Improving Long-term Prognosis for Survivors of

Mechanical Ventilation in Patients With AIDS

With PCP and Acute Respiratory Failure
Five-Year Follow-up of Intensive Care Unit Discharges
Cory Franklin, MD; Yaakov Friedman, MD; Terrence Wong, MD; Tzyy-Chyn Hu, RN, MS

Background: Before 1987, the hospital survival of pa- lated on the basis of the life-table analysis. Median sur-
tients with acquired immunodeficiency syndrome, Pneu- vival was estimated by means of the product-limit method.
mocystis carinii pneumonia, and acute respiratory fail-
ure receiving mechanical ventilation was less than 15%. Results: During the 5-year follow-up of the 47 sub-
Hospital survival has improved since then, but con- jects, 31 patients died, 12 were unavailable for follow\x=req-\
cerns have been raised that the post\p=m-\hospitaldischarge up, and four were still alive at the end of the cutoff. The
survival of these patients remains extremely poor. This cumulative survival rate at 1 year was 80% (95% confi-
study evaluated the long-term survival of patients dis- dence interval, 92% to 68%); at 2 years, 49% (95% con-
charged alive after an acute episode of acute respiratory fidence interval, 65% to 34%); at 3 years, 18% (95% con-
failure caused by P carinii pneumonia. fidence interval, 32% to 4%); and at 4 years, 6% (95%
confidence interval, 17% to 0%). Median survival time
Methods: A prospective cohort study was conducted for for all subjects was 602 days (1.65 years), and the long-
the 5-year period from May 1987 through May 1992 in est survival time for a single patient was 1774 days (4.86
an urban teaching hospital. Forty-seven patients dis- years).
charged from the hospital after receiving mechanical ven-
tilation and/or continuous positive airway pressure for Conclusions: Post\p=m-\hospitaldischarge survival of pa-
acquired immunodeficiency syndrome, P carinii pneu- tients with acquired immunodeficiency syndrome, P ca-
monia, and acute respiratory failure were followed up from rinii pneumonia, and acute respiratory failure has im-
their initial intensive care unit admission until death or proved dramatically in the past decade. Patients can
termination of the study to measure the long-term sur- undergo intubation and mechanical ventilation with the
vival and cumulative probability of survival of the study hope of reasonable long-term survival.
cohort. Actuarial life-table analysis was performed, and
long-term cumulative probability of survival was calcu- (Arch Intern Med. 1995;155:91-95)

PNEUMOCYSTIS
CARlNíi pneu¬ tilation, which had been common during
monia (PCP) remains the the early years of the AIDS epidemic,
most common opportunis¬ dropped off precipitously despite in¬
tic infection in patients with creases in the numbers of hospital admis¬

acquired immunodeficiency sions for PCP.6 Questions were raised re¬

syndrome (AIDS), occurring within 1 year garding the cost and ethical propriety of
in 18% of patients with baseline CD4 admitting patients with PCP and ARF to
counts less than 0.20X 109/L who
(200/µ _) the intensive care unit (ICU). It was sug¬
fail to receiveadequate prophylaxis.1The gested that in most situations these pa¬
most severe manifestation of PCP is acute tients be excluded from receiving aggres¬
respiratory failure (ARF), where the or¬ sive life-sustaining care.4,7·8
ganism causes a form of lung injury clini¬ Beginning in 1988, several medical
cally indistinguishable from adult respi¬ centers reported improved hospital sur¬
ratory distress syndrome.2 vival for these patients.9"11 Since PCP with
Before 1987, the prognosis for pa¬ ARF was no longer considered uniformly
tients seropositive for human immunode¬
From the Department of
Medicine, Cook County ficiency virus (HIV) who had PCP and de¬
Hospital, Chicago, Ill, and veloped ARF was grim, with hospital See Patients and Methods
University of Health Sciences/ survival rates of less than 15%.3"5 Owing
on next page
The Chicago Medical School, to the poor response of these patients, the
North Chicago. use of intensive care for mechanical ven-

