Reasons for Social Work Referrals in an Urban Safety-Net Population: A Natural Language Processing and Market Basket Analysis Approach

Background: Encouraged by multiple federal policies, healthcare organizations are assuming greater responsibility for patients' social needs. This study describes the individual and co-occurring social needs that lead to a referral to social workers in primary care. Methods: In a secondary data analysis of a longitudinal cohort, we used natural language processing (NLP) to categorize reasons for social work referral documented in electronic health records referral orders (n = 9,473) from a federally qualified health center (2011–2016) in the United States, using a literature-derived classification scheme. We used market basket analysis (MBA) to identify co-occurring social needs. Results: The most frequent needs leading to a social work referral were financial (25%), pregnancy (25%), behavioral health (16%), and family/social support (9%) needs. The most frequently co-occurring needs are pregnancy with language limitation (support = 0.07; confidence = 0.78); behavioral health with family/social support (support = 0.03; confidence = 0.28); and financial with behavioral health (support = 0.025; confidence = 0.14). Conclusion: The diversity of reasons for social work referrals signifies the complexities of social needs among patients and the potential role for social workers in addressing these needs. A clearer understanding of patients’ social needs helps inform social work staffing decisions and the development of effective intervention packages to address patients’ social needs.