Moderators of Exercise Effects on Depressive Symptoms in Multiple Sclerosis_ A Meta-regression

Matthew P. Herring, PhD,1,2 Karl M. Fleming, MSc,1 Sara P. Hayes, PhD,2,3Robert W. Motl, PhD,4 Susan B. Coote, PhD2,3

Context: This study examined the extent to which patient and trial characteristics moderate the effects of exercise on depressive symptoms among people with multiple sclerosis.

Evidence acquisition: Twenty-four effects were derived from 14 articles published before August 2016 located using Google Scholar, MEDLINE, PsycINFO, PubMed, and Web of Science. Trials involved 624 people with multiple sclerosis and included both randomization to exercise training or a non-exercise control condition and measurement of depressive symptoms at baseline and at mid- and/or post-intervention. Hedges’ d effect sizes were computed, study quality was assessed, and random effects models were used for all analyses. Meta-regression quantified the extent to which patient and trial characteristics moderated the estimated population effect. Analyses were completed in September 2016 and updated in February 2017.

Evidence synthesis: Exercise training significantly reduced depressive symptoms by a heteroge- neous mean effect Δ of 0.55 (95% CI1⁄40.31, 0.78, po0.001). Significant improvement in fatigue moderated the overall effect (β1⁄40.37, pr0.03). Significantly larger antidepressant effects resulted from trials in which exercise significantly improved fatigue (Δ1⁄41.04, 95% CI1⁄40.53, 1.55, k1⁄48) compared with no significant improvement in fatigue (Δ1⁄40.41, 95% CI1⁄40.21, 0.60, k1⁄414, z1⁄42.91, pr0.004).

Conclusions: Exercise significantly improves depressive symptoms among people with multiple sclerosis. Exercise-induced improvements in fatigue significantly moderated exercise effects on depressive symptoms. Future trials may benefit from focusing on using exercise to concurrently improve depressive symptoms and fatigue as a symptom cluster.

Am J Prev Med 2017;53(4):508–518 & 2017 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.

CONTEXT

Depressive symptoms are among the most com- mon and debilitating comorbidities experienced by people with multiple sclerosis (PwMS). Prevalence estimates have ranged between 27%1 and 50%,2 and successful treatment by traditional therapy (i.e., cognitive behavioral therapy or selective serotonin reuptake inhibitors) remains limited and elusive.3 Phys- ical inactivity is pervasive among PwMS and can con- tribute to physical and mental comorbidities, including elevated depressive symptoms.4 On the other hand, the salutary benefits of exercise are well established, and evidence supports the positive effects of exercise on depressive symptoms among primarily healthy adults5and adults with a diverse range of chronic illnesses.6Exercise training further improves symptoms among

adults with a depressive disorder,7,8 and improvements are comparable to other standard treatments.8

The positive effects of exercise training on physical and mental health among PwMS, including improved depres- sive symptoms, are well established.6,9–12 Previous meta- analytic reviews have supported the beneficial effects of exercise training on depressive symptoms among PwMS.

From the 1Department of Physical Education and Sport Sciences, Uni- versity of Limerick, Limerick, Ireland; 2Health Research Institute, Uni- versity of Limerick, Limerick, Ireland; 3Department of Clinical Therapies, University of Limerick, Limerick, Ireland; and 4Department of Physical Therapy, University of Alabama at Birmingham, Birmingham, Alabama

508 Am J Prev Med 2017;53(4):508–518 & 2017 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.

The aggregate effect sizes indicated small to moderate magnitude improvements in response to exercise ranging from 0.22 to 0.37.6,13,14 However, given issues of bias in the process of meta-analysis, recent literature has questioned the utility of focusing on an aggregated effect size alone.7,15

To that end, there is a critical need to identify potentially important independent and collective sources of variability in the overall effect of exercise on depressive symptoms among PwMS as a guide for future interventions. Patient and trial characteristics, particularly modifiable characteristics includ- ing disease severity, duration and dose of the exercise intervention, and fatigue, may be critical moderators of the effects of exercise training on depressive symptoms among PwMS, but have remained relatively understudied. Although exercise effects on fatigue have been well studied among PwMS,16,17 the possibility that significant improvements in fatigue may influence exercise effects on depressive symp- toms has not been studied in quantitative syntheses. Given the strong correlation between depressive symptoms and self- reported fatigue, and the well-established positive effects of exercise on both fatigue and depressive symptoms, inves- tigation of how exercise-induced changes in fatigue may moderate exercise-induced changes in depressive symptoms among PwMS is warranted. This further aligns with the idea of a symptom cluster, whereby management of one symptom, for example, depressive symptoms, might yield an improve- ment in a second symptom, namely fatigue, or vice versa.

Thus, the meta-regression analysis reported here used the results from RCTs to estimate the population effect size for exercise effects on depressive symptoms among PwMS, and to examine the extent to which patient and trial characteristics of theoretic or practical importance, includ- ing disease severity, features of the exercise intervention,6and exercise-induced reductions in fatigue, account for significant variation in the estimated population effect.

