Sleep problems are common in spinal cord injury (SCI) and multiple sclerosis (MS). However, the degree to which sleep problems differ between these distinct clinical populations has not been evaluated. In this study, we examined self-reported sleep problems in individuals with SCI and those with MS, and compared these clinical groups on the Medical Outcomes Study Sleep Scale (MOS-SS).
Participants were 1,677 individuals (SCI = 581; MS = 1,096) enrolled in longitudinal study of self-reported health outcomes in SCI and MS. Univariate and multivariate analysis of covariance examined group differences on global sleep problems and domain-specific subscales of the MOS-SS.
Individuals with SCI reported an average of 30 fewer min of sleep per night (Sleep Quantity subscale) and significantly greater difficulty initiating and maintaining sleep (Sleep Disturbance subscale) compared to individuals with MS. However, groups did not differ on global sleep problems (Sleep Problems Index 9).
Although global sleep problems are more common in SCI and MS than in the general population, these groups exhibit differing sleep problem profiles, and thus may require unique treatment approaches to address the specific domains of sleep affected. For individuals with SCI, an additional focus on increasing sleep quantity and reducing sleep disruptions may be warranted.
Fogelberg DJ, Hughes AJ, Vitiello MV, Hoffman JM, Amtmann D. Comparison of sleep problems in individuals with spinal cord injury and multiple sclerosis. J Clin Sleep Med 2016;12(5):695–701.
Sleep problems are an important public health concern affecting an estimated 50 to 70 million adults in the United States.1 Sleep problems can substantially diminish health-related quality of life and are associated with poor physical and mental health outcomes, including depression, chronic pain, fatigue, obesity, hypertension, endocrine dysfunction, heart disease, cognitive dysfunction, and stroke.2–7 Sleep problems are especially prevalent in individuals living with neurological conditions, including spinal cord injury (SCI) and multiple sclerosis (MS).8–13
There are a number of neurological factors associated with SCI and/or MS that may contribute to sleep problems in these populations.10,12,14 In SCI, traumatic and nontraumatic injuries can sever key neural signaling pathways responsible for regulating and maintaining sleep.10,15 Similarly in MS, demyelination and neurodegeneration16–19 throughout the brain and spinal cord can damage neural structures and axonal projections that may be involved in sleep and sleep related disorders such as periodic limb movement20 and sleep related breathing disorders.21 For both SCI and MS, damage to the brain or cervical spinal cord can decrease secretion of melatonin, a neurohormone important for initiating sleep and maintaining circadian sleep-wake cycles.15,22 SCI and MS can also lead to autonomic disturbances that may adversely affect sleep.23,24 Damage to descending inhibitory neural pathways can cause spasticity and disruptive movements, including restless legs and periodic limb movement syndromes.24–26 These disruptive movements are common in SCI27 and MS25 and have the potential to delay sleep and interfere with sleep maintenance. Problems with bladder and bowel function are also common in SCI and MS and can lead to frequent nocturnal awakenings.28,29 Additionally, weakened breathing muscles, alterations in breathing mechanics, infrequent shifts in sleeping position, and antispasmodic medications (e.g., baclofen) are all associated with sleep related breathing problems (e.g., sleep apnea, snoring), which are particularly common in SCI and can also affect people with MS.9,30
Current Knowledge/Study Rationale: Individuals with neurological conditions such as spinal cord injury (SCI) and multiple sclerosis (MS) are at increased risk for developing sleep problems. However, as SCI and MS typically involve disparate disease courses and etiologies, it is plausible that sleep problem profiles may differ between these groups. The current study sought to compare SCI and MS groups on global and subscale scores of the MOS-SS, a validated, self-report measure of sleep.
Study Impact: Our findings show that although SCI and MS groups report similar overall levels of sleep problems, individuals with SCI report achieving significantly fewer hours of sleep and greater levels of sleep disruption. Results suggest that interventions aimed at increasing sleep duration may be more helpful for individuals with SCI, and that fatigue and daytime somnolence management interventions may be more appropriate for individuals with MS.
