Lew TWK, Kwek T, Tai D, Earnest A, Loo S, Singh K, Kwan KM, Chan Y, Yim CF, Bek SL, Kor AC, Yap WS, Chelliah YR, Lai YC, Goh S. Acute Respiratory Distress Syndrome in Critically Ill Patients With Severe Acute Respiratory Syndrome. JAMA. 2003;290(3):374-380. doi:10.1001/jama.290.3.374
Author Affiliations: Department of Anaesthesiology (Drs Lew, Kwek, Loo, Singh, Kwan, Chan, Bek, Chelliah, and Lai), Medical Intensive Care Unit, Department of General Medicine (Drs Tai and Goh), Clinical Research Unit (Mr Earnest), and Department of Respiratory Medicine (Drs Kor and Yap), Tan Tock Seng Hospital, Singapore; Department of Anaesthesia, Alexandra Hospital, Singapore (Dr Yim).
Caring for the Critically Ill Patient Section Editor: Deborah J. Cook, MD, Consulting Editor, JAMA.
Context Severe acute respiratory syndrome (SARS) is an emerging infectious disease
with a 25% incidence of progression to acute lung injury (ALI)/acute respiratory
distress syndrome (ARDS) and mortality exceeding 10%.
Objective To describe the clinical spectrum and outcomes of ALI/ARDS in patients
with SARS-related critical illness.
Design, Setting, and Patients Retrospective case series of adult patients with probable SARS admitted
to the intensive care unit (ICU) of a hospital in Singapore between March
6 and June 6, 2003.
Main Outcome Measures The primary outcome measure was 28-day mortality after symptom onset.
Results Of 199 patients hospitalized with SARS, 46 (23%) were admitted to the
ICU, including 45 who fulfilled criteria for ALI/ARDS. Mortality at 28 days
for the entire cohort was 20 (10.1%) of 199 and for ICU patients was 17 (37%)
of 46. Intensive care unit mortality at 13 weeks was 24 (52.2%) of 46. Nineteen
of 24 ICU deaths occurred late (≥7 days after ICU admission) and were attributed
to complications related to severe ARDS, multiorgan failure, thromboembolic
complications, or septicemic shock. ARDS was characterized by ease of derecruitment
of alveoli and paucity of airway secretion, bronchospasm, or dynamic hyperinflation.
Lower Acute Physiology and Chronic Health Evaluation II scores and higher
baseline ratios of PaO2 to fraction of inspired oxygen were associated
with earlier recovery.
Conclusions Critically ill patients with SARS and ALI/ARDS had characteristic clinical
findings, high rates of complications; and high mortality. These findings
may provide useful information for optimizing supportive care for SARS-related
Severe acute respiratory syndrome (SARS) is a new disease that emerged
in November 2002, was initially described in March 2003, and is thought to
be caused by a novel coronavirus (SARS CoV).1 In
Singapore, the index case was a traveler who was believed to have been infected
in Hong Kong. This patient was admitted to our hospital on March 1, 2003,
and subsequently infected 19 health care workers, patients, and their contacts.2,3 By June 9, 2003, a total of 206 probable
SARS cases had been diagnosed in Singapore. To date, more than 8000 individuals
have been infected with SARS worldwide, with two thirds of cases reported
in China alone.4 Toronto is the only city in
North America with a major outbreak of SARS.5
Approximately 25% of patients with SARS are likely to progress to severe
respiratory failure. In Hong Kong, the case-fatality rate is 13.2% for patients
younger than 60 years and is 43% for those aged 60 years or older.6 In Hong Kong and Canada, outcome data on critically
ill SARS patients are available either from very early in the course of the
outbreaks or from patients who were treated in many separate intensive care
units (ICUs).4,7 However, to date,
no single center has reported a sufficiently large cohort of critically ill
patients to accurately characterize the clinical spectrum and outcome of SARS-related
Early in the SARS outbreak in Singapore (ie, on March 22, 2003), Tan
Tock Seng Hospital, a 1200-bed acute care general hospital where the national
Communicable Disease Centre is located, was designated for the intake and
solitary isolation of all suspected and probable SARS cases. Except for 5
patients who were too ill to be transferred to Tan Tock Seng Hospital or had
only postmortem diagnoses of probable SARS, all critically ill SARS patients
were treated in a single dedicated SARS ICU. This report describes the clinical
characteristics and outcomes of 46 critically ill patients with probable SARS
treated during a 13-week period in this dedicated SARS ICU.