EVIDENCE ACQUISITION

The current systematic review with meta-analysis and meta- regression analysis was conducted in accordance with the recom- mendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) group.18

Data Sources and Searches

Articles published before August 2016 were located using Google Scholar, MEDLINE, PsycINFO, PubMed, and Web of Science. Keywords used included combinations of exercise, physical activity, depression, depressive, multiple sclerosis, randomized trial, andrandomized controlled trial. Supplemental searches of the articles retrieved were performed manually.

Study Selection/Inclusion Criteria

Inclusion criteria included: (1) English-language peer-reviewed publications; (2) adults aged Z18 years with a formal diagnosis of

MS; (3) randomized allocation to either an exercise intervention or a non-active control condition that lacked exercise training; and (4) a measure of depressive symptoms measured at baseline and at mid- and/or post-intervention. Investigations were excluded that (1) included exercise as one part of a multicomponent intervention but did not include the additional component in a comparison condition; or (2) compared exercise only with an active treatment.

Figure 1 provides a flowchart of study selection. Two authors (SH, KF) independently screened and then assessed the full text articles for eligibility. Where a disagreement over the eligibility of a study occurred, a third author (MH, SC) reviewed the paper and discussions occurred until consensus.

Data Extraction and Quality Assessment

Data were extracted from included articles into a file in SPSS, version 22.0, by three authors (SH, KF, MH) and cross-checked for accuracy. Extracted data included participant and trial character- istics, adherence and compliance, and exercise effects on outcomes of logical, theoretical, and/or prior empirical relation to depressive symptoms or exercise effects among PwMS. The quality of the included studies was assessed using the Physiotherapy Evidence Database checklist.19

Study Characteristics

Twenty-four effects were derived from 14 studies of 624 PwMS.20–33Depressive symptoms were not the primary outcome in any of the included trials. The most frequently investigated primary outcomes included fatigue,20,23,25,29 aerobic fitness,21,28 and aspects of walking performance (i.e., distance, speed, time).22,24,26 The mean age was 44.0 (SD=6.6) years. The mean percentage of women was 75% (SD=15%). Mean reported disease duration was 9.8 (SD=4.2) years, and mean baseline Expanded Disability Status Scale (EDSS) score was 3.4 (SD=1.2). Exercise training consisted of three (SD=1) sessions per week, 51 (SD=14) minutes per session, and 11 (SD=6) weeks in duration. Based on reported methods, PwMS were prescribed 122 (SD=38) minutes of exercise per week. The mean exercise training adherence rate was 85% (SD=15%) of prescribed sessions; adherence was reported for 16 of 24 (67%). Compliance with exercise prescription was not reported for any effects. The most frequently used measure of depressive symptoms was the Beck Depression Inventory (k=9).20,22,24,25,29 Though this review did not focus on exercise training effects among patients with a diagnosis of a depressive disorder, ten of 24 effects (41.7%) included depressive symptom scores high enough to suggest mild to moderate depres- sion based on cut scores commonly used for clinical screening.34–37

Effect Size Calculation

Hedges’ d effect sizes were calculated by subtracting the mean change in the comparison condition from the mean change in the exercise condition, and dividing this difference by the pooled SD of baseline scores.38 Effect sizes were adjusted for small sample size bias and calculated such that a larger reduction of depressive symptoms among exercising PwMS resulted in a positive effect size.38 Two-way (effects X raters) intraclass correlation coefficients for absolute agreement were calculated to examine inter-rater reliability for effect sizes. The initial intraclass correlation coef-ficients were 40.90; discrepancies were resolved by consensus, resulting in identical effects across extractors.

October 2017

Herring et al / Am J Prev Med 2017;53(4):508–518 509

510 Herring et al / Am J Prev Med 2017;53(4):508–518

Figure 1. Flowchart of study selection.MS, multiple sclerosis.

Data Synthesis and Analysis

Data analyses were completed in September 2016 and updated in February 2017. Meta-regression was used as the overall analysis of moderator effects. This technique reduces the probability of Type I error by computing concurrent estimates of independent effects by multiple moderators on the variation in effect size across trials. Random effects models were used with macros (SPSS MeanES, MetaReg) to aggregate mean effect size delta (Δ) and to test variation in effects according to moderator variables.38,39 Consis- tency and heterogeneity were evaluated with the I2 statistic and Qstatistic, respectively.40 If sampling error accounted for less than 75% of the observed variance, heterogeneity was indicated.38 The number of unpublished or unretrieved studies of null effect that would diminish the significance of observed effects to p40.05 was estimated as fail-safe Nþ.41

Primary Moderators

To provide focused research hypotheses about variation in effect size,42 five primary moderators were selected on the basis of logical, theoretical, or prior empirical relations to depressive symptoms and/or exercise effects on depressive symptoms among PwMS: age, disease severity (baseline EDSS), exercise program length, exercise session duration, and whether or not exercise significantly improved fatigue, based on calculated Hedges’ d and associated 95% CI (not based on statistical significance reported by authors in the original article).

