Abstract
Background:
Examining the associations of Social Determinants of Health (SDOH) with postoperative delirium in older adults will broaden our understanding of this potentially devastating condition. We explored the association between SDOH factors and incident postoperative delirium.
Methods:
A retrospective study of a prospective cohort of patients enrolled from June 18, 2010 to August 8, 2013 across two academic medical centers in Boston, Massachusetts. Overall, 560 older adults age ≥70 years undergoing major elective non-cardiac surgery were included in this analysis. Exposure variables included income, lack of private insurance, and neighborhood disadvantage. Our main outcome was incident postoperative delirium, measured using the Confusion Assessment Method long form.
Results:
Older age (odds ratio, OR: 1.01, 95% confidence interval, CI: 1.00, 1.02), income <20,000 a year (OR: 1.12, 95% CI: 1.00, 1.26), lack of private insurance (OR: 1.19, 95% CI: 1.04, 1.38), higher depressive symptomatology (OR: 1.02, 95% CI: 1.01, 1.04), and the Area Deprivation Index (OR: 1.02, 95% CI: 1.01, 1.04) were significantly associated with increased risk of postoperative delirium in bivariable analyses. In a multivariable model, explaining 27% of the variance in postoperative delirium, significant independent variables were older age (OR 1.01, 95% CI 1.00, 1.02), lack of private insurance (OR 1.18, 95% CI 1.02, 1.36), and depressive symptoms (OR 1.02, 95% CI 1.00, 1.03). Household income was no longer a significant independent predictor of delirium in the multivariable model (OR:1.02, 95% CI: 0.90, 1.15). The type of medical insurance significantly mediated the association between household income and incident delirium.
Conclusions:
Lack of private insurance, a social determinant of health reflecting socioeconomic status, emerged as a novel and important independent risk factor for delirium. Future efforts should consider targeting SDOH factors to prevent postoperative delirium in older adults.
Keywords: Private insurance, delirium, social determinants of health, clinical outcomes in hospitalization
INTRODUCTION
Social determinants of health (SDOH) are the constellation of contextual, environmental, social, and financial factors present throughout a person’s life1 that influence health. SDOH factors are receiving increased attention, first as major sources of health disparities in the United States (US)2 and second as factors that influence clinical outcomes even years after initial exposure.3 Understanding mechanisms by which SDOH factors affect health in older adults will enhance efforts to eliminate health disparities and align with the goals of precision medicine.4
SDOH factors have been studied in the context of chronic conditions, including neurodegenerative disorders and aging.5–7 Their influence relative to acute conditions, such as incident postoperative delirium, has yet to be well-investigated. Incident postoperative delirium (referred to as delirium from here onward), common among hospitalized older adults, is an acute alteration in cognitive functioning, predominantly involving diminished attention that includes fluctuations in symptoms throughout its duration (typically days or weeks) and represents a distinct change from baseline.8 The literature exploring individual-level or patient-related risk factors for delirium is rich with factors such as advancing age, frailty, increased disease burden, and pre-existing cognitive impairment shown to be significant risk factors.9,10 Research focused on identifying SDOH factors that increase delirium risk may guide the development of novel interventions that target contextual-level factors to prevent delirium.
We developed a conceptual framework of SDOH for delirium in older adults.11 By reviewing previous frameworks and published literature and engaging a multidisciplinary panel of experts in neurology, gerontology, geriatrics, neuropsychology, epidemiology, and nursing, we classified SDOH factors across seven domains: 1. Demographic; 2. Social and Economic; 3. Well-being: Emotional and physical; 4. Social Capital; 5. Social Engagement; 6. Built and Social Environment; and 7. Lifecourse Factors.11 Our framework addressed an important gap in the literature by integrating elements relevant to delirium in older adults.
To empirically evaluate our framework, we have examined the association of specific SDOH factors with delirium. We found a significant association between the Area Deprivation Index (ADI), a neighborhood disadvantage marker, and delirium severity in the Successful Aging after Elective Surgery (SAGES) cohort.12,13 Additionally, proxies of childhood socioeconomic status (e.g., parental education, parental occupation) and literacy levels were associated with an increased risk for delirium, and these associations persisted even after controlling for other well-established risk factors.14–16 Taken together, our previous work suggests that SDOH factors confer vulnerability for delirium above and beyond other well-accepted risk factors.
The present study examines variables representing the different domains from our conceptual framework in the SAGES study.11,13 To capture our à priori hypotheses, our approach was guided and informed by a directed acyclic graph (DAG),17 a nonparametric diagram that describes hypothesized observed and unobserved causal relationships between exposure and outcome variables. We first examined bivariable associations between SDOH factors with delirium. Next, following the DAG, we examined how household income was associated with delirium. We further explored whether other markers of socioeconomic status, including neighborhood disadvantage and type of medical insurance, mediated the association between household income and delirium.
METHODS
Design
Patients enrolled in the SAGES study, a prospective observational cohort study designed to examine risk factors for incident postoperative delirium, were included. Details about the SAGES study have been previously reported.13 Briefly, English-speaking patients age ≥70 years at the time of enrollment and scheduled for major (non-cardiac) surgery at one of two Harvard-affiliated hospitals with an anticipated hospital stay greater than two days were eligible to participate. Patients with evidence of dementia, a history of delirium or hospitalization within three months of the baseline interview, severe deafness or blindness, a terminal condition, or a history of alcohol withdrawal were excluded.
Setting and Participants
Written informed consent was obtained following procedures approved by the institutional review board of Beth Israel Deaconess Medical Center, with ceded review from Brigham and Women’s Hospital and Hebrew SeniorLife, the study coordinating center. The SAGES study enrolled 560 older adults between June 18, 2010, and August 8, 2013, for elective orthopedic [n (%)] 454 (81%), gastrointestinal 71 (13%), or vascular 35 (6%) surgery.
Data Sources
SDOH factors reflect the baseline status of patients before surgery. An in-person interview was completed in patients’ homes before surgery and included questions about their personal background, medical histories, and their daily functioning.13,18 All patients also completed comprehensive neuropsychological evaluation.19 A medical chart review was completed within 30 days after surgery. All study interviews were conducted by trained bachelor’s and master’s-level research associates who underwent semiannual standardization.13 A trained study physician completed all medical record reviews.
