Table 1.

Characteristics of studies and alerts included in the systematic review

Study	Study type	Intervention	Setting*	Country	Outcomes measured	Alert functionality**	Improved primary outcome (Yes/No)	False positive alerts reported*** (Yes/No)	Note
Bhardwaja et al. 2010 [33]	RCT	Drug-lab alert (Renal)	HMO	US	Proportion of medication errors in drug selection or dosing of targeted drugs	Medication decision guide printed in lieu of the prescription label.	Yes	Yes	Pharmacists were trained to ensure effective communication of the reason for alerts and the rationale for drug changes to prescribers and patients. All activities were documented electronically.
Raebel et al. 2007 [34]	RCT	Drug-pregnancy alert	HMO	US	Primary - proportion of pregnant women dispensed a FDA category D or X medication. Secondary - total number of first dispensing of targeted medications	Prescription label not printed until pharmacist intervened.	Yes	Yes	False positive alerts led to early cancellation of study. Specific intervention guideline and patient counselling script were developed.
Raebel et al. 2007 [20]	RCT	Drug-age alert	HMO	US	Proportion of first dispensing of medications on the targeted medication list	Prescription label not printed until pharmacist intervened.	Yes	Yes	Intervention guideline and patient counselling script were developed. Pharmacists were required to complete an intervention note in the system before being able to print the prescription label. Notes were reviewed retrospectively.
Humphries et al. 2007 [21]	Before-After study (no control)	Drug interaction alert (Contraindicated)	HMO	US	Proportion of patients co-dispensed two critically interacting drugs	Prescription label not printed, pharmacist must consult with the prescriber.	Yes	…	Pharmacists could bypass the alert. They documented their activities electronically. Decision support guide was developed to aid them in interpreting and resolving critical alerts. Scripted conversations were used to explain to patients the reason for alerts and the rationale for medication changes.
Mansour et al. 2010 [14]	Before-After study (no control)	Drug-lab alert	Large teaching hospital (504-bed)	US	Proportion of hospitalised patients requiring treatment for hyperkalaemia	Pop-up alert about patient’s last reported potassium level	Yes	…	Pharmacists documented their response to the alert electronically. They could override the alert but were required to provide a reason for doing so. Number of alerts was retrieved through a report generated by the information technology pharmacist. Inconsistent response to the alert led to the development of the hyperkalaemia treatment guideline.

* HMO – Health Maintenance Organisation, FDA – Food and Drug Administration, RCT – Randomised Controlled Trial.** Alert generated when a trigger medication order was entered. *** Ellipses indicate false positive alerts were not reported.