Importance: Health-related social needs (HRSN) data are used in referral and treatment decisions, in population health management strategies, and in health services research. However, evidence suggests HRSN data are at risk for bias. Objective: To identify and classify sources of bias in HRSN data and the implications for usage for patient care and population health. Design, Setting, and Participants: In this qualitative study, key informant interviews with patients and health care professionals in Indiana and Florida (recruited using multiple recruitment methods and snowball sampling) were conducted from January to May 2025. Key informants in Indiana were primarily associated with a public safety-net system including federally qualified health centers, or a multihospital system with services statewide. In Florida, key informants were associated with a large academic medical center, with some dual-affiliated with a US Department of Veterans Affairs hospital. Health care professionals had the titles such as physician, social worker, and community health worker. Main Outcomes and Measures: Data collection occurred via video or telephone call. Interviews followed a semistructured interview guide grounded in a framework describing sources of potential bias in health data. Participants were asked about HRSN data collection practices and experiences, documentation practices, responses to HRSN data collection, and how, in their own words, they defined food insecurity, housing instability, financial strain, and transportation barriers. Thematic analysis followed a consensus coding approach. Results: A total of 20 patients and 20 health care professionals were recruited (40 informants total; 22 aged 40-64 years 42.5%; 27 female 67.5%). Participants described variation in HRSN data collection and differing availability of organizational resources that contributed to sampling bias. Patients and professionals reported detection bias was possible because HRSNs could be intentionally sought during visits or not collected at all. Concerns about stigma or embarrassment, power distance, and privacy could result in nonresponse bias. Health care professionals and patients could all offer slightly different, or nuanced, definitions of different HRSNs. These more expansive or restrictive definitions could lead to misclassification bias. Conclusions and Relevance: In this qualitative study, both patients and health care professionals described opportunities for bias in HRSN data collection and documentation. These findings suggest that, while HRSN data are potentially valuable to patient care, their fitness for use in organizational decision-making, research, and health policy may need improvement.
Vest et al. (Fri,) studied this question.