Social Determinants of Health Variables in the SAGES Cohort
SDOH factors in our analyses were organized according to the domains from our previously developed framework: 11
Demographic Factors.
We included age, race, and sex. Age (years) was calculated using the patient’s date of birth. Patients were asked to self-identify race and ethnicity according to US Census categories. Racial categories were dichotomized as White vs non-White. Sex was determined by self-identification.13
Socioeconomic Factors.
Formal education was self-reported as years completed,13 and dichotomized for analysis (<12 vs. ≥12 years). Literacy level was assessed by the age-standardized Wechsler Test of Adult Reading [WTAR] score.20 To account for household wealth, patients were asked to select the option that best described the total combined income for themselves and all family members residing in their household for the year prior to the interview: A. <$20,000, B. >$20,000 to <50,000, C. >50,000 to <$75,000, >$75,000 to <100,000, and E. >100,000. Household income was dichotomized as <$20,000 vs. ≥$20,000 to most closely approximate poverty thresholds in 2010, 2011, 2012, and 2013 (years of study recruitment). According to the US Census Bureau, the weighted average poverty thresholds for a family of one or two persons with a 65 years and older householder were: A. $10,458 and $13,194 for 2010, B. $10,788 and $13,609 for 2011, C. $11,011 and $13,892 for 2012, and D. $11,173 and $14,095 for 2013.21 Insurance information was extracted from the medical records and coded as the presence of 1. Federal or state-sponsored insurance (Medicare alone or Medicare with Medicaid/MassHealth) or 2. Private insurance (employer-sponsored programs or private Medicare supplemental plans).22 All patients in SAGES had at least one type of medical insurance. The lack of private insurance was considered a vulnerability, potentially limiting access to healthcare.
Well-Being Factors.
Depressive symptoms were assessed using the 15-item Geriatric Depression Scale (GDS),23 with scores ranging from 0 to 15 (higher scores worse). Health-related quality of life was assessed using the 12-item Short Form Survey (SF12),24 with composite scores on the mental and physical health portions ranging from 0 to 100 (lower scores worse). Tobacco use was self-reported, and active smoking at baseline was considered a risk factor. Body mass index was calculated based on self-reported height and weight.13 Hearing and vision were assessed using the Whisper Test25 and Snellen test,26 respectively, and were dichotomized. Impairments in hearing or vision were considered risk factors.
Social Capital Factors.
Details about a patient’s living situation (i.e., living alone vs. with others) and marital status (i.e., never married, married, or living with a partner, widowed, divorced, separated) were self-reported.13 Living alone and being unmarried were considered risk factors.
Activities and Social Engagement Factors.
The Cognitive Activity Scale (CAS) indicated whether participants engaged in cognitively stimulating activities.27 Baseline activity level at study enrollment was used to reflect the “current level” of social involvement. CAS scores range from 0 to 110, with higher scores indicating increased social involvement. Participation across three categories of activities (monthly) was derived from the Social Activities Scale from the Epidemiologic Studies of the Elderly (EPESE):28 attending religious services, visiting with friends or relatives, and participating in group or community activities.
Built Environment Factors.
The Area Deprivation Index (ADI) incorporates 17 US Census indicators of resources (e.g., plumbing, telephone, etc.) to rate neighborhood disadvantage. Scores range from 1 to 100 (100= most disadvantaged),29 with the highest 5th percentile used to indicate neighborhoods in the disadvantaged range. The walkability index uses 7 US Census indicators of neighborhood density to produce a score from 1 to 20 (20= higher levels of walkability),30 with a score <10.5 indicating below average walkability.
Lifecourse Factors.
Primary language, country of origin, and parental education were collected by participant report. Paternal and maternal education levels were reported as the highest grade of formal education completed by their parents,13 and dichotomized as <12 vs. ≥12 years.
Incident Postoperative Delirium
Delirium status was assessed daily following surgery using the Confusion Assessment Method (CAM), long-form, a widely accepted approach with high sensitivity (94%), specificity (89%), and inter-rater reliability (kappa=0.92).31 The CAM is rated based on patient interviews, including a brief cognitive assessment.31 Combined with a validated chart review method,32 patients were classified as delirious if either CAM or chart criteria were met during hospitalization.
Statistical Methodology
The analytic plan, guided by the DAG (Figure 1),17,33 tested the hypothesized relationships. In terms of observed relationships, our DAG included variables from six out of the seven domains in our SDOH framework.11 In accordance with DAG processes,17 unobserved variables included: 1. Unobserved genetic or epigenetic factors that may influence vulnerability to delirium and 2. Systemic factors (e.g., racism, ageism) that may lead to inequitable access to healthcare. Ultimately, the DAG reflects our hypothesis that the type of medical insurance and characteristics of the built environment would mediate the association between household income and delirium.
Figure 1.
Directed Acyclic Graph (DAG) of SDOH Variables and Delirium
The figure illustrates our assumptions and hypothesis about the association between selected SDOH factors and delirium incidence. SDOH= Social determinants of health; GDS= Geriatric depression scale; WTAR= Wechsler Test of Adult Reading; ADI= Area Deprivation Index; Insurance= lack of private insurance; U= Unobserved variables; U1= Genetic or epigenetic predisposition; U2= Systemic factors leading to inequity (e.g., racism). Solid green lines reflect hypothesized observed causal relationships between exposure and outcome variables. Dashed red lines reflect hypothesized unobserved causal relationships between exposure and outcome variables.
Complete-case data were used to calculate descriptive statistics for the SAGES cohort and are presented for the entire sample and by delirium status (See Table 1). Distributions of all variables were assessed for normality and graphically examined to identify outliers. Missing data occurred at rates of <1% to 11% of our patients: household income (n=62; 11%), country of origin (n=32; 6%), parental education (n=10; 2%), body mass index (n=6; 1%), the Geriatric Depression Scale (n=2; <1%), vision (n=2; <1%), hearing (n= 1; <1%), and participation in social activities (n= 1; <1%). The missing data were handled using multiple imputations by chained equations,34 based on five cycles of imputation. Imputed data were used for bivariable and multivariable regression analyses. Bivariable associations between SDOH variables and delirium were estimated using logistic regression models (odds ratios and 95% confidence intervals) for all SDOH factors. Next, multivariable logistic regression models were used to estimate the association between household income and delirium and whether this association remained robust when other SDOH variables were incorporated into the model separately (series models) and then together (full model). Model fit and was assessed using Nagelkerke R2.
