Racial and Ethnic Disparities in Health-Related Outcomes in Crohn’s Disease: Results From the National Health and Wellness Survey

Abstract Background Crohn’s disease (CD) is a chronic inflammatory condition affecting the entire gastrointestinal tract that is associated with significant humanistic, clinical, and economic burdens. Few studies have assessed the association between CD severity and patient-reported outcomes (PROs), healthcare resource utilization (HCRU), and medical costs; even fewer have examined differences in disease outcomes among patients of various racial/ethnic groups. Methods In this cross-sectional study, sociodemographic data, PROs, and economic outcomes for participants with self-reported CD were collected from the National Health and Wellness Survey (2018–2020). Multivariable analyses were used to assess the association of CD severity and race/ethnicity with health-related quality of life (HRQoL), work productivity and activity impairment (WPAI), HCRU, and medical costs. Results Analyses included 1077 participants with CD (818 non-Hispanic White, 109 non-Hispanic Black, and 150 Hispanic). Participants with self-reported moderate/severe CD reported significantly worse HRQoL and WPAI, greater HCRU, and higher medical costs than those with self-reported mild CD. Non-Hispanic Black participants reported better HRQoL and fewer healthcare provider visits than non-Hispanic White participants. There were no significant differences in PROs between non-Hispanic White and Hispanic groups. Interactions between race/ethnicity and CD severity emerged for some, but not all groups: Specifically, non-Hispanic Black participants with moderate/severe CD reported greater absenteeism and more gastroenterologist visits than non-Hispanic Black participants with mild CD. Conclusions Participants with moderate/severe CD reported worse PROs, greater HCRU, and higher medical costs than those with mild CD. Additionally, racial/ethnic differences were found across several HCRU and economic outcomes. Further research is needed to better understand factors contributing to burden among patients with varying CD severity across racial/ethnic groups.


Introduction
Crohn's disease (CD), a type of inflammatory bowel disease (IBD), is a chronic inflammatory condition of the gastrointestinal (GI) tract that is characterized by relapsing and remitting symptoms of abdominal pain, diarrhea, bowel obstruction, weight loss, and fatigue. 1 Among patients with multiple relapses, CD can quickly progress from a mild to moderate condition, and eventually cause severe penetrating or stricturing disease. 2 Despite the development of new treatments for CD, including biologics and immunomodulators, up to 75% of patients with CD require surgery throughout their lifespan, with many requiring several surgical procedures. 3he CD has debilitating effects on patients' healthrelated quality of life (HRQoL), which may be attributed to impairments in the ability to participate in social and economic/vocational opportunities and diminished psychological well-being. 1 Most patients with CD develop symptoms before the age of 30, coinciding with their primary reproductive and working years. 1 Feelings of social isolation and stress due to the unpredictable nature of symptoms are common and patients with CD have a higher-than-average incidence of depression and anxiety. 4Symptoms of CD can negatively influence work productivity, leading to higher rates of unemployment and work disability relative to the general population. 5CD is also associated with high rates of healthcare resource utilization (HCRU), including emergency room visits, hospitalizations, medical procedures, and surgeries. 6Although medication and hospitalization are the main drivers of cost for patients with CD, indirect costs due to lost productivity can also be high.A recent study estimated total annualized medical costs for CD patients ranging from $16 711 to $66 027. 6[9][10] From 1970 to 2010, the incidence of IBD increased by 39% among White individuals and by 134% among nonwhite individuals globally. 11In the United States, the prevalence of IBD has historically been higher among White individuals than Black individuals. 12However, recent studies have reported similar incidence rates of IBD among White and Black individuals. 13The increasing incidence of CD among marginalized groups in the United States has brought forth awareness of differences in the disease course and patient journey. 14As such, understanding racial/ethnic differences is crucial for improving health equity and ensuring timely and adequate treatment of disease.
Patient-reported outcomes (PROs) are increasingly used in clinical studies as they emphasize the patient experience of disease burden. 15As IBD is an under-recognized condition in minority patient groups, PROs can allow data collection from a diverse patient population.Particularly, considering self-reported severity is crucial for understanding disease burden, as symptoms identified as most problematic by patients may not be considered the cardinal symptoms assessed in clinical practice. 16Furthermore, the impact of IBD symptoms on patients' lives is often underestimated. 17,18here are currently limited studies on PROs in the context of CD severity and race/ethnicity among patients with CD, and even fewer studies that compare more than 2 race/ethnicity groups simultaneously.Therefore, the objective of this study was to evaluate the relationship of race/ethnicity and CD severity with PROs of depression, anxiety, HRQoL, and work productivity and activity impairment (WPAI), and health economic outcomes such as HCRU, medical costs, and indirect costs among non-Hispanic White, non-Hispanic Black, and Hispanic patients with mild or moderate/severe CD.

