Abstract
Objective
Investigate the relationship between socioeconomic status (SES) and race with self-reported fatigue, depression, and anxiety levels in multiple sclerosis (MS).
Methods
Cross-sectional review of the MS Partners Advancing Technology and Health Solutions (MS PATHS) database for adults with MS in the United States. We evaluated race and socioeconomic status (available markers: insurance, employment status, or level of education) as predictors of fatigue, depression, and anxiety sub- scores of the Neuro-QoL (Quality of life in neurological disorders), with particular interest between Caucasians/ whites (CA) and African Americans/blacks (AA). Multivariate linear regression models included as covariates age, sex, disability status, smoking status, body mass index, and disease-modifying therapy.
Results
7,430 individuals were included; compared to CA, AA tended to be younger, more female-pre-dominant, and had a higher level of disability. AA had completed slightly less education, had a higher level of Medicaid coverage or uninsured status, and had higher rates of unemployed or disabled status. In the univariate model, markers oflower SES, by whichever definition we used, correlated with worse affective symptoms. In the multivariate model stratified by race, CA showed similar trends. In contrast, in AA, only lower SES by employment status was correlated with worse affective symptoms. In both CA and AA, moderate and severe level of disability correlated with worse affective symptoms.
Conclusion
SES and race may influence affective symptoms reported by individuals with MS. The reasons for the correlation are likely multifactorial. Longitudinal studies should strive to identify factors associated with risk of affective symptoms in MS that may be modifiable.
Keywords: Multiple sclerosis, Epidemiology, Health outcomes, Quality of life
1. Introduction
Socioeconomic status (SES) is a measure of an individual’s social standing in relation to others, which includes access to resources, opportunity, privilege, amongst others. (Galobardes et al., 2006) Studies have linked lower SES to the development of various health conditions and with increased disability and mortality. (Isaacs and Schroeder, 2004) Similar to the effects of lower SES, African American race has been associated with increased mortality and higher level of chronic diseases. (Cunningham et al., 2017) Disentangling the effects of SES and race is difficult in the United States (US). (Sohn, 2017; Williams et al., 2016; Williams et al., 2010) How SES and race, alone or in combination, may affect health outcomes is important to consider when working toward addressing healthcare disparities. (Williams et al., 2016; Williams et al., 2010)
The effect of SES on outcomes in multiple sclerosis (MS) is also gaining interest. Recent studies have demonstrated that lower SES is associated with higher risk of disability progression in the UK and Canada (Harding et al., 2019) and a single-center, US cohort. (Briggs et al., 2019) A prior study in MS (Marrie et al., 2006) showed that accounting for SES might explain some of differing outcomes between Caucasians and African Americans. Most prior studies have focused upon physical disability and associated outcomes. (Briggs et al., 2019; Marrie et al., 2006; Kister et al., 2010; Weinstock-Guttman et al., 2003; Kaufman et al., 2003) Affective symptoms are important to study, as mental comorbidities are common in MS (Marrie et al., 2009) and associated with worse outcomes, including greater long-term disability and mortality. (Marrie et al., 2008; Mohr et al., 1997; Feinstein, 2002; McKay et al., 2018)
There is a lack of studies evaluating how race and SES relate to patient-reported outcomes encompassing affective symptoms in MS. Thus, we aimed to investigate the relationship between markers of SES and race, and self-reported measures of affective disorders in a large population of US-based people with MS.
2. Methods
2.1. Standard protocol approvals, registrations, and patient consents
Institutional Review Board approval was granted by all US sites including Johns Hopkins University School of Medicine. Participants provided informed consent or an authorization to use their data when an institution granted a waiver of written informed consent.
2.2. Design
Cross-sectional observational study of individuals with MS in the US.
2.3. Study population
This study utilized data available from the Multiple Sclerosis Partners Advancing Technology and Health Solutions (MS PATHS) network, which is a network of seven MS centers in the US and three in Europe, comprised of patients with confirmed clinically isolated syndrome (CIS) or MS, and is supported by Biogen. US centers were exclusively included in this study since the social implications of race likely differ dramatically between countries. The US centers include the Cleveland Clinic, Cleveland Clinic Nevada, New York University, Ohio Health Research Institute, Johns Hopkins University, University of Rochester, and Washington University in St. Louis. Data are collected using an iPad-based system during participants’ routine clinical care activities at each center. Participants enter demographic characteristics including age, sex, ethnicity, race (American Indian/Alaska Native, Asian, Black or African American [AA], Native Hawaiian/Other Pacific Islander, White or Caucasian American [CA]), as well as MS characteristics including age at symptom onset, disease subtype, and disease modifying therapy (DMT) use. Clinical and lifestyle information, including height and weight (used to calculate body mass index [BMI]) and smoking status, is also available. Participants complete the MS Performance Test (MSPT), which is an iPad-based adaptation of the MS Functional Composite (MSFC) and includes assessment of walking speed, manual dexterity, vision, and processing speed, in addition to a health questionnaire and the Neuro-QoL (see details of the latter below). (Rudick et al., 2014)
2.4. Defining markers of socioeconomic status
Measurable indicators of SES include income, education, employment, amongst others. (Galobardes et al., 2006) Three markers of SES were available, including education, employment status, and type of medical insurance. Education was grouped into the following categories: (1) Less than or equal to 12 years (high school diploma or equivalent), (2) 13–15 years (some college or vocational), (3) 16 years (college graduate), and (4) 17 years or more (graduate degrees). Employment status was grouped into three categories: (1) Full-time, homemaker, student, or retired, (2) Part-time, and (3) Unemployed or disabled. Medical insurance was grouped into two categories: (1) Medicaid or uninsured and (2) Non-Medicaid (Medicare, Private, and Military).
