Table 2.

Adjusted patient outcomes associated with admission to hospital during first 90 days of implementation of electronic health records (EHRs) compared with prior 90 days*

Outcomes	Control hospitals (n=399)			Study hospitals (n=17)		Adjusted difference-in-differences change in outcome for study hospitals v controls (95% CI)†	P value‡
	Adjusted pre-EHR rate (95% CI)†	Adjusted post-EHR rate (95% CI)†		Adjusted pre-EHR rate (95% CI)†	Adjusted post-EHR rate (95% CI)†
30day mortality (%)	7.25 (7.12 to 7.40)	7.29 (7.14 to 7.45)		7.02 (6.46 to 7.75)	7.49 (6.94 to 8.07)	0.43 (−0.11 to 0.99)	0.12
30 day readmissions (%)	18.2 (17.9 to 18.5)	17.7 (17.4 to 17.9)		18.9 (17.9 to 19.9)	18.1 (17.3 to 18.9)	−0.28 (−1.19 to 0.55)	0.57
PSI-90 rate (events per 1000 admissions)	7.32 (6.79 to 7.95)	7.25 (6.73 to 7.85)		9.38 (7.39 to 11.9)	10.3 (8.1 to 13.0)	0.96 (−0.80 to 2.69)	0.28

PSI-90=patient safety for selected indicators 90.

*Hospitals newly implementing EHRs defined as those with less than a basic inpatient EHR before implementation, as defined in Jha et al 2009.

†Absolute percentage changes in each outcome attributable to EHR implementation were simulated using a marginal standardization approach (see supplementary eMethods).

‡Estimated from difference-in-differences model comparing change for each period compared with baseline period (1-90 days pre-implementation) between hospitals implementing EHR and control hospitals in same hospital referral region as study hospital. Models adjusted for age, sex, race, original reason for Medicare eligibility, major diagnostic category for admission, hospital referral region, and length of stay (for PSI-90 outcome only). All analyses used robust variance estimators to account for clustering of admissions within hospitals.