Study Design
This retrospective, cross-sectional study leveraged data from the 2018, 2019, and 2020 US National Health and Wellness Survey (NHWS).The NHWS is a self-administered, internet-based survey administered to a nationally representative sample of adults (ages ≥ 18 years; ~N = 75 000 per year). 19Potential participants were recruited through opt-in e-mails, co-registration with panel partners, e-newsletter campaigns, banner placements, and affiliate networks.All panelists agreed to take part as a panel member and completed an in-depth demographic registration profile.A quota sampling technique ensured that the demographic composition (ie, age, gender, and race/ethnicity) of the NHWS sample was representative of the adult population in the United States.The NHWS was reviewed by Pearl Institutional Review Board (Indianapolis, IN) and granted exemption status.All participants provided their informed consent electronically.

Study Population
Participants were included in the analytic sample if they met the following inclusion criteria: ≥18 years old, resident of the United States, self-reported receipt of a physician diagnosis of CD, self-reported Hispanic ethnicity or non-Hispanic White or non-Hispanic Black/African American race and had complete data for outcomes relevant to the study (Figure 1).Participants who reported a concomitant diagnosis of ulcerative colitis or another race/ethnicity were excluded from the study sample.

Sociodemographic and Clinical Characteristics
Self-reported sociodemographic and clinical characteristics collected from the NHWS were as follows: Sex, age, marital status, education level, employment status, household income, health insurance, CD severity, body mass index, alcohol use, cigarette smoking, exercise, current prescription medication use, sleep problems/symptoms, comorbidity index, patient activation measure (PAM) score, 20 and previous clinical trial participation.Self-reported severity of condition was captured as participants' rating of their CD as either mild, moderate, or severe.Participants who used a prescription medication for CD during the study period rated their CD severity while on their medication, whereas participants who did not use prescription medication rated their CD severity in a general sense (ie, not specific to any symptoms).Self-reported race and ethnicity were categorized into 3 mutually exclusive groups: Non-Hispanic White, non-Hispanic Black, or Hispanic.Patient activation (or engagement) was assessed using the PAM, a non-disease-specific measure that assesses an individual's knowledge and confidence in managing their health. 20The questionnaire includes 13 items with a total score of 0 to 100, where higher scores indicate greater Burbage et al 3 levels of engagement.Individuals can also be grouped into 4 levels of activation (ie, level 1: 0-47, level 2: 47.1-55.1,level 3: 55.2-72.4,and level 4: 72.5-100).Previous clinical trial participation data were collected from a non-disease-specific survey question; response options included "yes" or "no."

Study Outcomes
Outcomes of interest included PROs of depression and anxiety, HRQoL, labor force participation, WPAI, and economic outcomes of HCRU and medical costs.Detailed descriptions of the study outcomes and how they were calculated are included in Supplementary Materials.Depression severity was assessed using the patient health questionnaire-9 (PHQ-9) and anxiety severity was assessed using the generalized anxiety disorder assessment (GAD-7).HRQoL was assessed using 2 summary scores of the medical outcomes study 36-item short form survey instrument (SF-36v2): physical component summary (PCS) and mental component summary (MCS). 21The SF-6D and the EuroQol 5-dimension health questionnaire (EQ-5D) instruments were used to assess health state utilities, the latter of which consists of the EQ-5D-5L utility index and EQ visual analog scale (VAS). 22abor force participation was derived from NHWS data through coding employment status as currently in the labor force (ie, full-time employed, part-time employed, selfemployed, or not unemployed but looking for work) or not currently in the labor force (ie, retired, disabled, homemaker, student, or not employed and not looking for work).Work productivity was assessed using the WPAI questionnaire. 23nly participants who reported a work status of full-time, part-time, or self-employed provided data for absenteeism, presenteeism, and overall work impairment.HCRU included the number of self-reported visits to any healthcare provider (HCP), gastroenterologists (GE), and emergency rooms (ER), and hospitalizations over the past 6 months.Direct medical costs were imputed using data from the region-specific Medical Expenditure Panel Survey and included costs of an average HCP visit, ER visit, and hospitalization. 24Indirect costs were those associated with work productivity impairment and were calculated using estimated wages/salaries for each participant with data from the US Bureau of Labor Statistics.This cost analysis approach has been used in prior research in the NHWS. 25,26