2.5. Inclusion criteria
Adult individuals (18 years or older) who received care in one of the US centers with a diagnosis of MS were included. Individuals with at least one demographic or SES data and completed the Neuro-QoL questionnaire were included. When several visits were available, the first completed Neuro-QoL questionnaire was selected.
2.6. Health-related quality oflife in neurologic disease (Neuro-QoL)
At each clinic visit, participants reported health-related quality of life via the Neuro-QoL for 12 domains. Specifically, the Neuro-QoL computer adaptive test short-form (Neuro-QoL 2019) was used, with the sub-scores for fatigue, depression, and anxiety as the primary outcome measures of interest. The NeuroQoL has been validated in persons with MS. (Miller et al., 2016) The scoring generates a T-score with a mean of 50 and a standard deviation of 10. The fatigue measurement utilized a clinical reference population, while the anxiety and depression measurements utilized a general population reference sample. (Neuro-QoL 2019) A higher T-score represents more of the concept being measured, thus in our measurements of interest, a higher T-score represents worse symptoms.
2.7. Statistical analysis
Our statistical analysis was designed with the primary objective of examining whether race and markers of SES are associated with fatigue, depression, and anxiety in people with MS. Chi-square tests were used to compare demographics and clinical characteristics between the two major racial groups (CA and AA). We then assessed the associations between race and the SES measures with fatigue, depression, and anxiety using linear models adjusting for age, sex, self-reported MS disease subtype and Patient Determined Disease Steps (PDDS) level (i.e., disability status). We performed supplementary analyses with the adjustment for additional confounders of smoking status, BMI, and DMT used. DMT was divided into injectable (including glatiramer acetate and interferons), oral (teriflunomide, fingolimod, and dimethyl fumarate), infusion therapies (natalizumab, ocrelizumab, rituximab, and alemtuzumab), no therapy, or therapy not listed. Missing information was relatively uncommon (<5% for all covariates). We imputed missing values using multiple imputation by chained equations (van Buuren, 2007) with 50 imputations. All analyses were conducted within each dataset of the imputed sets and the overall estimate and standard error were calculated by using the Rubin’s rules. (Rubin, 2004) We additionally compared the demographic characteristics between the complete information and the imputed dataset. No significant differences were found (Supplemental Table 1), which suggests that the data were likely to be missing at random and the multiple imputation method was appropriate to use.
We assessed the effect modification by race for each of the SES measures on fatigue, depression and anxiety using cross-product terms and likelihood ratio tests. Since results suggested significant effect modification by race, we stratified analyses by racial groups (CA and AA). The stratified multivariate models were performed on both the imputed data and the complete cases dataset, and no significant differences were observed. We additionally conducted sensitivity analysis to model without PDDS, and we found that the main effects of SES were not significantly changed by removing this covariate.
All statistical analyses were conducted using R programming (version 3.6.1). Missing data imputation was performed using the “mice” package. (van and Groothuis-Oudshoorn, 2011) General linear hypotheses testing was performed using the R function “glht” (multcomp package). (Hothorn et al., 2008) All P-values were reported at the 0.05 significance level.
2.8. Data availability
MS PATHS data is currently only accessible to Biogen or participating healthcare institutions in the MS PATHS program.
3. Results
A total of 11,878 individuals were enrolled in MS PATHS across US and Europe as of September 14, 2018, 9309 (78%) of whom were residing in the US. Based on the abovementioned inclusion criteria, 7430 (80%) individuals with MS in the US were included in this study, 5504 (74%) of whom were female (Table 1). Age range was 18–85 years old, with a mean of 48 years old (SD: 12.68 years). The majority of participants were CA (79%), and the largest minority group was AA (12%).
Table 1.
Baseline characteristics of participants.
| Characteristic | Whole sample, n = 7430 | Caucasian/white, n = 5846 | African American/black, n = 924 | p value |
|---|---|---|---|---|
| Age, mean ± SD; y | 48.50 ± 12.68 | 49.49 ± 12.46 | 45.02 ± 12.44 | < 0.001 |
| Female, n (%) | 5504 (74.10) | 4262 (72.90) | 752 (81.39) | < 0.001 |
| Race, n (%) | 5846 (78.70) | |||
