To design and implement two point-of-care clinical decision support alerts for pediatric blood pressure (BP)-- one for triage personnel and one for providers-- to improve the recognition of abnormal BP measurements in children ages 3-17 years.
Two point-of-care clinical decision alerts were built in the Epic electronic health record (EHR) to assess BP measurements at pediatric visits throughout a large tertiary healthcare system in Ohio. One alert was aimed at triage personnel, and the other alert was aimed at visit providers. The underlying logic evaluated the relationship of the entered BP to absolute thresholds and (if available) height, age and sex norms. The triage alert fired repeatedly upon entry of an elevated BP, and gave instructions for effectively rechecking the measurements. The provider alert fired if the last recorded triage BP was elevated at the time that the provider entered the chart, or if the provider entered a new high BP. Exclusion criteria suppressed the provider alert if the patient had a prior diagnosis of hypertension. The provider alert displayed historical BP data, diagnosis guideline links, an order set, and other acknowledgment options. Three months of data were collected and analyzed using Microsoft Excel.
A total of 13,854 visits were evaluated using the alerts. There were 2352 visits with an initially high BP: 671 visits had only a triage alert; 1665 visits had both a triage and a provider alert; and 16 visits had only a provider alert. Overall, the BP was normal on recheck in 886 (38%) of the visits. Subanalysis of triage visits showed that 534 (22%) ended the triage process with an ultimately normal BP; in the other 137, the patient did not see a provider on that day. Subanalysis of the 1665 visits where the patient saw both a nurse and a provider showed that 349 (21%) of the visits ended with a normal BP; in the remaining 1316 (79%) of visits the last BP entered was still high. The provider alert was acknowledged in 503 of 1681 provider alert visits (30%), a nephrology referral entered in 28 visits, a diagnosis of elevated BP in 4 visits, and a diagnosis of hypertension in 3 visits. Accounting for visits with duplicates among these outcomes, the provider alert had an impact in 425 (28%) additional visits. Thus, overall, triage and provider behavior was impacted in 66% of visits where an alert fired.
Clinical decision support has the potential to impact the recognition of abnormal BP values and improve the diagnosis of hypertension and pre-hypertension for pediatric patients. These alerts impacted the quality of pediatric BP data in the EHR and led to a combined 66% recognition rate of high BP measurements in pediatric patients.