Accepted Manuscript The effect of exercise on depressive symptoms in adults with neurological disorders: A systematic review and meta-analysis Brynn C. Adamson, MS, Ipek Ensari, EdM, Robert W. Motl, PhD PII:
S0003-9993(15)00009-X
DOI:
10.1016/j.apmr.2015.01.005
Reference:
YAPMR 56081
To appear in:
ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION
Received Date: 28 October 2014 Revised Date:
5 January 2015
Accepted Date: 5 January 2015
Please cite this article as: Adamson BC, Ensari I, Motl RW, The effect of exercise on depressive symptoms in adults with neurological disorders: A systematic review and meta-analysis, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2015), doi: 10.1016/j.apmr.2015.01.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
RI PT
EXERCISE AND NEUROLOGICAL DISORDERS
SC
The effect of exercise on depressive symptoms in adults with neurological disorders: A systematic review and meta-analysis
M AN U
Brynn C. Adamson, MS, Ipek Ensari, EdM, Robert W. Motl, PhD
TE D
University of Illinois Urbana-Champaign, Department of Kinesiology and Community Health, Urbana, IL.
EP
Corresponding author: Brynn C. Adamson, MS, Department of Kinesiology and Community Health, University of Illinois, 233 Freer Hall, Urbana, IL 61801, USA, phone: (217) 244-1191, fax: (217) 244-7332, e-mail:
[email protected] AC C
Requests for reprints should be addressed to the corresponding author. We certify that no party having a direct interest in the results of the research supporting this article has or will confer a benefit on us or on any organization with which we are associated AND, if applicable, we certify that all financial and material support for this research (eg, NIH or NHS grants) and work are clearly identified in the title page of the manuscript.
1
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
1 2 3 4 5 6
The effect of exercise on depressive symptoms in adults with neurological disorders: A systematic review and meta-analysis
7
neurological disorders. We searched CINAHL, the Cochrane Register of Controlled Clinical
8
Trials, EMBASE, ERIC, MEDLINE, PsycINFO, PubMed, and SPORTDiscus with the last
9
search performed in May 2014.
Abstract
SC
RI PT
Objective: To review and quantify the effect of exercise on depression in adults with
Study Selection: We included randomized controlled trials conducted in adults diagnosed with a
11
neurological disorder. We included studies which compared an exercise intervention group with
12
a control group and used depression as an outcome measure.
13
Data Extraction: Depression data were extracted independently by 2 authors. Methodological
14
quality was assessed independently by 2 authors.
15
Data Synthesis: Forty-three full-length articles were reviewed and 26 trials met our inclusion
16
criteria. These trials represented 1,324 participants with 7 different neurological disorders;
17
Alzheimer’s disease (n=4), migraine (n=1), multiple sclerosis (n=13), Parkinson’s disease (n=2),
18
spinal cord injury (n=1), stroke (n=2) and traumatic brain injury (n=3). Depression measure data
19
were extracted and effect sizes were computed for 23 trials. Results from a meta-analysis yielded
20
an overall effect size of 0.28 (SE=0.07, CI=0.15-0.41, p=0.00) favoring a reduction in depression
21
outcomes following an exercise intervention compared with control. Of note, interventions
22
which met physical activity guidelines yielded an overall effect of 0.38 compared with 0.19 for
23
studies which did not meet physical activity guidelines.
24
Conclusions: This review provides evidence that exercise, particularly when meeting physical
25
activity guidelines, can improve depressive symptoms in adults with neurological disorders.
