RESUMO
Objective: To determine the prevalence of individual-level social risk factors documented in unstructured data from electronic health records (EHRs) and the relationship between social risk factors and adverse clinical outcomes. Study setting: Inpatient encounters for adults (≥18 years) at the University of Virginia Medical Center during a 12-month study period between July 2018 and June 2019. Inpatient encounters for labor and delivery patients were excluded, as well as encounters where the patient was discharged to hospice, left against medical advice, or expired in the hospital. The study population included 21,402 inpatient admissions, representing 15,116 unique patients who had at least one inpatient admission during the study period. Study design: We identified measures related to individual social risk factors in EHRs through existing workflows, flowsheets, and clinical notes. Multivariate binomial logistic regression was performed to determine the association of individual social risk factors with unplanned inpatient readmissions, post-discharge emergency department (ED) visits, and extended length of stay (LOS). Other predictors included were age, sex, severity of illness, location of residence, and discharge destination. Results: Predictors of 30-day unplanned readmissions included severity of illness (OR = 3.96), location of residence (OR = 1.31), social and community context (OR = 1.26), and economic stability (OR = 1.37). For 30-day post-discharge ED visits, significant predictors included location of residence (OR = 2.56), age (OR = 0.60), economic stability (OR = 1.39), education (OR = 1.38), social and community context (OR = 1.39), and neighborhood and built environment (OR = 1.61). For extended LOS, significant predictors were age (OR = 0.51), sex (OR = 1.18), severity of illness (OR = 2.14), discharge destination (OR = 2.42), location of residence (OR = 0.82), economic stability (OR = 1.14), neighborhood and built environment (OR = 1.31), and education (OR = 0.79). Conclusions: Individual-level social risk factors are associated with increased risk for unplanned hospital readmissions, post-discharge ED visits, and extended LOS. While individual-level social risk factors are currently documented on an ad-hoc basis in EHRs, standardized SDoH screening tools using validated metrics could help eliminate bias in the collection of SDoH data and facilitate social risk screening.
RESUMO
The conventional statistical methods to detect group differences assume correct model specification, including the origin of difference. Researchers should be able to identify a source of group differences and choose a corresponding method. In this paper, we propose a new approach of group comparison without model specification using classification algorithms in machine learning. In this approach, the classification accuracy is evaluated against a binomial distribution using Independent Validation. As an application example, we examined false-positive errors and statistical power of support vector machines to detect group differences in comparison to conventional statistical tests such as t test, Levene's test, K-S test, Fisher's z-transformation, and MANOVA. The SVMs detected group differences regardless of their origins (mean, variance, distribution shape, and covariance), and showed comparably consistent power across conditions. When a group difference originated from a single source, the statistical power of SVMs was lower than the most appropriate conventional test of the study condition; however, the power of SVMs increased when differences originated from multiple sources. Moreover, SVMs showed substantially improved performance with more variables than with fewer variables. Most importantly, SVMs were applicable to any types of data without sophisticated model specification. This study demonstrates a new application of classification algorithms as an alternative or complement to the conventional group comparison test. With the proposed approach, researchers can test two-sample data even when they are not certain which statistical test to use or when data violates the statistical assumptions of conventional methods.