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 06/20/2015

 Table 1. Characteristics of Study Subjects (n=47)
PATIENTS AND METHODS No. (%) of
Variable Patients
The study was initiated in the medical ICU of a large Age, y
inner-city public hospital. Beginning on May 1,1987, =:40 y 12(25.5)
all patients with AIDS who had PCP and ARF as de¬ <40 y 35 (74.5)
fined previously19 and who received mechanical ven¬ Sex
tilation and/or continuous positive airway pressure and M 46 (97.9)
survived to hospital discharge were enrolled. All these F 1 (2.1)
patients were closely followed up in the HIV outpa¬ Ethnic group
tient clinic after hospital discharge. Telephone con¬ White 17 (36.2)
tact was attempted for patients who did not keep their Black 23 (48.9)
routine clinic appointments. Survival was measured in Hispanic 7(14.9)
Transmission category
days from the date of medical ICU admission to death.
Deaths were ascertained on the basis of death certifi¬ Homosexual 24(51.1)
cate, hospital records, or information provided from
Injecting drug user 12 (25.5)
Heterosexual 9(19.1)
relatives or telephone contacts. The cutoff date of en¬
Other 2 (4.2)
rollment for this report was May 31, 1992, with con¬
tinued follow-up through July 31, 1993. This yielded
a total of 47 effective cases for final analysis.
Actuarial life-table analysis was performed to ana¬ mosexual (n=24 [51.1%]), and less than 40 years old
lyze disease progression. Long-term cumulative prob¬ (n=35 [74.5%] ). The median age of the group was 36 years
ability of survival was calculated on the basis of the (mean age±SD, 36.47±9.34 years).
life-table analysis. Median survival time was esti¬
mated by the product-limit method. Gehan's Gener¬ Among the 47 study subjects, there were 31 con¬
alized Wilcoxon Test and log-rank test were used to firmed deaths (66%). Four patients (8.5%) were still alive
test homogeneity and trend of subgroups. Confi¬ at the time of cutoff of follow-up, and 12 (25.5%) could
dence intervals were calculated by Greenwood's for¬ not be contacted at some point and were considered un¬
mula and normal approximation. available for follow-up at that point. Overall cumulative
survival rate was 80% at 1 year (95% confidence inter¬
val, 92% to 68%), 49% at 2 years (95% confidence inter¬
val, 65% to 34%), 18% at 3 years (95% confidence inter¬
fatal,ICU admissions again rose, despite a decrease in val, 32% to 4%), and 6% at 4 years (95% confidence
the number of diagnoses of PCP owing to antiviral therapy interval, 17% to 0%).
and PCP prophylaxis.12"14 This prompted a second wave Observed median survival time for all study sub¬
of ethical and cost analysis. Although acute survival had jects was 602 days (1.65 years). The longest survival
improved, the long-term prognosis of these patients was time for a single patient was 1774 days (4.86 years).
thought to be poor. This perpetuated the question of Table 2 shows the subgroup median survival time
whether or not patients with AIDS who had PCP and ARF and cumulative probability of survival for 1, 2, and 3
were appropriate candidates for ICU and life-sustaining years. The survival curves for the entire cohort are
therapy. depicted in Figure 1. Hispanic patients had the long¬
Questions regarding ICU use and chronic diseases, est median survival (734 days), followed by blacks
such as AIDS, involve issues of medical futility, ration¬ (559 days) and whites (545 days). Median postadmis¬
ing, and justice.15"17 Brown and Sprung,18 in addressing this sion survival time for patients entered in the earliest
issue, stated, "... good medical ethics should be based on year of the study (1987) was shorter than for those
sound medical knowledge and up-to-date data. Changes entered in the next 2 years (1988 and 1989), with sur¬
in treatment policy based on the latest and most reliable vivals of 382, 518, and 599 days, respectively,
data are no less an ethical duty than a medical one." Along although the subgroup comparisons and trends were
these lines, as part of our ongoing surveillance of HIV- not statistically significant (Figure 2).
infected patients who have received ICU care and me¬
chanical ventilation, we report our 5-year follow-up of AIDS COMMENT
survivors of PCP with ARF. This series includes a previ¬
ously described group followed up for up to 3 Vi years along The initial optimism over the improved hospital sur¬
with a group of new patients discharged during the ex¬ vival of patients with AIDS with PCP and ARF has been
panded follow-up interval.19 To our knowledge, this re¬ tempered by concern that long-term survival for these
port represents the longest follow-up for the largest pub¬ patients was poor and specifically that 1-year survival was
lished series of AIDS survivors with PCP and ARF at a single less than 10%.13 Wächter and associates6 reported in 1986
medical center in either North America or Europe. that median survival for five patients who survived an
episode of mechanical ventilation was 7 months. Since
RESULTS that time, Friedman et al19 demonstrated that 74% of the
patients with AIDS with PCP who survived hospitaliza-
Table 1 lists the characteristics of the study subjects. tion for ARF were alive at 1 year. The current updated
The cohort was predominantly male (n=46 [97.9%]), ho- data indicate a 1-year survival of 80%, median postad-