Despite the prevalence and scope of sleep problems in SCI and MS, research has only recently begun to focus on the assessment of sleep problems in these populations. Self-report sleep measures provide subjective estimations of sleep quantity, number of awakenings, and sleep latency, which are essential to the interpretation of actigraphy, and other objective sleep measures and can be easily used across time unlike polysomnography. Moreover, self-report measures can provide supplemental information on individuals' perceptions of other important aspects of sleep such as sleep quality and sleep related daytime dysfunction.
In a recent study31 we examined two self-report tools for assessing sleep difficulties in a related sample of individuals with SCI and MS, which included the Medical Outcomes Study Sleep Scale (MOS-SS),32,33 a commonly used self-report measure that provides a global index score for overall sleep problems, as well as six subscales that capture a number of qualitative and quantitative aspects of sleep. In that study, mean scores on the MOS-SS global index were significantly worse for SCI and MS cohorts relative to the published general population norms.31 However, differences between SCI and MS cohorts were not investigated. Given the differing disease courses and etiologies, it is possible that sleep problems may differ between these groups and require unique treatment foci.
Examination of MOS-SS subscales in the aforementioned study suggested that although SCI and MS share a number of clinical features that can negatively affect sleep, these groups may exhibit different profiles of sleep problems. For example, although both groups endorsed significantly poorer sleep overall relative to the general population, the MS sample reported longer average sleep durations. Currently, it is unclear whether SCI and MS groups significantly differ from each other on sleep quantity and other aspects of sleep. Additional analyses are needed to better understand shared and unique aspects of sleep problems in these populations.
To extend our previous findings, the objective of this study was to compare the large SCI and MS samples on the global index and subscales of the MOS-SS. Such information will contribute to more effective, targeted, and empirically supported interventions for sleep problems in these neurorehabilitation populations.
Participants with SCI were recruited through the Northwest Regional Spinal Cord Injury Model System at the University of Washington (Seattle, WA) and the Shepherd Center at the Virginia Crawford Research Institute (Atlanta, GA). Participants with MS were recruited through the Greater Northwest Chapter of the National MS Society (Seattle, WA). All participants were at least 18 y of age at enrollment and reported a definitive diagnosis of either SCI or MS.
Data for this study were collected during the baseline assessment of a longitudinal study of self-reported health outcomes in SCI and MS. All procedures were approved by the University of Washington Human Subjects Division (Institutional Review Board). Participants were recruited through web and print advertisements and through mailed invitations (SCI = 2,408; MS = 7,806). Recruitment procedures yielded a total of 837 eligible SCI (34.8% response rate) and 1,597 eligible MS (20.5% response rate) respondents. Eligible participants were mailed an initial survey to be completed and returned online or on paper. Nonresponders received a reminder letter after 1 mo, and a telephone reminder after an additional 2 w. Completed paper surveys were checked for missing data upon return and participants were called up to a total of four times to collect missing data. If contact was not made, a brief letter was sent asking the participant to contact the study office.
