Dinah Foer, MD

Pronouns:

She/Her/Hers

Rank:

Instructor

Institution:

BWH

Department:

Medicine

Authors:

Dinah Foer, MD,* David M. Rubins, MD, MA, Vi Nguyen, BA, Li Zhou, MD, PhD, David W. Bates, MD, MSc.

Principal Investigator:

Dinah Foer

Utilization of Electronic Health Record Gender Demographic Fields: A Metadata Analysis

Increasingly, EHR data are being used for clinical research study and patient-oriented interventions to address women’s health, sex and gender differences. At MGB, EHR data are also linked to MGB Biobank samples, which may be used for translational and mechanistic study. Data quality regarding gender and sex should therefore be of paramount concern to researchers. While there is a strong literature on gender field implementation best practices, a post-implementation, longitudinal assessment of gender demographic field use is lacking, highlighting a knowledge gap for research and care of gender diverse populations. Our study addresses this gap. In doing so, we identified potential data quality concerns and system-level vulnerabilities in gender demographic data that may impact research efforts to incorporate sex and gender as research variables as well as patient safety. More broadly, the user-level trends may inform system interoperability initiatives and provider- and patient-level education and training regarding sex and gender.

Overview

Despite federally mandated collection of gender demographic data in the electronic health record (EHR) concerns about data quality persist. EHR metadata represent a novel data source for studying “big data” collection in clinical settings. Our objective was to evaluate longitudinal gender demographic field utilization.

 

Methods

Patients ≥18 years of age in MGB who had ≥1 value change to a gender demographic field between 1/8/2018 (all fields first available) and 1/1/2022 were included in this retrospective cohort study. MGB’s Epic EHR has three gender demographic fields: Gender Identity, Legal Sex and Sex Assigned at Birth (SAB). Gender demographic metadata were extracted. We quantified field completion rates and values, identified user types, and analyzed change over time.

 

Results

2,355,913 patients met cohort inclusion criteria. All had a Legal Sex, 45.4% had a Gender Identity, and 44.8% had a SAB. Legal Sex was changed for 67.5% of patients, followed by SAB (44.5%), and Gender Identity (42.1%). Users included patients, clinical providers, administrative staff, and automated updates via HL7 interface messages (communications between systems, i.e. other EHRs). Administrative users made the most changes (66.8%), patients made 36% and providers (N=6,992) made the fewest; provider use varied by subspecialty. Administrators made most changes to Gender Identity values, except when the change was made to Questioning/Unsure, or Genderqueer/Queer, which were more frequently made by providers or patients. Interface-initiated changes were most likely to be subsequently changed to a different value.

 

Discussion

Key findings include 1) administrative users make most field changes, 2) provider use varies by specialty, and 3) HL7 interfaces frequently update MGB’s fields—though have the highest rate of subsequent changes, suggestive of incorrect data transmission or more scrutiny to changes made by the interface. These findings raise questions about gender demographic data quality and have implications for EHR-based research and clinical decision support.