To reduce seizure burden, children on the KD must maintain a low level of carbohydrate intake in order to maintain ketosis. This makes KD patients vulnerable to prescribing errors in the hospital setting. Computerized clinical decision support (CDS) can mitigate medication errors but has not been studied in children on the KD. 

A CDS alert was developed through formative and summative usability testing that warned prescribers when they entered an order for a carbohydrate-containing medication for KD patients. The alert was implemented using a quasi-experimental design with sequential crossover from control to intervention at two tertiary care pediatric hospitals within a single health system. Study participants included all KD patients admitted between January and October 2019. The primary outcome was carbohydrate-containing medication orders per patient-day. CDS utilization metrics included alert firing and acceptance rates.

There were 143 KD patient admissions totaling 714 patient-days during the study period. Following alert implementation, carbohydrate-containing medication orders declined from 0.69 to 0.31 orders per patient-day (absolute rate reduction 0.38, 95% CI 0.28-0.48). This corresponded to an estimated 104 erroneous orders prevented. The alert fired 114 times and was accepted (i.e., the order was removed) 70 times for an overall acceptance rate of 61%.

Implementation of a CDS alert at order entry resulted in a sustained reduction in carbohydrate prescribing errors for hospitalized KD patients without a significant increase in alert burden. The number of errors prevented was substantially higher than the number of alert acceptances, supportive of a learning effect; ordering clinicians exposed to the alert were less likely to attempt a future inappropriate order in the future. This study illustrates how computerized CDS developed with user-centered design principles can improve patient safety for children on the KD.