Accuracy of Patient Demographics: Prehospital and Electronic Health Record vs Self-Report

The purpose of this study is to assess accuracy of demographic data between patient self-report (SR), hospital electronic health record (EHR), and emergency medical services (EMS) electronic patient care report (ePCR). Race, in particular, has been shown to impact health outcomes, especially for minority patients1. No known literature exists quantifying accuracy of patient demographics in an EMS setting, which may be the first interaction a patient has with the healthcare system.

SR demographics were collected from 499 patients (ages 18-99) in the emergency department (ED) of an urban, level 1 trauma center using a 30-question survey consisting of multiple-choice and short-answer questions. Patients were asked their preferred language, sex at birth, gender identity, race, ethnicity, and religious affiliation. Accuracy comparisons were made between three sources: SR, the hospital EHR, and the EMS ePCR. Some data were not included in either EHR or EMS ePCR and excluded from agreement calculations.

Compared to SR, language was accurate for 97.8% of patients in their EHR. Agreement for sex assigned at birth for SR versus EHR and SR versus EMS ePCR was 99.6% and 99.6%, respectively. Concordance of gender identity was 98.6% for SR versus EHR and 98.7% for SR versus EMS ePCR. White and Black patients had the highest concordance comparing SR to EHR (99.0%; 98.3%) and SR to EMS ePCR (98.9%; 97.6%). Agreement for Latino patients was 63% (SR versus EHR) and 46.2% (SR versus EMS ePCR). Asian and Native patients (total n=12) had the lowest concordance for SR versus EHR (weighted average=58.3%) and SR versus EMS ePCR (weighted average=8.3%). Overall percent agreement for religious affiliation for SR versus EHR was 77.2%.

Preferred language, sex assigned at birth, and gender had the highest concordance between self-report compared to EHR and EMS ePCR. Race and ethnicity accuracy was least accurate for Latino, Asian, and Native patients. Religious affiliation was moderately accurate between SR and EHR. Further research should aim to explore the basis for medical record inaccuracy and re-evaluate concordance with LGBTQ+, Latino, Asian, and Native populations as these were underrepresented in our study but the most discordant. References: 1. PMID: 33280534