Comparison of Clinical Criteria for the Acute Respiratory Distress Syndrome with Autopsy Findings

Context Do all patients who meet clinical criteria for the acute respiratory distress syndrome (ARDS) have diffuse alveolar damage? Contribution Of 382 patients who had autopsies after dying in an intensive care unit, 127 met clinical criteria for ARDS. The sensitivity and specificity of clinical criteria for identifying patients with diffuse alveolar damage at autopsy were 75% and 84%, respectively. People with extrapulmonary risk factors, such as sepsis syndrome, more often had findings of diffuse alveolar damage than did those with only pulmonary risk factors, such as pneumonia. Implications In severely ill patients, clinical criteria and pathologic findings for ARDS are not closely linked. The Editors In 1967, Ashbaugh and colleagues (1) reported a syndrome characterized by the acute onset of tachypnea, hypoxemia, and loss of lung adherence that occurred after various stimuli. Over the subsequent 2 decades, the acute respiratory distress syndrome (ARDS) remained very loosely defined even in clinical trials. In a systematic overview of the incidence of and risk factors for ARDS, Garber and colleagues (2) found that 51% of 83 articles identified did not specifically define this syndrome. In 1992, however, the AmericanEuropean Consensus Conference on ARDS was formed with the goal of bringing clarity and uniformity to the definition of acute lung injury and acute respiratory distress syndrome (3). The resultant AmericanEuropean Consensus Conference definition of ARDS requires the acute onset of bilateral chest radiographic infiltrates with hypoxemia but without left atrial hypertension (3). Diffuse alveolar damage is the histopathologic finding that corresponds to the clinical entity of ARDS. Katzenstein and colleagues (4) described diffuse alveolar damage in 1976, and their criteria remain in use today. We are unaware, however, of any study that has systematically analyzed the correlation between clinical criteria for ARDS and histologic findings. This may be important because several diseases with different histopathologic findings, such as bacterial pneumonia, pulmonary hemorrhage, and bronchiolitis obliteransorganizing pneumonia, may present with a clinical picture identical to that of ARDS. In addition, ARDS associated with pulmonary risk factors may differ from that associated with extrapulmonary risk factors, a concept introduced by the AmericanEuropean Consensus Conference (3). Several studies have documented differences in the radiologic pattern (5), the respiratory mechanics (6), and the response to prone position (7) between pulmonary and extrapulmonary cases of ARDS. These findings suggest that we are faced with 2 different clinical entities or that the AmericanEuropean Consensus Conference definition operates differently depending on the origin of ARDS. To interpret the results of both observational and interventional research in ARDS, we must be confident that these studies are truly examining patients with ARDS and diffuse alveolar damage. We therefore need a measure of the validity of the widely used AmericanEuropean Consensus Conference definition. We evaluated the criterion validity of the definition of ARDS proposed by the AmericanEuropean Consensus Conference (3) by using autopsy findings as the reference standard. Second, we analyzed whether the validity of this definition was modified according to the presence of pulmonary or extrapulmonary risk factors. Methods Patients We included all patients who died in the intensive care unit of the Hospital Universitario de Getafe, Madrid, Spain, between June 1991 and December 2002 and whose relatives gave informed consent to perform a clinical autopsy. We approached the families of all patients who died except those who became organ donors (because donated organs were not available for autopsy) and those whose autopsies were legally mandated (because these autopsies were performed outside the hospital and we could not access their results). The institutional ethics committee approved the study. Clinical Criteria for ARDS After a patient's death, 2 intensivists, who were blinded to the autopsy findings, reviewed clinical charts together. They recorded the following data for each patient: age, sex, date of admission to intensive care unit, date of initiation of mechanical ventilation, principal diagnosis, risk factor for ARDS (pulmonary [pneumonia, aspiration, near-drowning, inhalational injury, or lung contusion] or extrapulmonary [sepsis syndrome, multiple trauma, several blood transfusions, shock, or pancreatitis]), date when all AmericanEuropean Consensus Conference criteria were first met, date of death, and clinical condition at the time of the death. If a patient had 2 or more risk factors for ARDS, we recorded the risk factor that presented first. The clinical condition at the time of death was considered to be shock with hypoxemia, shock without hypoxemia, or hypoxemia without shock. Shock