| 924 (12.40) | ||||
| 660 (8.90) | ||||
| Age at onset, mean ± SD; y | 32.80 ± 11.38 | 33.29 ± 11.23 | 32.10 ± 11.60 | 0.07 |
| Age at diagnosis, mean ± SD; y | 35.74 ± 11.16 | 36.33 ± 10.98 | 34.58 ± 11.43 | < 0.001 |
| Body Mass Index (BMI), mean ± SD; kg/m2 | 29.11 ± 7.15 | 28.92 ± 7.05 | 30.67 ± 7.65 | 0.03 |
| Smoking status, n (%) | < 0.001 | |||
| Never | 4212 (56.69) | 3227 (55.20) | 605 (65.48) | |
| Former | 1898 (25.55) | 1546 (26.45) | 182 (19.70) | |
| Current | 1033 (13.90) | 835 (14.28) | 113 (12.23) | |
| Unreported | 287 (3.86) | 238 (4.07) | 24 (2.60) | |
| Employment status, n (%) | < 0.001 | |||
| Group 1 | ||||
| Full time | 3242 (43.6) | 2600 (44.47) | 389 (42.10) | |
| Retired | 816 (11.00) | 682 (11.68) | 80 (8.66) | |
| Homemaker | 336 (4.50) | 283 (4.84) | 17 (1.84) | |
| Student | 144 (1.90) | 85 (1.45) | 25 (2.71) | |
| Group 2: | ||||
| Part time | 592 (8.00) | 479 (8.19) | 52 (5.63) | |
| Group 3: | ||||
| Disabled | 1778 (23.90) | 1357 (23.21) | 262 (28.35) | |
| Unemployed | 385 (5.20) | 261 (4.46) | 78 (8.44) | |
| Other | 114 (1.50) | 86 (1.47) | 18 (1.95) | |
| Unreported | 23 (0.30) | 12 (0.21) | 3 (0.32) | |
| Insurance, n (%) | < 0.001 | |||
| Medicaid/Uninsured | 642 (8.70) | 394 (6.74) | 152 (16.45) | |
| Other | 6749 (90.80) | 5429 (92.87) | 765 (82.79) | |
| Unreported | 39 (0.50) | 23 (0.67) | 7 (0.76) | |
| Education, n (%) | < 0.001 | |||
| High school or lower (≤ 12 years) | 1772 (23.85) | 1417 (24.24) | 264 (28.57) | |
| Post-high school (13–15 years) | 2177 (29.30) | 1636 (27.98) | 328 (35.50) | |
| College (16 years) | 1603 (21.57) | 1320 (22.58) | 156 (16.88) | |
| Graduate (> 16 years) | 1878 (25.28) | 1473 (25.20) | 176 (19.05) | |
| MS type, n (%) | 0.006 | |||
| Relapsing | 4619 (62.20) | 3671 (62.80) | 539 (58.33) | |
| Progressive | 2412 (32.50) | 1850 (31.65) | 335 (36.26) | |
| Unreported | 399 (5.30) | 325 (5.56) | 50 (5.41) | |
| Patient-Determined Disease Steps (PDDS), n (%) | <0.001 | |||
| Mild (0–1) | 3952 (53.19) | 3167 (54.17) | 445 (48.16) | |
| Moderate (2–5) | 2665 (35.87) | 2060 (35.24) | 350 (37.88) | |
| Severe (> 5) | 743 (10.00) | 563 (9.63) | 119 (12.88) | |
| Unreported | 70 (0.94) | 56 (0.96) | 10 (1.08) | |
| Current Disease Modifying Therapy, n (%) | < 0.001 | |||
| Injectable | 1683 (22.65) | 1379 (23.59) | 182 (19.70) | |
| Oral | 2218 (29.85) | 1779 (30.43) | 243 (26.30) | |
| Infusion | 1496 (20.13) | 1087 (18.59) | 243 (26.30) | |
| No therapy | 1508 (20.30) | 1194 (20.42) | 177 (19.16) | |
| Not listed | 511 (6.88) | 393 (6.72) | 79 (8.55) | |
| Unreported | 14 (0.188) | 14 (0.24) | 0 | |
| Neuro-QoL, T-score (SD) | ||||
| Fatigue | 50.08 ± 9.87 | 49.80 ± 9.76 | 50.45 ± 10.05 | < 0.001 |
| Depression | 47.05 ± 8.05 | 46.68 ± 7.94 | 48.49 ± 8.33 | < 0.001 |
| Anxiety | 50.66 ± 9.39 | 50.26 ± 9.30 | 51.78 ± 9.68 | < 0.001 |
Relative to CA individuals, AA individuals tended to be younger and female. AA individuals also had a higher level of disability based on PDDS, lower level of attained education, higher levels of Medicaid or uninsured status, and higher rates of either unemployed or disabled employment status (Table 1). In addition, AA had higher levels of fatigue, depression, and anxiety in univariate models (Table 1).
In the univariate model, all markers of lower SES were associated with worse scores in all outcomes (fatigue, depression, and anxiety) (Table 2). When employment status was used as the SES measure, the group of unemployed or disabled individuals had worse scores in all outcomes (mean difference for fatigue: 6.93 points, 95% confidence interval [CI] 6.43–7.43; depression: 5.51 points, 95% CI 5.10–5.92; anxiety: 5.43 points, 95% CI 4.95–5.92). When insurance status was used, individuals with Medicaid or uninsured status, compared to all other insurance types, showed worse scores in all outcomes (mean difference for fatigue: 4.19 points, 95% CI 3.36–5.01; depression: 4.17 points, 95% CI 3.50–4.84; anxiety: 4.40 points, 95% CI 3.62–5.19). When education was used as the SES measure, relative to individuals with high school or equivalent level or lower, individuals with advanced degrees demonstrated better scores in all outcomes (mean dif- ference for college graduates, fatigue: −3.64 points, 95% CI −4.31 to −2.96; depression: −3.13 points, 95% CI −3.68 to −2.57; anxiety: −2.91 points, 95% CI −3.56 to −2.26. For individuals with graduate degrees, fatigue: −4.24 points, 95% CI −4.89 to −3.58; depression −3.39 points, 95% CI −3.92 to −2.85; anxiety −3.53 points, 95% CI −4.16 to −2.90). In analyses assessing for effect modification, there were interactions between race and each SES marker (Table 3). As a result, we proceeded with analyses stratified by race.
Table 2.
