The information provided by electronic health record (EHR) data may provide sufficient risk prediction capabilities without the inclusion of social factors, according to a study recently published in JAMA Network Open. Social factors, such as neighborhood socioeconomic status, are helpful in risk prediction, but EHR data may be reliable enough to single-handedly power population risk assessments.
This cohort study included 90,097 participants in a training set, of whom 57,507 were women and whose age was a mean of 47.2±17.7 years. It also included a testing set of 122,812 participants of whom 75,517 were female and whose mean age was 46.2±17.9 years. In the 1-year period before the 7-year study period began, participants had ≥1 health care experience and residence. Census tract data was merged with participant data, allowing the study researchers to establish associations between neighborhood socioeconomic status and likelihood of adverse outcomes, which included stroke, myocardial infarction, influenza, asthma, hospitalization, and trips to emergency departments. Neighborhood socioeconomic status was used as a variable to gauge its effect on risk prediction of adverse outcomes compiled from the electronic health record. To determine whether it affected risk prediction power, machine-learning procedures were employed.
Living in a neighborhood with lower socioeconomic status was associated with less time to use of emergency department services. It was also associated with less time elapsed before hospitalization due to adverse outcomes. Neighborhood socioeconomic status showed moderate, variable predictive value. However, adding it to EHR variables did not improve predictive performance connected with any outcome.
Limitations to this study included the use of a single geographical region, a single institution’s EHR data, a single parameterization of neighborhood socioeconomic status, and the potential for different results among different modeling algorithms.
The study researchers concluded that “the social environment is associated with health outcomes. However, these results suggest that information about the environment in which a person lives may not contribute much more to population risk assessment than is already provided by EHR data. Although this result does not mean that integrating social determinants of health into the EHR has no benefit, researchers may be able to use EHR data alone for population risk assessment.”
Bhavsar NA, Gao A, Phelan M, Pagidipati NJ, Goldstein BA. Value of neighborhood socioeconomic status in predicting risk of outcomes in studies that use electronic health record data. JAMA Netw Open. 2018;1(5):e182716.