AC C
EP
TE D
M AN U
10
26 1
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
Key Words: meta-analysis, neurological disorder, depression, mood, exercise
RI PT
Abbreviations: MS (multiple sclerosis), AD (Alzheimer’s disease), BDI (Beck Depression Inventory), HADS (Hospital Anxiety and Depression Scale), SCI (Spinal Cord Injury), CBT (Cognitive Behavior Therapy), PEDro (Physiotherapy Evidence Database), ES (effect size), CI (confidence interval), PAGs (physical activity guidelines), MHR (maximum heart rate), RPE (rating of perceived exertion), VO2 max (maximal oxygen consumption), MADRS (Montgomery-Asberg Depression Rating Scale), BDI-II (Beck Depression Inventory II), CES-D (Center for Epidemiologic Studies Depression Scale), CSDD (Cornell Scale for Depression in Dementia), GDS (Geriatric Depression Scale), IDS-SR (Inventory of Depressive Symptomology-Self Report), MDI (Major Depression Inventory), POMS (Profile of Mood States).
SC
27 28 29 30 31 32 33 34 35 36 37 38 39
M AN U
40
Depression and depressive symptoms are common in the general population of adults,
41
and are particularly common in adults with neurological disorders (i.e., diseases or disorders of
43
the nervous system). Lifetime prevalence of major depressive disorder in the U.S. population is
44
19.1%, and the rate is even higher among adults with neurological disorders. For example,
45
lifetime prevalence for depression is approximately 50% in both multiple sclerosis1 (MS) and
46
Alzheimer’s disease2,3 (AD). Studies of depressive symptoms report higher scores on depressive
47
outcomes in populations with neurological disorders than in the general population. For example,
48
one community-based study compared depressive symptoms based on the Beck Depression
49
Inventory (BDI) among persons with Parkinson’s disease and two age-matched control groups,
50
one composed of healthy older adults and one composed of older adults with diabetes mellitus.
51
The mean score on the BDI was 12.8 among participants with Parkinson’s disease, 7.9 in
52
participants with diabetes, and 5.9 among healthy older adult controls4. In a large web-based
53
survey of depression in 4178 people with MS, the mean Hospital Anxiety and Depression Scale
54
(HADS) score was 7.6, whereas the mean score in a UK population reference group was only
55
3.75.
AC C
EP
TE D
42
2
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
56
Depression and its symptoms may be associated with various adverse consequences in persons with neurological disorders. Such consequences include loss of independence, and even
58
mortality. For example, depression has been associated with a higher rate of premature mortality
59
in Alzheimer’s patients 2. In recent spinal cord injury (SCI) patients undergoing rehabilitation,
60
depression has been associated with fewer functional improvements, decreased mobility and
61
decreased functional independence6,7. Persons with migraine and comorbid depression
62
experience decreased quality of life, and restriction of activities 8,9. Depression has been
63
associated with decreased functional recovery poststroke10. Depression further has been
64
associated with poor treatment compliance in people with MS11.
SC
M AN U
65
RI PT
57
Pharmacological treatment for depression in populations with neurological disorders yields conflicting results. Antidepressant medications have improved depressive symptoms and
67
other outcomes in persons with stroke10. By comparison, one review of pharmacological
68
treatment of depression in MS reported a lack of consistent evidence for improvement of
69
depression following drug treatment12. One large-scale, double-blind trial compared the effects
70
of two different antidepressant medications (sertraline and mirtazapine) on depressive outcomes
71
in clinically depressed persons with AD, and reported no clinical effectiveness of the drugs
72
compared with a placebo13. Other forms of treatment such as cognitive behavior therapy (CBT)
73
have similarly produced equivocal results in several populations. For example, one meta-analysis
74
of depression treatment methods following SCI reported positive reactions to CBT, but the
75
included trials lacked the methodological rigor necessary to attribute depression improvement to
76
CBT14. One systematic review of both pharmacological and psychosocial treatment of depression
77
following TBI concluded that there is insufficient evidence regarding the effects of
78
antidepressant medication and psychotherapeutic methods15. Interestingly, in migraine patients,
AC C
EP
TE D
66
3
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
79
antidepressants have been used successfully not only for treatment of comorbid depression but in
80
migraine prophylaxis as well12. Adverse side effects of antidepressant medication are commonly
81
reported especially in the MS and AD populations12,13. Exercise is defined as any planned, repetitive bodily movement done with the goal of
RI PT
82
increasing fitness and is an effective behavioral approach for reducing depressive symptoms and
84
treating major depressive disorder in otherwise healthy adults16,17. When comparing exercise
85
with a control group, one meta-analysis of 23 trials yielded an effect size of −0.82 in favor of
86
exercise reducing depressive symptoms16. Another meta-analysis of 17 trials with clinically
87
depressed participants reported that the exercise groups had a 1 standard deviation reduction in
88
depression scores compared with control groups17. There is additional evidence that exercise
89
training is comparable with antidepressant medication (sertraline) in its effects on depression and
90
depressive symptomology18. Similarly, a Cochrane review of the effects of exercise training on
91
depression reported no differential improvement in depressive symptoms for those receiving
92
exercise and those receiving CBT16.