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 06/20/2015

 Table 2. Length of Survival and Cumulative Probability of Survival4
Survival Probabilities (95% Confidence Interval), %
Median
Patient Category Survival, d 1y 2y 3y
Sex
M 588 79(67-91) 48 (32-64) 19(4-34)
F •
••t
Ethnic group
White 545 74(52-96) 48 (20-76) 32(1-64)
Black 559 77(60-95) 44 (22-67) 12 (0-30)
Hispanic 734 100(...t) 69(34-100) 17 (0-47)
Transmission category
Homosexual 502 73 (55-92) 31 (11-52) 16(0-34)
Injecting drug user 799 83(61-100) 71 (42-99) 30 (0-69)
Heterosexual 765 88(67-100) 74(42-100) 18 (0-50)
Other ...t
Year entered study
1987 382 57(21-94) 29 (0-62) ...t
1988 518 80(45-100) 40 (0-83) 40 (0-83)
1989 559 83 (62-100) 50 (22-78) 8 (0-24)
Age, y
a40 458 82(59-100) 55 (20-89) ...t
<40 601 79(65-93) 48 (30-66) 15 (1-29)
*Data for patients with acquired immunodeficiency syndrome with Pneumocystis carinii pneumonia and acute respiratory failure, Cook County Hospital
Medical Intensive Care Unit, Chicago, III, May 1, 1987, through May 31, 1991.
\Data do not have variation for calculation.

mission survival of 602 days, and maximum survival of infection and may also be a factor in increasing sur¬
4.86 years. We anticipate that as greater numbers of pa¬ vival.28·30
tients with AIDS who have PCP survive ARF and are fol¬ In identifying improved survival, two important di¬
lowed up for longer periods, these figures will continue agnostic artifacts, lead-time bias and length bias, must
to improve. Survival of at least 6 years has already been be considered.31·32 Lead-time bias confounds all studies
reported for one PCP, and ARF.20
survivor of AIDS, of survival with AIDS and refers to an artifactual, rather
Harris21 and Lemp et al22 both demonstrated im¬ than an actual, increase in survival because the diagno¬
proved survival trends for American patients with AIDS sis is made earlier in some patients than in others (in this
initially diagnosed as having PCP as the 1980s pro¬ case AIDS may first be established by the presence of PCP

gressed. During the same period, Whitmore-Overton et with ARF). Lead-time bias is unlikely to account com¬
al23 observed an improvement in survival in English pa¬ pletely for the coincidental improvements described in
tients with AIDS including, but not limited to, patients different locations. Length bias may artificially improve
presenting with PCP. Danish researchers also docu¬ survival because AIDS is not a single disease entity; rather,
mented a similar trend of increasing survival over time it is a diagnosis established when an HIV-infected per¬
for patients with AIDS in whom PCP was diagnosed.24 son develops a qualifying condition. Those HIV-
In a large-scale follow-up of 28 Danish patients with AIDS infected patients with PCP and ARF as their qualifying
who received mechanical ventilation because of PCP, condition for AIDS could conceivably have a long pre-
Nielsen et al23 observed a dramatic improvement in sur¬ clinical-but-detectable phase and thus be more easily iden¬
vival after 1986, with hospital survival of 50% and me¬ tified than other HIV-infected patients. This could ex¬
dian postepisode survival of 20.5 months, figures simi¬ plain why patients whose AIDS diagnosis is established
lar to those reported here. by PCP with ARF might have improved survival com¬
While these studies strongly suggest that survival pared with those with other qualifying conditions (dif¬
trends for patients with AIDS who have PCP have gen¬ ferent opportunistic infections, HIV-related malignant
erally improved in both the United States and Europe, neoplasms, or neurologic conditions), but it would not
the reasons for the improvement are less clear. Corti- explain the specific temporal improvement in this group.
costeroids are thought to have played a role in the A final possibility in explaining improved survival
improvement in immediate hospital survival.19·20·26 is the experience or learning curve phenomenon, in which
Most authorities believe that nonreporting of deaths is centers with extensive experience taking care of pa¬
not a major factor, since reporting and follow-up are tients with AIDS show better survival results.33 This has
reasonably complete and have generally improved in been demonstrated through variations in survival in pa¬
the past several years.27 The impact of zidovudine on tients with AIDS in different centers and may explain the
long-term survival of patients with AIDS who have tendency to improved survival in a single medical cen¬
PCP is assumed to have played a part, although there ter.19·25
is some controversy surrounding its effect.28·29 Prophy¬ This study did not allow us to identify a specific rea¬
laxis for PCP reduces the incidence of PCP during HIV son for the improved long-term survival of our patients.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 06/20/2015

