A 9-year, single-institution, retrospective review of death rate and prognostic factors in adult respiratory distress syndrome

Abstract

Objective: To evaluate, at a single institution, the adult respiratory distress syndrome (ARDS) death rate in critically ill ventilated surgical/trauma patients and to identify the factors predicting death in these patients.

Summary background data: The prognostic features affecting mortality at the onset of ARDS have not been clearly defined. Defining rare characteristics would be valuable because it would allow for better stratification of patients in clinical trials and more appropriate utilization of constrained resources in ICU environments.

Methods: A retrospective analysis of 980 ventilated surgical and trauma intensive care unit patients from January 1990 to December 1998 was performed at Rhode Island Hospital. One hundred eleven adult intensive care unit patients with ARDS were identified using the criteria of Lung Injury Score more than 2.50 and the definition from the American-European Consensus Conference. Slightly more than half were trauma patients, 57% were men, and the median age was 59 years. The overall death rate was 52%. Patients were segregated by admission date to the intensive care unit (before or after January 1, 1995). Severity of illness was measured by the Revised Trauma Score for trauma patients and the Acute Physiology and Chronic Health Evaluation III for surgical patients. The Multiple Organ Dysfunction Score was determined on the day of onset of ARDS for all patients. Other recorded variables were age, sex, intensive care unit length of stay, length and mode of ventilation, presence or absence of tracheostomy, ventilation variables of peak and mean airway pressures, lung injury scores, elective versus emergency surgery, and presence or absence of pneumonia.

Results: There was a significant decrease in the ARDS death rate from the period 1990 to 1994 to the period 1995 to 1998. The major reason for the decline was a reduction in the posttraumatic ARDS death rate. Lung-protective ventilation strategies were used more frequently in the second period than in the first, and the death rate was significantly decreased in trauma patients in the second period when lung-protective ventilation modes were used. Predictors of death at the onset of ARDS were advanced age, Multiple Organ Dysfunction Score of 8 or more, and Lung Injury Score of 2.76 or more.

Conclusion: In this single-institution series, the death rate from ARDS declined from 1990 to 1998, primarily in posttraumatic patients, and the decrease is related to the use of lung-protective ventilation strategies. Based on this patient population, the authors developed a statistical model to evaluate important prognostic indicators (advanced age, organ system and pulmonary dysfunction measurements) at the onset of ARDS.

Figure 1. Total yearly mortality rate of patients with adult respiratory distress syndrome in surgical and trauma patients from 1990 to 1998.

Figure 2. Mortality rates in surgical patients, trauma patients, and overall, comparing the period 1990 to 1994 with the period 1995 to 1998. #P = .0001, * P = .001.

Figure 3. Logistic equation for the prediction of the likelihood of death at the onset of adult respiratory distress syndrome with the prognostic variables of age older than 75 years (A), Multiple Organ Dysfunction Score of 8 or more (M), Lung Injury Score 2.76 to 3.0 (L2.76–3.0), and Lung Injury Score > 3.0 (L > 3.0). Log-odds death (L) = −1.712 + (2.061 * age older than 75) + (1.527 * Multiple Organ Dysfunction Score ≥ 8) + (0.91 * Lung Injury Score 2.76 to 3.0) + (1.742 * Lung Injury Score > 3.0). The log-odds (L) is converted to a probability by the formula 1/(1 + e^-L).

Figure 4. Comparing the mortality rate of patients with ARDS based on cause between the periods 1990 to 1994 (group 1) and 1995 to 1998 (group 2): infection (inf), multiple trauma (mult), and other causes (oth), and direct (dir) and indirect (indir) injury groups. * P < .050.