Univariate model. Neuro-QoL score change and 95th percentile confidence interval for correlations between socioeconomic status and fatigue, depression, and anxiety scores.
| Fatigue |
Depression |
Anxiety |
||||
|---|---|---|---|---|---|---|
| Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | |
| Employment status | ||||||
| Full time, retired, student and homemaker | Reference | |||||
| Part time | 1.69 (0.85, 2.53) | < 0.001 | 0.87 (0.18, 1.56) | 0.01 | 0.99 (0.18, 1.81) | 0.02 |
| Disabled, unemployed | 6.93 (6.43, 7.43) | <0.001 | 5.51 (5.10, 5.92) | < 0.001 | 5.43 (4.95, 5.92) | <0.001 |
| Other | 4.45 (2.63, 6.26) | < 0.001 | 2.99 (1.50, 4.47) | < 0.001 | 3.75 (1.99, 5.51) | < 0.001 |
| Medicaid/Uninsured | 4.19 (3.36, 5.01) | < 0.001 | 4.17 (3.50, 4.84) | < 0.001 | 4.40 (3.62, 5.19) | < 0.001 |
| Education | ||||||
| High school or lower | Reference | |||||
| Post-high school | −0.27 (−0.90, 0.36) | 0.40 | −1.26 (−1.78, −0.75) | < 0.001 | −0.90 (−1.51, −0.30) | 0.003 |
| College | −3.64 (−4.31, −2.96) | < 0.001 | −3.13 (−3.68, −2.57) | < 0.001 | −2.91 (−3.56, −2.26) | < 0.001 |
| Graduate | −4.24 (−4.89, −3.58) | < 0.001 | −3.39 (−3.92, −2.85) | < 0.001 | −3.53 (−4.16, −2.90) | < 0.001 |
Table 3.
Neuro-QoL score change and 95th percentile confidence interval for adjusted associations between socioeconomic status and fatigue, depression, and anxiety scores. All values were adjusted for age, sex, disease subtype, and PDDS.
| Fatigue |
Depression |
Anxiety |
||||
|---|---|---|---|---|---|---|
| Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | |
| White/Caucasian | ||||||
| Employment status | ||||||
| Full time, retired, student and homemaker | Reference | |||||
| Part time | 0.98 (0.15, 1.80) | 0.02 | 0.59 (−0.11, 1.29) | 0.10 | 0.37 (−0.46, 1.20) | 0.38 |
| Disabled, unemployed | 3.97 (3.40, 4.55) | <0.001 | 3.37 (2.89, 3.86) | <0.001 | 3.47 (2.89, 4.00) | <0.001 |
| Other | 3.69 (1.86, 5.53) | < 0.001 | 2.55 (0.99, 4.12) | 0.001 | 3.23 (1.38, 5.09) | <0.001 |
| Medicaid/Uninsured | 2.23 (1.35, 3.10) | <0.001 | 3.12 (2.38, 3.86) | <0.001 | 3.02 (2.14, 3.90) | <0.001 |
| Education | ||||||
| High school or lower | Reference | |||||
| Post-high school | 0.19 (−0.42, 0.81) | 0.54 | −0.84 (−1.36, −0.31) | 0.002 | −0.58 (−1.20, −0.05) | 0.07 |
| College | −2.05 (−2.71, −1.40) | <0.001 | −1.88 (−2.44, −1.32) | <0.001 | −1.91 (−2.58, −1.25) | <0.001 |
| Graduate | −2.61 (−3.25, −1.97) | <0.001 | −2.02 (−2.57, −1.47) | <0.001 | −2.40 (−3.04, −1.75) | <0.001 |
| Patient-Determined Disease Steps (PDDS) | ||||||
| Mild (0–1) | Reference | |||||
| Moderate (2–5) | 6.35 (5.77, 6.93) | < 0.001 | 3.50 (3.01, 4.00) | < 0.001 | 4.25 (3.66, 4.83) | < 0.001 |
| Severe (> 5) | 3.84 (2.90, 4.77) | < 0.001 | 2.87 (2.08, 3.67) | < 0.001 | 1.45 (0.51, 2.40) | 0.003 |
| Black/African American | ||||||
| Employment status | ||||||
| Full time, retired, student and homemaker | Reference | |||||
| Part time | −0.70 (−3.40, 1.99) | 0.61 | −0.21 (−2.45, 2.04) | 0.86 | −0.26 (−2.89, 2.36) | 0.84 |
| Disabled, unemployed | 1.83 (0.37, 3.29) | 0.01 | 2.98 (1.77, 4.19) | < 0.001 | 2.43 (1.01, 3.85) | 0.001 |
| Other | −0.29 (−4.67, 4.08) | 0.89 | 0.35 (−3.28, 3.97) | 0.85 | 0.13 (−4.12, 4.38) | 0.95 |
| Medicaid/Uninsured | −0.025 (−1.67, 1.62) | 0.98 | 0.44 (−0.95, 1.83) | 0.53 | 0.81 (−0.81, 2.42) | 0.33 |
| Education | ||||||
| High school or lower | Reference | |||||
| Post-high school | 0.20 (−1.32, 1.72) | 0.79 | −1.16 (−2.43, 0.12) | 0.08 | −0.83 (−2.44, 0.78) | 0.43 |
| College | −0.35 (−2.21, 1.50) | 0.71 | −1.31 (−2.87, 0.25) | 0.10 | 0.029 (−1.89, 1.95) | 0.61 |
| Graduate | −0.63 (−2.46, 1.20) | 0.50 | −1.72 (−3.26, −0.19) | 0.03 | −1.23(−3.07, 0.61) | 0.52 |
| PDDS | ||||||
| Mild (0–1) | Reference | |||||
| Moderate (2–5) | 5.89 (4.38, 7.40) | < 0.001 | 3.96 (2.69, 5.22) | < 0.001 | 5.15 (3.67, 6.63) | < 0.001 |
| Severe (> 5) | 3.17 (0.88, 5.45) | 0.007 | 2.63 (0.71, 4.54) | 0.007 | 2.05 (−0.18, 4.28) | 0.07 |
Interaction p-values: For employment: fatigue (p < 0.001), depression (p = 0.65), anxiety (p = 0.42). For Medicaid/uninsured: fatigue (p = 0.03), depression (p = 0.001), anxiety (p–0.04). For education: fatigue (p = 0.02), depression (p = 0.83), anxiety (p = 0.02).