 1.00- * 1.00

0.75 0.75

"E 0.50 'g 0.50

200 400 600 800 1000 1200 1400 1600 1800 200 400 600 800 1000 1200 1400 1600 1800
Days of Survival After ICU Admission Days of Survival After ICU Admission

Figure 1. Survival curves in patients with acquired immunodeficiency Figure 2. Survival curves in patients with acquired immunodeficiency
syndrome with Pneumocystis carinii pneumonia and acute respiratory syndrome with Pneumocystis carinii pneumonia and acute respiratory
failure after mechanical ventilation. ICU represents intensive care unit. failure after mechanical ventilation according to the year in which the
patient entered the study. ICU represents intensive care unit.

We agree with Wächter et al12 that demographic, clini¬ sonably constant and we found no obvious differences
cal, and therapeutic variables do not completely explain between patients who entered the study at different in¬
why patients with AIDS, PCP, and ARF who receive me¬ tervals.
chanical ventilation are experiencing better ICU sur¬ A final caveat regarding the decision to institute
vival. They speculated that a change in either the HIV or life-support in patients with ARF and PCP is that,
the infected hosts has made PCP and ARF a less virulent because approximately half the patients die within a
process, and that this remains the leading possibility for short period, short-term mortality must be factored
improved hospital outcomes. Others have also alluded into the long-term prognosis for any individual
to an attenuation in virulence of the HIV virus.34 patient. itself, this should not be a deterrent
In and of
The main limitation of our data is that they repre¬ to ICU and mechanical ventilation.35 Currently,
care
sent a selected sample of patients with AIDS who have hospital survival of patients with PCP with ARF
PCP and ARF, excluding those who elected not to un¬ ranges between 18% and 55%, essentially equivalent to
dergo mechanical ventilation and ICU care, either be¬ survival from other noncardiogenic causes of respira¬
cause they suffered from other chronic comorbid con¬ tory failure with a similar degree of lung injury.36
ditions (eg, HIV-related malignant neoplasms, neurologic Recent data suggest that survival from HIV-related res¬
diseases, or chronic intercurrent infections) or simply be¬ piratory failure carries a survival comparable with that
cause they did not wish to undergo ICU care. The pa¬ of non-HIV-related respiratory failure.37
tients we described were primarily patients without co- In conclusion, the long-term survival for ICU sur¬
morbid conditions who were experiencing first-time vivors with AIDS, PCP, and ARF has improved dramati¬
episodes of PCP. Hospital surveillance indicated that this cally since initial reports in the mid-1980s. Combined
represented most of the patients with AIDS who had ARF with increases in recovery from ARF demonstrated at a
and PCP in our hospital. (Few patients with ARF and PCP, number of centers, these figures suggest that in HIV-
but no comorbidity, refused life support. Some patients infected patients without serious comorbid conditions,
who refused ICU care may have also refused diagnostic ARF caused by PCP is a manageable complication with
procedures to confirm PCP.) Current literature on PCP reasonable short- and long-term survival. The occur¬
and ARF suggests that this is also the experience at other rence of ARF secondary to PCP should not be consid¬
centers dealing with large numbers of HIV-infected pa¬ ered an end-stage complication of AIDS; in most cases it
tients.12·25 Nevertheless, long-term survival figures in pa¬ can be approached as a potentially treatable intercur¬
tients with ARF must be viewed with caution, because rent condition. Future studies defining the role of such
in some patients long-term survival may be related to co¬ factors as CD4 levels, antiretroviral therapy, and second¬
existing comorbidity. ary PCP prophylaxis undoubtedly will help to explain
A second limitation of the data is inherent in the these findings more completely. Our intent in this study
methods of this type of follow-up. By definition, the co¬ was to document that patients with AIDS who have PCP
hort studied was not a single prospectively identifiable who are willing to undergo intubation and mechanical
group. Entry into the study was predicated on survival ventilation for ARF and subsequently survive the acute
from ARF. Because not all patients entered the study at episode may now experience survival measurable in years,
the same time, outside factors affecting AIDS survival in contrasted with weeks or months, as previously de¬
the 5-year period could have influenced the type of pa¬ scribed.
tients who survived and became eligible for follow-up (eg,
greater numbers of healthy survivors early in the sur¬ Accepted for publication April 29,1994.
veillance period could skew the results toward greater Reprint requests Medical
toIntensive Care Unit, Cook
long-term survival). We do not believe this played a ma¬ County Hospital, 1835 W Harrison St, Chicago, IL 60612
jor role, because the acute mortality for ARF was rea- (Dr Franklin).