Primary Moderator Analysis

Each of the moderators were coded according to planned contrasts43 among its levels. Primary moderators were included in mixed-effects multiple linear regression analysis with maximum likelihood estimation,38,39 adjusting for non-independence of multiple effects contributed by single studies42 and baseline depressive symptom severity.6 Tests of the regression model (Q[R]) and its residual error (Q[E]) were reported. Significant moderators in the regression analyses were decomposed using a random effects model to compute mean effect sizes and 95% CIs.39

Secondary Moderators

Secondary moderators were selected for descriptive, univariate analyses. Random effects models were used to calculate mean effect sizes (Δ) and 95% CIs for continuous and categorical variables.39 In addition, each continuous and categorical moder- ator was included in random effects univariate meta-regression analysis with maximum-likelihood estimation.38,39 Definitions for each moderator and associated levels are presented in Appendix Table 1 (available online).

EVIDENCE SYNTHESIS

Critical information for each of the 14 included studies was summarized using the Template for Intervention Description and Replication framework44 and is available

www.ajpmonline.org

Herring et al / Am J Prev Med 2017;53(4):508–518 511Table 1. Quality of Included Trials Assessed Using the PEDro Scale

PEDro, Physiotherapy Evidence Database.

in Appendix Tables 2 and 3 (available online). The quality of each of the included studies is presented inTable 1. Physiotherapy Evidence Database scores ranged from 4 to 7 with a mean of 5.79 (SD1⁄40.80).

Twenty-one of 24 effects (87.5%) were larger than zero. Following exercise training, mean scores for five of 24 (20.8%) effects suggested remission based on a mean score below suggested clinical cut scores.34–37 Based on a frequently used response threshold of Z50% reduction

in baseline score,45 reductions for four of 24 effects (16.7%) indicated significant response, with a mean reduction from baseline of 66.3%; however, the mean percentage reduction from baseline across all effects was approximately 26.5%. A forest plot of effects is presented in Figure 2. The mean effect size Δ was 0.55 (95% CI1⁄40.31, 0.78, z1⁄44.59, po0.001). The overall effect was heterogeneous (QT[23]1⁄454.82, po0.001). Sampling error accounted for 44.6% of the observed variance.

Figure 2. Forest plot of Hedges’ d effect sizes.October 2017

512 Herring et al / Am J Prev Med 2017;53(4):508–518

The effect was moderately consistent across studies (I21⁄459.9%, 95% CI1⁄449.6%, 68.1%). The fail-safe number of effects was 238, suggesting that 238 unpublished or unretrieved studies of null effect would be necessary to diminish the significance of the reported effect top40.05.

Overall Model

The overall meta-regression model was significantly related to effect size (QR[7]1⁄420.26, pr0.005, R21⁄40.47;QE[14]1⁄422.61, p1⁄40.07). A significant improvement in fatigue (β1⁄40.37, z1⁄42.21, pr0.03) accounted for signifi- cant variation in the overall effect of exercise on depressive symptoms. Significantly larger improvements in depressive symptoms resulted from trials in which exercise significantly improved fatigue (Δ1⁄41.04, 95% CI1⁄40.53, 1.55, k1⁄48) compared with trials in which fatigue was not significantly improved (Δ1⁄40.41, 95% CI1⁄40.21, 0.60, k1⁄414, z1⁄42.91, pr0.004). Age (β1⁄40.32), disease severity (β1⁄40.05), exercise program length (β1⁄40.26), and exercise session duration (β1⁄40.13) were not significantly related to effect size (all p40.15).

Univariate Meta-regression Analyses

Univariate meta-regression analyses (Table 2) showed that sex (β1⁄4 0.60, po0.001), age category (β1⁄40.45,pr0.01), exercise frequency (β1⁄40.38, pr0.044), signifi- cant improvement in fatigue (β1⁄40.52, po0.004), and total minutes of exercise prescribed per week (β1⁄40.42,po0.021) were significantly associated with exercise effects on depressive symptoms. Larger effects were derived from trials in which:

1. PwMS were aged 30–39 years (Δ1⁄40.95, 95% CI1⁄40.29, 1.61, k1⁄47) compared with PwMS aged 40–59 years (Δ1⁄40.34, 95% CI1⁄40.18, 0.51, k1⁄417, z1⁄42.57, pr0.01);
2. only female participants were included (Δ1⁄41.36, 95% CI1⁄40.44, 2.27, k1⁄44) compared with mixed samples of men and women (Δ1⁄40.37, 95% CI1⁄40.21, 0.54, k1⁄420,z1⁄4 3.77, po0.001);
3. session frequencies of Z3 days per week were used (Δ1⁄40.81, 95% CI1⁄40.43, 1.20, k1⁄412) compared with session frequencies of r2 days per week (Δ1⁄40.36, 95% CI1⁄40.13, 0.59, k1⁄411, z1⁄42.02, pr0.044); and
4. fatigue was significantly improved (Δ1⁄41.04, 95% CI1⁄40.53, 1.55, k1⁄48) compared with no significant improvement in fatigue (Δ1⁄40.41, 95% CI1⁄40.21, 0.60,z1⁄42.91, po0.004).
   Disease duration (β1⁄40.13, p40.53), baseline disease

severity (EDSS; β1⁄40.21, p40.34), exercise mode (β1⁄40.27, p40.16), whether or not baseline scores met

clinical cut scores indicative of mild to moderate depres- sion (β1⁄40.19, p40.08), depressive symptom measure used (β1⁄4 0.02, p40.93), type of control condition (β1⁄40.32, p40.10), program length (β1⁄40.32, p40.08), exercise session duration (β1⁄4 0.06, p40.77), social environment of the exercise session (β1⁄40.32, p40.08), blinded allocation (β1⁄4 0.36, pZ0.06), exercise adher- ence (β1⁄40.26, p40.40), a significant change in fitness (β1⁄40.01, p40.95), a significant change in the trial primary outcome (β1⁄40.35, p40.07), and change in depressive symptoms among control conditions (β1⁄40.35, p40.06) were not significantly associated with exercise effects on depressive symptoms.