Table 1.
SDOH Factors in the SAGES Study
| SDOH Framework Domain | Risk Factor | Entire Sample N= 560 |
Delirium N= 134 |
No Delirium N= 426 |
|---|---|---|---|---|
| Demographic | Age, mean (SD) | 76.7 (5.2) | 77.4 (5.1) | 76.4 (5.2) |
| Race, n (%) | ||||
| White | 518 (93) | 121 (90) | 397 (93) | |
| Non-White | 42 (7) | 13 (10) | 29 (7) | |
| Sex/Gender, N (%) | ||||
| Male | 234 (42) | 53 (40) | 181 (42) | |
| Female | 326 (58) | 81 (60) | 245 (58) | |
| Education, n (%) | ||||
| ≥ 12 years | 536 (96) | 127 (95) | 409 (96) | |
| < 12 years | 24 (4) | 7 (5) | 17 (4) | |
| Wechsler Test of Adult Reading, mean (SD) | 37.7 (9.9) | 35.7 (9.9) | 38.3 (9.9) | |
| Socioeconomic | Current Household Income, n (%), N=498* | |||
| ≥ $20,000 | 438 (88) | 96 (72) | 342 (80) | |
| < $20,000 | 60 (12) | 21 (16) | 39 (9) | |
| Medical Insurance, n (%) | ||||
| Private insurance (i.e., employer plan or Medicare supplement) |
523 (93) | 119 (89) | 404 (95) | |
| Federal or State-sponsored Insurance (Medicare alone, n=22; or Medicaid/MassHealth, n=15) | 37 (7) | 15 (11) | 22 (5) | |
| Geriatric Depression Scale, mean (SD), N=558† | 2.5 (2.5) | 3.0 (2.8) | 2.3 (2.4) | |
| GDS > 6** | 50 (9%) | 18 (13%) | 32 (8%) | |
| SF-12 Mental Health Composite T-score, mean (SD) | 60.8 (11.3) | 59.5 (12.1) | 61.2 (11.0) | |
| SF-12 Physical Health Composite T-score, mean (SD) | 39.7 (12.4) | 37.9 (11.9) | 40.2 (12.6) | |
| Well-being | Tobacco use, n (%) | |||
| No use | 534 (95) | 128 (96) | 406 (95) | |
| Active Use | 26 (5) | 6 (4) | 20 (5) | |
| Body Mass Index, mean (SD), N=554§ | 28.5 (5.5) | 29.3 (5.2) | 28.3 (5.6) | |
| Hearing, n (%), N=559‡ | ||||
| No hearing impairment | 377 (67) | 85 (63) | 292 (69) | |
| Hearing Impairment | 182 (33) | 49 (37) | 133 (31) | |
| Vision, n (%), N=558† | ||||
| No visual impairment | 555 (99) | 131 (98) | 424 (96) | |
| Visual impairment | 3 (1) | 2 (1) | 1 (<1) | |
| Social Capital | Marital Status, n (%) | |||
| Married | 332 (59) | 79 (59) | 253 (59) | |
| Non-married | 228 (41) | 55 (41) | 173 (41) | |
| Living Situation, n (%) | ||||
| Living with someone | 393 (70) | 95 (71) | 298 (70) | |
| Living alone | 167 (30) | 39 (29) | 128 (30) | |
| Activities and Social Engagement | Cognitive Activities Scale at 70 years, mean (SD) | 49.8 (9.9) | 48.9 (9.3) | 50.0 (10.1) |
| Physical and Social Activities, N=559‡ | ||||
| Attended religious service > 1 per month | 245 (44) | 67 (50) | 178 (42) | |
| Did not attend religious services >1 per month | 314 (56) | 67 (50) | 247 (58) | |
| Visited friends/relatives >1 per month | 449 (80) | 109 (81) | 340 (80) | |
| Did not visit friends/relatives >1 per month | 110 (20) | 25 (19) | 85 (20) | |
| Participated in group or community activities >1 per month | 243 (43) | 55 (41) | 188 (44) | |
| Did not participate in any group or community activities >1 per month |
316 (57) | 79 (59) | 237 (56) | |
| Built Environment | ADI, mean (SD) | 16.1 (14.9) | 18.4 (18.8) | 15.4 (13.3) |
| ADI >44 (cutoff used for disadvantage in SAGES), n (%) | 26 (5) | 12 (9) | 14 (14) | |
| Walkability Index, mean (SD) | 11.3 (4.1) | 11.4 (4.1) | 11.2 (4.2) | |
| Walkability Index <10.5, n (%) | 238 (42) | 59 (44) | 179 (42)- | |
| Lifecourse Factors | Language, n (%)|| | |||
| English as primary language | 527 (94) | 124 (93) | 403 (95) | |
| English as a second language | 32 (6) | 9 (7) | 23 (5) | |
| Country of origin, n (%), N=528¶ | ||||
| US | 513 (97) | 124 (93) | 389 (91) | |
| Non-US | 15 (3) | 6 (4) | 9 (2) | |
| Maternal Education, n (%), N=550# | 11.2 (3.6) | 11.0 (3.8) | 11.2 (3.6) | |
| ≥ 12 years | 347 (63) | 80 (61) | 267 (64) | |
| < 12 years | 203 (37) | 50 (38) | 153 (36) | |
| Paternal Education, n (%), N=550# | 11.5 (4.2) | 10.6 (4.3) | 11.8 (4.1) | |
| ≥ 12 years | 330 (60) | 70 (54) | 260 (62) | |
| < 12 years | 220 (40) | 60 (46) | 160 (38) |
ADI= Area Deprivation Index; SF-12= 12-Item Short Form Survey.
Missing values for household income [n (%)] = 62 (11%).
Missing values for the Geriatric Depression Scale and vision= 2 (<1%) each.
Missing values for hearing, participation in religious services and social activities, visiting friends= 1 (<1%) each.
Missing values for body mass index= 6 (1%).
Missing values for language= 1 (<1%).
Missing values for country of origin= 32 (6%).