Data Analysis
For analyses, due to sample size limitations with the severe CD category (Table 1), moderate and severe CD were grouped into 1 category to evaluate disease severity as an effect modifier.Descriptive statistics (frequencies, percentages for categorical variables; means with standard deviations [SD]) were used to characterize the CD cohort and better understand the distribution of key variables.Bivariate analyses were used to examine sociodemographic and clinical characteristics across CD severity groups (ie, mild and moderate/severe disease) and race/ethnicity groups (ie, non-Hispanic White, non-Hispanic Black, and Hispanic) among patients with self-reported physician-diagnosed CD and inform covariate selection for multivariable analyses (Supplementary Tables S1 and S3).
To assess associations between CD severity and race/ethnicity with outcomes of interest, multivariable analyses were conducted by constructing generalized linear regression models using the outcome-appropriate distribution and link functions (ie, linear model for SF-36 and EQ-5D outcomes, binary logistic regression model for labor force participation, and negative binomial distribution with log-link for all other outcomes).The same covariates were included in each regression model for all outcomes, enabling comparisons of the magnitude of the association for each independent variable.Models controlled for age (continuous; set to mean = 45.19 years), sex (male [reference group (ref)] or female), marital status (single/never married/decline to answer or married/ living with a partner [ref]), educational attainment (less than a college degree/declined to answer or college graduate or higher [ref]), household income (<$25 000, $25 000 to <$50 000, $50 000 to <$100 000, or $100 000 + [ Separate covariate-adjusted models were used to assess whether the relationship between CD severity and specific outcomes depended on race/ethnicity.Heterogeneity of the association was determined using an interaction term between race/ethnicity and CD severity, which was added to each covariate-adjusted model.The statistical significance of the interaction term was assessed using type III analyses (the F test for linear models and likelihood ratio test for logistic or negative binomial models).For models where the interaction term was statistically significant, interactions were probed via stratification, and covariate-adjusted estimates were interpreted in the context of the interaction term.All analyses were conducted using SAS 9.4.

Sociodemographic and Clinical Characteristics
This study included 1077 participants with a self-reported diagnosis of CD from a physician.Of these participants, 60.2% reported having mild disease and 39.8% reported having moderate/severe disease; furthermore, 76.0% identified as non-Hispanic White, 13.9% as Hispanic, and 10.1% as non-Hispanic Black.There was a lower proportion of female participants in the Hispanic group (36.7%) than in the non-Hispanic White group (51.5%;P = .003)(Table 1).Mean age was lower among non-Hispanic Black (36.84  [SD: 14.39]) and Hispanic participants with CD (34.05 [SD: 11.05]) than among non-Hispanic White participants (48.34 [SD: 16.48]; both P < .001).Severe CD was reported by a higher proportion of non-Hispanic Black participants (13.8%) than Hispanic (11.3%) and non-Hispanic White participants (6.7%; P = .032),and mild disease was reported by a greater proportion of non-Hispanic White participants (62.8%) than Hispanic (55.3%) and non-Hispanic Black participants (46.8%;P = .004).Interestingly, only the non-Hispanic Black participant group had a greater proportion of moderate/severe disease than mild disease.A greater proportion of non-Hispanic Black participants had a household income <$25 000 (29.4%) than non-Hispanic White (12.5%;P < .001)and Hispanic (10.7%;P = .001)participants.Although a similar proportion of participants across race/ethnicity groups had commercial insurance, a lower proportion of non-Hispanic White participants were not insured (7.1%) than non-Hispanic Black (14.7%;P = .022)and Hispanic (14.0%;P = .016)participants.Patient activation, on average, was lower among non-Hispanic Black (56.90 [SD: 13.95]) and Hispanic (58.16 [SD: 13.76]) participants than non-Hispanic White participants (62.31 [SD: 12.30]; P < .001and P = .001,respectively; Table 1).Notably, 72.6% of the study population had a PAM level of 3 (ie, taking action and gaining control) or 4 (ie, maintaining behaviors and pushing further).Previous participation in any clinical trial was significantly higher among non-Hispanic Black (33.9%) and Hispanic participants (45.3%) than among non-Hispanic White participants (20.0%;P = .003and P < .001,respectively).3D, Supplementary Table S1).Multivariable results for outcomes by disease severity largely aligned with previous bivariate analyses (Supplementary Table S2).