The results of univariate moderator analyses for each primary and secondary moderators are presented inTable 2. For each level of each moderator, the number of effects (k), mean effect size (Δ) and 95% CI, and contrast p-value are provided.

DISCUSSION

Exercise training significantly improved depressive symptoms among PwMS by a moderate effect size Δ of 0.55, providing support for the efficacy of exercise to treat this prevalent and debilitating symptom among PwMS.1,2These findings are important, given the limited amount of evidence to support the success of traditional treat- ments for elevated depressive symptoms among PwMS, including cognitive behavioral therapy46 and pharmaco- therapy.3 Indeed, a recent systematic review and position stand called for new efforts on alternative approaches for managing depression in MS.3

The magnitude of this effect is slightly larger than previously reported effects of exercise on depressive symptoms among PwMS,6,13,14 due in part to a different sample of included studies, including two studies from which moderate20 and very large29 effects were derived, respectively, and the calculation of Hedges’ d and variance estimates used in random effects models.38,39The magnitude of this effect is consistent with the positive effects of exercise training on other related symptoms among PwMS, including fatigue,17 anxiety,12quality of life,47 and mobility.48 Moreover, the presentfindings are consistent with the antidepressant effects of exercise reported from reviews of clinically depressed patients,7,8 chronically ill adults,6 women with antenatal depression,49 and older adults.50

Regarding the clinical meaningfulness of the currentfindings, the review reported herein did not focus on exercise training effects among patients with a diagnosis of a depressive disorder. Nonetheless, 41.7% of effects came from symptom scores high enough to suggest a mild to moderate clinical elevation based on cut scores

www.ajpmonline.org

Herring et al / Am J Prev Med 2017;53(4):508–518 513Table 2. Summary of Univariate Moderator Analysis

October 2017

Note: Boldface indicates statistical significance (po0.05).
BDI, Beck Depression Inventory; CES-D, Center for Epidemiological StudiesDepression Scale; EDSS, Expanded Disability Status Scale; HADS-D, Hospital Anxiety and Depression ScaleDepression Scale; IDS-SR, Inventory of Depressive SymptomatologySelf Report; POMS-D, Profile of Mood StatesDepression; MDI, Major Depression Inventory; MS, multiple sclerosis; N/A, not applicable.

www.ajpmonline.org

commonly used for clinical screening.34–37 Although baseline depressive symptoms were not significantly associated with the overall effect of exercise training, the mean effect derived from studies in which baseline scores were indicative of mild to moderate depression (Δ=0.79) was non-significantly larger than effects from studies in which baseline scores were below suggested clinical cut scores (Δ=0.36). Following exercise training, mean scores for 20.8% of effects suggested remission based on a mean score below suggested clinical cut scores. Based on a frequently used response threshold ofZ50% reduction in baseline score,45 reductions for 16.7% indicated significant response, with a mean reduction from baseline of 66.3%; however, the mean percentage reduction from baseline across all effects was approximately 26.5%. Thus, the current findings suggest that the moderate-sized improvements reported here may also be clinically meaningful.

Meta-regression analysis was conducted to examine the extent to which potentially important patient and trial characteristics accounted for significant variation in the estimated population effect; this has not been systematically undertaken in previous meta-analyses. Univariate meta-regression showed that larger improve- ments in depressive symptoms were derived from trials in which patients were aged 30–39 years compared to 40–59 years, trials in which only female participants were included compared with mixed samples, trials in which session frequencies of 3 or more days per week were used, and trials in which fatigue was significantly improved compared to no significant improvement in fatigue. One potential limitation regarding interpretation of these univariate findings is that only four effect sizes were available for female-only samples and adherence was generally not well reported.

Importantly, while concurrently considering variation associated with other plausible sources of variability, including age, disease severity, exercise program length, exercise session duration, and baseline depressive symp- tom severity, a significant improvement in fatigue accounted for significant variation in the overall effect of exercise on depressive symptoms, such that signifi- cantly larger improvements in depression resulted from trials in which exercise significantly improved fatigue compared with trials in which fatigue was not signifi- cantly improved. These findings are consistent with previous evidence suggesting that treatment for depres- sion was associated with reductions in the severity of fatigue among PwMS.51 Potentially important partici- pant and trial characteristics within studies in which fatigue was significantly improved plausibly influenced these findings. Notably, though all comparisons were statistically non-significant (all p40.07), compared with

studies in which fatigue was not significantly improved, studies in which fatigue was significantly improved included a higher percentage of women (83.6% [SD1⁄418.6%] vs 71.3% [SD1⁄412.1%]), younger patients (40.7 [SD1⁄47.1] years vs 46.0 [SD1⁄46.2] years), patients with a lower disease duration (8.8 [SD1⁄44.3] years vs 11.0 [SD1⁄44.3] years), and patients with a lower baseline EDSS score (3.4 [SD1⁄41.0] vs 3.8 [SD1⁄41.2]). Moreover, though compliance cannot be rigorously evaluated based on reported information, these studies prescribed patients significantly more total minutes of exercise per week during the intervention (148.1 [SD1⁄436.2] minutes vs 106.1 [SD1⁄433.0] minutes, pr0.012). In addition, for six of the eight effects for which fatigue was significantly improved, a primary outcome of the trial was fatigue.