Missing values for parental education= 10 (2%).
GDS Cutoff >6 has specificity 60% and sensitivity 66%.
Sensitivity Analyses
Given that type of surgery has been shown to affect incidence of delirium in postoperative setting, the bivariable association between surgery type and delirium was also examined using logistic regression model. Delirium incidence among patients who underwent elective orthopedic surgery [n (%)] 454 (81%) was compared to delirium incidence among patients who underwent gastrointestinal 71 (13%), or vascular 35 (6%) surgeries.
Following the DAG, we used mediation models to quantify the direct effects of neighborhood deprivation, as measured by the ADI, and insurance type, as measured by the presence or absence of private insurance, on the association between household income and delirium using linear regression (Figure 2). The indirect effects were estimated using the product-of-coefficients method.35 Statistical analyses were performed using R version 3.6.1. (R Foundation for Statistical Computing).36 An alpha level of 0.05 was used to indicate statistical significance.
Figure 2.
Path Diagram for the Mediation Models
The figure illustrates the associations between household income and incident delirium mediated by neighborhood disadvantage (red) (i.e., Area Deprivation Index, ADI) and the association between household income (purple) and delirium incidence (green) mediated by insurance type (i.e., no private insurance). ADI= Area Deprivation Index. Path effects indicated by unadjusted beta coefficients derived from linear regression; *p<.05
RESULTS
Social Determinants of Health in the SAGES Cohort
The cohort included 26% who were ≥ 80 years old and 6% who were ≥ 85 years old at baseline. While the racial diversity of the SAGES cohort was limited to 7% non-White (See Table 1), SAGES is comparable to state-level racial representation in Massachusetts by census data.37 Self-reported ethnoracial categories were: American Indian/Alaska Native (n=0), Asian/Native Hawaiian or Pacific Islander (n=5), Black or African American (n=29), White (n=518), other (n=4), or unknown (n=4). Seven (1%) patients self-identified as Hispanic/Latina(o). In the SAGES cohort study, 4% have less than a high school education, 12% have income at the poverty level, and 7% have no private insurance. Significant depressive symptoms were present in 9%,38 while 7% and >60% of the sample reported decreased quality of life by the SF12 Mental Health and the Physical Health composite scores, respectively.39 Overall, 5% of the cohort reported active smoking, 35% were obese by BMI, and 33% and 1% were assessed to have hearing and visual impairments, respectively. In SAGES, 41% of the participants were unmarried, and 30% lived alone. On the Cognitive Activity Scale, 28 (5%) did not have substantial active participation in cognitively stimulating activities (less than once a month). Regarding social engagement, 20% to 57% did not participate in any social activities at least once a month. English was not the first language for 6%, and 3% were born outside the US. Patients reported that 40% of their fathers and 37% of their mothers completed <12 years of education.
SDOH Variables and Incident Delirium
Incident delirium developed in 134 (24%) of the cohort. Bivariable associations between SDOH variables and incident delirium are shown in Table 2. Baseline household annual income <$20,000 (16% versus 9%; OR: 1.12, 95% CI: 1.00, 1.26) and lack of private insurance (11% versus 5%; OR: 1.19, 95% CI: 1.04, 1.38) were significantly associated with a higher risk of delirium. Additionally, older age (OR: 1.01, 95% CI: 1.00, 1.02), higher GDS scores (OR: 1.02, 95% CI: 1.01, 1.04), and higher ADI (OR: 1.00, 95% CI: 1.00, 1.01) were significantly associated with increased risk of delirium.
Table 2.
Association of SDOH Variables with Delirium in the SAGES Cohort (N=560): Bivariable Analyses*
| SDOH Framework Domain | Vulnerability Factors | OR | 95% CI | p value |
|---|---|---|---|---|
| Demographic | Age (years), continuous | 1.01 | 1.00, 1.02 | 0.04 |
| Non-White Race | 1.08 | 0.94, 1.23 | 0.27 | |
| Female Sex | 1.02 | 0.95, 1.10 | 0.55 | |
| Socioeconomic | Education <12 years | 1.06 | 0.89, 1.26 | 0.54 |
| Lower Wechsler Test of Adult Reading (WTAR) score, continuous† | 1.05 | 1.01, 1.08 | 0.01 | |
| Household Income <$20,000 a year | 1.12 | 1.00, 1.26 | 0.04 | |
| No private insurance | 1.19 | 1.04, 1.38 | 0.01 | |
| Well-being | Geriatric Depression Scale (GDS), continuous | 1.02 | 1.01, 1.04 | <0.01 |
| Lower SF12 Mental Health Composite T-score, continuous | 1.01 | 1.00, 1.02 | 0.14 | |
| Lower SF12 Physical Health Composite T-score, continuous | 1.01 | 1.00, 1.02 | 0.07 | |
| Active smoking | 0.99 | 0.84, 1.17 | 0.92 | |
| Body Mass Index, continuous | 1.01 | 0.99, 1.02 | 0.08 | |
| Hearing Impairment | 1.04 | 0.97, 1.13 | 0.26 | |
| Vision Impairment | 1.49 | 0.90, 2.45 | 0.12 | |
| Social Capital | Non-married | 0.99 | 0.93, 1.07 | 0.93 |
| Lives alone | 0.99 | 0.92, 1.06 | 0.84 | |
| Activities and Social Engagement | Lower Cognitive Activities Scale score at enrollment, continuous | 1.01 | 1.00, 1.01 | 0.24 |
| Attend religious services <1 per month‡ | 1.06 | 0.99, 1.14 | 0.10 | |
| Visited friends/relatives <1 per month‡ | 1.02 | 0.93, 1.11 | 0.74 | |
| Participate in any group or community activity <1 per month‡ | 0.98 | 0.91, 1.05 | 0.52 | |
| Built Environment | Area Deprivation index, continuous | 1.00 | 1.00, 1.01 | 0.04 |
| Lower Walkability Index, continuous | 1.00 | 1.00, 1.02 | 0.70 | |
| Lifecourse Factors | English as a Second Language | 1.05 | 0.89, 1.22 | 0.57 |
| Born outside of the US | 1.06 | 0.89, 1.25 | 0.55 | |
| Maternal Education <12 years | 1.02 | 0.95, 1.10 | 0.62 | |
| Paternal Education <12 years | 1.07 | 0.99, 1.15 | 0.08 |
SDOH= Social determinants of health; OR= Odds ratio; CI= Confidence interval; SF12= 12-item Short Form Survey; ADI= Area Deprivation Index
OR presented per 10-unit change in WTAR score.
