Professional Documents
Culture Documents
98464
98464
Thomas Bice, MD; Guangxi Li, MD; Michael Malinchoc, MS; Augustine S. Lee, MD; Ognjen Gajic, MD
Objective: To determine risk factors for development of recur- 2.93) per 100,000 person years. The median time to development of
rent acute lung injury. the second episode was 264 days (interquartile range 80 – 460 days),
Design: A population-based case-control study. with a mortality of 47% during the episode. The history of gastro-
Setting: The study was conducted in Olmsted County, MN, from esophageal reflux disease was highly prevalent in patients who
1999 to 2008. developed recurrent acute lung injury: 15 of 19 patients (79%) com-
Patients: Using a validated electronic screening protocol, in- pared to 5 of 19 (26%) matches with a single episode of acute lung
vestigators identified intensive care patients with acute hypox- injury (p ⴝ .006) and 8 of 19 (42%) matches without acute lung injury
emia and bilateral pulmonary infiltrates. (p ⴝ .016). Other exposures were similar between the cases and the
Interventions: None. two matched controls.
Measurements and Main Results: The presence of acute lung Conclusions: Recurrent acute lung injury is not a rare phenom-
injury was independently confirmed according to American-Euro- enon in the intensive care unit and may continue to increase with
pean Consensus Conference criteria. Recurrent acute lung injury improvements in survival following acute lung injury. Gastro-
cases were subsequently matched (1:1:1) with two controls (single esophageal reflux disease was identified as an important risk
acute lung injury and no acute lung injury) on age, gender, duration factor for recurrent acute lung injury and may suggest an impor-
of follow-up, and predisposing conditions. Risk factors evaluated tant role of gastric aspiration in the development of this syn-
included gastroesophageal reflux disease, alcohol consumption, drome. (Crit Care Med 2011; 39:000 – 000)
smoking, chronic opioid use, and transfusions. We identified 917 KEY WORDS: epidemiology; incidence; recurrent acute lung in-
patients with acute lung injury, 19 of which developed a second jury; acute lung injury; acute respiratory distress syndrome; gas-
episode, yielding a frequency of 2.02 (95% confidence interval 1.10 – troesophageal reflux disease
A cute lung injury (ALI) is a However, survival has improved to the Rochester, MN. No other critical care services
common and important cause point that there have been case reports of are available in the surrounding communities.
of acute respiratory failure that recurrent ALI (15–18). Cases and Case Finding. The healthcare
typically develops after severe A systematic study of this phenome- system utilizes an integrated electronic med-
injury or illness (1–5). The incidence of non has not been reported. The charac- ical record system. Combined with multiple
ALI reported on prior studies ranges from teristics of such patients may point to comprehensive, prospectively collected clini-
cal research databases (Rochester Epidemiol-
1.5 to 78.9 per 100,000 person years (6). specific environmental or genetic factors
ogy Project, Mayo Clinic Life Sciences System,
Despite recent advances in our under- important in the development of this Acute Physiology and Chronic Health Evalua-
standing of the pathophysiology and complication. Savici and Katzenstein (17) tion database), the research environment en-
treatment of ALI, it is not known why identified a history of gastroesophageal ables easy and extensive access to detailed
some patients develop this syndrome, and reflux disease (GERD) and chronic pain clinical information from this patient popula-
mortality has remained high (6 –14). treated with opioids in a series of patients tion (19). A previously validated screening tool
with diffuse alveolar damage. We sought (“acute respiratory distress syndrome screen-
to identify the incidence and risk factors ing tool”) with excellent negative predictive
*See also p. 000. for recurrent ALI in Olmsted County, MN. value (0.99) (20) screened all patients who
From the Division of Pulmonary and Critical Care received intensive care services from 1999 to
Medicine (TB, GL, MM, ASL, OG), Mayo Epidemiology 2008. The screening tool was designed to iden-
and Translational Research in Intensive Care, and Division MATERIALS AND METHODS tify all patients who met the following criteria
of Biomedical Informatics and Biostatistics (MM), Mayo within a single 24-hr period.
