Validity
We took a sample of 20 de-identified University of Miami records and asked the Health IT team to reidentify them. Our sample was randomly selected from the highest two deciles of SCD risk since this was the group in which we mostly wanted to determine the use of CV testing. We compared the data in our enterprise data environment with the EHR to assess the validity of the components of the SCD score and if an echocardiogram or stress test was done. We defined accuracy if the same value of the CV risk factor or the CV testing was done on the same date when comparing the medical record and the data on the research environment.