The Tan Tock Seng Hospital research ethics committee approved this study.
We reviewed all probable SARS ICU cases diagnosed according to the prevailing
World Health Organization definition at the time of admission from March 6
to June 6, 2003.8 Patients were transferred
to the ICU if they developed signs and symptoms of respiratory failure, arterial
oxygen saturation (SaO2) of less than 92% despite oxygen therapy
of at least 50% fraction of inspired oxygen (FIO2), or if they
presented to another hospital with acute respiratory failure and had contact
history that was suspicious of SARS.
Patients were treated in a 36-bed ICU. All beds were in individual rooms
that had individual air-conditioning systems with negative pressure airflow.
A strict personal protection protocol that met or exceeded prevailing World
Health Organization guidelines9,10 was
put in place by the first week of the outbreak and was strictly followed by
health care workers.
Patients underwent intubation and mechanical ventilation if they deteriorated
clinically or could not maintain more than 90% SaO2 with spontaneous
ventilation despite maximal oxygen therapy. A low-tidal-volume lung-protective
strategy was used for ventilation, with volume or pressure control ventilation
targeting tidal volumes at 6 mL/kg of predicted body weight and plateau pressures
of less than 30 cm H2O. Positive end-expiratory pressure (PEEP),
FIO2, and ventilator rates were then titrated to keep PaO2 greater than 55 mm Hg (oxygen saturation as measured by pulse oximetry
>88%-90%), with normal pH and PaCO2, if possible. Patients undergoing
mechanical ventilation were sedated with a combination of an opioid, benzodiazepine,
and/or propofol. Atracurium or pancuronium were used for neuromuscular paralysis
to facilitate ventilation when indicated. Standard cardiovascular support
to maintain euvolemia, nutritional support, prophylactic and culture-directed
antibiotic therapy, continuous renal replacement therapy in renal failure,
and euglycemia therapy were administered. Prophylaxis against deep vein thrombosis
involved use of stockings, calf sequential compression devices, and subcutaneous
low-molecular-weight heparin once daily unless contraindicated.
Antimicrobial therapy was not protocol based, but most patients had
been treated with either levofloxacin or a combination of a macrolide and
intravenous cephalosporin prior to ICU admission. Antiviral agents were initially
used (ribavirin was administered to 96 patients and oseltamivir to 6 patients)
but were subsequently discontinued because of lack of clinical efficacy.2 An immunomodulation regimen, combining intravenous
pulse methylprednisolone (200 mg) and high-dose intravenous immunoglobulin
(0.4 g/kg of body weight) administered once daily for 3 consecutive days,
was administered within 24 hours of ICU admission to 16 eligible critically
ill probable SARS patients with acute lung injury (ALI) or acute respiratory
distress syndrome (ARDS) but without any evidence of bacterial or fungal infections
(absence of positive microbiologic cultures, marked leukocytosis, or elevated
serum procalcitonin levels).
Data on the hospital cohort were drawn from a computerized database
maintained by the hospital's Communicable Disease Centre and administered
by a clinical epidemiologist. A group of trained physicians collected and
regularly updated demographic, clinical, and treatment characteristics in
a standardized format. Intensive care unit–related data in this database
were predefined by the clinical director of the SARS ICU (D.T.) and regularly
reviewed for accuracy and consistency. Additional information used in this
study was collected from chart review by a second team of 4 physicians, all
of whom had spent at least 14 days or more managing cases in the SARS ICU.
Detailed physiological data was also obtained manually from our ICU Clinical
Information System (CareVue 2000, version I.2, Philips Medical Systems, Andover,
Mass) and entered into a computerized database. Pilot data were initially
reviewed by the 2 senior investigators (T.W.K.L. and T.K.K.) for consistency
and refinement of definitions before the full data set was collected.
The main outcome measure was mortality at 28 days after symptom onset
for patients in the cohort. Deaths were defined as early (<7 days after
ICU admission) or late (≥7 days) and were classified based on clinical
diagnoses of the proximate cause of death, with postmortem evidence when available
(n = 5).
