Redesigning Clinical Decision Support for Retinopathy of Prematurity Screening After Alert Failure

Area of Science:

• Pediatric ophthalmology
• Clinical informatics
• Healthcare quality improvement

Background:

• Retinopathy of prematurity (ROP) is a leading cause of preventable childhood blindness in premature infants.
• Current screening guidelines target infants with gestational age < 31 weeks or birthweight ≤ 1500g.
• Timely ROP screening during post-birth readmissions presents significant logistical challenges.

Purpose of the Study:

• To evaluate the performance of an interruptive clinical decision support (CDS) alert for ROP screening in readmitted premature infants.
• To analyze how data from the alert's performance informed the transition to a non-interruptive surveillance dashboard.

Main Methods:

• An interruptive alert was implemented for patients aged 1-365 days requiring ROP screening during hospital readmissions.
• Alert metrics, user responses, patient demographics, and ophthalmology consultations were extracted from the hospital's data warehouse.
• Statistical process control charts analyzed ophthalmology consultation rates before, during, and after alert implementation.

Main Results:

• The alert triggered 3309 times across 2194 patient encounters.
• Only 43% of encounters resulted in an "Accept and place order" response, with just 11% leading to an ophthalmology consult.
• A significant portion of consultations (34%) occurred despite alternative user responses, indicating alert inefficiency.

Conclusions:

• An analysis of the underperforming interruptive alert led to its deimplementation and replacement with a targeted, non-interruptive surveillance dashboard.
• This transition highlights the critical need to align CDS tools with clinical workflows, data availability, and user decision-making processes.
• Effective governance is essential for successful CDS implementation and optimization in healthcare settings.