Study Outcomes by Race/Ethnicity-Main Effects
In multivariable analyses adjusting for potential confounding variables, non-Hispanic Black participants with CD reported lower PHQ-9 (depression) scores than non-Hispanic White participants (6.70 [SE: 0.54] vs. 8.43 [SE: 0.24]; P = .008),indicating less depressive symptomology, whereas Hispanic participants had PHQ-9 scores that were comparable to non-Hispanic White participants (Figure 4A, Supplementary Table S3).There were no significant differences in GAD-7 (anxiety) scores between race/ethnicity groups (P > .05; Figure 4A, Supplementary Table S3).Estimated mean MCS scores were also significantly higher among non-Hispanic Black compared with non-Hispanic White participants, suggesting better mental health in the context of HRQoL (43.48   4B, Supplementary Table S3).There were no differences in other HRQoL measures across race/ethnicity groups (Figure 4B and 4C, Supplementary Table S3).
Labor force participation was comparable across race/ethnicity groups (Supplementary Table S3), and WPAI outcomes did not differ relative to race/ethnicity groups (Figure 4D, Supplementary Table S3).The estimated mean number of HCP visits reported over the past 6 months was significantly lower among non-Hispanic Black participants (5.15 [SE: 0.60]) than non-Hispanic White participants (6.99 [SE: 0.28]; P = .015;Figure 5A, Supplementary Table S3); there was no difference between Hispanic and non-Hispanic White participants.Moreover, there were no significant differences between race/ethnicity groups in the estimated mean number of ER visits or hospitalizations over the past 6 months (Figure 5B, Supplementary Table S3), though non-Hispanic Black participants reported a greater number of ER visits than non-Hispanic White and Hispanic participants.In line with these findings, direct and indirect medical costs were comparable across race/ethnicity groups (Figure 5C and 5D, Supplementary Table S3).Bivariate analyses by race/ethnicity are presented in Supplementary Table S4.

Interaction Between Race/Ethnicity and Disease Severity
There was a significant interaction between race/ethnicity and CD severity on our multivariable models evaluating absenteeism and GE visits (Supplementary Table S5).In the main effects model, absenteeism was unrelated to both race/ethnicity and CD severity.However, race/ethnicity significantly moderated the relationship between CD severity and absenteeism (LRT X 2 (2) = 7.11, P = .029),such that CD severity was significantly associated with absenteeism, but only among non-Hispanic Black participants.As such, non-Hispanic Black participants with moderate/severe CD reported significantly more absenteeism over the past 7 days than non-Hispanic Black participants with mild CD (30.82% [SE: 15.34] vs. 9.61% [SE: 4.80]; RR: 3.21 [SE: 1.54]; P = .015;Figure 6A).
Race/ethnicity also moderated the relationship between CD severity and GE visits over the past 6 months (LRT X 2 (2) = 7.65, P = .022).The estimated mean number of GE visits was significantly higher among participants with moderate/severe CD than among those with mild CD for non-Hispanic White (  6B).