We also classified the patients' clinical course of ARDS, based on clinical
observation, into 3 groups: those who survived without need for mechanical
ventilator support (early recovery), those who required mechanical ventilation
for 14 days or less (intermediate recovery), and those who required mechanical
ventilation for more than 14 days (late survival). We compared these groups
according to the following characteristics: age and sex, patient demographics,
Acute Physiology and Chronic Health Evaluation (APACHE) II score, time from
illness onset to ICU admission, baseline (lowest initial) ratio of PaO2 to FIO2 as recorded on admission, time from illness onset
to requirement of mechanical ventilation, earliest time to reversal of oxygenation
shunt and cessation of mechanical ventilation, and peak and time to peak serum
lactate dehydrogenase level, a previously described marker of severity.2,7
We also examined treatment characteristics for principal ventilator
mode used, effectiveness in meeting ventilation targets, maximum PEEP used,
management of sedation and muscle paralysis, incidence and treatment of complications,
and specific ventilator and caregiver precautions for prevention of spread
of aerosolized droplets.
Continuous variables were compared among the 3 survival groups and the
nonsurvivors using the Kruskal-Wallis test. Pairwise comparisons were made
using the Mann-Whitney test. Categorical variables were compared using the χ2 test or the Fisher exact test whenever appropriate. Nonparametric
tests were chosen because of the small sample size in each group. The Kaplan-Meier
method was used for length of stay in the ICU because there were censored
cases (ie, patients still in the ICU as of the last follow-up date). The log-rank
test was used to compare length of stay in the ICU. We performed a logistic
regression analysis to identify predictors of early and intermediate recovery
vs late survival and death. Starting from the most significant variable in
the univariate analysis, the log-likelihood ratio test was used to determine
whether inclusion of a new variable improved the fit of the multivariate model.
All tests were conducted at the P<.05 level of
significance, and data analysis was carried out using Stata version 6.0 software
(Stata Corp, College Station, Tex).
Of 199 patients with probable SARS treated at our hospital, 46 (23%)
were admitted to the SARS ICU. Table 1 shows
the comparison of the characteristics of the ICU and non-ICU cohorts. Complete
data were available for 149 of 153 non-ICU patients. There were 4 direct transfers
from ICUs of other hospitals. Patients admitted to the ICU were older and
more likely to have elevated lactate dehydrogenase levels. The main indication
for ICU admission for nonventilated patients was hypoxemia (97%). One patient
was admitted with posthypoxic encephalopathy after hypoxic cardiopulmonary
Forty-five patients met the criteria for either ALI (PaO2/FIO2 ≤300 mm Hg) or ARDS (PaO2/FIO2 ≤200 mm
Hg).11 The overall 28-day post–symptom
onset mortality rate was 20 (10.1%) of 199 (95% confidence interval [CI],
6.2%-15.1%), and the overall mortality at 13 weeks was 27 (13.6%) of 199 (95%
CI, 9.1%-19.1%) for the entire hospital cohort. For SARS patients admitted
to the ICU, the 28-day post– symptom onset mortality was 17 (37%) of
46 (95% CI, 23.2%-52.5%) and mortality at 13 weeks was 24 (52.2%) of 46 (95%
CI, 36.9%-67.1%). Three patients who died in the hospital declined ICU admission.
Five patients died within the first 7 days of ICU admission. All but 1 of
the early deaths had significant preexisting comorbidities. The proximate
causes of ICU deaths are shown in the Box. The majority of deaths (75%) occurred late in the course of the
disease from complications related to severe ARDS, multiorgan failure, thromboembolic
complications, or septicemic shock.
Early (<7 Days in ICU; n = 5)
Dilated cardiomyopathy (1)Cardiac failure with septicemic shock
(1)Ventricular fibrillation and end- stage renal failure (1)Biliary
peritonitis and acute-on- chronic renal failure (1)ARDS with bacterial
pneumonia and pulmonary embolism (1)
Late (≥7 Days in ICU; n = 19)
End-stage renal failure (1)Late ARDS with multiorgan failure
(7)Late ARDS with intractable hypoxia (single-organ failure)(2)
Acute myocardial infarction (1)Postanoxic brain ischemia (1) Massive
cerebrovascular accident (2)ARDS with acute pulmonary embo lism (3)Septicemic
*ICU indicates intensive care unit; ARDS, acute respiratory distress
syndrome. Autopsies were performed in 5 of these patients.
Table 2 shows the survival
group (n = 22) divided into 3 subgroups based on clinical course: early recovery
without mechanical ventilation, intermediate recovery with mechanical ventilation
for 14 days or less, and late survival with mechanical ventilation for more
than 14 days. In the late survival group, 4 patients had been discharged from
the ICU and 2 remained in critical condition. The groups differed in age (P = .03), APACHE II score (P =
.002), baseline PaO2/FIO2 ratio values, (P<.001), and peak lactate dehydrogenase levels (P = .02).
