Evaluating construct validity of computable acute respiratory distress syndrome definitions in adults hospitalized with COVID-19: an electronic health records based approach

Table 3 Associations between EHR-Berlin definition and clinical outcomes

	ARDS-	ARDS+	P	r	Unadjusted OR or β (95% CI)
	N = 594	N = 171
Hospital mortality	49 (8%)	73 (43%)	< 0.001	0.39	8.28 (5.43, 12.82)
Respiratory support-free days	28 (27–28)	0 (0–18)	< 0.001	-0.62	-16.67 (-18.16, -15.18)
Ventilator-free days	28 (28–28)	1 (0–21)	< 0.001	-0.61	-16.19 (-17.69, -14.70)
Ordinal score ≤ 5 at day 14	498 (84%)	30 (18%)	< 0.001	-0.59	0.05 (0.03, 0.07)

ARDS- and ARDS + columns indicate N (%) for categorical outcomes (mortality and ordinal score) and median (interquartile range) of continuous outcomes (respiratory support- and ventilator-free days) by patients who were not and who were classified as EHR-Berlin ARDS, respectively. P values are for Chi square tests for categorical outcomes and Wilcoxon rank sum tests for continuous outcomes. r indicates Pearson correlation coefficient between EHR-Berlin classification and each outcome. Odds ratio (OR) calculated with logistic regression for categorical outcomes, beta (β) calculated with linear regression for continuous outcomes

ISSN: 1471-2466