Though the current findings are, to the authors’knowledge, the first findings regarding this relationship in response to exercise among PwMS, the finding that improved fatigue is linked to improved depressive symptoms is not surprising. The association between depressive symptoms and fatigue is well documented,52including among PwMS.53 Similar neurobiological adap- tations plausibly underlie exercise effects on fatigue and depressive symptoms, including altered serotonin syn- thesis and metabolism.54–58 This is consistent with the notion of fatigue and depressive symptoms as part of a symptom cluster, and that treatment of one symptom is likely associated with improvement in a secondary out- come. This also suggests that targeting multiple, co- occurring symptoms might yield a larger improvement than targeting a single symptom.

Researchers continue to strive to understand the optimal frequency, intensity, time, and type of exercise for MS, and the findings of this meta-regression can inform the design of future studies comparing type and dose of exercise. The present findings showed that larger effects were derived from trials in which participants exercised 3 or more days per week and which prescribed higher total minutes of exercise per week. This is consistent with a systematic review of exercise for depression that found significant effects from studies using 3 days per week of aerobic exercise59 and with systematic reviews that found larger antidepressant effects from trials in which patients met physical activity guidelines.6,60 However, no other exercise parameters, including mode, session duration, and program length, were significantly related to the effect of exercise training on depressive symptoms among PwMS. The evidence on which these findings are based is somewhat limited. However, these initial findings suggest that the benefits of exercise training for depressive symptoms among PwMS may not depend on specific exercise parameters and provide support for PwMS to focus on meeting and

October 2017

Herring et al / Am J Prev Med 2017;53(4):508–518 515

516 Herring et al / Am J Prev Med 2017;53(4):508–518

surpassing suggested exercise guidelines61 for overall symptom benefits.

Limitations

A strength of this paper is that it adds to previous systematic reviews to give an updated estimate of the effect of exercise for depressive symptoms in PwMS. Additionally, through advanced statistical analysis using meta-regression, the authors have significantly advanced understanding of the patient and trial characteristics associated with variability in the effect of exercise on depressive symptoms. It is also important to note that the purpose of the meta-regression was to examine patient characteristics and features of exercise that could be modified to optimize the effect of exercise on depressive symptoms among PwMS. The purpose was not to test whether those factors might help to explain the effect of exercise. That purpose would require that trials assess plausible mediators of exercise effects, which was not the case in the included trials. A limitation is that not all studies reported the primary patient, exercise, and trial factors of interest; compliance to the prescribed exercise intervention; information regarding antidepressant use or engagement in psychotherapy; or provided informa- tion regarding long-term follow-up. In addition, though the overall effect of exercise did not significantly vary based on disease duration or baseline EDSS, given a mean reported disease duration of 9.8 (SD1⁄44.2) years and mean baseline EDSS score of 3.4 (SD1⁄41.2), the currentfindings may be limited by disease duration, disability status, and type of MS, such that the current results may not yet be broadly generalizable across MS.

Clinical Implications

This paper confirms the positive effects of exercise on depressive symptoms for PwMS and offers an evidence- based alternative with a moderate effect to the management of this prevalent and debilitating symptom. Further, the present findings suggest that significantly larger antidepres- sant effects may result from interventions where fatigue is also improved.

Future Research

Future studies should provide detailed descriptions of the participants and exercise parameters to enable future pooling of data to allow further exploration of the sources of variability in exercise effects on depressive symptoms among PwMS. Additionally, future trials should focus on clinically depressed PwMS, examine depression as a primary outcome, and investigate the comparative effec- tiveness of exercise with other empirically supported antidepressant treatments, including cognitive behavioral therapy and pharmacotherapy. The efficacy of exercise as

an augmentation, for first-line treatment approaches with limited success, should be investigated. Finally, the current findings highlight the need for future multi- factorial exercise trials in which critical symptoms among PwMS, including fatigue, depressive symptoms, anxiety symptoms, pain, and sleep, are concurrently targeted with exercise training.

CONCLUSIONS

Exercise training significantly improves depressive symp- toms among PwMS. Exercise-induced improvements in fatigue significantly moderated exercise effects on depressive symptoms. Future trials may benefit from focusing on using exercise to concurrently improve both depressive symptoms and fatigue as a symptom cluster.

ACKNOWLEDGMENTS

No financial disclosures were reported by the authors of this paper.

SUPPLEMENTAL MATERIAL

Supplemental materials associated with this article can be found in the online version at https://doi.org/10.1016/j. amepre.2017.04.011.

