Edwell April, Huang Jia Xin, Bongiovanni Tasce, Pantell Matthew
Department of Pediatrics, University of California, San Francisco.
Department of Surgery, University of California, San Francisco.
JAMA Netw Open. 2025 Feb 3;8(2):e2461079. doi: 10.1001/jamanetworkopen.2024.61079.
Behavioral flags in the electronic health record (EHR) may introduce bias and perpetuate structural racism and discrimination. Descriptions of differences in the way that markers of behavioral risk are communicated will help clarify the inequities that pediatric patients and their families experience in the hospital.
To assess whether racially and socioeconomically marginalized pediatric patients and families are more likely than their counterparts to be assigned a behavioral flag in their EHR.
DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study used EHR data for pediatric patients (aged <18 years) hospitalized from June 2012 to July 2021 across care settings at the University of California, San Francisco health care facilities, an academic quaternary care hospital system that includes 2 pediatric inpatient facilities. The analysis was completed from December 29, 2022, to November 22, 2024.
The primary outcome of interest was any of the following behavioral flags placed in a patient's EHR: witnessed substance abuse, history of inappropriate behavior, security, violent behavior, dismissal from practice, and child protective services (CPS) hold. The primary variables were patients' race, ethnicity, insurance status, and primary language.
Of 55 865 pediatric encounters (52.2% among males; median patient age at the first encounter, 3 years [IQR, 0-12 years]), 236 (0.4%) had behavioral flags. Compared with encounters among patients who identified as White, encounters among patients who identified as Black or African American were more likely to have a behavioral flag (incidence rate ratio [IRR], 2.07; 95% CI, 1.32-3.25). Behavioral flags were also more likely among encounters of individuals with government insurance compared with those with private insurance (IRR, 2.60; 95% CI, 1.85-3.65). Black or African American patients younger than 1 year (IRR, 3.53; 95% CI, 1.80-6.91) and aged 1 to 7 years (IRR, 2.87; 95% CI, 1.34-6.15) had a higher likelihood of flag placement compared with their White counterparts.
This cohort study found significant inequities in incidence of behavioral flags in the EHR among racially and socioeconomically marginalized pediatric patients. This finding was most pronounced for Black or African American patients younger than 8 years, suggesting that this phenomenon may be a response to Black families rather than specific patient behavior.
电子健康记录（EHR）中的行为标识可能会引入偏见，并使结构性种族主义和歧视长期存在。对行为风险标志物传达方式差异的描述将有助于阐明儿科患者及其家庭在医院所经历的不平等现象。
评估在种族和社会经济方面处于边缘地位的儿科患者及其家庭在其电子健康记录中被赋予行为标识的可能性是否高于其他患者。
设计、设置和参与者：这项回顾性队列研究使用了2012年6月至2021年7月期间在加利福尼亚大学旧金山分校医疗设施住院的儿科患者（年龄<18岁）的电子健康记录数据，该医疗设施是一个学术性四级医疗系统，包括2个儿科住院设施。分析于2022年12月29日至2024年11月22日完成。
感兴趣的主要结局是患者电子健康记录中出现以下任何一种行为标识：目睹药物滥用、不当行为史、安保问题、暴力行为、被停止诊疗、儿童保护服务（CPS）扣留。主要变量是患者的种族、民族、保险状况和主要语言。
在55865次儿科诊疗中（男性占52.2%；首次诊疗时患者的中位年龄为3岁[四分位间距，0-12岁]），236次（0.4%）有行为标识。与认定为白人的患者的诊疗相比，认定为黑人或非裔美国人的患者的诊疗更有可能有行为标识（发病率比[IRR]，2.07；95%置信区间，1.32-3.25）。与有私人保险的人相比，有政府保险的人在诊疗中也更有可能出现行为标识（IRR，2.60；95%置信区间，1.85-3.65）。与白人同龄人相比，1岁以下（IRR，3.53；95%置信区间，1.80-6.91）和1至7岁（IRR，2.87；95%置信区间，1.34-6.15）的黑人或非裔美国患者被赋予行为标识的可能性更高。
这项队列研究发现，在种族和社会经济方面处于边缘地位的儿科患者中，电子健康记录中行为标识的发生率存在显著不平等。这一发现对于8岁以下的黑人或非裔美国患者最为明显，表明这种现象可能是对黑人家庭的一种反应，而非特定患者行为所致。
