We utilized the NHIS 2023, an annual cross-sectional nationally representative household survey of the U.S. civilian noninstitutionalized population administered by the National Center for Health Statistics (NCHS) in collaboration with the Centers for Disease Control and Prevention (CDC).⁹ The NHIS captures self-reported demographic, health, household, and socioeconomic data from a representative sample of U.S. residents. For 2023, the survey included 29,522 adults and 7,692 children from 30,670 households across all 50 states and the District of Columbia, of whom 66% were NHW individuals; by incorporating complex, multistage sampling, it can provide estimates on noninstitutionalized, U.S. population. As the data used in this study were publicly available and de-identified, our research was exempt from Institutional Review Board review.

We identified adult individuals (aged 18 and above) with IBD in the NHIS 2023 Sample Adult Interview dataset, if they responded affirmatively to either of the questions: “Have you ever been told by a doctor or other healthcare professional that you had Ulcerative Colitis?” or “Have you ever been told by a doctor or other healthcare professional that you had Crohn’s Disease?” To augment self-reported diagnosis, we conducted sensitivity analyses limiting to a subset of patients who besides self-reporting a diagnosis of ulcerative colitis (UC) or Crohn’s disease (CD), also answered ‘yes’ to “Have you ever had a colonoscopy or sigmoidoscopy?” (88% of those who self-reported a diagnosis of IBD).

We examined social risk across six domains: food insecurity, financial hardship, housing instability, transportation needs, employment, and discrimination. Since there is no standard, validated and universally accepted questionnaire for assessing social risk, we classified social risk across each domain based on functional definitions from questionnaires in NHIS, as detailed in Table 1.^{6, 10, 11} For example, adults were classified as being food insecure based on the U.S. Department of Agriculture Economic Research Service’s 10-item questionnaire, which focuses on being worried about food running out, inability to afford a balanced meal, cutting portions or skipping meals due to costs, etc. or based on whether they used food stamps in the last 12 months.¹² Financial hardship was ascertained based on difficulty paying medical bills or foregoing medical care due to costs. Housing instability was based on whether individuals reported difficulty paying for housing or if they lived in government-subsidized low-income housing.⁸ We were unable to cover social risk factor domains related to social support, household utilities and interpersonal violence due to lack of relevant questions.

In the absence of clear guidance on quantifying social risk beyond the need to measure it across multiple domains, we defined individual social risk burden was defined based on the number of domains (out of six) for which the respondent was classified as adverse similar to prior studies: 0/6 domains, no risk; 1/6 domains, mild risk; 2/6 domains, moderate risk; and ≥ 3/6 domains, severe risk.¹³

Our primary outcome was the proportion of adults experiencing unplanned healthcare utilization defined as emergency department (ED) use and/or hospitalization in the preceding 12 months. Our secondary outcome was prevalence of cost-related medication non-adherence. Cost-related medication non-adherence was defined based on affirmative to any of the following four survey questions: (1) “Did you need but could not afford prescription medicines over the past 12 months?”, (2) “Did you take less medication to save money over the past 12 months?”, (3) “Did you skip medication doses to save money over the past 12 months?”, (4) “Did you delay refilling medications to save money over the past 12 months?”⁸

We collected data on the following covariates: age, sex, race/ethnicity, body mass index (BMI), tobacco use, marital status, education, employment history in the past 12 months, family size, family income (calculated based on ratio of family income to federal poverty limit from the Census Bureau: high-income [≥4.00], middle-income [2.00–3.99], low-income [1.00–1.99]) and below poverty line [<1.00]), insurance status (private insurance or Medicare vs. Medicaid/Children’s Health Insurance Program/uninsured), and self-reported medical comorbidities (e.g., diabetes, hypertension, coronary artery disease, stroke, heart attack, chronic obstructive pulmonary disease [COPD], arthritis, and positive screens for depression and/or anxiety).

We examined the prevalence and burden of social risk across multiple domains, overall and by racial and ethnic minority groups. We used the chi-square test to compare categorical variables across racial and ethnic minority groups and Kruskal-Wallis test to compare continuous variables. We calculated the weighted mean of each variable and the corresponding 95% confidence interval (CI). We accounted for the complex survey design of the NHIS by using survey-specific descriptive statistics to provide weighted national estimates. Subsequently, we examined the association between burden of social risk and outcomes in survey-weighted logistic regression models, both unadjusted and adjusted for covariates including age, sex, BMI, smoking status, major comorbidities including diabetes, hypertension, hyperlipidemia, coronary artery disease, COPD/emphysema/bronchitis, arthritis, and positive screening for depression and/or anxiety. To examine whether race/ethnicity, income and insurance type modify this association between social risk factors and outcomes, we conducted two additional analyses: (1) additionally adjusted for race/ethnicity (Hispanics vs. non-Hispanic Blacks vs. NHWs), and (2) additionally adjusted for family income and insurance status. We also examined univariate association between race/ethnicity, family income and insurance status with outcomes. Of note, since NHIS is a cross-sectional survey, we were unable to ascertain a causal association between the exposure and outcome. As noted earlier, sensitivity analyses were performed by restricting the sample to patients with IBD with a history of colonoscopy or sigmoidoscopy. All statistical analyses were implemented in R version 4.2.2. (R Project for Statistical Computing), with statistical significance set at the 5% level.