Discussion
This study aimed to assess PROs, including HRQoL and WPAI, and economic outcomes such as HCRU and annualized medical costs among 1,077 participants with CD relative to disease severity and race/ethnicity.Adjusted analyses showed that moderate/severe CD was associated with significantly worse outcomes than mild CD.We also found evidence to suggest that non-Hispanic Black race was associated with higher HRQoL scores with respect to depression and SF-36 MCS than non-Hispanic White race, and non-Hispanic Black race was associated with fewer HCP visits over the past 6 months than non-Hispanic White race.Finally, the relationship between CD severity and absenteeism, as well as the number of GE visits over the past 6 months, depended on race/ethnicity.
Participants with moderate/severe CD reported significantly worse HRQoL scores across all metrics measured.Importantly, the differences in SF-36 PCS (−4.84 [SE 0.56]) and EQ-5D-5L (−0.08 [SE 0.01]) between participants with moderate/severe CD and those with mild CD were clinically meaningful, highlighting the impact of CD symptom progression to a moderate/severe state.Participants with moderate/severe CD also reported greater work productivity Multivariable analyses of patient-reported outcomes of (A) PHQ-9 and GAD-7, (B) MCS, PCS, and EQ VAS, (C) SF-6D and EQ 5D, and (D) WPAI by Crohn's disease severity.Note: Lower PHQ-9 and GAD-7 scores and higher MCS, PCS, EQ VAS, SF-6D, and EQ-5D scores indicate better HRQoL.Error bars represent standard error.Presenteeism was only assessed among participants who worked > 0 hours over the last 7 days.PHQ-9, GAD-7, and WPAI outcomes were modeled using a log link with a negative binomial distribution; exp(β) = rate ratio.MCS, PCS, EQ VAS, SF-6D, and EQ-5D were modeled using an identity link with a normal distribution; β = estimated mean difference.Models control for age (continuous; set to mean = 45.19 years), gender (male [ref]; female), marital status (single/never married/decline to answer; married/living with a partner [ref]), educational attainment (less than a college degree/declined to answer; college graduate or higher [ref]), household income (<$25 000; $25 000 to <$50 000; $50 000 to <$100 000; $100 000 + impairment than those with mild CD, suggesting that more severe CD symptoms can significantly impact patients' productivity and ability to remain in the workforce.Moderate/ severe CD was also associated with higher HCRU, including HCP visits, ER visits, and hospitalizations; accordingly, participants with moderate/severe CD accumulated $22 421 more in annualized direct medical costs and $3584 more in annualized indirect costs than participants with mild CD.][9][10] Furthermore, these results highlight the importance of widespread access to education, early management, and adequate treatment options to delay and/or prevent the progression of CD.
In our study, a greater proportion of non-Hispanic Black participants reported severe CD than non-Hispanic White participants.Additionally, non-Hispanic Black participants reported significantly fewer HCP visits over the past 6 months than non-Hispanic White participants.There were no statistically significant differences in any PROs between Hispanic and non-Hispanic White participants.Though not significant, non-Hispanic Black participants reported more ER visits and Hispanic participants reported more hospitalizations over the past 6 months than non-Hispanic White participants.Numerous studies establish that marginalized populations disproportionally encounter obstacles to receiving satisfactory care in healthcare systems. 27,28Delayed diagnosis among nonwhite patients is a primary concern, as IBD has traditionally been viewed as a disease of European populations. 14,29n fact, some studies have reported that non-Hispanic Black IBD patients have longer symptom durations and are older at the time of diagnosis than non-Hispanic Black individuals. 14,305][36] Altogether, these issues can contribute to more severe disease and worse health outcomes among non-Hispanic Black individuals.
Consistently, in this study, the non-Hispanic Black cohort was the only group with a greater proportion of individuals with moderate/severe CD than mild CD.These issues may help explain our results of a greater proportion of non-Hispanic Black individuals with severe disease, as well as the trend toward more ER visits among the non-Hispanic Black cohort and more hospitalizations among the Hispanic cohort compared with non-Hispanic White participants.
Other potential contributors to greater disease severity, fewer HCP visits, and more ER visits observed among non-Hispanic Black participants may include decreased engagement with personal health, lack of access to healthcare resources, and distrust of the healthcare system. 37Indeed, Figure 4. Multivariable analyses of patient-reported outcomes of (A) PHQ-9 and GAD-7, (B) MCS, PCS, and EQ VAS, (C) SF-6D and EQ 5D, and (D) WPAI by race/ethnicity.Lower PHQ-9 and GAD-7 scores and higher MCS, PCS, EQ VAS, SF-6D, and EQ-5D scores indicate better HRQoL.Error bars represent standard error.Presenteeism was only assessed among participants who worked > 0 hours over the last 7 days.PHQ-9, GAD-7, and WPAI outcomes were modeled using a log link with a negative binomial distribution; exp(β) = rate ratio.MCS, PCS, EQ VAS, SF-6D, and EQ-5D were modeled using an identity link with a normal distribution; β = estimated mean difference.Models control for age (continuous; set to mean = 45.19 years), gender (male [ref]; female), marital status (single/never married/decline to answer; married/living with a partner [ref]), educational attainment (less than a college degree/declined to answer; college graduate or higher [ref]), household income (<$25 000; $25 000 to <$50 000; $50 000 to <$100 000; even among our engaged study population, non-Hispanic Black participants had lower patient activation scores than other participants.Previous studies have shown that non-Hispanic Black populations in the United States have lower rates of commercial health insurance than non-Hispanic White populations, and non-Hispanic Black patients with IBD are more likely to report concerns about the cost of treatment than non-Hispanic White patients. 29,33Results from this study corroborate previous findings, with a greater proportion of non-Hispanic Black participants reporting a household income <$25 000 and lacking commercial health insurance compared with non-Hispanic White participants.Additionally, a higher proportion of non-Hispanic Black participants were single/unmarried than non-Hispanic White participants, which may influence financial stability and health insurance coverage.Out-of-pocket costs and the requirement to pay health insurance deductibles may also impede access to healthcare.Furthermore, recent evidence suggests that treatment with anti-TNF agents has a significantly lower therapeutic response among non-Hispanic Black patients with IBD than among non-Hispanic White patients, which may be due to the underrepresentation of racial/ethnic diversity in clinical trials. 38As anti-TNFs are the standard first-line treatment for many patients with moderate or severe IBD, the difference in therapeutic response may partially contribute not only to deferred healthcare access, but also to potential distrust of the healthcare system should concerns regarding lack of treatment effectiveness be dismissed.As such, as evidenced in this study, non-Hispanic Black participants may face financial barriers to healthcare access which prevent them from using primary and/or preventative care for their condition, instead turning to emergency care when their symptoms cannot be otherwise managed.Improved access and increasing racial/ethnic diversity of clinical trial participants may inform practice and offer potential solutions to some of the issues enumerated here. Notably, in moderation analyses conducted to assess the relationship between CD severity and race on outcomes, the rate of workplace absenteeism and number of GE visits among non-Hispanic Black participants with moderate/severe CD was 3.21 and 3.46 times that of non-Hispanic Black participants with mild disease, respectively.In contrast, CD severity did not moderate any of the study outcomes among non-Hispanic White or Hispanic participants.Greater absenteeism among non-Hispanic Black participants with moderate/severe CD may be linked to experiencing more debilitating symptoms and the increased number of GE visits.Individuals who work in occupations that require an onsite presence or with inflexible schedules may find it impossible to work when their CD symptoms are severe.In a 2021 US Labor Force survey, employed non-Hispanic Black individuals were more likely than non-Hispanic White individuals to work in service occupations (21.5% vs. 15.0%) and production, transportation, and material moving (17.8% vs. 12.1%). 