Clinic, Rochester, MN; Division of Pulmonary Medicine After receiving institutional review board (1) Qualifying arterial blood gas analysis: ra-
(GL), Guang An Men Hospital, China Academy of Chinese
approval, we performed a population-based, tio of the partial pressure of oxygen to the frac-
Medical Science, Beijing, China; and Department of Crit-
ical Care Medicine (ASL), Division of Pulmonary Medicine, retrospective case-control study of Olmsted tion of inspired oxygen (PaO2/FIO2) of ⬍300.
Mayo Clinic, Jacksonville, FL. County, MN, residents (ⱖ18 yrs) from 1999 to (2) Qualifying chest radiograph report: free
Supported, in part, by grant LM10468Z-01 from 2008. The demographics of Olmsted County text Boolean query containing words—
the National Institutes of Health, Bethesda, MD. residents are typical of a suburban community “edema” or “bilateral” and “infiltrate.”
Dr. Bice and Dr. Li received funding from the NIH. Trained critical care fellows reviewed digi-
in the Midwestern United States. The popula-
The other authors have not disclosed any potential
conflicts of interest. tion was 124,277, consisting largely of middle- tal chest radiographs, hemodynamic monitor-
For information regarding this article, E-mail: class Caucasians, with minorities representing ing data (pulmonary artery occlusion pressure
gajic.ognjen@mayo.edu 13% of the population according to 2000 U.S. of ⱕ18 cm H2O or central venous pressure of
Copyright © 2011 by the Society of Critical Care census reports. Because of its geographic iso- ⬍15 cm H2O in the absence of pulmonary
Medicine and Lippincott Williams & Wilkins lation, critical care services are provided ex- hypertension), echocardiography results (E/E⬘
DOI: 10.1097/CCM.0b013e31820edf91 clusively by the two Mayo Clinic hospitals in ⬍15), brain natriuretic peptide levels (⬍250
Table 2. Associations with recurrent acute lung injury and single acute lung injury
Exposure Frequency
Discordant Concordant
Recurrent Acute Single Acute Pairsa Pairsa
Lung Injury, Lung Injury,
Variable n (%) n (%) ⫹/⫺ ⫺/⫹ ⫹/⫹ ⫺/⫺ Odds Ratiob p
17 of the 19 patients had died by the time DISCUSSION associated mortality of 44% (6). Unfortu-
of this study (89%), with a median sur- nately, aspiration in both clinical practice
vival after the second episode of 22 (in- With the improvement of ALI survival and research is limited by its clinical def-
terquartile range 4 –371) days. in this population-based study, we ob- inition of a “witnessed aspiration” event,
There was a significantly higher preva- served a significant number of patients raising the concern that many aspiration
lence (79%) of GERD in the recurrent ALI who had more than one episode of ALI. events that lead to ALI/acute respiratory
group than in the single ALI group (26%, GERD was identified as the most impor-
distress syndrome are unrecognized. In
p ⫽ .006) or the no ALI group (42%, p ⫽ tant risk factor for recurrent ALI, poten-
those with witnessed gastric aspiration,
.016) (Tables 2 and 3). Witnessed or sus- tially suggesting an important role of gas-
pected aspiration was more commonly doc- tric aspiration in the development of this we observed a “dose-dependent” pattern
umented in patients with recurrent ALI. syndrome. from 5% in those who had no ALI to 16%
The chronic use of acid suppression did not ALI represents a specific injury pat- in those with a single episode of ALI to
modify the risk of recurrent ALI. No signif- tern resulting from a variety of insults in finally 37% in those who had recurrent
icant differences were observed in chronic a susceptible host. Aspiration is an im- ALI. This is consistent with our under-
opioid use, alcohol consumption, smoking, portant recognized mechanism for ALI standing of aspiration and its potential to
or transfusions. found to be a risk factor in 11% with an cause lung injury, but it further under-
Discordant Concordant
Exposure Frequency Pairsa Pairsa
Variable Recurrent ALI, n (%) No ALI, n (%) ⫹/⫺ ⫺/⫹ ⫹/⫹ ⫺/⫺ Odds Ratiob p
scores that it is a risk factor for recurrent prevention of upper intestinal bleeding ment in survival from ALI as a result of
episodes of ALI. (30). There is increasing concern as to advances in supportive care (3, 35, 36).
