.Coniervative

. U.8. age-adjusted

Years of follow-up

-1�-- Tracheotomy (N.Til

FIGURE 1 . Comparison of survival curves of OSAS patients within afive-year follow-up study. The two patient groups are those treated

surgically with tracheostomy (N = 71) (upper curve) vs those treated

conservatively with weight-loss recommendation (N = 127) (lowercurve). The US age-adjusted survival curve for the general popula-lion (middle curve) is also presented. There is a significant difference

between the two patient group survival curves at five years. Despite

the fact that patients treated with tracheostomy had more severe

obstructive sleep apnea syndrome, no vascular death was noted at

five-year follow-up.

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

CHEST I 94 I 6 I DECEMBER, 1988 1203

alternative. At that time, when OSAS was less com-

monly recognized, patients were usually more severely

affected than those seen today. However, the A! range

at entry was between 5.3 and 130.0 for the total group,

and 12.0 and 110.0 for the “death at five-year follow-

up” group. No oxygen saturation measurements were

available, as the first reliable ear-oximeter came on

the market iii 1979.

Patients who accepted tracheostomy as treatment

were more severely affected, based upon symptom list

and Al. All had been excessively sleepy and conse-

quently at risk while driving. Their improvement

following tracheostomy may explain why no driving or

industrial accident-related death was seen in our

follow-up study. Our analysis has been very conserva-

tive overall. We have performed many of our calcula-

tions not on the total number of deaths (with a

mortality rate of 11 per 100 patients per five years),

but only on the vascular deaths. This was done due to

the previous findings of a significant relationship

between heavy snoring and hypertension, 1214 heavy

snoring and myocardial infarction and cerebrovascular

accidents, 15. 16 and OSAS and monitored cardiovascular

changes during sleep.

Our conservative approach, for example, excluded

from the vascular death analysis the patient with Shy-

Drager syndrome who died en route to the hospital

after respiratory arrest following progressively longer

repetitive obstructive apneas. On the other hand, a

patient with Shy-Drager syndrome in the tracheos-

tomy group survived at five-year follow-up.

It must be emphasized that there were no significant

differences at entry when hypertension, cardiovascular

disease, and associated chronic obstructive pulmonary

disease (COPD) were assessed in the two conserva-

tively vs surgically treated patient groups, indicating a

similar initial terrain. If BMI was higher (p<0.02) in

the tracheostomy group, subjects were mildly older in

the conservatively treated and in the vascular death

subgroup (mean=49±11 vs 53±11 vs 58± 12.7

years).

This may be an issue. The age factor could be

interpreted in several ways: (a) older patients may

have had a mild or moderate OSAS for a longer time

with correspondingly greater chance of cardiovascular

or other system impairment; (b) other factors not

analyzed in our study may be associated with aging,and when combined with mild to moderate OSAS,

may play a role in the mortality rate of the conserva-

tively treated patients.

We have used two different approaches for mortality

rate and survival curve calculations. When calculating

CI for survival time analysis, the Kaplan-Meier prod-

uct limit technique is the recognized approach. How-

ever, the formulas for CI with small sample sizes and

for p close to 0 or 1, as in the present case when we

considered vascular deaths, must be interpreted with

caution. We therefore performed calculations for mor-

tality rates and 95 percent CI following Schoenberg,9

as we have a Poisson-type distribution. Our results are

thus always presented with a conservative approach.

In a similar vein, odds ratios for vascular death can be

calculated if we agree to add one fictitious death to

the surgical group. The calculated odds ratio is then

4. 7 for vascular death. Independent of the approach

taken, the death risk is always higher in the conser-

vatively treated group, for whom there is also a higher

chance of vascular death at five-year follow-up.

Our findings indicate that moderate OSAS has a

mortality risk and that this syndrome may be effec-

tively treated, as in the past, by tracheostomy. How-

ever, new methods of treatment-nasal continuous

positive airway pressure, uvulopalatopharyngoplasty,

and maxillofacial surgery-now exist, and their use-

fulness will be shown by future follow-up studies. Our

report should be associated with the very recent

publication of He et al.� It brings complementary

information, as we were able to perform a more in-

depth investigation with documentation of the cause

of death in each case.

ACKNOWLEDGMENTS: This work was supported by grant AG06066 from the National Institute of Aging Dr Partinen wassupported by U.S. Public Health Service International ResearchFellowship 1 FO 5TW03648-O1. We thank Boyd Hayes for providingtechnical assistance and Alison Grant for editing the manuscript.

