Loyola University Maryland

Emerging Scholars

Katherine Dunn, Jagriti (Jackie) Bhattarai, Ph.D., Abbey J. Hughes, Ph.D.

Title: Insomnia and Fatigue in Multiple Sclerosis: A Subgroup Meta-analysis Comparing the PSQI, ISI, and MOS

Authors and Institutions: KM Dunn, J Bhattarai, AJ Hughes

Johns Hopkins University School of Medicine, Dept. of Physical Medicine and Rehabilitation

Loyola University Maryland, Department of Psychology

Introduction: Insomnia is reported in over half of individuals with multiple sclerosis (MS) and is thought to be a contributing factor to fatigue in this population. However, associations between insomnia and fatigue vary widely across studies. The present study featured meta-analytic moderation analyses to examine whether associations between insomnia and fatigue in MS differ as a function of the insomnia measure employed. Three insomnia measures were compared with respect to their associations with fatigue: the Pittsburgh Sleep Quality Index (PSQI), the Insomnia Severity Scale (ISI), and the Medical Outcome Survey (MOS) Sleep scale.

Methods: Studies were identified using broad search terms (“sleep,” “fatigue,” and “MS”) in five electronic databases (PubMed, PsycINFO, EMBASE, Cochrane and CINAHL). Of the 1,576 articles screened, 22 (n = 2,553) provided sufficient data to obtain bivariate correlations, odds ratios, and/or standardized mean differences for the relationship between insomnia and fatigue. Effect sizes were transformed to a common metric (Fischer’s z), and weighted by variance. Categorical moderation analyses were performed to assess mean effect sizes between and within each insomnia measure.

Results: Pearson’s r correlations between measures of insomnia and fatigue ranged from .07 to .72, with a mean of .37 (95% CI [.29, .45]). Subgroup analyses yielded moderate effects for the PSQI (Zr = .40, 95% CI [.31, .49], p < .001) and ISI (Zr = .44, 95% CI [.28, .61], p < .001), and small effects for the MOS (Zr = .17, 95% CI [.05, .28], p = .003). Moderation analyses revealed overall differences in effect sizes between the PSQI, ISI, and MOS (Q = 22.48, p < .001), with the PSQI and ISI yielding greater associations with fatigue than the MOS (PSQI: Q = 21.16, p < .001; ISI: Q = 15.43, p = .001).

Conclusion: Associations between insomnia and fatigue in MS differ according to which insomnia measure is used. In terms of clinical utility, the PSQI and ISI may more reliably predict fatigue in MS relative to the MOS. Given that analyses were limited to global insomnia measures, future research is needed to determine if different aspects of insomnia (e.g., onset, maintenance, early morning waking) are differentially associated with fatigue.

Support: JB is supported by the NMSS Mentor-Based Fellowship (MB 0032). AH is supported by the National Institutes of Health (K23HD086154).