Self-reported sleep was assessed using the MOS-SS.33 The MOS-SS consists of 12 Likert-type items assessing six dimensions of sleep problems: Sleep Disturbance (i.e., difficulty initiating or maintaining sleep), Snoring, Respiratory Problems, Sleep Quantity, Sleep Adequacy, and Daytime Somnolence. The MOS-SS instructs respondents to rate their sleep during the previous 4-w period. Scores for the Sleep Disturbance, Snoring, Respiratory Problems, Sleep Adequacy, and Daytime Somnolence dimensions range from 0 to 100. For all subscales, except Sleep Adequacy and Sleep Quantity, higher scores indicate worse sleep. The Sleep Quantity dimension is reported as average nightly sleep time. The MOS-SS also yields a global sleep index score, the Sleep Problems Index 9, which draws from 9 of the 12 items. The MOS-SS demonstrates good psychometric properties, and has been evaluated in large population studies, both with nationally representative32 and chronically ill samples.34 Those studies also provided normative scores for referenced groups.32–34
Self-reported demographics and clinical variables were collected from each participant. Variables included age (in years), time since diagnosis/injury (in years), SCI severity including level and whether the injury was complete/incomplete (coded as C1-C4 complete and incomplete, C5-C8 complete and incomplete, T1-S4/5 complete and incomplete) or MS type (relapsing remitting, secondary progressive, primary progressive, or progressive relapsing), sex, and highest level of education achieved (coded as “High School Graduate or Less,” “Some College/College Graduate,” or “Graduate School or Higher”), race and ethnicity, marital status (married/committed partnership, divorced/separated, never married, widowed), living situation (living alone, with family/friends, in a nursing/adult family home), and hours of weekly caregiver assistance. For those with MS, the mobility section of the self-report version of the Expanded Disability Status Scale (EDSS) was used as a measure of disease severity.35 The self-report version correlates highly with the physician scored version,36 EDSS scores were stratified into minimal severity (≤ 4.0), intermediate severity (4.5–6.5), and advanced severity (≥ 7.0).
Distributions and residual plots of demographics, clinical variables, and MOS-SS scores were examined to assess for assumptions of normality and skewness. Simple descriptive statistics were calculated to summarize demographics and clinical characteristics (Table 1). Given that individuals with SCI were more likely to be male and younger at the time of injury relative to individuals with MS, age, sex, and time since diagnosis/injury were selected as covariates to be included in the primary analyses. For the primary analyses, a one-way analysis of covariance (ANCOVA) was performed to compare groups on the MOS-SS Sleep Index 9. Additionally, a one-way multivariate analysis of covariance (MANCOVA) was conducted to identify differences between SCI and MS groups on the six MOS-SS subscales. If the multivariate effect for diagnostic group based on Wilks Λ was significant, follow-up univariate effects were examined to determine which MOS-SS variables differed as a function of diagnostic group.
Demographics and clinical characteristics for spinal cord injury and multiple sclerosis groups
Demographics and clinical characteristics for spinal cord injury and multiple sclerosis groups
Completed data were obtained from 1,677 participants (SCI = 581; MS = 1,096). Descriptive statistics and relevant group differences on demographics and clinical variables are presented in Table 1. The majority of participants lived in the community (99% with SCI and 97.4% with MS) and required less than 40 h of caregiver assistance per week (74.5% with SCI and 93% with MS). Consistent with prior research, individuals with SCI were younger, had less time since injury/ diagnosis, and were more likely to be male than individuals with MS (all ps < 0.01). In addition, individuals with SCI had less education, were more likely to be nonwhite and Hispanic, less likely to be married, live alone, and require more hours of caregiver assistance than those with MS.
Age, sex, and time since diagnosis/injury also emerged as significant covariates on ANCOVA (age: F(1, 1672) = 23.48, η2 = 0.01, p < 0.001; sex: F(1, 1672) = 17.37, η2 = 0.01, p < 0.001; time since injury/diagnosis: F(1, 1672) = 5.35, η2 = 0.01, p < 0.01) and MANCOVA (age: F(6, 1667) = 11.06, Wilks Λ = 0.96, η2 = 0.04, p < 0.001; sex: F(6, 1667) = 9.91, Wilks Λ = 0.97, η2 = 0.03, p < 0.001; time since injury/diagnosis: F(6, 1667) = 2.98, Wilks Λ = 0.99, η2 = 0.01, p < 0.01) analyses.