Physical and social activities questions were selected from the Social Activities Scale of EPESE, see text for details.
Sensitivity Analyses
Incident postoperative delirium did not differ by surgery type. Relative to patients who underwent orthopedic surgeries, patients who underwent vascular (OR: 1.4, 95% CI: 0.8, 2.3) or gastrointestinal (OR: 1.1, 95% CI: 0.7, 17) surgeries did not exhibit significantly higher rates of delirium.
Examining the Association between Household Income, Other SDOH, and Delirium
We further examined the association between household income and delirium. Logistic regression models (Table 3) showed that household income of <$20,000 a year predicted delirium (OR: 1.12, 95% CI: 1.00, 1.26, R2=.20). Model prediction improved when age (OR: 1.01, 95% CI: 1.00, 1.02; R2=.21), WTAR (OR: 0.96, 95% CI: 0.93, 0.99; R2=.22), lack of private insurance (OR: 1.16, 95% CI: 1.00, 1.34; R2=.22), GDS scores (OR: 1.02, 95% CI: 1.01, 1.03; R2=.22), ADI (OR: 1.00, 95% CI: 1.00, 1.01; R2=.22), and paternal education (OR: 1.06, 95% CI: 0.99, 1.15; R2=.22) were added individually, but not with the addition of sex or education.
Table 3.
Association of SDOH Factors with Incident Delirium in SAGES (N= 560): Series and Full Multivariable Models*
| SDOH Framework Predictors | Series Models OR (95% CI) |
Full Multivariable Model | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Household Income |
1.12
(1.00, 1.26) |
1.11 (0.99, 1.25) |
1.12 (0.99, 1.25) |
1.12
(1.00, 1.26) |
1.09 (0.98, 1.23) |
1.09 (0.97, 1.23) |
1.09 (0.98, 1.23) |
1.10 (0.99, 1.24) |
1.12 (0.99, 1.25) |
1.02 (0.90, 1.15) |
| Age | -- |
1.01
(1.00, 1.02) |
-- | -- | -- | -- | -- | -- | -- |
1.01
(1.00, 1.02) |
| Sex | -- | -- | 1.02 (0.95, 1.10) |
-- | -- | -- | -- | -- | -- | 1.02 (0.94, 1.09) |
| Education | -- | -- | -- | 1.06 (0.89, 1.26) |
-- | -- | -- | -- | -- | 0.96 (0.79, 1.15) |
| WTAR † | -- | -- | -- | -- |
0.96
(0.93, 0.99) |
-- | -- | -- | -- | 0.97 (0.93, 1.01) |
| No private insurance | -- | -- | -- | -- | -- |
1.16
(1.00, 1.34) |
-- | -- | -- |
1.18
(1.02, 1.36) |
| GDS | -- | -- | -- | -- | -- | -- |
1.02
(1.01,1.03) |
-- | -- |
1.02
(1.00,1.03) |
| ADI | -- | -- | -- | -- | -- | -- | -- | 1.00 (1.00,1.01) |
-- | 1.00 (0.99, 1.00) |
| Paternal Education | -- | -- | -- | -- | -- | -- | -- | -- | 1.06 (0.99, 1.15) |
1.04 (0.97, 1.12) |
| Nagelkerke R2 | 0.20 | 0.21 | 0.20 | 0.20 | 0.22 | 0.22 | 0.22 | 0.21 | 0.22 | 0.27 |
OR- Odds ratios; CI- Confidence intervals; WTAR- Wechsler Test of Adult Reading; GDS- Geriatric depression scale; ADI- Area Deprivation Index. Results from a full multivariable logistic regression model.
OR presented per 10-unit change in WTAR score.
In the full multivariable logistic regression model, household income <$20,000 was no longer significant (OR: 1.02, 95% CI: 0.90, 1.15). However, older age (OR: 1.01, 95% CI: 1.00, 1.02), lack of private insurance (OR: 1.18, 95% CI: 1.02, 1.36), and higher GDS scores (OR: 1.02, 95% CI: 1.00, 1.03) remained statistically significant independent predictors of incident delirium.
Figure 2 illustrates the direct effects of ADI and insurance type on delirium. Overall, ADI and lack of private insurance were independent predictors of delirium. We found that 19% of the association between household income and delirium was significantly mediated by the type of insurance (indirect effect estimate= 0.03; 95% CI= 0.01, 0.06). On the other hand, 12% of the association between household income and delirium was mediated by ADI (indirect effect estimate= 0.02; 95% CI= −0.01, 0.04); however, this mediation effect did not achieve statistical significance.
DISCUSSION
Our results characterize SDOH factors in SAGES, a retrospective study of a prospective cohort of patients 70+ undergoing elective surgery. SDOH vulnerabilities were identified across all domains: demographic (26% ≥ age 80); socioeconomic (12% <$20,000 household income; 7% lack of private insurance); well-being (depressive symptomatology in 2%, SF-12 Physical Health Composite in >60%, hearing impairment in 33%); social engagement (lack of social activities in up to 57%); social capital (41% unmarried); built environment (high ADI in 5%); and lifecourse factors (40% of fathers with < high school education). In bivariable models, older age, low household income, lack of private insurance, depressive symptomatology, and ADI were significant predictors of delirium. In a multivariable analysis, older age, lack of private insurance, and depressive symptomatology were independently associated with delirium. While older age and depression are well-described risk factors for delirium, our results also suggest that type of insurance is both an independent predictor for delirium, as well as a significant mediator of the association between household income and incident delirium. For example, individuals with private insurance are more likely to have consistent care and receive preventive treatment such as vaccinations and annual wellness visits,40–44 fall into a higher socioeconomic status, and/or achieve higher education with a healthier lifestyle,43,45,46 suggesting numerous potential pathways. Household income serves as a proxy for financial stability.47,48 Having a household income below the poverty threshold is an important social determinant of health associated with poor outcomes.45,48 In addition, based on our mediation analyses, lack of private insurance may have a more proximate effect than ADI. This study provides unique and novel insights into the potential pathways by which SDOH variables may influence delirium.
