Clinical Implications
• Antidepressants are currently the accepted treatment of choice for major depressive disorder, but a fair percentage of patients do not achieve remission with first-line drugs, requiring combination/augmentation treatments that increase the risk of side effects and discontinuation.
• A number of studies carried out in the past showed that exercise may reduce depressive symptoms, but systematic reviews and meta-analysis findings, while generally positive, continued to be inconclusive and debated, and focused on the efficacy of exercise as a sole treatment for depression.
• A systematic analysis of the literature suggests that exercise should be regarded as an adjunctive strategy with antidepressants, rather than a stand-alone treatment per se, reflecting more closely daily clinical practice.
Background
Major depressive disorder (MDD) is currently ranked 11th among the leading causes of disease burden and is ranked as the second leading cause of years lived with disability globally.Reference Murray, Vos and Lozano1 It is projected over the next 20 years to be the second leading cause of disability worldwide and the leading cause of disability in high-income nations.Reference Mathers and Loncar2 Antidepressants are currently the accepted treatment of choice for MDDReference Hollon, Thase and Markowitz3 and are among the most commonly prescribed drugs in Western countries, with a point prevalence of yearly use of 6% in the community in France,Reference Gasquet, Nègre-Pagès and Fourrier4 4.7% in Italy,Reference Carta, Aguglia and Bocchetta5 4.9% in the UK,Reference Hamer, Batty, Seldenrijk and Kivimaki6 and 10.1% in U.S. households.Reference Olfson and Marcus7
Newer antidepressants, such as selective serotonin reuptake inhibitors (SSRIs) and serotonin–norepinephrine reuptake inhibitors (SNRIs), have become the first-line treatment for MDD because of a lower risk of severe side effects, higher tolerability and safety of use, and an efficacy similar to that of older antidepressant drugsReference Gartlehner, Hansen and Thieda8; nevertheless, only 47% of patients respond, and only 33% of patients achieve remission in first-line of treatment with citalopram, one of the most frequently used SSRIs.Reference Trivedi, Rush and Wisniewski9 In the sequenced treatment alternatives to relieve depression (STAR*D) trial, the cumulative remission rate of treatment-resistant unipolar major depression, after 4 treatment levels, increased to 67%,Reference Rush, Trivedi and Wisniewski10 meaning that fewer than 50% of treatment-resistant patients do not achieve remission. Combination therapies or augmentation strategies (with lithium salts, triiodothyronine, bupropion, lamotrigine, psychostimulants, and atypical antipsychotic drugs) significantly increase the risk of drug–drug interactions, side effects, and drop-outs, and around one-third of treatment-resistant patients continue to be resistant.Reference Thase, Howland and Friedman11–Reference Al-Harbi13
All of these issues emphasize the need for nonpharmacologic, complementary, and adjunctive therapies, which should be both effective and safe for depressed patients, and free of drug–drug interactions. Exercise is potentially one such therapy.
In the last 30 years, a number of studies have shown that physical activity may reduce depressive symptoms in healthy populations,Reference Blumenthal, Emery and Madden14–Reference King, Taylor and Haskell17 in patients diagnosed with MDD,Reference Klein, Greist, Gurman and Neiberyer18–Reference Singh, Clements and Singh22 and those affected by depressive comorbidity in association with physical conditions, such as Parkinson's disease,Reference Goodwin, Richards, Taylor, Taylor and Campbell23 Alzheimer's disease,Reference Williams and Tappen24 multiple sclerosis,Reference Motl and Pilutti25 cardiovascular diseases,Reference Ades and Coello26 cancer,Reference Newton and Galvão27 and metabolic syndrome–related disorders.Reference Pedersen and Saltin28
Depression negatively impacts quality of life (QoL), a dimension concerning patients’ overall perspectives about their physical, social, and psychological status.Reference Kennedy, Eisfeld and Cooke29 Physical activity has been shown to improve the perceived physical QoL in depressed patients,Reference Carta, Hardoy and Pilu30–Reference Patten, Williams, Lavorato and Bulloch32 and high doses of exercise have been found to be related to larger improvements both in mental and physical aspects of QoL.Reference Martin, Church, Thompson, Earnest and Blair33