REFERENCES

1. Marrie RA, Reingold S, Cohen J, et al. The incidence and prevalence of psychiatric disorders in multiple sclerosis: a systematic review. Mult Scler. 2015;21(3):305–317. https://doi.org/10.1177/1352458514564487.
2. Feinstein A. Multiple sclerosis and depression. Mult Scler. 2011;17 (11):1276–1281. https://doi.org/10.1177/1352458511417835.
3. Minden SL, Feinstein A, Kalb RC, et al. Evidence-based guideline: assessment and management of psychiatric disorders in individuals with MS Report of the Guideline Development Subcommittee of the American Academy of Neurology. Neurology. 2014;82(2):174–181.https://doi.org/10.1212/WNL.0000000000000013.
4. Motl RW, McAuley E, Snook EM. Physical activity and multiple sclerosis: a meta-analysis. Mult Scler. 2005;11(4):459–463. https://doi. org/10.1191/1352458505ms1188oa.
5. Conn VS. Depressive symptom outcomes of physical activity inter- ventions: meta-analysis findings. Ann Behav Med. 2010;39(2):128–138.https://doi.org/10.1007/s12160-010-9172-x.
6. Herring MP, Puetz TW, O’Connor PJ, Dishman RK. Effect of exercise training on depressive symptoms among patients with a chronic illness: a systematic review and meta-analysis of randomized con- trolled trials. Arch Intern Med. 2012;172(2):101–111. https://doi.org/ 10.1001/archinternmed.2011.696.
7. Schuch FB, Vancampfort D, Richards J, Rosenbaum S, Ward PB, Stubbs B. Exercise as a treatment for depression: a meta-analysis adjusting for publication bias. J Psychiatr Res. 2016;77:42–51.https://doi.org/10.1016/j.jpsychires.2016.02.023.
8. Cooney GM, Dwan K, Greig CA, et al. Exercise for depression.Cochrane Database Syst Rev. 2013;9:CD004366. https://doi.org/10. 1002/14651858.CD004366.pub6.
9. Coote S, Hughes L, Rainsford G, Minogue C, Donnelly A. A pilot randomized trial of progressive resistance exercise augmented by neuromuscular electrical stimulation for people with multiple sclerosis who use walking aids. Arch Phys Med Rehabil. 2015;96(2):197–204.https://doi.org/10.1016/j.apmr.2014.09.021.

10. Garrett M, Hogan N, Larkin A, Saunders J, Jakeman P, Coote S. Exercise in the community for people with minimal gait impairment due to MS: an assessor-blind randomized controlled trial. Mult Scler. 2013;19(6):782–789. https://doi.org/10.1177/1352458512461966.
11. Coote S, Garrett M, Hogan N, Larkin A, Saunders J. Getting the balance right: a randomised controlled trial of physiotherapy and exercise interventions for ambulatory people with multiple sclerosis.BMC Neurol. 2009;9(1):34. https://doi.org/10.1186/1471-2377-9-34.
12. Herring MP, O’Connor PJ, Dishman RK. The effect of exercise training on anxiety symptoms among patients: a systematic review. Arch Intern Med. 2010;170(4):321–331. https://doi.org/10.1001/archinternmed. 2009.530.
13. Ensari I, Motl RW, Pilutti LA. Exercise training improves depressive symptoms in people with multiple sclerosis: results of a meta-analysis.J Psychiatr Res. 2014;76(6):465–471. https://doi.org/10.1016/j.jpsy- chores.2014.03.014.
14. Dalgas U, Stenager E, Sloth M, Stenager E. The effect of exercise on depressive symptoms in multiple sclerosis based on a meta‐analysis and critical review of the literature. Eur J Neurol. 2015;22(3):443-e34.https://doi.org/10.1111/ene.12576.
15. Ekkekakis P. Honey, I shrunk the pooled SMD! Guide to critical appraisal of systematic reviews and meta-analyses using the Cochrane review on exercise for depression as example. Ment Health Phys Act. 2015;8:21–36. https://doi.org/10.1016/j.mhpa.2014.12.001.
16. Andreasen A, Stenager E, Dalgas U. The effect of exercise therapy on fatigue in multiple sclerosis. Mult Scler. 2011;17(9):1041–1054. https: //doi.org/10.1177/1352458511401120.
17. Pilutti LA, Greenlee TA, Motl RW, Nickrent MS, Petruzzello SJ. Effects of exercise training on fatigue in multiple sclerosis: a meta-analysis.Psychosom Med. 2013;75(6):575–580. https://doi.org/10.1097/PSY. 0b013e31829b4525.
18. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009;151(4):264–269. https://doi.org/10.7326/0003-4819- 151-4-200908180-00135.
19. Maher CG, Sherrington C, Herbert RD, Moseley AM, Elkins M. Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther. 2003;83(8):713–721.
20. Ahmadi A, Arastoo AA, Nikbakht M, Zahednejad S, Rajabpour M. Comparison of the effect of 8 weeks aerobic and yoga training on ambulatory function, fatigue and mood status in MS patients. Iran Red Crescent Med J. 2013;15(6):449–454. https://doi.org/10.5812/ircmj. 3597.
21. Briken S, Gold S, Patra S, et al. Effects of exercise on fitness and cognition in progressive MS: a randomized, controlled pilot trial. Mult Scler. 2014;20(3):382–390. https://doi.org/10.1177/1352458513507358.
22. Cakit BD, Nacir B, Genç H, et al. Cycling progressive resistance training for people with multiple sclerosis: a randomized controlled study. Am J Phys Med Rehabil. 2010;89(6):446–457. https://doi.org/ 10.1097/PHM.0b013e3181d3e71f.
23. Dalgas U, Stenager E, Jakobsen J, et al. Fatigue, mood and quality of life improve in MS patients after progressive resistance training. Mult Scler. 2010;16(4):480–490. https://doi.org/10.1177/1352458509360040.
24. Dettmers C, Sulzmann M, Ruchay‐Plössl A, Gütler R, Vieten M. Endurance exercise improves walking distance in MS patients with fatigue. Acta Neurol Scand. 2009;120(4):251–257. https://doi.org/ 10.1111/j.1600-0404.2008.01152.x.
25. Hebert JR, Corboy JR, Manago MM, Schenkman M. Effects of vestibular rehabilitation on multiple sclerosisrelated fatigue and upright postural control: a randomized controlled trial. Phys Ther. 2011;91(8):1166–1183. https://doi.org/10.2522/ptj.20100399.

