TABLE 5.

POOLED TRAUMA AND METAANALYSIS RESULTS OF RS315952C–ACUTE RESPIRATORY DISTRESS SYNDROME ASSOCIATION ACROSS THREE POPULATIONS, WITH CLINICAL ADJUSTMENT

Stage	n	Unadjusted OR (95% CI)	P Value	Adjusted OR (95% CI)	P Value (Adjusted)
I	224	0.37 (0.22–0.62)	1.9 × 10−4	0.39 (0.23–0.66)	4.0 × 10−4
II	778	0.67 (0.52–0.88)	2.3 × 10−3	0.66 (0.49–0.88)	6.0 × 10−3
I/II pooled	1002	0.62 (0.53–0.80)	2.3 × 10−5	0.66 (0.53–0.82)	9.9 × 10−5
III	2063	0.83 (0.71–0.96)	0.015	0.82 (0.70–0.97)	0.021
Metaanalysis	3065	0.81 (0.72–0.91)	1.1 × 10−5	—	4.2 × 10−5

Definition of abbreviations: CI = confidence interval; OR = odds ratio.

Subject-level data were available for stages I and II, which allowed for a pooled trauma association with acute respiratory distress syndrome (ARDS). Stage III results were then metaanalyzed with the pooled trauma results as enacted in METAL (40), to obtain unadjusted and adjusted metaanalysis P values. We estimated the pooled OR by chi-square testing using absolute genotype counts in subjects with ARDS (78/467/689) versus those without ARDS (non-ARDS; 179/705/927). Hospital of the University of Pennsylvania trauma population was adjusted for age, injury severity score (ISS), modified Acute Physiology and Chronic Health Evaluation (APACHE) III scores, blunt mechanism of injury, and packed red blood cell transfusion during the first 24 hours after trauma. The stage II (Harborview) population was adjusted for age, sex, ISS, APACHE II score, and blunt mechanism of injury. The Massachusetts General Hospital population was adjusted for age, sex, modified APACHE III scores, and clinical risk factors (bacteremia, sepsis without shock, sepsis with shock, pneumonia, trauma, aspiration, multiple transfusion). The IL1RN SNP rs315952 demonstrated a significant association with decreased ARDS risk that was robust to adjustment for clinical covariates in each stage. Modified APACHE scores reflect removal of the oxygenation component, as hypoxemia was collinear with a diagnosis of ARDS.