In the CA group, we found that lower SES, by any definition, correlated with worse Neuro-QoL scores in fatigue, depression, and anxiety (Table 3). When employment status was used as the SES measure, the group of unemployed or disabled individuals had worse scores in all outcomes (mean difference for fatigue: 3.97 points, 95% CI 3.40–4.55; depression: 3.37 points, 95% CI 2.89–3.86; anxiety: 3.47 points, 95% CI 2.89–4.00). When insurance status was used, CA individuals with Medicaid or those that were uninsured, compared to all other insurance types, showed worse scores in all outcomes (mean difference for fatigue: 2.23 points, 95% CI 1.35–3.10; depression: 3.12 points, 95% CI 2.38–3.86; anxiety: 3.02 points, 95% CI 2.14–3.90). When education was used as a SES measure, relative to individuals with high school or equivalent level or lower, CA individuals with college or more advanced degrees demonstrated better scores in all outcomes (mean differenze for college graduates, fatigue: −2.05 points, 95% CI −1.40 to −2.71 points; depression: −1.88 points, 95% CI −1.32 to −2.44 points; anxiety: −1.91 points, 95% CI −1.25 to −2.58 points; for graduate degree, fatigue: −2.61 points, 95% CI −1.97 to −3.25 points; depression: −2.02 points, 95% CI −1.47 to −2.57 points; anxiety: −2.40 points, 95% CI −1.75 to −3.04 points). Both moderate and severe disability correlated with worse scores (For moderate disability, fatigue: 6.35 points, 95% CI 5.77–6.93; depression: 3.50 points, 95% CI 3.01–4.00; anxiety: 4.25 points, 95% CI 3.66–4.83; for severe disability, fatigue: 3.84 points, 95% CI 2.90–4.77; depression: 2.87 points, 95% CI 2.08–3.67; anxiety: 1.45 points, 95% CI 0.51–2.40).
These findings are in contrast to the AA group where significantly worse Neuro-QoL scores were found for fatigue, depression, and anxiety when employment was used as the predictor (For disabled or un- employed status, fatigue: 1.83 points, 95% CI 0.37–3.29; depression: 2.98 points, 95% CI 1.77–4.19; anxiety: 2.43 points, 95% CI 1.01–3.85). When insurance status was used, no differenze in outcomes was seen. When education was used, only graduate level or higher education correlated with better scores (−1.72 points, 95% CI −0.19 to −3.26). Similar to the CA group, both moderate and severe disability correlated with worse scores (For moderate disability, fatigue: 5.89 points, 95% CI 4.38–7.40; depression: 3.96 points, 95% CI 2.69–5.22; assnxiety: 5.15 points, 95% CI 3.67–6.63; for severe disability, fatigue: 3.17 points, 95% CI 0.88–5.45; depression: 2.63 points, 95% CI 0.71–4.54).
After adjustments for additional confounders of smoking status, BMI, and DMT, we did not see a meaningful change in the results (Table 4).
Table 4.
Neuro-QoL score change and 95th percentile confidence interval for adjusted associations between socioeconomic status and fatigue, depression, and anxiety scores. All values were adjusted for age, sex, disease subtype, PDDS, smoking status, BMI, and disease modifying therapy group.
| Fatigue |
Depression |
Anxiety |
||||
|---|---|---|---|---|---|---|
| Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | Estimate (95% CI) | p-value | |
| White/Caucasian | ||||||
| Employment status | ||||||
| Full time, retired, student and homemaker | Reference | |||||
| Part time | 1.02 (0.21, 1.84) | 0.01 | 0.58 (−0.11, 1.28) | 0.10 | 0.35 (−0.48, 1.17) | 0.41 |
| Disabled, unemployed | 3.50 (2.93, 4.07) | <0.001 | 2.99 (2.50, 3.47) | <0.001 | 3.11 (2.53, 3.69) | <0.001 |
| Other | 3.45 (1.63, 5.27) | < 0.001 | 2.24 (0.69, 3.79) | 0.005 | 2.91 (1.06, 4.75) | 0.002 |
| Medicaid/Uninsured | 1.51 (0.64, 2.39) | 0.007 | 2.48 (1.74, 3.23) | < 0.001 | 2.36 (1.48, 3.24) | < 0.001 |
| Education | ||||||
| High school or lower | Reference | |||||
| Post-high school | 0.42 (−0.19, 1.04) | 0.18 | −0.60 (−1.12, −0.07) | 0.002 | −0.35 (−0.97, 0.27) | 0.27 |
| College | −1.36 (−2.03, −1.70) | < 0.001 | −1.22 (−1.79, −0.65) | < 0.001 | −1.29 (−1.96, −0.61) | 0.0002 |
| Graduate | −1.76 (−2.42, −1.11) | < 0.001 | −1.23 (−1.80, −0.67) | < 0.001 | −1.66 (−2.33, −0.99) | < 0.001 |