Patients in the early recovery group had an abbreviated course of ALI
and their FIO2 requirements peaked at a median of day 8 of illness,
which coincided with the median day of admission to the ICU. Patients in the
intermediate recovery group had improvement in oxygenation and pulmonary compliance
after a median of 5 days of mechanical ventilation and were extubated within
a median of 4 days of improvement (Figure
1). Patients in the late survival group had a protracted and severe
course of ARDS, required the most interventions and treatment, and had the
In univariate analysis, age, APACHE II score, and baseline PaO2/FIO2 ratio were associated with early/intermediate recovery.
In multivariate analysis, only APACHE II scores and baseline PaO2/FIO2 ratio were independently associated with early/intermediate recovery
(Table 3). The odds ratio of recovery
decreased by 0.87 (95% CI, 0.78-0.97) for every 1-unit increase in APACHE
II score after adjusting for baseline PaO2/FIO2 ratio
(P = .02). The odds ratio of recovery increased by
1.02 (95% CI, 1.00-1.04) for every 1-unit increase in baseline PaO2/FIO2 ratio after adjusting for APACHE II score (P =
Pressure control ventilation was used in 22 patients, volume control
in 17 patients, and airway pressure release ventilation in 1 patient. Seven
patients underwent ventilation in the prone position for periods ranging from
4 to 9 h/d. Three of these 7 patients had significant improvements in SaO2 and 1 survived; there were 3 survivors overall from this group. Respiratory
tract secretions were minimal and bronchospasm was encountered in only 3 patients
who underwent ventilation.
Tidal volumes, minute ventilation, plateau pressures, and PEEP use were
not significantly different between the intermediate recovery and late survival/nonsurvival
groups on day 1 (data not shown). The mean tidal volumes used were low (6.46
and 6.94 mL/kg of predicted body weight), with plateau pressures kept to less
than 30 cm H2O and pH and PaCO2 levels within normal
limits. On day 7, the late survival and nonsurvival groups had higher mean
minute ventilation (P = .02) coupled with higher
mean PaCO2 and higher mean plateau pressures (P = .03), indicative of worsening ARDS in that group. PEEP use was
generally high on day 1 (>12 cm H2O) with a trend toward decreasing
requirements by day 7 in the intermediate recovery group, consistent with
earlier recovery in that group.
We noted a low incidence of dynamic hyperinflation and auto-PEEP in
SARS-related ARDS. In 9 patients undergoing ventilation in whom auto-PEEP
was measured, this ranged from 1 to 5 cm H2O. Two patients received
inhaled nitric oxide, up to 30 ppm, as rescue treatment for intractable hypoxia.
High-frequency oscillatory ventilation was attempted in 1 patient as a rescue
mode after progressive deterioration. These 3 patients did not survive.
Regular lung recruitment maneuvers were necessary as SARS-related ARDS
was characterized by hyposecretory airways, probably predisposing them to
alveolar derecruitment and desaturation. Twenty-eight of 39 patients who underwent
ventilation required muscle paralysis. The longest duration of paralysis was
Twelve patients developed positive blood cultures and 24 had positive
lung aspirate cultures. Eight patients developed pneumothorax and 1 developed
pneumomediastinum. Nine patients developed acute renal failure and required
continuous renal replacement therapy (continuous veno-veno hemofiltration
or hemodiafiltration). Four patients developed ischemic strokes. There were
11 episodes of deep vein thrombosis and 7 episodes of proven or suspected
pulmonary embolism in the cohort. Four patients had both deep vein thrombosis
and pulmonary embolism. Two patients received streptokinase for massive pulmonary
embolism. There were improvements noted in hemodynamic parameters and pulmonary
shunting but one of these patients died after 24 hours and the other died
subsequently from progression of ARDS.
In our cohort of critically ill patients with SARS and ALI/ARDS, 28-day
mortality was 37% and overall ICU mortality was 52.2% after 13 weeks.
We found that critically ill SARS patients with ARDS experienced a potentially
fatal but in some cases self-limiting clinical course. A third of patients
recovered early, generally within 14 days of illness. However, the majority
of patients underwent a protracted course of ARDS, accompanied by complications
of severe hypoxia, multiorgan failure, thromboembolic complications, and sepsis.