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 06/20/2015

 19. Friedman Y, Franklin C, Freels S, Well MH. Long-term survival of patients with
REFERENCES
AIDS, Pneumocystis carinii pneumonia, and respiratory failure. JAMA. 1991;
266:89-92.
1. Phair J, Munoz A, Detels R, et al. The risk of Pneumocystis carinii pneumonia 20. Wachter RM, Luce JM, Hopewell PC. Critical care of patients with AIDS. JAMA.
among men infected with human immunodeficiency virus type 1. N Engl J 1992;267:541-547.
Med. 1990;322:161-165. 21. Harris JE. Improved short-term survival of AIDS patients initially diagnosed
2. Stokes DC, Hughes WT, Alderson PO, King RE, Garfinkel DJ. Lung mechanics, with Pneumocystis carinii pneumonia, 1984 through 1987. JAMA. 1990;263:
radiography and 67 Ga scintigraphy in experimental Pneumocystis cariniipneu- 397-401.
monia. Br J Exp Pathol. 1986;67:683-693. 22. Lemp GF, Payne SF, Neal D, Temelso T, Rutherford GW. Survival trends for
3. Murray JF, Felton CP, Garay SM, et al. Pulmonary complications of the ac- patients with AIDS. JAMA. 1990;263:402-406.
quired immunodeficiency syndrome. N Engl J Med. 1984;310:1682-1688. 23. Whitmore-Overton SE, Tillett HE, Evans BG, Allardice GM. Improved survival
4. Rosen MJ, Cucco RA, Teirstein AS. Outcome of intensive care in patients with from diagnosis of AIDS in adult cases in the United Kingdom and bias due to
the acquired immunodefiency syndrome. J Intens Care Med. 1986;1:55-60. reporting delays. AIDS. 1993;7:415-420.
5. Murray JF, Garay SM, Hopewell PC, et al. NHLBI workshop summary: pulmo- 24. Pedersen C, Gerstoft J, Tauris P, et al. Trends in survival of Danish AIDS pa-
nary complications of the acquired immunodeficiency syndrome: an update. tients from 1981 to 1989. AIDS. 1990;4:1111-1116.
Am Rev Respir Dis. 1987;135:504-509. 25. Nielsen TL, Guldager H, Pedersen C, Mathiesen L, Nielsen JO. The outcome of
6. Wachter RM, Luce JM, Turner J, Volberding P, Hopewell PC. Intensive care of mechanical ventilation in patients with an AIDS-associated primary episode of
patients with the acquired immunodeficiency syndrome: outcome and chang- Pneumocystis carinii pneumonia. Scand J Infect Dis. 1991;23:37-41.
ing patterns of utilization. Am Rev Respir Dis. 1986;134:891-896. 26. Montaner JS, Russell JA, Lawson L, Ruedy J. Acute respiratory failure sec-
7. Lo B, Raffin TA, Cohen NH, et al. Ethical dilemmas about intensive care for ondary to Pneumocystis carinii pneumonia in the acquired immunodeficiency
patients with AIDS. Rev Infect Dis. 1987;9:1163-1167. syndrome: a potential role for systemic corticosteroids. Chest. 1989;95:881\x=req-\
8. Wachter RM, Cooke M, Hopewell PC, Luce JM. Attitudes of medical residents 884.
regarding intensive care for patients with the acquired immunodeficiency syn- 27. Piette J, Mor V, Fleishman J. Patterns of survival with AIDS in the United States.
drome. Arch Intern Med. 1988;148:149-152. Health Serv Res. 1991;26:75-95.
9. El-Sadr W, Simberkoff MS. Survival and prognostic factors in severe Pneu- 28. Graham NMH, Zeger SL, Park LP, et al. The effects on survival of early treat-
