Help-Seeking Patterns Among Socially Isolated Patients in Primary Care

We conducted a retrospective analysis of electronic health records from a systematic social needs screening program implemented across a large integrated healthcare system from August 2017 to December 2024. This study was considered exempt by the University of Michigan institutional review board, and informed consent was not needed because the data were anonymous. The study population included all primary care patients aged ≥ 18 years who completed social needs screening during routine visits. Social isolation was assessed using a validated single-item screener: “Do you feel isolated from others?” administered with three timeframes: past 12 months, currently, and past 4 weeks. Response options were binary (yes/no). Patients were classified as socially isolated if they responded “yes” to any timeframe. The primary outcome was interest in assistance, assessed with two questions: (1) “Do you want to be contacted by the Guest Assistance Program for help on any of your responses?” and (2) “Do you want to get connected with resources for isolation?” (y/n). Demographic and clinical characteristics, and co-occurring social needs (categorized as 0, 1–2, or ≥ 3) were also examined. Among socially isolated patients, adjusted multivariable logistic regression identified predictors of requesting assistance, calculating adjusted odds ratios (AOR) with 95% confidence intervals. Among 110,995 patients, 25,067 (22.6%) screened positive for social isolation. Of patients with complete assistance data ( n = 60,034), only 1,860 of 21,256 socially isolated patients (8.8%) requested assistance. Socially isolated patients were more likely to be female (63.9% vs. 56.3%), single (37.7% vs. 25.2%), or divorced/widowed (11.0% vs. 7.3%). Social isolation rates were higher among Non-Hispanic Black patients (8.7% vs. 6.6%) and lower among Asian patients (5.9% vs. 9.2%). Isolated patients had higher rates of Medicaid coverage (10.9% vs. 5.8%), severe obesity (BMI ≥ 35: 20.6% vs. 14.8%), and co-occurring social needs (Table 1 ). Among socially isolated patients, several groups were more likely to request assistance: older adults (ages 65–74: AOR 1.36, 95% CI 1.09–1.70; ≥ 75: AOR 1.75, 95% CI 1.36–2.25), unmarried patients (single: AOR 1.58, 95% CI 1.38–1.81; divorced/widowed: AOR 1.48, 95% CI 1.23–1.77), non-white patients (Non-Hispanic Black: AOR 1.75, 95% CI 1.49–2.07; Hispanic: AOR 1.47, 95% CI 1.11–1.94; Asian: AOR 1.84, 95% CI 1.46–2.30), and publicly insured patients (Medicare: AOR 1.42, 95% CI 1.19–1.69; Medicaid: AOR 1.56, 95% CI 1.32–1.83) (Table 2 ). The strongest predictor was co-occurring social needs. Patients with 1–2 needs (AOR 3.18, 95% CI 2.78–3.62) and ≥ 3 needs (AOR 9.10, 95% CI 7.67–10.80) were more likely to request assistance than those with no needs. While nearly one in four patients experience social isolation, only 10% request assistance. The most critical finding is that patients with multiple co-occurring social needs were nine times more likely to seek help, suggesting that isolation alone may be perceived differently than isolation combined with material hardships. This pattern means that socially isolated patients without other material needs are most likely to remain undetected. Contrary to expectations, traditionally vulnerable populations were more likely to seek help when isolated, likely because they experience isolation alongside other pressing needs. Healthcare systems should implement proactive strategies to capture patients who screen positive for isolation but lack other co-occurring social needs. Current passive screening approaches may systematically miss these patients who appear demographically advantaged and lack obvious material hardships. Study limitations include the retrospective design limiting causal inference, potential unmeasured confounding, and limited generalizability from a single healthcare system. These findings underscore the need for healthcare systems to redesign screening approaches to identify and engage socially isolated patients who may not self-advocate, addressing a significant challenge affecting nearly one-quarter of patients.