After controlling for age, sex, and time since diagnosis/injury, no significant differences were observed between groups on mean MOS-SS Sleep Index 9 scores (F(1, 1672) = 0.46, p = 0.50). However, the multivariate effect of diagnostic group was significant (F(6, 1667) = 5.19, Wilks Λ = 0.98, η2 = 0.02, p < 0.001), indicating a significant difference between SCI and MS groups on at least one of the six MOS-SS sub-scales. Follow-up univariate tests showed that SCI and MS groups differed significantly with regard to the Sleep Disturbance (F(1, 1672) = 4.93, p = 0.03) and Sleep Quantity (F(1, 1672) = 14.60, p < 0.001) dimensions (see Table 2).
Multivariate analysis of covariance and analysis of covariance results comparing spinal cord injury and multiple sclerosis cohorts.
Multivariate analysis of covariance and analysis of covariance results comparing spinal cord injury and multiple sclerosis cohorts.
Mean scores on the sleep disturbance subscale, adjusted for covariates, were significantly greater for the SCI group, indicating greater difficulty initiating or maintaining sleep than the MS group. Mean scores on the Sleep Quantity dimension, adjusted for covariates, were significantly lower for the SCI group, indicating fewer hours of sleep than the MS group. Cohen d effect sizes were calculated to assess the magnitude of these differences. Effect sizes were small, at 0.12, for Sleep Disturbance and moderate, at 0.30, for Sleep Quantity. As indicated in Table 2, no significant group differences emerged for Snoring, Respiratory Problems, Sleep Adequacy, or Daytime Somnolence (all ps > 0.05). Effect sizes were small for all nonsignificant MOS-SS variables.
To our knowledge, the current study is the first to compare self-reported sleep problems between individuals with SCI and individuals with MS using the MOS-SS, a validated measure of self-reported sleep commonly used in neurorehabilitation populations. Although a previous analysis of data collected from the same longitudinal study found that SCI and MS groups evidenced poorer sleep on the MOS-SS relative to the general population, additional research examining differences in sleep profiles between SCI and MS were indicated to better inform clinical interventions for these groups. Group differences were examined after controlling for age, sex, and time since injury/diagnosis.
Examination of the six dimensions of the MOS-SS found two statistically significant differences that may be relevant for designing and implementing clinical interventions to improve sleep in these populations. First, individuals with SCI reported significantly greater levels of sleep disturbance (i.e., difficulty initiating or maintaining sleep) than individuals with MS. However, given the small effect size and the lack of agreed-upon thresholds for clinically meaningful differences on this MOS-SS subscale, it remains unclear whether group differences on the Sleep Disturbance dimension reflect a clinically meaningful difference. Nonetheless, the current results support the notion that difficulties initiating and maintaining sleep may be prevalent among individuals with SCI and thus may be an area of additional clinical focus. Second, individuals with SCI reported an average sleep duration of 6.6 hours per night— substantially fewer hours than were reported by the MS group. Although an optimal sleep duration for individuals with SCI has not been established, the average amount of nighttime sleep reported by the SCI group fell short of the recommended minimum of 7 h of nighttime sleep for healthy adults to reduce health risks associated with chronic inadequate sleep.37 In addition to statistical significance, the effect size for this group difference (0.30) was moderate in strength.
Despite achieving longer sleep durations than individuals with SCI, individuals with MS reported similar levels of daytime somnolence, sleep adequacy, and overall sleep problems. Results suggested that the additional sleep achieved by individuals with MS did not necessarily improve other aspects of their sleep and daily functioning, and that sleep may not be sufficiently restorative in MS. The observed profile is consistent with increased rates of daytime fatigue and sleepiness in MS,38 and may be related to use of either prescribed or over-the-counter sleep medications, both of which are highly prevalent in this population.14,39 Thus, although sleep quantity is an important aspect of sleep, future interventions for both of these groups, and particularly for individuals with MS, may benefit from a focus on strategies to change sleep quality.