We identified 572 adults who self-reported a diagnosis of UC or CD in NHIS 2023. Using survey weights, this corresponded to approximately 4.1 million (95% CI, 3.6–4.5 million) adults living with IBD in the United States, of whom 2.9 million report UC, and 1.2 million report CD (Table 2). Using an alternative definition of IBD based on self-reported UC or CD and receipt of colonoscopy or sigmoidoscopy, the estimated prevalence of IBD is 2.75 million.

Mean age of patients with IBD was 56 years (95% CI, 54–58y), of whom 57% were female and 24% reported use of immunosuppressive therapies. Approximately 81% identified as NHW (estimated 3.3 million [95% CI, 2.9–3.6 million]), 12.9% as Hispanic (estimated 0.53 million [95% CI, 0.34–0.71 million]), and 6.6% as non-Hispanic Blacks (estimated 0.27 million [95% CI, 0.17–0.37 million]). Hispanic and Non-Hispanic Black individuals reported the highest prevalence of obesity (46% and 44%, respectively), and non-Hispanic Blacks individuals reported the highest prevalence of hypertension (68%) and diabetes (46%). Approximately 35% (95% CI, 30–40%) adults with IBD reported emergency department visit, and 17% (95% CI, 13–21%) reported hospitalization within the preceding 12 months; 14% (95% CI, 11–18%) experienced cost-related medication non-adherence.

The burden of social risk in patients with IBD was substantial (Table 3). Overall, 22% (95% CI, 18–27%) reported food insecurity, including 15% who reported using food stamps in the last 12 months; 28% (95% CI, 23–32%) reported financial hardship, including 10% who delayed medical care due to costs and 9% who did not get medical care due to costs. Approximately 14% (95% CI, 10–17%) patients reported housing instability, and 9% (95% CI, 6–12%) reported delaying care due to transportation barrier. Overall, 8.7% reported having less than high school education and were either unemployed or, if employed, were in a job which did not provide paid sick leave. Approximately 41% (95% CI, 36–46%) reported discrimination with frequent encounters where they were treated with less respect (30%), received poor service (11%), were treated as not smart (21%) or felt threatened or harassed (9%). Overall, 15% (95% CI, 11–19%) adults experienced severe burden of social risk (across 3 or more domains), 17% (95% CI, 13–21%) experienced moderate burden, 32% (95% CI, 27–37%) experienced mild burden and 36% (95% CI, 31–41%) experienced no burden of social risk (Table 3). Findings were similar on sensitivity analysis focusing only patients with self-reported diagnosis of UC or CD and previous receipt of colonoscopy and/or sigmoidoscopy (eTable 1).

There were significant racial and ethnic disparities in the burden of social risk (Figure 1, Table 3). Overall, non-Hispanic Black adults with IBD and Hispanic adults with IBD had the highest burden of social risk, with 58% and 52% reporting moderate to severe burden of social risk, respectively, compared with 28% of NHW adults. While there were significant differences in prevalence of social risk across the domains of food insecurity, financial hardship, housing instability and education and employment, no differences were observed in the prevalence of transportation needs and discrimination. Among racial and ethnic minority groups, Non-Hispanic Black adults experienced higher burden of food insecurity, housing instability and discrimination, whereas Hispanic adults faced higher burden of financial hardship and adversities in education and employment.

Social risk burden was significantly associated with unplanned healthcare utilization including hospitalizations and ED visits. Compared to patients with IBD with no social risk factors, patients with severe burden of social risk had significantly higher risk of hospitalization and/or ED visits, after adjusting for key confounding variables (Table 4). This trend remained significant after additional adjusting for race and ethnicity, and additionally adjusting for family income and health insurance status. Similar findings were observed for the risk of cost-related medication non-adherence (Table 4). Race and ethnicity were not independently associated with risk of hospitalization and/or ED visits, nor with cost-related medication non-adherence.