| Patient-Determined Disease Steps (PDDS) | ||||||
| Mild (0–1) | Reference | |||||
| Moderate (2–5) | 6.06 (5.49, 6.63) | < 0.001 | 3.31 (2.82, 3.80) | < 0.001 | 4.12 (3.53, 4.70) | < 0.001 |
| Severe (> 5) | 3.61 (2.68, 4.54) | < 0.001 | 2.70 (1.90, 3.49) | < 0.001 | 1.36 (0.42, 2.30) | 0.003 |
| Black/African American | ||||||
| Employment status | ||||||
| Full time, retired, student and homemaker | Reference | |||||
| Part time | −0.88 (−3.23, 1.47) | 0.46 | −0.42 (−2.39, 1.55) | 0.77 | −0.53 (−2.83, 1.78) | 0.65 |
| Disabled, unemployed | 1.50 (0.023, 2.97) | 0.05 | 2.68 (1.45, 3.91) | < 0.001 | 2.10 (0.66, 3.54) | 0.004 |
| Medicaid/Uninsured | −0.61 (−2.27, 1.05) | 0.47 | 0.013 (−1.38, 1.41) | 0.99 | 0.32 (−1.31, 1.95) | 0.70 |
| Education | ||||||
| High school or lower | Reference | |||||
| Post-high school | 0.18 (−1.35, 1.70) | 0.82 | −1.01 (−2.30, 0.27) | 0.12 | −0.40 (−1.90, 1.10) | 0.60 |
| College | −0.054 (−1.94, 1.83) | 0.96 | −0.98 (−2.57, 0.61) | 0.23 | 0.98 (−0.87, 2.83) | 0.30 |
| Graduate | −0.17 (−2.04, 1.70) | 0.86 | −1.30 (−2.87, 0.28) | 0.11 | 0.022 (−1.81, 1.86) | 0.98 |
| PDDS | ||||||
| Mild (0–1) | Reference | |||||
| Moderate (2–5) | 5.83 (4.33, 7.34) | <0.001 | 4.13 (2.87, 5.39) | <0.001 | 5.34 (3.86, 6.82) | 0.06 |
| Severe (> 5) | 3.40 (1.14, 5.66) | 0.003 | 2.99 (1.10, 4.88) | < 0.001 | 2.42 (0.20, 4.63) | 0.03 |
4. Discussion and conclusions
This study demonstrates that markers of lower SES are correlated with worse self-reported scores in fatigue, depression, and anxiety in US-based individuals with MS treated at tertiary MS centers. We also demonstrate that there are race-based differences in these outcomes, and that the effects of SES differ between CA and AA.
In the univariate model, prior to stratification by race, disabled or unemployed status and Medicaid or uninsured status are correlated with worse self-reported levels of fatigue, depression, and anxiety, while a higher attained education level (college or greater) is correlated with better outcomes. This is in keeping with a recent study that identified lower median income and Medicaid or uninsured status as associated with higher depressive scores in individuals with MS. (Briggs et al., 2019) In addition, a few recent studies have demonstrated that lower SES is associated with a higher risk of MS disability progression in UK and Canada (Harding et al., 2019) and a Cleveland Clinic cohort. (Briggs et al., 2019) There is evidence for a biological basis for these differences, with studies indicating low SES associated with a proinflammatory phenotype. (Loucks et al., 2010; Tabassum et al., 2008; Miller et al., 2009) Modifiable features that may well influence MS disease progression, such as smoking and higher BMI, are generally over-represented in lower SES groups. (Pampel et al., 2010) In our study, adjustment for smoking status and BMI did not meaningfully change our results, suggesting that these modifiable features are likely not the only influencers of outcomes. These factors associated with disability progression may also be associated with worse affective symptomatology.
Similar to the findings of the univariate model, within CA, markers of lower SES (disabled or unemployed status and Medicaid coverage or uninsured status) are correlated with worse outcomes, while marker of higher SES (higher attained education) is correlated with better outcomes. Within AA, similar to CA, disabled or unemployed status correlated with worse self-reported levels of fatigue, depression, and anxiety. However, Medicaid or uninsured status was not associated with these outcomes. In addition, another difference is that higher attained education level (graduate degree or higher) is correlated with lower level of depression only. This indicates that contrary to what was seen within CA, SES may have less of an association with these out- comes in AA people with MS. The underlying mechanisms for this race- based difference are not explored in this study, but they are likely to be multi-factorial, and a number of potential explanations exist, which we explore below.