Mortality in this group was high despite maximal supportive therapy. This
pattern is consistent with that reported in the literature, in which mortality
in late ARDS is related primarily to the degree of other organ dysfunctions.12- 14
We modeled our ventilator strategy after the recommendations of the
National Institutes of Health ARDS Network,15 targeting
low tidal volumes and plateau pressures of less than 30 cm H2O
and titrating PEEP according to FIO2. ARDS in SARS was generally
severe, with poor compliance and requirement for high PEEP to maintain adequate
oxygenation. In patients with progressive disease, we found it increasingly
difficult to maintain adequate lung recruitment and had to compromise plateau
pressures, exceeding the limit of 30 cm H2O to keep tidal volumes
at 6 mL/kg of predicted body weight. The mean PEEP levels used in our series
were higher than those reported by others (8-10 cm H2O) using similar
low-tidal-volume ventilation strategies.15- 17 This
may indicate greater severity of ARDS encountered in SARS or may merely reflect
local preference for higher PEEP over higher FIO2.Although viral
infections such as respiratory syncytial virus and rhinovirus have been reported
to be associated with asthma exacerbations,18 bronchospasm,
dynamic hyperinflation, and auto-PEEP were not common findings in SARS.
We suggest, like others,7,19 that
progression to a severe and protracted course of ARDS is most likely related
to patients' hyperimmune response to SARS. This is similar to ARDS associated
with sepsis or the systemic inflammatory response syndrome. Our impressions
were supported by postmortem pathologic findings of diffuse alveolar damage
and pulmonary fibroproliferation, similar to findings from Hong Kong.20
Previous studies of ARDS patients treated with high-dose methylprednisolone
did not show lowering of serum complement levels or improvement in outcome.21,22 The benefit of using steroid therapy
(alone) in ARDS thus remains controversial. Immunomodulation also requires
exclusion of bacterial suprainfection. This may be difficult in SARS because
of routine use of prophylactic antibiotics and where bronchoalveolar lavage
or open-circuit plugged telescoping catheter sampling is not routinely performed.
Among the subgroup of patients in our study who received the combined immunoglobulin
and methylprednisolone regimen daily for 3 days, an interim analysis of the
first 15 patients treated revealed an adjusted hazard ratio for mortality
of 0.41 (95% CI, 0.14-1.23; P = .11) in the treated
group compared with those who did not receive treatment (n = 30). There was
also a trend toward earlier recovery in the treated group. However, conclusions
about the efficacy of this therapy await a final analysis of the data.
Differences in the virulence of different SARS Co-V strains and viral
load may play a role in determining patients more likely to develop a severe
ARDS course. This differentiation awaits more precise and quantitative diagnostic
tools. Clinically, poor baseline PaO2/FIO2 ratios and
higher APACHE II scores were the only predictors of late survival or death.
This is consistent with more severe disease presentation. We are also studying
cytokine levels for potential value as predictors. Attempts to identify patients
who may progress to a severe course of ARDS, and to target such patients for
early aggressive intervention, are supported in a recent review and in an
early-outcome study on ARDS.23,24
Although we have presented ICU management of ARDS in SARS as a single
retrospective case series, management of this disease had not been standardized
from the beginning but evolved rapidly with experience and an improved understanding
of its characteristics. This evolution also included an initial period of
profound fear and emotional distress experienced by health care workers working
with potentially fatal nosocomial transmission.25 However,
after implementation of specific, rigorous protective measures, there were
no known nosocomial transmission of SARS to the 211 health care workers in
our ICU for the period of March 17 to June 9, 2003. After the initial 2 weeks
of operation, when powered air-purifying respirator units became available,
their use was mandated for procedures in the ICU involving actual or potential
ventilator circuit disconnections or "splash" dissemination of body fluids.
These procedures included manual lung recruitment, ventilator tubing changes,
thoracocentesis, tracheostomies, and interventional radiological procedures.
Neither bronchoscopic procedures nor bronchoalveolar lavage was carried out
in the 46 patients. In our experience, ensuring the safety of health care
workers and providing timely psychological peer support were critical in maintaining
staff morale and to secure the delivery of appropriate care. Although no health
care worker became ill with SARS in our ICU, we cannot rule out the possibility
of subclinical infections.
In this series of critically ill patients with SARS and ARDS, the treatment
of patients who progressed to severe and protracted ARDS was challenging and
associated with high mortality.