Associations among stressors across the lifespan, disability, and relapses in adults with multiple sclerosis

Abstract Introduction Stress and adversity during childhood, adolescence, and adulthood could impact the present and future health and well‐being of people with multiple sclerosis (PwMS); however, a lifespan approach and nuanced stressor data are scarce in this nascent area of research. Our aim was to examine relationships among comprehensively measured lifetime stressors and two self‐reported MS outcomes: (1) disability and (2) relapse burden changes since COVID‐19 onset. Methods Cross‐sectional data were collected from a nationally distributed survey of U.S.‐based adults with MS. Hierarchical block regressions were used to sequentially evaluate contributions to both outcomes independently. Likelihood ratio (LR) tests and Akaike information criterion (AIC) were used to evaluate additional predictive variance and model fit. Results A total of 713 participants informed either outcome. Most respondents (84%) were female, 79% had relapsing remitting multiple sclerosis (MS), and mean (SD) age was 49 (12.7) years. Childhood (R 2 = .261, p < .001; AIC = 1063, LR p < .05) and adulthood stressors (R 2 = .2725, p < .001, AIC = 1051, LR p < .001) contributed significantly to disability, above and beyond prior nested models. Only adulthood stressors (R 2 = .0534, p < .001; AIC = 1572, LR p < .01) significantly contributed above the nested model for relapse burden changes since COVID‐19. Conclusions Stressors across the lifespan are commonly reported in PwMS and could contribute to disease burden. Incorporating this perspective into the “lived experience with MS” could facilitate personalized health care by addressing key stress‐related exposures and inform intervention research to improve well‐being.


INTRODUCTION
Multiple sclerosis (MS), an immune-mediated inflammatory disease, is the leading cause of nontraumatic disability in young adults and is the second highest with respect to direct medical expenses associated with a chronic illness in the United States (Hartung, 2021). The potential link between childhood stress and MS disease burden has been a burgeoning area of research; however, relationships between stress and MS pathogenesis and progression remain unclear due to mixed and insufficient evidence (Polick et al., 2022). For example, there are two available studies regarding disability. Spitzer et al. (2012) did not find a relationship between childhood trauma and adult disability in people with multiple sclerosis (PwMS). Although Horton et al. (2022) did initially identify higher odds of walking aid dependence in PwMS who experienced four or more adverse childhood experiences (ACEs), after adjusting for multiple testing comparisons, the significance of this finding became nonsignificant. Childhood physical and sexual abuse have been significantly associated with higher relapse rates in PwMS, but these relationships have only been explored in one study (Spitzer et al., 2012).
Limitations to the measurement of ACEs, which is count based and lacks nuance to assess severity, and to study designs (e.g., limited sample size, covariates) likely contribute to these conflicting results. For example, none of the studies in a recent systematic review (n = 12) controlled for disease-modifying therapies, and only two studies accounted for smoking/substance use (Goodwin & Stein, 2004;Riise et al., 2011), which can both alter disability and relapses (Polick et al., 2022). Yet, evidence suggests that traumatic childhood stressors including abuse, neglect, and household dysfunction could alter brain development, cause aberrant inflammatory responses, increase the perception of adult stress, and lead to maladaptive lifestyle behaviors that increase the risk of MS or disease progression (e.g., smoking, obesity) (Albott et al., 2018;Centers for Disease Control & Prevention, 2022;McEwen, 2017). Acute stress responses can progress to chronic stress, through amygdala-related pathways that sensitize to threat detection over the lifespan (McEwen, 2017). Thus, to inform efforts that address disease burden and treatment strategy, it is important to expand beyond an exclusive focus on childhood to incorporate adult stressors.
Few studies have used a lifespan approach to ascertain the cumulative effect of stressors on MS, and to the best of our knowledge, none have focused on physical symptoms of disability or relapse burden (Polick et al., 2022). The purpose of this study is to characterize relationships among comprehensively measured lifetime stressor exposures (e.g., child, adult, severity, expanded criteria), and two clinically relevant MS outcomes: disability severity and relapse burden changes since COVID-19 onset. Likert scale) of 55 stressors over the lifespan (Slavich & Shields, 2018).