was defined as a systolic blood pressure persistently below 90 mm Hg during the 6 hours before death, and hypoxemia was defined as arterial oxygenation saturation persistently below 85% during the final 6 hours. We used a case report form that was specifically designed to define and uniformly capture the variables of interest. A diagnosis of ARDS was established when all criteria defined by the AmericanEuropean Consensus Conference were met: acute onset, evidence on chest radiographs of airspace changes in all 4 quadrants, ratio of Po 2 to inspired fraction of oxygen of less than 200, and pulmonary artery wedge pressure less than 18 mm Hg or no clinical evidence of left atrial hypertension (3). Pathologic Criteria for Diffuse Alveolar Damage We used a predefined protocol, described previously (8), for the pathologic examination and clinicalpathologic correlation. Postmortem study was performed within 12 hours of death. After removal from the thorax, the lungs were inflated with 10% formalin to a pressure of 35 cm of H2O and were fixed in block with 10% formalin. After 48 hours, the lungs were cut into slices 3 cm thick. We took samples for microscopic analysis from each pulmonary lobe and additional samples from areas with macroscopic injuries. Two pathologists independently analyzed each sample, and a third pathologist resolved any discrepancies. Criteria for diffuse alveolar damage were hyaline membranes plus at least 1 of the following: alveolar cell type I or endothelial cell necrosis, edema, organizing interstitial fibrosis, or prominent alveolar cell type II proliferation (4, 9). Histologic criteria for the diagnosis of acute pneumonia included the presence of intense neutrophilic infiltration in the interstitium and intra-alveolar spaces, particularly around terminal bronchioles. Alveoli had to be at least partially full with neutrophils, fibrinous exudates, and cellular debris to establish a histologic diagnosis of acute pneumonia. Alveolar hemorrhage was diagnosed when acute hemorrhage was observed in alveoli and airways, along with the presence of macrophages staining positive for hemosiderin. Statistical Analysis Data are expressed as means (SD), medians (interquartile ranges), or proportions (95% CIs), as appropriate. All P values are 2-sided. We performed 3 analyses to evaluate the validity of the clinical criteria of ARDS: including all patients, including patients with any risk factor for ARDS, and separating patients with risk factors according to pulmonary or extrapulmonary origin. We calculated operative indexes, including sensitivity, specificity, and likelihood ratios, according to standard criteria (10). We used chi-square tests to compare sensitivity and specificity (11) and the Cochran Q-test to compare likelihood ratios between the cohort with pulmonary risk factors and the cohort with extrapulmonary risk factors (12). We used SPSS 11.5 (SPSS Inc., Chicago, Illinois) and Meta-Disc for Windows (Ramn y Cajal Hospital, Madrid, Spain [13]) for statistical analysis. Role of the Funding Source The funding source had no role in the design, conduct, or analysis of the data or in the decision to submit the manuscript for publication. Results During the study period, 8157 patients were admitted to the intensive care unit and 1399 (17%) died. We subsequently excluded 104 organ donors and 90 patients with legally mandated autopsies. We obtained consent and performed a clinical autopsy in 406 (34%) of the remaining patients. Twenty-four (6%) of these patients were excluded from the analysis because the hospital clinical records were lost and the information needed to determine both risk factor and clinical definition status was therefore unavailable. Table 1 shows the patient demographic and clinical characteristics. Organ donors and patients who required legally mandated autopsy were younger than other patients, and more of them had trauma as a risk factor for ARDS. Organ donors were more likely to be admitted to the intensive care unit with a neurologic problem, and patients with legally mandated autopsy were more likely to be men. Patients whose relatives declined autopsy were similar to the patients included in the study. Study patients had a median duration of mechanical ventilation of 4 days (interquartile range, 1 to 11 days) and a median length of stay in the intensive care unit of 5 days (interquartile range, 1 to 12 days). Table 1. Characteristics of the Patients Who Died in the Intensive Care Unit during the Study A total of 127 patients (33%) met the AmericanEuropean Consensus Conference criteria. After first meeting these criteria, they had a median time to death of 3 days (interquartile range, 1 to 6 days). Meanwhile, 112 patients (29%) met the pathologic criteria for diffuse alveolar damage. Table 2 shows the agreement between the clinical and pathologic diagnoses of ARDS and the operative indexes calculated by using these data. In all patients, the sensitivity of the AmericanEuropean Consensus Conference criteria was 75% (95% CI