Implementation context for addressing social needs in a learning health system: a qualitative study

Abstract

This version may be subject to change during the production process.

Integration of social determinants of health information within the primary care electronic health record: a systematic review of patient perspectives and experiences

“…The literature search identified 2701 papers, 397 duplicates were removed, 2279 were removed at title and abstract screening, 18 were excluded at full text review, a further 7 were included at citation screening, 14 were included in the analysis (Figure 1). The 14 studies included: 1-1 interviews (n=8) (39)(40)(41)(42)(43)(44)(45)(46), surveys (n=4) (43,(47)(48)(49), focus groups (n=5) (41-43, 50, 51), and a systematic review (n=1) (52). 11 studies were based in USA (39-45, 47, 48, 50, 51), 2 in Canada (46,49), and 1 from UK (52) (See Supplementary Table 3).…”

“…Participants made connections between SDOH and health, for example, the effect of food insecurity on diet related illnesses, poor housing conditions on asthma, and SDOH related stress on health (40). Some participants described how they normally avoid talking about SDOH with their doctor (41,42), due to the stigma associated with asking for assistance (42), therefore they appreciated doctors initiating the discussion (41,42,51). Participants believed that by sharing SDOH information, the care team would gain a greater understanding of their health, the environment they live in (e.g.…”

A Fair Individualized Polysocial Risk Score for Identifying Increased Social Risk in Type 2 Diabetes

“…Addressing patients' unmet social needs in health care settings is a complex task due to 1) the insufficient SDoH records in EHRs (e.g., lack of use of Z codes for SDoHassociated diagnosis, 72 and extremely low utilization of existing SDoH screening surveys embedded in EHRs 17 ), 2) the concerns about the extra burden on providers 11,73,74 and potential harms on patients 20,22,23,75 , 3) the potential data bias associated with SDoH that exists within subpopulations (e.g., racial and ethnic minority groups 12 ), and 4) the observational natural of real-world EHR data (e.g., confounding and selection bias). 76 Our EHR-based iPsRS pipeline was carefully designed to overcome the abovementioned limitations.…”

“…Moreover, our model is designed to generate an initial iPsRS based on historical EHR data at the beginning of a medical encounter to guide targeted, in-person conversations between the patient and provider to collect additional SDoH information and update the iPsRS as needed, which has been carefully considered for its integration into existing clinical workflow to avoid potential harms to patients imposed by survey-type SDoH screenings and to promote patient-provider shared decision making on addressing patients' unmet social needs. 20,22,23,75 With applications of multiple XAI and causal learning techniques. e.g., SHAP 40 values to identify key predictors and causal structure learning 41,[59][60][61] to identify causal pathways, our iPsRS is able to generate interpretable outputs and has shown its ability to identify potential focal targets for intervention and policy programs to address patients' unmet social needs essential to their health outcomes.…”

Global developments in social prescribing

“…9 65-67 The complexity of social prescribing pathways makes evaluation difficult, 68 though technology-facilitated data standardisation and tracking aid this process through a Learning Health Systems approach using iterative feedback. 69 In particular, tying sociodemographic data of social prescribing participants to outcomes data can help elucidate what approaches to social prescribing work for whom and in what circumstances. 9 Interventions will likely need to be tailored for specific subpopulations (eg, older adults, persons with disabilities, immigrant communities, etc).…”