Self
Examples include physical/sexual abuse, neglect, poverty, divorce, neighborhood violence, and witnessing domestic abuse. Stressors occurring before age 18 were classified as childhood stressors. The STRAIN was validated with an autoimmune sample and has shown good test-retest reliability over time (Slavich & Shields, 2018). disability (Learmonth et al., 2013). The PDDS has been validated for PwMS and strongly correlates with clinical findings (Learmonth et al., 2013).
Hierarchical block ordinal logistic regression modeling in STATA was used to sequentially assess baseline demographic and covariate contributions to each outcome. Covariates shown in Tables 2 and 3 were tailored per each outcome, encompassing known risk factors (e.g., smoking) and factors hypothesized to be relevant (e.g., seasonality).
Next, we compared model fit of the base model to our second model, which included the base predictors plus childhood stressor predictors to determine if these early stressors contribute to MS outcomes over and above the base model. Finally, we added a third set of predictors evaluating the contributions of adult stressors to determine if they contributed additional predictive variance overall. Each successive model was compared to the prior model using likelihood ratio (LR) testing, and we report Akaike information criterion (AIC) as an index of relative model fit, with a lower number indicating better fit. Analyses were powered to detect 4% changes between models. Due to case-wise deletion, only participants with complete cases per outcome were included. It is important to note that this hierarchical block strategy assumes collinearity between similar stressors clustered together in model blocks to represent a latent concept (e.g., childhood stress), and thus may make individual stressor variables appear unremarkable. Therefore, interpretation is conducted at the model level to provide foundational evidence of relationships in this developing area of research.

Disability
The base model including demographic and MS history predictors contributed significantly to estimates of disability level (R 2 = .256, p < .001; model AIC = 1066) ( Table 2)

DISCUSSION
Our findings reveal associations among (1) childhood and adulthood stressors and MS disability level, and (2) adult stressors and MS relapse burden change since COVID-19. Contrary to previous studies (Horton et al., 2022;Spitzer et al., 2012), our findings support associations between childhood stressors and disability, possibly due to more comprehensive measurement (e.g., severity, more stressors). Our study found that birth seasonality impacted disability. Seasonality is more conventionally included in analyses of MS incidence risk and, until now, has not been included in studies assessing childhood or lifetime stressors and MS symptoms.
Variation in vitamin D absorption and its effects on immunomodulation during fetal immune system development in utero could help TA B L E 2 Final analytic hierarchical model of disability using ordinal logistic regression (n = 695).  (Pust et al., 2020), and psychiatric comorbidity (Wan et al., 2022). Further, ACE research has been growing in many adjacent areas of symptom severity such as functional neurological disorder (Paredes-Echeverri et al., 2023), Tourette's syn-drome (Yang et al., 2022), frequent headaches (Anto et al., 2021), and Parkinson's disease (Subramanian et al., 2023). A lifespan approach may similarly increase nuance in these areas as well as MS.

Limitations
Limitations include cross-sectional and subjective self-report data, which could be subject to recall bias, from a small portion of the 80,000 PwMS on the National MS Society listserv. Although a convenience sample, this approach had the advantage of reaching a large TA B L E 3 Final analytic hierarchical model of relapse burden change since COVID-19 onset, using ordinal logistic regression (n = 668). Abbreviations: AIC, Akaike information criterion; CI, confidence interval; LR, likelihood ratio; MS, multiple sclerosis; OR, odds ratio.
national U.S. sample of PwMS, as previous work has been conducted with smaller and/or clinic-based samples, so this is a noted departure.
This approach may also introduce selection bias toward people who have, and can navigate, internet access, which may not be inclusive of those with lower functionality or socioeconomic status.

CONCLUSIONS
This study highlights how stressors across the lifespan can help conceptualize when stressors matter most, and how nuanced measurement (e.g., count vs. severity) may play critical roles for MS disability level and relapse burden. These findings could help inform research design and analysis in this growing field to mitigate or prevent adverse outcomes associated with stressors across the lifespan. This work may also