The association between built environment (e.g., ADI) and delirium severity has been previously reported in this cohort.12 Bivariable models revealed a statistically significant association between ADI and delirium, and ADI mediated part of the association between household income and incident delirium. However, ADI did not remain an independent predictor of delirium once other SDOH factors were included in our multivariable model. The absence of statistical significance does not indicate the absence of an association. It is possible that the representation of ADI in our sample was too restricted or that power was limited. It is important to note that the average ADI score in Boston is high relative to most other US regions.
Our hypothesis-driven approach, which followed a DAG17 developed a priori, is a strength of this study. Another strength is the consideration of SDOH factors occurring both proximal and distant from our outcome, incident delirium. This is perhaps the most comprehensive examination of SDOH factors and delirium to date. Other strengths include our well-characterized cohort, the use of well-validated measures of delirium, and the high quality and completeness of the data collection.
Several caveats are important to mention. The cohort is predominantly well-educated, largely White, and from a discrete geographical area. Thus, generalizability to other samples requires further evaluation. The availability of private insurance is unique to the US. As such, these findings may not be generalizable to other parts of the world where basic insurance is provided to all citizens equitably. Despite limited racial diversity, our sample demonstrated diversity in many SDOH factors. Another limitation is that household income was missing in 11%. Studies show that under-reporting of SDOH factors is a frequent problem in epidemiologic studies and that this pattern is correlated with patients’ values.49 Efforts to collect SDOH data using culturally sensitive approaches are instrumental to advancing this work. Moreover, researchers will need to accommodate missing data in estimating these constructs.49 In this study, missingness was addressed through accepted multiple imputation approaches. Another limitation of the study relates to the operationalization of household income. In this study, patients were asked to select the income bracket that best captured their wealth, with <$20,000 a year being the lowest category. It is possible that within-group variability existed among patients selecting this category and that older adults experiencing severe financial constraints in our sample were not accurately represented in our results. Of note, patients in this bracket were low earners relative to peers enrolled in the SAGES study and to other earners in the state of Massachusetts (2010 Median income: $66,300; 2011 Median income: $65,100, 2012 Median income: $66,300, and 2013 Median income: $66,800).37 As such, designating them as “low-income” was appropriate relative to State and Federal poverty thresholds. Importantly, our cut-off of household income of <$20,000 per year appears to be robust, representing $3,000 to $8,000 below weighted average poverty thresholds for 65 years and older householders for 2010, 2011, 2012, and 2013.21 We examined the association between different surgeries and incident delirium. Overall, we found no significant difference in incident delirium between patients who underwent orthopedic, vascular, or gastrointestinal surgeries. Our sample was disproportionately comprised of older adults undergoing orthopedic surgery. It is possible that our findings reflect a lack of power to detect differences across surgical groups and not a lack of association. The use of psychoactive medications (e.g., narcotics, anticholinergics) has been associated with increased delirium risk50 but surgery type is a poor surrogate for this variable. Future studies should account for the contribution that perioperative and postoperative medication use has on delirium incidence.
Lack of private insurance, defined as having federal or state-sponsored insurance (Medicare alone or Medicare with Medicaid/MassHealth), was the most highly associated variable among those studied. Lacking a Medicare supplement plan may not be equivalent to lacking an employer sponsored insurance in terms of access to health care. Thus, ““lack of private insurance” is a crude proxy of access to care, given the nuances unique to health insurance and access to care in the US. A comprehensive evaluation of the interaction between access to care with delirium will be required in future work. Further, our study could not comprehensively assess the association between delirium and different types of public insurance, given the limited sample size in each category and the nature of the variables collected (e.g., patient-reported type of insurance without details about coverage). Relatedly, we were unable to examine differences according to the varying types of federal and state-sponsored insurance which may have identified distinct groups of patients (e.g., with pre-existing medical conditions, advanced illness, or institutionalization). We hope future studies will expand upon our findings by assessing these factors comprehensively in diverse populations.
CONCLUSIONS
In older adults scheduled for elective surgery, SDOH factors are prevalent, and independent risk factors for delirium include older age, lack of private insurance, and depressive symptoms. While age and depression are well-described risk factors, the lack of private insurance is a novel and robust risk factor that mediated the relationship between household income and delirium. The magnitude of the associations identified in these analyses are small, consistent with other studies measuring these factors.5,6 Nevertheless, this is one of the first efforts to comprehensively explore acute changes in cognition in context. Most importantly, our goal is to promote a paradigm shift, so that systematic and comprehensive assessment of SDOH factors in research allows for more nuanced examination of their contributions on health. We hope this work may provide a foundation for future work to address SDOH to improve clinical outcomes, such as delirium, in older adults. Targeting SDOH factors such as access to care and economic marginalization have been shown to improve health outcomes.51–55 Recognizing the contributions of SDOH factors to health may propel the development of innovative interventions designed to support, educate, and leverage existing resources to promote health.
Key Points
Lack of private insurance, older age, and depressive symptoms were significant independent predictors of incident postoperative delirium.
Combined with other SDOH factors, lack of private insurance, older age, and depressive symptoms explained 27% of the variance in incident postoperative delirium.
Why does this matter?
Lack of private insurance emerged as a robust and novel predictor of incident postoperative delirium.
Medical insurance is an important SDOH factor that may be a proxy for access to healthcare services.
Addressing SDOH factors may be important in perioperative settings to prevent incident delirium.
ACKNOWLEDGMENTS
Funding:
This manuscript was funded in part by the National Institute on Aging grants no. R24AG054259 (SKI), P01AG031720 (SKI). Dr. Franchesca Arias’ time was supported in part by an NIA Diversity Supplement to grant P01AG031720 (SKI) and by grant no. 2019-AARFD-644816 from the Alzheimer’s Association. Dr. Alegria’s time is funded in part by R01AG046149. Dr. Inouye holds the Milton and Shirley F. Levy Family Chair at Hebrew SeniorLife/Harvard Medical School.
Sponsor’s Role:
The funding sources was not involved in the design, analysis, or reporting of the results.