26. Learmonth Y, Paul L, Miller L, Mattison P, McFadyen A. The effects of a 12-week leisure centre-based, group exercise intervention for people moderately affected with multiple sclerosis: a randomized controlled pilot study. Clin Rehabil. 2012;26(7):579–593. https://doi.org/10.1177/ 0269215511423946.

27. Oken B, Kishiyama S, Zajdel D, et al. Randomized controlled trial of yoga and exercise in multiple sclerosis. Neurology. 2004;62(11): 2058–2064. https://doi.org/10.1212/01.WNL.0000129534.88602.5C.

28. Petajan JH, Gappmaier E, White AT, Spencer MK, Mino L, Hicks RW. Impact of aerobic training on fitness and quality of life in multiple sclerosis. Ann Neurol. 1996;39(4):432–441. https://doi.org/10.1002/ ana.410390405.

29. Razazian N, Yavari Z, Farnia V, et al. Exercising impacts on fatigue, depression, and paresthesia in female patients with multiple sclerosis.Med Sci Sports Exerc. 2016;48(5):796–803. https://doi.org/10.1249/ MSS.0000000000000834.

30. Romberg A, Virtanen A, Ruutiainen J. Long-term exercise improves functional impairment but not quality of life in multiple sclerosis.J Neurol. 2005;252(7):839–845. https://doi.org/10.1007/s00415-005- 0759-2.

31. Schulz K-H, Gold SM, Witte J, et al. Impact of aerobic training on immune-endocrine parameters, neurotrophic factors, quality of life and coordinative function in multiple sclerosis. J Neurol Sci. 2004;225 (1):11–18. https://doi.org/10.1016/j.jns.2004.06.009.

32. Skjerbæk AG, Næsby M, Lützen K, et al. Endurance training is feasible in severely disabled patients with progressive multiple sclerosis. Mult Scler. 2014;20(5):627–630. https://doi.org/10.1177/1352458513505351.

33. Sutherland G, Andersen MB, Stoové MA. Can aerobic exercise training affect health-related quality of life for people with multiple sclerosis?J Sport Exerc Psychol. 2001;23(2):122–135. https://doi.org/10.1123/ jsep.23.2.122.

34. Kendall PC, Hollon SD, Beck AT, Hammen CL, Ingram RE. Issues and recommendations regarding use of the Beck Depression Inventory.Cognit Ther Res. 1987;11(3):289–299. https://doi.org/10.1007/BF011 86280.

35. Weissman MM, Sholomskas D, Pottenger M, Prusoff BA, Locke BZ. Assessing depressive symptoms in five psychiatric populations: a validation study. Am J Epidemiol. 1977;106(3):203–214. https://doi. org/10.1093/oxfordjournals.aje.a112455.

36. Crawford J, Henry J, Crombie C, Taylor E. Normative data for the HADS from a large non‐clinical sample. Br J Clin Psychol. 2001;40 (4):429–434. https://doi.org/10.1348/014466501163904.

37. Rush AJ, Bernstein IH, Trivedi MH, et al. An evaluation of the quick inventory of depressive symptomatology and the hamilton rating scale for depression: a sequenced treatment alternatives to relieve depression trial report. Biol Psychiatry. 2006;59(6):493–501. https://doi.org/ 10.1016/j.biopsych.2005.08.022.

38. Hedges LV, Olkin I. Statistical Models for Meta-analysis. New York, NY: Academic Press, 1985.

39. Lipsey MW, Wilson DB. Practical Meta-analysis. Thousand Oaks, CA: Sage Publications, 2001.

40. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–560.https://doi.org/10.1136/bmj.327.7414.557.

41. Rosenberg MS. The file-drawer problem revisited: a general weighted method for calculating fail-safe numbers in meta-analysis. Evolution. 2005;59(2):464–468. https://doi.org/10.1111/j.0014-3820.2005.tb01004.x.