M AN U
TE D
93
SC
83
Multiple exercise intervention trials conducted in populations of adults with neurological disorders have measured depressive symptoms as an outcome. These trials, unfortunately, yield
95
conflicting results even when considering populations with the same neurological disorder. For
96
example, two separate 4-month long trials conducted among participants with Alzheimer’s
97
disease yielded effect sizes of −0.0619 and 0.3920. Both of these interventions consisted of
98
walking and a comprehensive set of strength, balance and flexibility exercises. There is similar
99
variability in effect sizes for trials conducted among participants diagnosed with MS 21,22.
100 101
AC C
EP
94
We undertook a systematic review and meta-analysis of exercise training and depression in adults with neurological disorders based on the evidence for improvement of depression and
4
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
depressive symptoms following exercise interventions in the general population, and the
103
conflicting evidence of its effects on depression in adults with neurological disorders. This
104
systematic review and meta-analysis serves to determine the efficacy of exercise interventions
105
for improving depressive outcomes in persons with various neurological disorders. We further
106
provide information of the safety of exercise training in neurological disorders based on reported
107
adverse events and the differences in efficacy of interventions which met or did not meet
108
Physical Activity Guidelines for Americans23.
111
Search Strategy
M AN U
110
SC
Methods
109
RI PT
102
We followed protocol outlined by the Meta-analysis of Observational Studies in Epidemiology(MOOSE)24 and the procedure outlined by McDonnell et al25 in a review of
113
aerobic exercise and cognitive function in the same population. We searched the following
114
databases: CINAHL, the Cochrane Register of Controlled Clinical Trials, EMBASE, ERIC,
115
MEDLINE, PsycINFO, PubMed, and SPORTDiscus with no date limits. The key terms searched
116
involved several possible acronyms for neurological conditions (e.g. "Nervous System
117
Diseases", "Autoimmune Diseases of the Nervous System") and several possible acronyms for
118
exercise/physical activity (e.g. "Exercise", "Physical Fitness", "Resistance Training"). Although
119
our aim was to include exercise interventions only (as opposed to physical activity
120
interventions), we used “physical activity” as a search term so as not to miss any trials that may
121
have included physical activity as a keyword. We further focused on exercise for consistency in
122
comparison with other meta-analyses in adults16,26 and persons with MS21,22. “Depression” or
123
“depressive symptoms” were not used as search words so as to avoid potentially excluding any
AC C
EP
TE D
112
5
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
124
eligible studies which did not use depression as a key term if it was not a primary outcome
125
measure of the study. See appendix 1 for an example of the search terms.
126
Following initial searches, PubMed was searched again with each condition name(i.e. Parkinson’s disease, multiple sclerosis, migraine etc.) as a medical subject heading (MeSH term)
128
with a combination of exercise related MeSH terms in order to find any other studies which were
129
not included in the first search. Reference lists of the studies which fit the following inclusion
130
criteria were also searched for any additional studies.
SC
131
RI PT
127
Following preliminary searching, irrelevant studies based on title and abstract were initially removed, as well as studies with no mention of measuring depression. Remaining
133
articles were reviewed to identify the articles that met our inclusion criteria. The last search was
134
performed in May 2014.
135
Inclusion Criteria
136
Types of publications: Randomized controlled trials which measured depressive symptoms as
137
one of the outcomes before and after an exercise program was delivered were included.