First, our study is powered to detect more of a correlation in CA due to the larger size of that racial subgroup. In addition, when we look at the demographics of CA versus AA, we see that the distribution of the sample within the various subgroups of the markers of SES assessed was less widely distributed in the AA group, again affecting the power to detect associations within the AA group. Alternatively, it is possible that the markers that we have used for SES are not true gauges of the underlying social stratification within the groups. (Isaacs and Schroeder, 2004) Another potential explanation for why SES does not appear as meaningful in AA compared to CA is the clinical differences between the two racial groups, which are not fully accounted for, that may factor into predicting the outcomes. Prior studies point toward a more aggressive MS disease course in AA, (Marrie et al., 2006; Kister et al., 2010; Weinstock-Guttman et al., 2003; Kaufman et al., 2003) and in our cohort, similar to previously published studies, AA tended to be younger, female, and have a higher level of disability. (Marrie et al., 2006; Kister et al., 2010; Weinstock-Guttman et al., 2003; Kaufman et al., 2003; Cree et al., 2004) An individual’s genetic admixture may also influence immune system reactivity and susceptibility to immune-mediated disease, such as a study that indicated that AA are more likely to have alleles that are associated with a pro-inflammatory response, (Nédélec et al., 2016) and inflammation has been proposed as a contributor to affective symptomatology in people with MS. (Rossi et al., 2017; Colasanti et al., 2016) However, the mechanisms for the abovementioned racial difference are likely more complex than these clinical variations. It is possible that additional stressors of being AA in the US supersede much of the SES effect (Williams et al., 2016) that we clearly see in the CA group. Notably, AA face significant challenges in healthcare, including access to services, cultural stigma, and systemic racism. (Beatty et al., 2003; Russell and Jewell, 1992; Saadi et al., 2017; Fabius et al., 2018)
There are some limitations of the study. Firstly, the study was cross- sectional rather than longitudinal. SES is dynamic throughout an individual’s life; therefore, it would be particularly helpful to assess how changes in SES (such as in employment status, insurance coverage, and such) may influence disability and quality of life. This would provide a stronger evidence for an association. Additional SES components, such as geographic or zip-code data and household income, would be beneficial in creating a more comprehensive measure of SES. In addition, separation of SES and race effects is quite difficult, as these are tightly linked in the US. (Sohn, 2017; Williams et al., 2016; Williams et al., 2010; Marrie et al., 2006) Even though our study attempted to evaluate their synergistic effects, we recognize that this may not be fully possible (though the large size of our study mitigates this to an extent). The correlation identified between lower SES and worse affective symptoms may be a bi-directional relationship; we did not evaluate the risk of lower SES conferred by the degree of affective symptomatology. Future longitudinal analyses would be better suited to address these issues. Additional limitations, which are inherent in such studies, include unidentified confounders and issues related to self-reported information. While a main strength of our study is the inclusion of a large cohort of individuals with MS, who are all receiving care in established MS clinics, it still may be that participants receiving care in these centers are different from those with MS who do not. Non-participants and non-responders may have contributed to selection bias in our study. Individuals from particularly low SES groups and those with more significant mental co-morbidities are likely to be more represented in these categories. (A need for greater reporting of socioeconomic status and race in clinical trials 2019; Gross et al., 2005; Svensson et al., 2012)
Identifying affective symptoms and mental comorbidities is important, as they are associated in longitudinal studies with worse disability and increased mortality in individuals with MS. (Marrie et al., 2009; McKay et al., 2018) Despite their prevalence in individuals with MS, these comorbidities are under-diagnosed and under-treated. (Marrie et al., 2009) Identifying individuals at risk is important in assessing for these comorbidities. Prior studies have shown that affective symptoms are particularly under-recognized in lower SES groups. (Marrie et al., 2009) SES is an important determinant of health out- comes, and incorporation of social stratification into future studies may help to allow for further exploration of the underlying mechanisms of these differences. (Isaacs and Schroeder, 2004; Williams et al., 2016; Marrie et al., 2008; Schroeder, 2007) Our study shows that among MS individuals, those in lower SES groups are at increased risk for affective symptomatology and potentially at risk for having mental comorbidities. In addition, we note that there are race-based differences in the presence of affective symptoms. Our study adds to the large body of evidence that SES and race are important determinants of health that should be recognized by clinicians, and importantly, that their role in MS outcomes should be addressed on a national level. (Williams et al., 2016; Schroeder, 2007)
Supplementary Material
Acknowledgments
Role of funding source
Study funding: The MS PATHS study is funded by Biogen. Biogen did not have a role in this analysis.
Footnotes
Declaration of Competing Interest
None.
Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.msard.2020.102010.
References
- Galobardes B, Shaw M, Lawlor DA, Lynch JW, 2006. Indicators of socioeconomic position (part 1). J. Epidemiol. Community Health 60, 7–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Isaacs SL, Schroeder SA, 2004. Class-the ignored determinant of the nation’s health. N. Engl. J. Med 351, 1137–1142. [DOI] [PubMed] [Google Scholar]
- Cunningham TJ, et al. , 2017. Vital signs: racial disparities in age-specific mortality among blacks or African Americans — United States, 1999–2015. MMWR Morb. Mortal. Wkly. Rep 66, 444–456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sohn H, 2017. Racial and ethnic disparities in health insurance coverage: dynamics of gaining and losing coverage over the life-course. Popul. Res. Policy Rev 36,181–201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Williams DR, Priest N, Anderson NB, 2016. Understanding associations among race, socioeconomic status, and health: patterns and prospects. Health Psychol. Off. J. Div. Health Psychol. Am. Psychol. Assoc 35, 407–411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Williams DR, Mohammed SA, Leavell J, Race, Collins C, 2010. socioeconomic status, and health: complexities, ongoing challenges, and research opportunities. Ann. N. Y. Acad. Sci 1186, 69–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harding KE, et al. , 2019. Socioeconomic status and disability progression in multiple sclerosis: a multinational study. Neurology 92, e1497–e1506. [DOI] [PubMed] [Google Scholar]
- Briggs FBS, Thompson NR, Conway DS, 2019. Prognostic factors of disability in relapsing remitting multiple sclerosis. Mult. Scler. Relat. Disord 30, 9–16. [DOI] [PubMed] [Google Scholar]
- Marrie RA, Cutter G, Tyry T, Vollmer T, Campagnolo D, 2006. Does multiple sclerosis-associated disability differ between races? Neurology 66, 1235–1240. [DOI] [PubMed] [Google Scholar]
- Kister I, et al. , 2010. Rapid disease course in African Americans with multiple sclerosis. Neurology 75, 217–223. [DOI] [PubMed] [Google Scholar]
- Weinstock-Guttman B, et al. , 2003. Multiple sclerosis characteristics in African American patients in the New York state multiple sclerosis consortium. Mult. Scler. Houndmills Basingstoke Engl 9, 293–298. [DOI] [PubMed] [Google Scholar]
- Kaufman MD, Johnson SK, Moyer D, Bivens J, Norton HJ, 2003. Severity and progression rate in African Americans compared with whites. Am. J. Phys. Med. Rehabil 82, 582–590. [DOI] [PubMed] [Google Scholar]
- Marrie RA, et al. , 2009. The burden of mental comorbidity in multiple sclerosis: fre- quent, underdiagnosed, and undertreated. Mult. Scler. Houndmills Basingstoke Engl 15, 385–392. [DOI] [PubMed] [Google Scholar]
- Marrie R, et al. , 2008. Comorbidity, socioeconomic status and multiple sclerosis. Mult. Scler. Houndmills Basingstoke Engl 14, 1091–1098. [DOI] [PubMed] [Google Scholar]
- Mohr DC, et al. , 1997. Treatment of depression improves adherence to interferon beta- 1b therapy for multiple sclerosis. Arch. Neurol 54, 531–533. [DOI] [PubMed] [Google Scholar]
- Feinstein A, 2002. An examination of suicidal intent in patients with multiple sclerosis. Neurology 59, 674–678. [DOI] [PubMed] [Google Scholar]
- McKay KA, et al. , 2018. Psychiatric comorbidity is associated with disability progression in multiple sclerosis. Neurology 90, e1316–e1323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rudick RA, et al. , 2014. The Multiple Sclerosis Performance Test (MSPT): an iPad-Based disability assessment tool. J. Vis. Exp. JoVE 10.3791/51318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Neuro-QoL. Available at: http://www.healthmeasures.net/explore-measurement-systems/neuro-qol (Accessed: 3rd June 2019), (2019).
