ABSTRACT
Objective
To assess the utility of using discrete primary care electronic health record (EHR) data to identify social risk referrals in a national network of community-based clinics.
Data Sources and Study Setting
Primary data were abstracted from the OCHIN network EHR (June 2016 to February 2022) of 1459 community-based clinics across the United States.
Study Design
Structured data elements included 913 commonly used smartphrases and 53 procedure codes that were considered potential indicators of social risk referrals. Using stratified purposive sampling, we compared these discrete data with clinical notes to assess concordance of social risk referral documentation, and of the prevalence, types, and outcomes of such referrals.
Data Collection/Extraction Methods
Smartphrases were classified into three categories (likely, possible, or unlikely to indicate a social risk referral); 50 chart notes were sampled for each of the 25 most frequently used smartphrases in each category, and for 53 of the most frequently used procedure codes. A total of 6104 chart notes were reviewed.
Principal Findings
In 59% of chart notes where discrete data suggested a social risk referral occurred, there was no documentation of this in the note. Primary domains addressed were food insecurity (38%), financial stress (18%) and housing needs (18%). Common referral activities included providing contact information (26%), help with assistance applications (17%), and direct provision of resources (16%). Documentation indicated the patient received resources in 29% of notes.
Conclusions
EHR documentation of social risk referrals in structured data fields is inconsistent. Further work should establish best practices, especially given emerging policies that tie payments to documentation of social risk screening and intervention provision. Community health centers may struggle to use data elements such as smartphrases and procedure codes to monitor and report on their social risk referrals until standardized coding practices are established and effectively implemented.