Dedication:
This work is dedicated to the memory of Joshua Bryan Inouye Helfand. The authors gratefully acknowledge the contributions of the patients, family members, nurses, physicians, research team, and Executive Committee members who participated in the Successful Aging after Elective Surgery (SAGES) study.
Footnotes
Conflict of Interest: The authors have no competing interests or conflicts to declare.
REFERENCES
- 1.Marmot M, Friel S, Bell R, Houweling TA, Taylor S, Commission on Social Determinants of Health. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet 2008.: 10.1016/S0140-6736(08)61690-6 [DOI] [PubMed] [Google Scholar]
- 2.Parikh RB, Jain SH, Navathe AS. The sociobehavioral phenotype: applying a precision medicine framework to social determinants of health. Am J Manag Care. 2019;25(9):421–3. [PubMed] [Google Scholar]
- 3.Swan JE, Aldridge A, Joseph V, Tucker JA, Witkiewitz K. Individual and community social determinants of health and recovery from alcohol use disorder three years following treatment. J Psychoactive Drugs. 2021;12:1–0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Shaibi GQ, Kullo IJ, Singh DP, et al. Returning genomic results in a Federally Qualified Health Center: the intersection of precision medicine and social determinants of health. Genet Med. 2020;22(9):1552–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Glymour MM, Manly JJ. Lifecourse social conditions and racial and ethnic patterns of cognitive aging. Neuropsychol Rev. 2008;18(3):223–54. [DOI] [PubMed] [Google Scholar]
- 7.Held ML, Jones A, Forrest-Bank S. Predictors of Latinx youth health and emotional well-being: Social determinants of health perspective. Journal Racial Ethn. 2020;7(6):1188–201. [DOI] [PubMed] [Google Scholar]
- 8.American Psychiatric Association. Diagnostic and statistical manual of mental disorders (DSM-5®) 2013. Washington, DC: American Psychiatric Publishing. [Google Scholar]
- 9.Tieges Z, Quinn T, MacKenzie L, et al. Association between components of the delirium syndrome and outcomes in hospitalised adults: a systematic review and meta-analysis. BMC Geriatrics. 2021;21(1):1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Inouye SK, Viscoli CM, Horwitz RI, Hurst LD, Tinetti ME. A predictive model for delirium in hospitalized elderly medical patients based on admission characteristics. Ann Intern Med. 1993;15;119(6):474–81. [DOI] [PubMed] [Google Scholar]
- 11.Arias F, Alegria M, Kind AJ, et al. A framework of social determinants of health for delirium tailored to older adults. JAGS. 2022;70(1):235–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Arias F, Chen F, Fong TG et al. Neighborhood-Level Social Disadvantage and Risk of Delirium Following Major Surgery. JAGS. 2020;68(12):2863–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Schmitt EM, Saczynski JS, Kosar CM, et al. The successful aging after elective surgery study: cohort description and data quality procedures. JAGS. 2015;63(12):2463–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Arias F, Chen F, Shiff H, et al. Parental Education and Delirium Risk after Surgery in Older Adults. Clin Gerontol. 2022; 25: 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Shiff HM, Arias F, Dufour AB, Carr D, Chen F, Gou Y, … & Inouye SK (2022). Paternal Occupation and Delirium Risk in Older Adults: A Potential Marker of Early-Life Exposures. Innov Aging. 2022; 6(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cizginer S, Marcantonio ER, Vasunilashorn S, et al. The cognitive reserve model in the development of delirium: the Successful Aging after Elective Surgery Study (SAGES). J Geriatr Pyschiatry Neurol. 2017; 30:337–345 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lipsky AM, Greenland S. Causal directed acyclic graphs. JAMA. 2022;327(11):1083–4. [DOI] [PubMed] [Google Scholar]
- 18.Katz S. Assessing self-maintenance: activities of daily living, mobility, and instrumental activities of daily living. JAGS.1983. [DOI] [PubMed] [Google Scholar]
- 19.Fong TG, Hshieh TT, Wong B… et al. Neuropsychological profiles of an elderly cohort undergoing elective surgery and the relationship between cognitive performance and delirium. JAGS. 2015;63(5):977–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Green RE, Melo B, Christensen B, Ngo LA, Monette G, Bradbury C. Measuring premorbid IQ in traumatic brain injury: an examination of the validity of the Wechsler Test of Adult Reading (WTAR). J Clin Exp Neuropsychol. 2008;30(2):163–72. [DOI] [PubMed] [Google Scholar]
- 21.Census Bureau US. “ Poverty Thresholds by Size of Family and Number of Children.” Current Population Survey Annual Social and Economic Supplement (CPS ASEC). https://www.census.gov/data/tables/time-series/demo/income-poverty/historical-poverty-thresholds.html
- 22.Brown DA, Himes BT, Kerezoudis P, et al. Insurance correlates with improved access to care and outcome among glioblastoma patients. Neuro-oncology. 2018;20(10):1374–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yesavage JA, Sheikh JI. Geriatric depression scale (GDS): recent evidence and development of a shorter version. J Clin Gerontol. 1986;5(1–2):165–73. [Google Scholar]
- 24.Ware JE Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Medical care. 1996;1:220–33. [DOI] [PubMed] [Google Scholar]
- 25.Macphee GJ, Crowther JA, McAlpine CH. A simple screening test for hearing impairment in elderly patients. Age Ageing. 1988;17:347–351. [DOI] [PubMed] [Google Scholar]
- 26.Collins SD, Britten RH. Variation in eyesight at different ages, as determined by the Snellen test. Public Health Reports. 1924;39(5). [Google Scholar]
- 27.Wilson R, Barnes L, & Bennett D. Assessment of lifetime participation in cognitively stimulating activities. J Clin Exp Neuropsychol. 2003;25(5), 634–642. doi: 10.1076/jcen.25.5.634.14572. [DOI] [PubMed] [Google Scholar]
- 28.Mendes de Leon CF, Glass TA, Berkman LF. Social engagement and disability in a community population of older adults: the New Haven EPESE. Am J Epidemiol. 2003;157(7):633–42. [DOI] [PubMed] [Google Scholar]
- 29.Kind AJ, Jencks S, Brock J, et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalizations: an analysis of Medicare data. Ann Intern Med. 2014;161(11):765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Cerin E, Saelens BE, Sallis JF, Frank LD. Neighborhood Environment Walkability Scale: validity and development of a short form. MSSE. 2006;38(9):1682. [DOI] [PubMed] [Google Scholar]
- 31.Inouye SK, van Dyck CH, Alessi CA, Balkin S, Siegal AP, Horwitz RI. Clarifying confusion: the confusion assessment method: a new method for detection of delirium. Ann Intern Med. 1990;113(12):941–8. [DOI] [PubMed] [Google Scholar]
- 32.Saczynski JS, Kosar CM, Xu G, et al. A tale of two methods: chart and interview methods for identifying delirium. JAGS. 2014;62(3):518–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Glymour MM, Spiegelman D. Evaluating public health interventions: 5. Causal inference in public health research—do sex, race, and biological factors cause health outcomes? Am J Public Health. 2017;107(1):81–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? IJMPR. 2011;20(1):40–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Rijnhart JJ, Lamp SJ, Valente MJ, MacKinnon DP, Twisk JW, Heymans MW. Mediation analysis methods used in observational research: a scoping review and recommendations. BMC Med Res Methodol. 2021;21(1):1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Thomas RJ. Data analysis with R statistical software: a guidebook for scientists. Eco-explore; 2015 [Google Scholar]
- 37.Dugan E, Porell F, Silverstein N, Palombo R, Mann S. Massachusetts Healthy Aging Data Report: Community Profiles. Pubished in 2014.