mocystis carinii pneumonia requiring mechanical ventilation. Am Rev Respir ment of human immunodeficiency virus infection. N Engl J Med. 1992;326:
Dis. 1988;137:1264-1267. 1037-1042.
10. Friedman Y, Franklin C, Rackow E, Weil MH. Improved survival in patients 29. Aboulker JP, Swart AM. Preliminary analysis of the Concord trial. Lancet. 1993;
with acquired immunodeficiency syndrome, Pneumocystis carinii pneumonia 341:889-890.
and severe respiratory failure. Chest. 1989;96:862-866. 30. Leoung GS, Feigal DW Jr, Montgomery AB, et al. Aerosolized pentamidine for
11. Efferen LS, Nadarajah D, Palat DS. Survival following mechanical ventilation prophylaxis against Pneumocystis carinii pneumonia: the San Francisco com-
for Pneumocystis pneumonia in patients with the acquired immunodeficiency munity prophylaxis trial. N Engl J Med. 1990;323:769-775.
syndrome: a different perspective. Am J Med. 1989;87:401-404. 31. Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenon: stage mi-
12. Wachter RM, Russi MB, Bloch DA, Hopewell PC, Luce JM. Pneumocytis gration and new diagnostic techniques as a source of misleading statistics for
carinii pneumonia and respiratory failure in AIDS: improved outcomes and survival in cancer. N Engl J Med. 1985;312:1604-1608.
increased use of intensive care units. Am Rev Respir Dis. 1991;143:251\x=req-\ 32. Black WC, Welch HG. Advances in diagnostic imaging and overestimates of
256. disease prevalence and the benefits of therapy. N Engl J Med. 1993;328:1237\x=req-\
13. Wachter RM, Luce JM. Intensive care for patients with Pneumocystis carinii 1243.
pneumonia and respiratory failure: are we prepared for our new success? Chest. 33. Rothenberg R, Woelfel M, Stoneburner R, Milberg J, Parker R, Truman B. Sur-
1989;96:714-715. vival with the acquired immunodeficiency syndrome. N Engl J Med. 1987;317:
14. Sprung CL, Steinberg A. Acquired immunodeficiency syndrome and critical care. 1297-1302.
Crit Care Med. 1990;18:1300-1302. 34. Ewald PW. The evolution of virulence. Sci Am. 1993;268:86-93.
15. Schneiderman LJ, Jecker NS, Jonsen AR. Medical futility: its meaning and ethi- 35. Veatch L, Stapleton JT. Mechanical ventilation for Pneumocystis carinii pneu-
cal implications. Ann Intern Med. 1990;112:949-954. monia in AIDS. Chest. 1990;97:1499.
16. Englehardt HT, Rie MA. Intensive care units, scarce resources, and conflicting 36. Suchyta MR, Clemmer TP, Orme JF Jr, Morris AH, Elliott CG. Increased sur-
principles of justice. JAMA. 1986;255:1159-1164. vival of ARDS patients with severe hypoxemia (ECMO criteria). Chest. 1991;
17. Franklin C, Friedman Y. Acquired immunodeficiency syndrome and critical care. 99:951-955.
Crit Care Med. 1991;19:1218-1219. 37. Friedman Y, Franklin C, Freels S. Comparison of mortality in HIV positive and
18. Brown J, Sprung CL. Ethical considerations in the treatment of AIDS patients HIV negative patients with hypoxemic respiratory failure. Chest. 1993;104:
in the intensive care unit. Crit Care Clin. 1993;9:115-123. 100S.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 06/20/2015