Our ability to control for relevant covariates assists us in understanding diagnostic-specific differences that exist between individuals with SCI and MS. For example, men generally report greater problems with snoring and respiratory-related sleep problems (e.g., sleep apnea) than women.40 Without controlling for unequal proportions of men and women in SCI and MS groups, group differences on the Snoring dimension of the MOS-SS would be misleading. After controlling for sex, group differences on the Snoring dimension were nonsignificant. Thus, snoring does not appear to be a sleep related problem unique to individuals with SCI, but rather males in both groups. Clinical approaches to reducing snoring (e.g., weight loss, smoking cessation, adjusting sleep position) are likely to be helpful for both of these diagnostic groups, particularly males in these groups. Additionally, the use of the MOS-SS was another strength of this study. The MOS-SS has been previously evaluated in SCI and MS,11,41 and has been shown to effectively discriminate these groups from the general population.31
Data in the current study came from community-based self-report surveys. Thus, results may be sensitive to selection and response biases. However, the use of self-report surveys allowed for the inclusion of a larger sample from a wider range of geographic areas. Self-reported and objective sleep measures systematically differ from one another,42,43 and we did not collect objective measures of sleep in this study. The MOS-SS does not include a measure of daytime sleep, which may be present in both of the clinical groups that were surveyed. We also gathered limited data on medical comorbidities and medication use, both of which may affect sleep. Second, this study was cross-sectional, which limits conclusions regarding causal inferences. However, this limitation was outweighed by the benefits offered by the large sample available at the first time point of our longitudinal study. Attrition at subsequent time points was subject to some bias, as indicated in another cross-sectional assessment conducted at the fourth time point.14 Third, although the MOS-SS has previously been used to examine sleep in those with SCI and MS, it does not yield sufficiently detailed information about sleep problems to enable definitive differentiation between specific sleep disorders (e.g., insomnia, circadian rhythm disorders, movement disorders, or sleep related breathing disorders) that may have contributed to the sleep difficulties described by participants. Proper diagnosis of such disorders would be key to determining the appropriate clinical intervention for any given individual. However, use of the MOS-SS can help screen for symptoms suggestive of these disorders to inform more definitive evaluations and treatment strategies.
In summary, the current study suggests important differences between individuals with SCI and MS in types of sleep problems experienced, but additional work is needed to further characterize sleep profiles, ideally using both self-report and objective measures of sleep such as polysomnography and actigraphy, as well as identify potential mediating and moderating variables. Although other clinical features may be relevant to sleep problems (e.g., depression, pain) these were specifically not included in the current analyses. Because sleep is a diagnostic symptom of depression, inclusion of a depression measure would add redundancy to the model and introduce additional bias. In addition, research has increasingly demonstrated that pain and depression may be the consequence of sleep problems. Thus, interventions aimed at targeting sleep symptoms may help improve both depression and fatigue. For the current study, we chose to focus primarily on the types of sleep symptoms, rather than causes or consequences, of sleep problems in SCI and MS and further research in these areas is needed.
The increased recognition of sleep's critical role in maintaining health, function, and quality of life has highlighted the importance of developing a better understanding of sleep in neurorehabilitation populations such as SCI and MS. Although global sleep problems, as measured by the Sleep Problems Index 9, were the same in these two groups and worse than the normative population, individuals with SCI reported achieving less sleep and having greater difficulty initiating and maintaining sleep. Thus, individuals with SCI or MS may require different treatment approaches for sleep problems, depending on aspects of sleep affected. Understanding these differences can help frame future research and can inform the development of more effective treatments for sleep problems in these populations
This was not an industry supported study. The contents of this article were developed under grants from the Department of Education (NIDRR H133B031129 & H133B080025) and the National Institutes of Health (NIAMS 5U01AR052171 and NICHD K01HD076183). The contents of this paper do not represent the view of the National Institutes of Health or the Department of Education. In addition, the work reported in this manuscript was supported by a mentor-based fellowship grant from the National Multiple Sclerosis Society (MB 0026). The authors have indicated no financial conflicts of interest. The research reported in this manuscript was completed at the University of Washington.