Table 2.

A summary of the key findings of the studies.

Author, year	Description of the used method (n = sample size)	Reported outcomes of hospitalization
Blalock et al. (2021) 8	The authors examined veterans at high risk for hospitalization through a mail survey on SDOH. The research team used eleven self‐reported items known to impact hospital admission and to be sensitive to the intervention to classify participants based on their social risk using latent class analysis. (n = 4684)	Applying latent class analysis, five subgroups were identified: “minimal SDOH vulnerabilities” (8% hospitalization rate), “poor/fair health with few SDOH vulnerabilities” (12% hospitalization rate), “social isolation” (10% hospitalization rate), “multiple SDOH vulnerabilities” (12% hospitalization rate), and “multiple SDOH vulnerabilities without food or medication insecurity” (10% hospitalization rate). The “Multiple SDOH vulnerabilities” subgroup showed a higher risk of 180‐day hospitalization than those with “minimum SDOH vulnerabilities” (OR: 1.53).
Canterberry et al. (2022) 15	A sample population of Medicare beneficiaries was surveyed for health‐related social needs. The data were linked to medical claims, and a regression model was applied to assess the association between social needs and healthcare utilization. (n = 56,155)	Compared to those without social needs (SN), those with any SN (OR: 1.53), one SN (OR: 1.35), two SN (OR: 1.68), three SN (OR: 1.57), four SN (OR: 1.82), and five or more SN (OR: 2.12) had higher odds of avoidable hospital stays.
Foster et al. (2020) 9	Children (0–18 years) eligible for Supplemental Security Income and Medicaid were included in the research. Multivariable hurdle Poisson regression was used to assess the connection between SDOH and one‐year hospital and ED utilization. (n = 226)	The ED visit rate was 55% (mean: 1.5 per year). The incidence of hospitalization was 20% (mean: 0.4 per year). Patients with a history of “unaddressed housing insecurity” (rate ratio: 1.55) or a “safety concern” (rate ratio: 2.04). had a higher annual ED usage rate among persons who had >0 ED Visits or Hospitalizations.
Jones et al. (2022) 10	6000 patients from seven primary care clinics were surveyed. 1748 were matched to medical claims. A two‐part model was used to assess the impact of SDOH on healthcare utilization. A modified logistic regression model was used to estimate risk ratios for cumulative SDOH variables and self‐reported chronic illnesses. (n = 1748)	Three or more SDOH needs were associated with an increased incidence of ED (aRR: 1.61) and inpatient (aRR: 1.76) visits.
MacCarthy et al. (2020) 11	A retrospective analysis of Medicaid beneficiaries utilizing a combination of patient‐reported SDOH and Medicaid claims. By latent class analysis, participants were divided into four social risk classes. (n = 8943)	With each higher (worse) social risk class, the adjusted log relative rates of both primary care visits and visits to the ED were higher. Participants who were “unemployed and had many social risks” (the highest social risk class) had a log relative primary care treatable rate of 39% and a log relative need for ED care rate of 29%, after adjusting for age, gender, and severity of illness.
Rogers et al. (2020) 12	Social needs were screened among a population of predicted high healthcare utilizers. Latent class analysis was applied to categorize the participants based on their reported SDOH. (n = 2,533)	Participants were separated into four social risk classes based on latent class analysis. Class 1 consisted of people with four or more self‐reported risks, and class 4 consisted of participants with no self‐reported risks. Despite having a lower Charlson comorbidity score, class 1 patients had considerably more total inpatient visits than class 4 patients (1.5 vs. 1.1, p < 0.001).
Wray et al. (2021) 13	A cross‐sectional study assessed the association of each SDOH and cumulative SDOH burden with hospitalization using a patient‐reported SDOH survey. (n = 55,186)	Hospitalized participants reported greater educational deficits (67%), economic instability (33%), food insecurity (14%), lack of community (14%), less access to healthcare (6%), and more social isolation (34%) compared with nonhospitalized individuals.
Zulman et al. (2020) 14	Veterans who had a 1‐year risk of hospitalization or death in the 75th or higher percentile were eligible to participate. The major outcomes of interest were all‐cause hospitalization 90 and 180 days after completion of the SDOH survey. (n=4,685)	Based on the Akaike information criterion, the regression model with survey‐based covariates and electronic health records‐based covariates predicted hospital admission at 90 days and 180 days more accurately than restricted models with only electronic health records‐based covariates.

Abbreviations: aRR, adjusted risk ratios; ED, emergency department; SDOH, social determinants of health; SN, social need.