138
Language limits were set to English and no date limits were set.
139
Sample: Adults 18 years of age and older who were diagnosed with any neurological disorder
140
were included. The following conditions were present in the searches: Alzheimer’s disease or
141
dementia, fibromyalgia, migraine, MS, Parkinson’s disease, rheumatoid arthritis, SCI, and
142
traumatic brain injury. We did not include fibromyalgia or rheumatoid arthritis as they are not
143
considered by the National Institute of Neurological Disorders to be conditions neurological in
144
nature and thus were excluded based on diagnosis27. The databases we searched use the age
145
filter of Adult (19+ years of age), but upon removing the age filter, no additional studies were
146
located that included those aged 18+.
AC C
EP
TE D
M AN U
132
6
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
Characteristics of the exercise program: no limits as to the type of exercise were set, therefore
148
the following exercise modalities were included: aerobic exercise, resistance training, balance
149
training, yoga, and others involving a combination. To be included, the intervention had to
150
compare at least one group receiving only an exercise intervention to a control group. Some
151
studies used several different exercise modalities and compared them to each other and to a
152
control group. Studies of this sort were included. No limits concerning frequency, intensity or
153
duration of the exercise intervention were set.
154
Outcomes: Studies must have included at least one measure of depressive symptoms or a clinical
155
diagnosis of depression measured before and after the exercise intervention.
156 157 158
Data Collection and Analysis
159
were made after both authors reached agreement. A third reviewer was consulted on studies in
160
question.
M AN U
SC
RI PT
147
TE D
Potentially relevant studies were reviewed by 2 reviewers and decisions on inclusion
For descriptive purposes, the quality of studies included in the meta-analysis was
162
determined using the 11-item Physiotherapy Evidence Database (PEDro) scale28. Two reviewers
163
independently scored the quality of the included trials. This scale contains 11 questions about
164
randomization, allocation, comparability of baseline scores in different groups, blinding, dropout
165
rate, and statistical analyses used (including intention to treat, between-groups measures and
166
measures of variability). The PEDro scale has been reported to be a valid and reliable measure of
167
methodologic quality28. Additionally, we searched the text of the included articles for
168
information about adverse events which occurred during the interventions.
AC C
EP
161
169
We computed effect sizes (ESs) expressed as Cohen’s d 29. To do this, we computed the
170
mean change of depressive symptom outcome measures from before to after the intervention of
7
ACCEPTED MANUSCRIPT EXERCISE AND NEUROLOGICAL DISORDERS
the exercise training group and subtracted the mean change of the control group. The resulting
172
difference in mean change between groups was then divided by the pooled baseline standard
173
deviation (SD) for the experimental and control groups. We contacted authors of studies with
174
unreported depression outcome measure data via email to supply missing information. The ESs
175
were calculated so that a positive ES indicated an improvement in mean depression scores after
176
exercise training, whereas a negative ES indicated a worsening of mean depression scores in the
177
exercise-training group compared with control. Separate ESs were calculated for each outcome
178
measure of depression presented as well as for each separate exercise training condition if more
179
than one was included. We completed the analysis using a single ES per study (i.e., an average
180
ES when there was more than one ES computed by the software) as multiple ESs from the same
181
study are not independent 30. The lack of independence biases the standard error (SE) for judging
182
the significance of the overall ES and multiple ESs from one study bias the overall ES
183
disproportionately compared with studies that have a single ES 30.
TE D
M AN U
SC
RI PT
171
The ESs along with the associated standard errors (SE) were entered into the
185
Comprehensive Meta-analysis software (Version 2.0, Biostat, Englewood, New Jersey). We
186
computed the overall or mean ES using a random-effects model. This model assumes that the
187
samples come from populations with different ESs and the true effect differs between studies 31.
188
We computed a 95% confidence interval (CI) around the mean ES. An overall Q value and I²
189
value were calculated to test for homogeneity of variance among ESs. The Q value is a measure
190
of variance among the ESs and a statistically significant (p