With a lack of a proven standard for whether the empirical use of potent acid The incidence of recurrent ALI in our
identifying gastric to pulmonary aspira- suppression may increase the risk for population was 2.02 per 100,000 person
tion, we looked for other signals that may pneumonia (31, 32) and Clostridium dif- years. When affected with a second epi-
indicate that silent aspiration occurs ficile infection (33, 34). However, there sode of ALI, mortality was high at 47%,
more frequently as a cause of ALI and are no data on how acid suppression with a median survival of 22 days. No
acute respiratory distress syndrome. might modify the risk for ALI, whether previous study has reported the incidence
Mechanistically, for gastric to pulmonary favorably or unfavorably. Although we ex- of recurrent ALI. Cely et al (15) reported
aspiration to occur, there must first be pect the acid suppression would confer a ten cases of recurrent ALI among their
reflux of gastric contents. This may occur protective effect, its concomitant in- series of veterans with ALI, with a similar
acutely as can be seen in the critically ill crease in risk for pneumonia and subse- mortality of 50% and a median time to
(e.g., emesis and aspiration in an ob- quent lung injury may be unfavorable. recurrence of 192 days. They also re-
tunded patient), but more commonly it is The limited sample size did not allow us ported a higher frequency of 16% per
seen in the chronic condition of GERD. to evaluate these risks in a multivariate year of developing a recurrent episode of
GERD could act as the necessary prereq- fashion to formally assess for confound- ALI. This difference may be explained by
uisite risk for aspiration, and further- ing and interaction. Additionally, the ret- the differences in study populations. Our
more, because it is chronic, it may be a study subjects were from a regional pop-
rospective nature of the study did not
risk for repeated episodes of silent aspi- ulation in the Midwest, while Cely’s study
allow us to quantify adherence or actual
ration and ALI. Our data support this population was from a Veterans Affairs
use of medications.
plausible mechanism showing GERD as Medical Center. Their study population
Only a limited number of reports are
the strongest risk factor for recurrent tended to be older, and they also included
available on this unique group of patients
ALI. With a prevalence of GERD of 20% referred patients who tend to have a
with recurrent ALI without an identifi-
in our population in Olmsted County and higher severity of illness than in a popu-
worldwide (28, 29), the at-risk population able cause (15–18). Savici and Katzen- lation-based study. Interestingly, both of
may be significant. Notably, the risk for stein (17) described six cases of patholog- our studies noticed the recurrent epi-
ALI as a result of having GERD was ically confirmed diffuse alveolar damage sodes usually happened around 200 days
higher than for aspiration itself. How- without a clear causative agent, as iden- after the initial one. This finding needs to
ever, this is not unexpected as aspiration tified in these other reports. Interest- be confirmed in a larger cohort in the
is often insensitively defined clinically ingly, five of their six patients were noted future.
and suggests that, potentially, many aspi- to have a history of GERD or a hiatal Our study is susceptible to the typical
ration events are unrecognized. This hernia, and they proposed that perhaps weakness of a case-control study utilizing
highlights the dire need for a more ob- this in combination with the use of existing clinical records, as well as the
jective determination of aspiration. chronic opiates seen in four of their cases limitations of a small sample size. We have
Chronic use of acid-suppressive med- predisposed the patients to an aspiration avoided some of the sampling bias inherent
ications did not alter the susceptibility of mechanism for lung injury. Our results to case-control studies by using incident
ALI development, but our study design support the possibility of an aspiration cases and drawing our sample from the
did not allow us to determine the timing mechanism, although we did not identify complete health-linkage records system of
of medication use when compared to the an independent risk with the chronic use our population in Olmsted County. There
initial insult. In the critically ill, acid of opiates. It is possible this risk is now are still some concerns toward measure-
suppression is typically utilized for the becoming more evident with the improve- ment bias since the exposures, including