42. Rosenthal R, DiMatteo MR. Meta-analysis: recent developments in quantitative methods for literature reviews. Annu Rev Psychol. 2001;52 (1):59–82. https://doi.org/10.1146/annurev.psych.52.1.59.

43. Rosenthal R. Meta-analytic Procedures for Social Research. London, UK: Sage, 1991. https://doi.org/10.4135/9781412984997.

44. Hoffmann TC, Glasziou PP, Boutron I, et al. Better reporting of interventions: Template for Intervention Description and Replication

October 2017

Herring et al / Am J Prev Med 2017;53(4):508–518 517

518 Herring et al / Am J Prev Med 2017;53(4):508–518

(TIDieR) checklist and guide. BMJ. 2014;348:g1687. https://doi.org/

10.1136/bmj.g1687.

45. Rush AJ, Trivedi MH, Ibrahim HM, et al. The 16-Item Quick Inventory
    of Depressive Symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): a psychometric evaluation in patients with chronic major depression. Biol Psychiatry. 2003;54(5):573–583.https://doi.org/10.1016/S0006-3223(02)01866-8.
46. Hind D, Cotter J, Thake A, et al. Cognitive behavioural therapy for the treatment of depression in people with multiple sclerosis: a systematic review and meta-analysis. BMC Psychiatry. 2014;14:5. https://doi.org/ 10.1186/1471-244X-14-5.
47. Motl RW, Gosney JL. Effect of exercise training on quality of life in multiple sclerosis: a meta-analysis. Mult Scler. 2008;14(1):129–135.https://doi.org/10.1177/1352458507080464.
48. Snook EM, Motl RW. Effect of exercise training on walking mobility in multiple sclerosis: a meta-analysis. Neurorehabil Neural Repair. 2009;23(2):108–116. https://doi.org/10.1177/1545968308320641.
49. Daley A, Foster L, Long G, et al. The effectiveness of exercise for the prevention and treatment of antenatal depression: systematic review with meta‐analysis. BJOG. 2015;122(1):57–63. https://doi.org/10.1111/ 1471-0528.12909.
50. Catalan-Matamoros D, Gomez-Conesa A, Stubbs B, Vancampfort D. Exercise improves depressive symptoms in older adults: an umbrella review of systematic reviews and meta-analyses. Psychiatry Res. 2016;244:202–209. https://doi.org/10.1016/j.psychres.2016.07.028.
51. Mohr DC, Hart SL, Goldberg A. Effects of treatment for depression on fatigue in multiple sclerosis. Psychosom Med. 2003;65(4):542–547.https://doi.org/10.1097/01.PSY.0000074757.11682.96.
52. American Psychiatry Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5s). Washington, DC: American Psychiatric Association, 2013.

53. Feinstein A, Magalhaes S, Richard J-F, Audet B, Moore C. The link between multiple sclerosis and depression. Nat Rev Neurol. 2014;10 (9):507–517. https://doi.org/10.1038/nrneurol.2014.139.

54. Chaouloff F. Effects of acute physical exercise on central serotonergic systems. Med Sci Sports Exerc. 1997;29(1):58–62. https://doi.org/ 10.1097/00005768-199701000-00009.

55. Davis JM, Bailey SP, Woods JA, Galiano FJ, Hamilton MT, Bartoli WP. Effects of carbohydrate feedings on plasma free tryptophan and branched-chain amino acids during prolonged cycling. Eur J Appl Physiol Occup Physiol. 1992;65(6):513–519. https://doi.org/10.1007/ BF00602357.

56. Dishman RK. Brain monoamines, exercise, and behavioral stress: animal models. Med Sci Sports Exerc. 1997;29(1):63–74. https://doi. org/10.1097/00005768-199701000-00010.

57. Melancon MO, Lorrain D, Dionne IJ. Exercise increases tryptophan availability to the brain in older men age 57–70 years. Med Sci Sports Exerc. 2012;44(5):881–887. https://doi.org/10.1249/MSS.0b013e31823ede8e.

58. Rethorst CD, Landers DM, Nagoshi CT, Ross JTD. Efficacy of exercise in reducing depressive symptoms across 5-HTTLPR genotypes.Med Sci Sports Exerc. 2010;42(11):2141–2147. https://doi.org/10.1249/ MSS.0b013e3181de7d51.

59. Stanton R, Reaburn P. Exercise and the treatment of depression: a review of the exercise program variables. J Sci Med Sport. 2014;17 (2):177–182. https://doi.org/10.1016/j.jsams.2013.03.010.

60. Adamson BC, Yang Y, Motl RW. Association between compliance with physical activity guidelines, sedentary behavior and depressive symptoms.Prev Med. 2016;91:152–157. https://doi.org/10.1016/j.ypmed.2016.08.020.

61. Latimer-Cheung AE, Ginis KAM, Hicks AL, et al. Development of evidence-informed physical activity guidelines for adults with multiple sclerosis. Arch Phys Med Rehabil. 2013;94(9):1829–1836.e7. https://doi. org/10.1016/j.apmr.2013.05.015.