Table 3.

Error data captured by the automated methods and how it is being used

Automated search methods identifying all errors
Paper	Descriptive information about error timings	Background error rate	Intervention error rate	Further relevant information
Used unique method of Koppel et al46
Koppel et al46	Orders stopped within 30 min were most likely to be deemed inappropriate	Not specified	No intervention
Method utilizing the descriptive discontinuation option native to the CPOE system (eg, “void,” “Error—erroneous entry”)
Kannampallil et al54	Not specified	490 “voids” per 100 000 orders (0.49%)	No intervention	Originally reported as per 1000 orders
Abraham et al41	Not specified	210 “voids” per 100 000 orders (0.21%)	No intervention	Originally reported as per 1000 orders
Hickman et al55	Not specified	450 “errors” per 100 000 orders (0.45%)	No intervention	Originally reported as per 1000 orders
Abraham et al26	Median time from medication ordering to its voiding was 0.38 h	Not specified	No intervention	Primarily a qualitative paper

Automated search methods identifying WPOE specifically
Paper	Descriptive information about error timings	Background error rate	Intervention error rate	Further relevant information
Used unique method of Levin et al48
Levin et al48	Median retract time (1 min) and reorder time (1 min) Highest risk of WPOE on Fridays, and 12:01 am to 6 am	64 per 100 000	No intervention	Originally reported as 0.064%
Used RAR method of Adelman et al22
Adelman et al22	Mean time to retraction of 1 min and 18 s	58 per 100 000 orders	1.5 per 1000 order sessions (control) 1.2 per 1000 sessions (intervention 1) 0.9 per 1000 sessions (intervention 2) (RAR rates)	WPOE calculated by dividing RAR events by PPV of RAR method by study authors. Ordering session—when providers selected a patient and then placed one or more orders.
Green et al23	Not specified	202 per 100 000 orders	4 months later—141 per 100 000 orders 2 years later—153 per 100 000 orders	Originally reported as per 1000 orders Intervention was a patient ID-verify alert
Adelman et al43	Not specified	45 per 100 000 orders	29 per 100 000 orders	WPOE calculated by dividing RAR events by PPV of RAR method by review team
Lombardi et al54	Not specified	6.125 “events” per week	4 “events” per week	Intervention was a patient ID-reentry function
Adelman et al51	Not specified	31.8 per 100 000 orders (single births) 50.3 per 100 000 orders (multiple births)	No intervention	WPOE calculated by dividing RAR events by PPV of RAR method by review team
Salmasian et al53	Not specified	142 per 100 000 orders	101 per 100 000 orders	WPOE calculated by dividing RAR events by PPV of RAR method by review team Intervention was displaying patient photographs within the EHR
Kern-Goldberger et al51	Not specified	60.6 per 100 000 order sessions (obstetrics) 32.2 per 100 000 order sessions (medical-surgical)	No intervention	WPOE calculated by dividing RAR events by PPV of RAR method by review team
Udeh et al52	Not specified	27.8 per 100 000 orders	21.1 per 100 000 orders	WPOE calculated by dividing RAR events by PPV of RAR method by review team
Unique method of Galanter et al49 coupled with RAR method of Adelman et al22
Galanter et al49	Not specified	Not specified	25 per 100 000 alerts	Originally reported as per 1000 orders Intervention is an indication-based CDS alert

Abbreviations: WPOE: wrong-patient order entry; RAR: retract-and-reorder; PPV: positive predictive value.