- Miller DM, et al. , 2016. Validating Neuro-QoL short forms and targeted scales with persons who have multiple sclerosis. Mult. Scler. Houndmills Basingstoke Engl 22, 830–841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- van Buuren S, 2007. Multiple imputation of discrete and continuous data by fully con- ditional specification. Stat. Methods Med. Res 16, 219–242. [DOI] [PubMed] [Google Scholar]
- Rubin Donald, 2004. Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc. [Google Scholar]
- van Buuren, S., Groothuis-Oudshoorn K, 2011. mice: multivariate imputation by chained equations in R. J. Stat. Softw 45, 1–67. [Google Scholar]
- Hothorn T, Bretz F, Westfall P, 2008. Simultaneous inference in general parametric models. Biom. J. Biom. Z 50, 346–363. [DOI] [PubMed] [Google Scholar]
- Loucks EB, et al. , 2010. Life course socioeconomic position is associated with in- flammatory markers: the Framingham offspring study. Soc. Sci. Med 198271, 187–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tabassum F, et al. , 2008. Effects of socioeconomic position on inflammatory and he- mostatic markers: a life-course analysis in the 1958 British birth cohort. Am. J. Epidemiol 167, 1332–1341. [DOI] [PubMed] [Google Scholar]
- Miller GE, et al. , 2009. Low early-life social class leaves a biological residue manifested by decreased glucocorticoid and increased proinflammatory signaling. Proc. Natl. Acad. Sci. U. S. A 106, 14716–14721. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pampel FC, Krueger PM, Denney JT, 2010. Socioeconomic disparities in health behaviors. Annu. Rev. Sociol 36, 349–370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cree BAC, et al. , 2004. Clinical characteristics of African Americans vs Caucasian Americans with multiple sclerosis. Neurology 63, 2039–2045. [DOI] [PubMed] [Google Scholar]
- Nédélec Y, et al. , 2016. Genetic ancestry and natural selection drive population differ- ences in immune responses to pathogens. Cell 167, 657–669.e21. [DOI] [PubMed] [Google Scholar]
- Rossi S, et al. , 2017. Neuroinflammation drives anxiety and depression in relapsing-re- mitting multiple sclerosis. Neurology 89, 1338–1347. [DOI] [PubMed] [Google Scholar]
- Colasanti A, et al. , 2016. Hippocampal neuroinflammation, functional connectivity, and depressive symptoms in multiple sclerosis. Biol. Psychiatry 80, 62–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beatty PW, et al. , 2003. Access to health care services among people with chronic or disabling conditions: patterns and predictors. Arch. Phys. Med. Rehabil 84, 1417–1425. [DOI] [PubMed] [Google Scholar]
- Russell K, Jewell N, 1992. Cultural impact of health-care access: challenges for im- proving the health of African Americans. J. Community Health Nurs 9, 161–169. [DOI] [PubMed] [Google Scholar]
- Saadi A, Himmelstein DU, Woolhandler S, Mejia NI, 2017. Racial disparities in neurologic health care access and utilization in the United States. Neurology 88, 2268–2275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fabius CD, Thomas KS, Zhang T, Ogarek J, Shireman TI, 2018. Racial disparities in Medicaid home and community-based service utilization and expenditures among persons with multiple sclerosis. BMC Health Serv. Res 18, 773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- A need for greater reporting of socioeconomic status and race in clinical trials | Annals of the Rheumatic diseases (2019). Available at: https://ard.bmj.com/content/63/12/1700 (Accessed: 27th September 2019). [DOI] [PMC free article] [PubMed]
- Gross CP, Filardo G, Mayne ST, Krumholz HM, 2005. The impact of socioeconomic status and race on trial participation for older women with breast cancer. Cancer 103, 483–491. [DOI] [PubMed] [Google Scholar]
- Svensson K, Ramírez OF, Peres F, Barnett M, Claudio L, 2012. Socioeconomic determinants associated with willingness to participate in medical research among a diverse population. Contemp. Clin. Trials 33, 1197–1205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schroeder SA, 2007. We can do better — improving the health of the American people. N. Engl. J. Med 357, 1221–1228. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
MS PATHS data is currently only accessible to Biogen or participating healthcare institutions in the MS PATHS program.