REFERENCES

1 Partinen M, Jamieson A, Guilleminault C. Mortality of patients

with obstructive sleep apnea syndrome: a follow-up study. Sleep

Res 1986; 15:153

2 MacGregor MI, Block AJ, Ball WC. Serious complication and

sudden death in the Pickwickian syndrome. Johns Hopkins Med

J 1970; 126-127:279-95

3 Guilleminault C. Natural history, cardiac impact, and long-term

follow-up of sleep apnea syndrome. In: Guilleminault C, Luga-

resi E, eds. Sleep/wake disorders: Natural history, epidemiology,

and long-term evolution. New York: Raven Press, 1983:107-25

4 Guilleminault C, Simmons FB, MottaJ, CummiskeyJ, Rosekind

M, Schroeder JS, et al. Obstructive sleep apnea syndrome and

tracheostomy: A long-term follow-up experience. Arch Intern

Med 1981; 141:985-88

5 Guilleminault C, ed Sleep and waking disorders: indications

and techniques. Menlo Park: Addison-Wesley, 1982

6 Kahn HA. An introduction to epidemiologic methods. New

York: Oxford University Press, 1983

7 Khosla T, Lowe FR. Indices ofobesity derived from body weight

and height. Br J Prey Soc Med 1967; 21:122-28

8 Dixon WJ, Brown MB, Engelman L, Frame JW, Jennrich RI,

Toporek JD. BMDP statistical software. Los Angeles: University

ofCalifornia Press, 19859 Schoenberg BS. Calculating confidence intervals for rates and

ratios. Simplified method utilizing tabular values based on the

Poisson distribution. Neuroepidemiol 1983; 2:257-65

10 Machin D, Gardner MJ. Calculating confidence intervals for

survival time analysis. Br Med J 1988; 296:1369-76

1 1 National Center for Health Statistics. Vital statistics ofthe United

States, 1979, Vol II, Mortality, Part A. DHHS Pub. No. (PHS)

84-1101 Public Health Service. Washington, DC: U.S. Govern-

ment Printing Office, 1984

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

Postgraduate Course: Practical RadiologyThe University of Arizona Health Sciences Center will present this sixth annual course

February 27-March 3 at Loews Ventana Canyon Resort, Tucson. For information, contact theOffice of Medical Education, Arizona Health Sciences Center, Tucson 85724 (602) 626-7832.

1204 Long-term Outcome in OSA (Partinen, Jamieson, Guilleminault)

12 Mondini 5, Zucconi M, Cirignotta F, Aguglia U, Lenzi PL,Zauli C, et al. Snoring as a risk factor for cardiac and circulatory

problems: an epidemiological study. In: Guilleminault C, Lu-

garesi E, eds. Sleep/wake disorders: Natural history, epidemi-ology, and long-term evolution. New York: Raven Press,

1983:107-25

13 Koskenvuo M, Kaprio J, Partinen M, Langinvaino H, Sarna S,

Heikkila K. Snoring as a risk factor ftr hypertension and angina

pectoris. Lancet 1985; i:893-95

14 Norton PG, Dunn EV Snoring as a risk factor fur disease: an

epidemiological survey. Br Med J 1985; 291:630-32

15 Partinen M, Palomaki H. Snoring and cerebral infarction. Lancet

1985; ii:1325-26

16 Koskenvuo M, Kaprio J, Telakivi T, Partinen M, Heikkila K,

Sarna S. Snoring as a risk factor for ischaemic heart disease and

stroke in men. Br Med J 1987; 294:16-9

17 Coccagna G, Mantovani M, Brignani F, Parchi C, Lugaresi E.

Enregistrement continu de la pression arterielle pulmonaire et

systemique pendant le sommeil dans les syndromes d�hypersom-

nie avec respiration periodique. Bull Eur Physiopathol Respir

1972; 8:1159-72

18 Tilkian AG, Guilleminault C, Schroeder JS, Lehrman KL,Simmons FB, Dement WC. Hemodynamics in sleep induced

apnea: studies during wakefulness and sleep. Ann Intern Med

1976; 85:714-19

19 Tolle FA, Judy W\� Yu PL, Markand ON. Reduced stroke

volume related to pleural pressure in obstructive sleep apnea.

J AppI Physiol 1983; 55:1718-2420 Guilleminault C, Moth J, Mthm F, Melvin K. Obstructive sleep

apnea and cardiac index. Chest 1986; 89:331-34

21 Shepard JW Jr. Gas exchange and hemodyamics during sleep.

Med Clin North Am 1985; 69:1243-64

22 He J, Kryger MH, Zorick FJ, Conway W, Roth T Mortality and

apnea index in obstructive sleep apnea. Experience in 385 male

patients. Chest 1988; 94:9-14

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from

DOI 10.1378/chest.94.6.1200 1988;94; 1200-1204Chest

M Partinen, A Jamieson and C GuilleminaultLong-term outcome for obstructive sleep apnea syndrome patients. Mortality.

 July 10, 2011This information is current as of

http://chestjournal.chestpubs.org/content/94/6/1200Updated Information and services can be found at:

Updated Information & Services

http://chestjournal.chestpubs.org/content/94/6/1200#related-urlsThis article has been cited by 76 HighWire-hosted articles:

Cited Bys

http://www.chestpubs.org/site/misc/reprints.xhtmlonline at: Information about reproducing this article in parts (figures, tables) or in its entirety can be foundPermissions & Licensing

http://www.chestpubs.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

Reprints

the right of the online article.Receive free e-mail alerts when new articles cite this article. To sign up, select the "Services" link to

Citation Alerts

slide format. See any online figure for directions. articles can be downloaded for teaching purposes in PowerPointCHESTFigures that appear in Images in PowerPoint format

 © 1988 American College of Chest Physicians by guest on July 10, 2011chestjournal.chestpubs.orgDownloaded from