39mployed non-Hispanic Black individuals made up 12% of all employed workers but were substantially over-represented among detailed occupational categories of transit and intercity bus drivers (35%), nursing assistants (33%), security guards and gambling surveillance officers (33%) and home health aides (32%). 39Individuals in such roles may lack flexibility in their work schedule to attend healthcare provider visits and to continue working despite CD symptoms. 39Regarding GE visits, 1 possibility for the large magnitude of difference observed among the non-Hispanic Black participants is that, as previously discussed, social and financial barriers could prevent non-Hispanic Black individuals from visiting GE clinics until their symptoms become severe.In contrast, non-Hispanic White patients who do not experience these barriers may receive GE care even when their symptoms are milder.This may explain why we only observed a small difference in the number of GE visits between non-Hispanic White participants with mild or moderate/severe CD.Overall, these findings suggest severe CD symptoms may negatively influence work productivity and increase GE visits among non-Hispanic Black patients to a greater degree than other groups, and further underscores the importance of improved access and early diagnosis and treatment among non-Hispanic Black patients with CD.
Interestingly, non-Hispanic Black participants reported better HRQoL than non-Hispanic White participants with respect to depression (PHQ-9) and mental health (SF-36 MCS), whereas no differences in HRQoL were observed between Hispanic and non-Hispanic White participants.Overcoming the barriers experienced by non-Hispanic Black patients with CD during their healthcare journey, including delayed diagnoses, limited access, and different responses to treatme nts, 14,[27][28][29][31][32][33] as well as the marginalization and discrimination that non-Hispanic Black individuals experience within and outside of the healthcare sphere, 37,40 may necessitate development/employment of a greater degree of adaptive social mechanisms to manage their health and well-being. Consitent with this idea, national studies have found that non-Hispanic Black individuals have a lower lifetime risk of psychiatric disorders, including depression and anxiety, than non-Hispanic White individuals, despite experiencing higher levels of social and financial adversity.41 As such, a number of studies have assessed factors that influence resilience to health conditions among non-Hispanic Black communities.42 However, it is important to recognize that in the context of equitable distribution of care, the need for patients to demonstrate or build resilience is a not desirable circumstance and may in fact indicate where improved equity and access to care is needed.It is also important to note that stigma surrounding mental illness among non-Hispanic Black communities may lead to apprehension in acknowledging mental health symptoms and seeking professional care; therefore, selfreported depressive and/or anxious symptomology among non-Hispanic Black participants could be underrepresented in this study.35 The pool of Hispanic participants surveyed may be heterogeneous with respect to ancestry and regional/cultural differences, and this may have limited the statistical power of analyses for the Hispanic group.Future analyses comparing non-Hispanic Black and Hispanic patients with CD may provide a better understanding of the differences between these 2 groups.
This study had several strengths including the use of selfreported disease severity and PROs, which provide patient perspectives on disease outcomes and are increasingly used in clinical studies over third-party perceptions of disease burden.The use of several different PROs to measure an outcome (eg, mental health) allowed us to capture a more holistic picture of health and well-being.Clinical trial study populations have not historically reflected the diversity and characteristics of the general/national disease population.As such, increasing racial/ethnic representation in PRO data collection can aid in understanding disease impacts on patient quality of life and help inform clinical care.The use of NHWS data, which is designed to be nationally representative of the US adult population with respect to age, sex, and race/ethnicity, in this study provides PRO data on a diverse group of patients with CD.Additionally, the sampling approach used by the NHWS ensures comprehensive capture of racial and ethnic groups, permitting comparisons between various races.The assessment of non-Hispanic Black and Hispanic groups allowed the identification of key differences between the 2 groups that would have been missed had these groups have been pooled together (ie, in a White vs. nonwhite comparison).The extensive roster of outcomes assessed in this study allows a robust contribution to the literature on a variety of PROs and economic outcomes, increases the validity of the study findings through consistency across measures, and contributes to a holistic understanding of patient burden in CD.
A key limitation of the study was that self-reported race and disease severity could not be independently confirmed since study information was collected via survey.However, the overall trend of participants with moderate/severe CD faring worse than participants with mild CD supports the validity of self-reported diagnosis data from the NHWS.Other limitations include the relatively small sample sizes for non-Hispanic Black and Hispanic groups, despite pooling data across 3 survey years, which may have limited the ability to detect significant associations across outcomes.With only 150 participants identifying as Hispanic, we were unable to compare any potential differences between Hispanic Whites and Hispanic Blacks due to insufficient power.As the survey was conducted in English in a virtual setting, certain populations could be underrepresented, including non-English speaking individuals, elderly groups, and those without reliable computer or internet access.Given that the NHWS is an internet-based convenience sample, our sample was generally more educated and affluent than adults in the general US population 44,45 ; this difference was particularly pronounced among the non-Hispanic Black and Hispanic participants in our study.As such, the generalizability of our findings may be limited to a subset of patients with CD.While there are important generalizability limitations, our findings play a role in bridging the evidence gap, contributing to a better understanding of differences at various levels.Given the paucity of data, analyzing NHWS data is a step forward and further underscores the significant complexity and nuances.Another limitation is that cost data may underestimate the financial impact in the current post-COVID era, as medical costs were calculated using 2018 data.Residual confounding may have biased multivariable models, despite attempts to adjust for other potential explanatory variables.Finally, due to the cross-sectional nature of the study, causal inference was not possible.

