Five studies evaluating CPOE (n greater than 18,259; the sample size in one study was not reported) and seven studies evaluating CDSSs (n=1,791) were included in the review. Of the CPOE studies, two were RCTs (one a crossover design) and three were observational studies. Of the CDSS studies, six were RCTs (one with a crossover design) and one was a non-randomised controlled trial.
CDSSs.
One RCT reported a statistically significant reduction in the rate of theophylline toxicity (18.9%, P=0.04) compared with controls. However, another RCT reported no significant difference in theophylline toxicity compared with controls. A decrease in adverse drug events (70%, P=0.02) with CDSSs was reported in the non-randomised controlled trial. One RCT reported a 17% greater pathogen susceptibility to an antibiotic regimen suggested by a computer versus a physician (P<0.001).
No statistically significant effect of CDSSs was reported for the following outcomes: bleeding complications (2 studies), over anticoagulation (1 study) and aminoglycoside toxicity (1 study).
CPOE.
One RCT reported a statistically significant improvement (25%, P<0.001) in ordering corollary medications with CPOE. The other RCT reported statistically significant decreases in inappropriate dose (13%, P<0.001) and inappropriate frequency of medication (24%, P<0.001). Observational studies reported statistically significant decreases in nonintercepted serious medication errors (86%, P<0.001) and medication errors (81%, P<0.001), and an improvement in five prescribing practices (P<0.001) with CPOE. One observational study reported decreases of 55% in nonintercepted serious medication errors and 17% in adverse drug events with CPOE, both of which were reported as not being statistically significant (P=0.37).