Racial, Ethnic, and Sex Differences in Social Risks and Social Needs Concordance Among Veterans.
Social screening initiatives often prioritize identifying social risks before offering support. The extent to which the emphasis on social risks contribute to overscreening and/or underdetection of needs, overall and across demographic subgroups, remains underexplored.
To examine the agreement between self-reported social risks and needs for support and variance in risk-need concordance across racial, ethnic, and sex subgroups.
This cross-sectional study examined responses to an online or mailed survey fielded between March 2 and May 9, 2023, to Veterans Health Administration patients who visited their primary care practitioner in January or February 2023. Survey weights were used to adjust for sampling frame and nonresponse. The data were analyzed between April 6 and December 15, 2025.
Intersection of race, ethnicity, and sex.
Risk-need sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and concordance across 12 domains (paying for basics, obtaining adult caregiving, obtaining childcare, finding or keeping work, paying for food, finding or keeping housing, getting transportation, accessing the internet at home, feeling isolated, feeling lonely, getting legal assistance, and getting additional education or job training) were measured using age-adjusted relative risk ratios (ARRRs) of risk-need discordance (vs concordance), controlling for a family-wise error rate of .05.
The analytic sample included 6596 respondents, representing 937 003 veterans after weighting (unweighted number [weighted percentage]: aged <65 years, 2992 [48.5%]; aged ≥65 years, 3604 [51.5%]; 1088 identifying as Black female [4.1%], 1140 as Black male [19.4%], 939 as Hispanic female [1.6%], 1279 as Hispanic male [11.3%], 802 as White female [5.3%], and 1348 as White male [58.4%] race or ethnicity and sex). Across domains, risk-need sensitivity ranged from 42% (95% CI, 34%-49%) for housing to 99% (95% CI, 98%-99%) for loneliness, and specificity ranged from 69% (95% CI, 66%-71%) for loneliness to 98% (95% CI, 97%-99%) for childcare. The PPV ranged from 27% (95% CI, 22%-33%) for housing to 69% (95% CI, 63%-75%) for legal issues, and NPV ranged from 93% (95% CI, 91%-94%) for housing to 99% (95% CI, 99%-99%) for both childcare and loneliness. In age-adjusted models, compared with White male veterans, Black male veterans had a significantly higher likelihood of need-without-risk discordance for paying for basics (ARRR, 3.95; 95% CI, 1.80-8.64), housing (ARRR, 2.67; 95% CI, 1.59-4.48), and adult caregiving (ARRR, 2.13; 95% CI, 1.30-3.48). Additionally, compared with White male veterans, the likelihood of need-without-risk discordance for loneliness was higher among White female (ARRR, 14.02; 95% CI, 2.85-68.95), Hispanic female (ARRR, 10.87; 95% CI 2.31-51.25), and Hispanic male (ARRR, 8.08; 95% CI, 2.47-26.39) veterans. Differences for risk-without-need discordance were not significant.
This cross-sectional study of Veterans Health Administration patients found that social risk questions were better at identifying veterans without a social need than those who needed support and underdetected need for support among Black male veterans. Improved survey methods are essential to accurately detect needs in diverse populations.
To examine the agreement between self-reported social risks and needs for support and variance in risk-need concordance across racial, ethnic, and sex subgroups.
This cross-sectional study examined responses to an online or mailed survey fielded between March 2 and May 9, 2023, to Veterans Health Administration patients who visited their primary care practitioner in January or February 2023. Survey weights were used to adjust for sampling frame and nonresponse. The data were analyzed between April 6 and December 15, 2025.
Intersection of race, ethnicity, and sex.
Risk-need sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and concordance across 12 domains (paying for basics, obtaining adult caregiving, obtaining childcare, finding or keeping work, paying for food, finding or keeping housing, getting transportation, accessing the internet at home, feeling isolated, feeling lonely, getting legal assistance, and getting additional education or job training) were measured using age-adjusted relative risk ratios (ARRRs) of risk-need discordance (vs concordance), controlling for a family-wise error rate of .05.
The analytic sample included 6596 respondents, representing 937 003 veterans after weighting (unweighted number [weighted percentage]: aged <65 years, 2992 [48.5%]; aged ≥65 years, 3604 [51.5%]; 1088 identifying as Black female [4.1%], 1140 as Black male [19.4%], 939 as Hispanic female [1.6%], 1279 as Hispanic male [11.3%], 802 as White female [5.3%], and 1348 as White male [58.4%] race or ethnicity and sex). Across domains, risk-need sensitivity ranged from 42% (95% CI, 34%-49%) for housing to 99% (95% CI, 98%-99%) for loneliness, and specificity ranged from 69% (95% CI, 66%-71%) for loneliness to 98% (95% CI, 97%-99%) for childcare. The PPV ranged from 27% (95% CI, 22%-33%) for housing to 69% (95% CI, 63%-75%) for legal issues, and NPV ranged from 93% (95% CI, 91%-94%) for housing to 99% (95% CI, 99%-99%) for both childcare and loneliness. In age-adjusted models, compared with White male veterans, Black male veterans had a significantly higher likelihood of need-without-risk discordance for paying for basics (ARRR, 3.95; 95% CI, 1.80-8.64), housing (ARRR, 2.67; 95% CI, 1.59-4.48), and adult caregiving (ARRR, 2.13; 95% CI, 1.30-3.48). Additionally, compared with White male veterans, the likelihood of need-without-risk discordance for loneliness was higher among White female (ARRR, 14.02; 95% CI, 2.85-68.95), Hispanic female (ARRR, 10.87; 95% CI 2.31-51.25), and Hispanic male (ARRR, 8.08; 95% CI, 2.47-26.39) veterans. Differences for risk-without-need discordance were not significant.
This cross-sectional study of Veterans Health Administration patients found that social risk questions were better at identifying veterans without a social need than those who needed support and underdetected need for support among Black male veterans. Improved survey methods are essential to accurately detect needs in diverse populations.
Authors
Russell Russell, Frank Frank, Purkayastha Purkayastha, McCoy McCoy, Leder Leder, Gordon Gordon, Lamba Lamba, Procario Procario, Moy Moy, Hausmann Hausmann
View on Pubmed