Conclusions
The study results demonstrate that patients with moderate/ severe CD experience greater symptom burden than those with mild CD due to poor HRQoL, increased work productivity loss and activity impairment, greater HCRU, and higher medical costs.Race/ethnicity was associated with some outcomes, with non-Hispanic Black participants reporting higher HRQoL and lower HCP visits than non-Hispanic White participants.Further, non-Hispanic Black participants with severe CD reported greater absenteeism and more GE visits than non-Hispanic Black participants with mild CD.These findings emphasize the need for more effective treatments to reduce disease severity and highlight the importance of considering individual race/ethnicity groups rather than limiting comparisons to non-Hispanic White and grouped non-non-Hispanic White populations.Further research is required to better understand the potential factors influencing the relationship between CD severity and race/ ethnicity among patients with CD.Studies like ours can aid in identifying racial/ethnic disparities in healthcare and support the development of resources and new approaches to reduce these inequities.

Figure 3 .
Figure 3. Multivariable analyses of (A) HCP visits, (B) ER visits and hospitalizations, (C) annualized direct medical costs, and (D) annualized indirect medical costs by Crohn's disease severity.Error bars represent standard error.Indirect costs were only calculated for participants in the labor force at the time of the survey and who had a valid response (ie, non-missing) for the number of hours worked over the past 7 days and the number of hours missed in the past 7 days (mild n = 425; moderate/severe n = 274).Outcomes were modeled using a log link with a negative binomial distribution; exp(β) = rate ratio.Models control for age (continuous; set to mean = 45.19 years), gender (male [ref]; female), marital status (single/never married/ decline to answer; married/living with a partner [ref]), educational attainment (less than a college degree/declined to answer; college graduate or higher [ref]), household income (<$25 000; $25 000 to <$50 000; $50 000 to <$100 000; $100 000 + [ref]), health insurance coverage (Medicare; Medicaid/VA/ CHAMPUS; uninsured; commercial/other types of insurance [ref]), weight status (obese; overweight; underweight/normal weight/ declined to answer [ref]), smoking status (current smoker; former smoker; never smoker [ref]), alcohol use (drinks alcohol; does not drink alcohol [ref]), and Charlson Comorbidity Index Score (continuous; set to mean = 1.14).Abbreviations: ER, emergency room; HCP, healthcare practitioner; USD, United States dollars.