- 38.Burke WJ, Roccaforte WH, Wengel SP. The short form of the Geriatric Depression Scale: a comparison with the 30-item form. Topics in Geriatrics. 1991;4(3),173–178. [DOI] [PubMed] [Google Scholar]
- 39.Gill SC, Butterworth P, Rodgers B, Mackinnon A. Validity of the mental health component scale of the 12-item Short-Form Health Survey (MCS-12) as measure of common mental disorders in the general population. Psychiat Research. 2007;152(1):63–71. [DOI] [PubMed] [Google Scholar]
- 40.McWilliams JM, Zaslavsky AM, Meara E, Ayanian JZ. Impact of Medicare coverage on basic clinical services for previously uninsured adults. JAMA. 2003;290(6):757–64. [DOI] [PubMed] [Google Scholar]
- 41.Wilper AP, Woolhandler S, Lasser KE, McCormick D, Bor DH, Himmelstein DU. Health insurance and mortality in US adults. Am J Public Health. 2009;99(12):2289–95 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Brown DA, Himes BT, Kerezoudis P, et al. Insurance correlates with improved access to care and outcome among glioblastoma patients. Neuro-oncology. 2018;20(10):1374–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Ochieng N, Biniek JF. Beneficiary Experience, Affordability, Utilization, and Quality in Medicare Advantage and Traditional Medicare: A Review of the Literature. DeNavas-Walt C, Proctor BD. Income and poverty in the United States: 2013. US Government Printing Office. Accessed on May 15, 2022. [Google Scholar]
- 44.Sherman BW, Sils B, Kamin L, Westrich K. Specialty drug and health care utilization vary by wage level in employer-sponsored health plans. Journal of Managed Care & Specialty Pharmacy. 2022. Aug;28(8):918–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Shankar A, McMunn A, Steptoe A. Health-related behaviors in older adults: relationships with socioeconomic status. Am J Prev Med. 2010;38(1):39–46. [DOI] [PubMed] [Google Scholar]
- 46.Deckers K, Cadar D, van Boxtel MP, Verhey FR, Steptoe A, Köhler S. Modifiable risk factors explain socioeconomic inequalities in dementia risk: evidence from a population-based prospective cohort study. J Alzheimer’s Dis. 2019;71(2):549–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Skolarus LE, Sánchez BN, Morgenstern LB, et al. Validity of proxies and correction for proxy use when evaluating social determinants of health in stroke patients. Stroke. 2010;41(3):510–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Frier A, Devine S, Barnett F, Dunning T. Utilising clinical settings to identify and respond to the social determinants of health of individuals with type 2 diabetes—a review of the literature. Health Soc Care Community. 2020;28(4):1119–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Westreich D, Greenland S. The table 2 fallacy: presenting and interpreting confounder and modifier coefficients. Am J Epidemiol. 2013;177(4):292–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Litaker D, Locala J, Franco K, Bronson DL, Tannous Z. Preoperative risk factors for postoperative delirium. General hospital psychiatry. 2001. Mar 1;23(2):84–9. [DOI] [PubMed] [Google Scholar]
- 51.Walter LA, Schoenfeld EM, Smith CH, Shufflebarger E, Khoury C, Baldwin K, Hess J, Heimann M, Crosby C, Sontheimer SY, Gragg S. Emergency department–based interventions affecting social determinants of health in the United States: a scoping review. Academic Emergency Medicine. 2021. Jun;28(6):666–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Yan AF, Chen Z, Wang Y, Campbell JA, Xue QL, Williams MY, Weinhardt LS, Egede LE. Effectiveness of social needs screening and interventions in clinical settings on utilization, cost, and clinical outcomes: a systematic review. Health Equity. 2022. Jun 1;6(1):454–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Mansfield LN, Carson SL, Sunku N, Troutt A, Jackson S, Santillan D, Vassar SD, Slaughter D, Kim G, Norris KC, Brown AF. Community-based organization perspectives on participating in state-wide community canvassing program aimed to reduce COVID-19 vaccine disparities in California. BMC Public Health. 2023. Dec;23(1):1–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Mishra S, Ma H, Moloney G, Yiu KC, Darvin D, Landsman D, Kwong JC, Calzavara A, Straus S, Chan AK, Gournis E. Increasing concentration of COVID-19 by socioeconomic determinants and geography in Toronto, Canada: an observational study. Annals of epidemiology. 2022. Jan 1;65:84–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Schrader CD, Robinson RD, Blair S, Shaikh S, Ho AF, D’Etienne JP, Kirby JJ, Cheeti R, Zenarosa NR, Wang H. Common step-wise interventions improved primary care clinic visits and reduced emergency department discharge failures: a large-scale retrospective observational study. BMC Health Services Research. 2019. Dec;19(1):1–1. [DOI] [PMC free article] [PubMed] [Google Scholar]