Automating Quality Metrics in the Era of Electronic Medical Records: Digital Signatures for Ventilator Bundle Compliance

Table 3

Sensitivity, specificity, and the concordance and discordance between electronic data extraction result and reference standard in derivation and validation cohort.

Item	Cohort	Number of patients	Sensitivity; (95% CI)	Specificity; (95% CI)	TP	TN	FP	FN
DVT prophylaxis	DC	542	91.7 (87.7–94.7)	93.5 (89.9–96.1)	243	259	18	22
	VC	100	100 (92.8, 100)	100 (92.8, 100)	50	50	0	0
Peptic ulcer prophylaxis	DC	542	94.1 (91.0–96.3)	96.6 (93.1–98.6)	317	198	7	20
	VC	100	100 (95.9, 100)	100 (95.9, 100)	89	11	0	0
Oral care	DC	542	100 (96.7–100)	99.8 (98.7–100)	111	430	1	0
	VC	100	100 (95, 100)	100 (87, 100)	73	27	0	0
Head of bed elevation	DC	542	96.5 (95.5–97.9)	50 (32.4–67.6)	490	17	17	18
	VC	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Sedation break	DC	254	100 (98.3–100)	87.9 (71.8–96.5)	221	29	4	0
	VC	73	94.4 (84.6, 98.8)	100 (83, 100)	51	20	0	3

CI: confidence interval; DC: derivation cohort; VC: validation cohort; DVT: deep vein thrombosis; TP: true positive; FP: false positive; TN: true negative; FN: false negative.
Because we could never achieve high specificity with the head of bed elevation in the derivation cohort, we did not attempt validation.
Only patients on sedation at any point during the test day were assessed for “sedation break.”