Figure 5 .
Figure 5. Multivariable analyses of (A) HCP visits, (B) ER visits and hospitalizations, (C) annualized direct medical costs, and (D) annualized indirect medical costs by Crohn's disease severity.Note: Error bars represent standard error.Indirect costs were only calculated for participants in the labor force at the time of the survey and who had a valid response (ie, non-missing) for the number of hours worked over the past 7 days and the number of hours missed in the past 7 days (non-Hispanic White n = 501; non-Hispanic Black n = 79; Hispanic n = 119).Outcomes were modeled using a log link with a negative binomial distribution; exp(β) = rate ratio.Models control for age (continuous; set to mean = 45.19 years), gender (male [ref]; female), marital status (single/never married/decline to answer; married/living with a partner [ref]), educational attainment (less than a college degree/declined to answer; college graduate or higher [ref]), household income (<$25 000; $25 000 to <$50 000; $50 000 to <$100 000; $100 000 + [ref]), health insurance coverage (Medicare; Medicaid/VA/CHAMPUS; uninsured; commercial/other types of insurance [ref]), weight status (obese; overweight; underweight/normal weight/ declined to answer [ref]), smoking status (current smoker; former smoker; never smoker [ref]), alcohol use [drinks alcohol; does not drink alcohol [ref]), and Charlson Comorbidity Index Score (continuous; set to mean = 1.14).Abbreviations: ER, emergency room; HCP, healthcare practitioner; USD, United States dollars.

Table 1 .
Sociodemographic characteristics of participants with self-reported Crohn's Disease.
a PAM scores range from 0 to 100 where higher scores indicate higher levels of activation.bPAM scores correlate to 1 of the 4 levels of patient activation: Level 1 = disengaged and overwhelmed; Level 2 = becoming aware but still struggling; Level 3 = taking action; Level 4 = maintaining behavior and pushing further.cDue to sample size limitations, moderate and severe CD were grouped into 1 category to allow for the evaluation of disease severity as an effect modifier.dThisrefers to all clinical trials, not just IBD-related trials.Sociodemographic characteristics were assessed using bivariate analyses.* α < 0.05 between non-Hispanic Black and non-Hispanic White participants.† α < 0.05 between Hispanic and non-Hispanic White participants.‡ α < 0.05 between Hispanic and non-Hispanic Black participants.Results were adjusted for multiple comparisons with the Bonferroni correction.Abbreviations: PAM, patient activation measure; SD, standard deviation.