It has been hypothesized that exercise could act on depression through a number of biological mechanisms similar to those reported for antidepressant drugs, such as an increase in the level of endorphin and serotonin, an enhancement of the neurotransmission mediated by norepinephrine, a decrease in the level of cortisol, and a promotion of neuropeptides related to neurotrophism, such as brain derived neurotrophic factor (BDNF), VGF, insulin-like growth factor 1 (IGF-1), and vascular endothelial growth factor (VEGF).Reference Helmich, Latini and Sigwalt34 Moreover, exercise exerts a beneficial effect on depressive symptoms by providing the possibility of diversion from negative thinking, a sense of purpose, and enhancement of social contact.Reference Searle, Calnan and Lewis35 In this sense, it seems reasonable to state that exercise can be considered a biopsychosocial treatment for MDD.Reference Lopresti, Hood and Drummond36
While a number of literature reviews and meta-analyses have been published recently on the efficacy of exercise on depressive disorder,Reference Blake31, Reference Searle, Calnan and Lewis35, Reference Blake, Mo, Malik and Thomas37–Reference Mura and Carta51 to the best of our knowledge no specific analysis has been conducted on the efficacy of exercise as an adjunctive treatment in patients who are undergoing antidepressant therapy. Moreover, the majority of the reviews that have been published sought to determine whether exercise could be regarded as an alternative to antidepressants, which is unlikely. Because antidepressants are a core treatment for MDD, and because they are often not fully effective, an adjunctive role for exercise in depression treatment is likely to be its most important role.
Objective
We carried out a systematic review of the literature to establish the current state of knowledge on the efficacy of exercise as an adjunctive treatment to antidepressant drug therapy.
Method
Identification of the studies
The search for relevant articles was carried out in PubMed/Medline, Google Scholar, and Scopus with the following keywords: “exercise,” “physical activity,” “physical fitness,” “depressive disorder,” “depression,” “depressive symptoms,” “add-on,” and “adjunction.” We included articles in all languages that were published from January 1980 to April 2013. The search was expanded to examine the bibliographies of the selected papers. In the cases when we needed further information, the authors of the articles were contacted and asked to clarify methodological concerns or to provide unpublished data.
Inclusion criteria
Studies were included in this review if they were randomized or non-randomized controlled trials, in which exercise as adjunctive with antidepressants was compared to standard treatments (including antidepressant drugs), no treatment, or placebo-control condition, in people of all ages with depression (diagnosed by any method) as defined by the authors. We excluded observational and cohort studies, those that compared different types of exercise without a “non-exercising” control group, those that measured outcomes immediately before and after a single bout of exercise, samples with psychiatric and medical comorbidity, and samples constituted of athletes.
Quality of studies
We assessed the risk of bias of studies by noting the concealment of allocation, the use of intention-to-treat (ITT) analysis, and blinding. Concealment was considered adequate if there was central randomization at a site remote from the study, computerized allocation in a locked file, sealed and opaque sequentially numbered envelopes, or other methods of ensuring adequate concealment. Concealment was considered inadequate if an open list or table of random numbers, open computer system, drawing of non-opaque envelopes, or unclear procedures were employed. Trials were defined as using ITT analysis if authors clearly declared that results were based on ITT analysis. For blinding, we distinguished between trials in which the main outcome was measured by a blinded assessor, and those in which the main outcome was measured either by the participants themselves or by a non-blinded assessor. Moreover, we considered the duration of the trial, if the sample included 20 subjects or more, and whether the study performed a follow-up assessment. We also considered the quality of assessment (ie, structured interview, observer-administered, or self-report questionnaire), both at the baseline and at the end-point of the trial.
Outcome measures
We considered the measure of depression declared by authors as the primary outcome measure. For a secondary outcome, we considered only quality of life assessment.
Data synthesis
Due to methodological heterogeneity of the studies identified, a meta-analysis was not conducted. Data synthesis was accomplished by detailed consideration of each study's characteristics and findings.
Results
Twenty-one full-text articles were retrieved by the search on PubMed. One paper was found through the supplementary search on Scopus; no additional articles were found either on Google Scholar or through searching the citation lists of the retrieved studies.
Nine papers were excluded because they did not fulfill our criteria. Figure 1 shows the process of inclusion of studies for review.
Figure 1 Process of inclusion of studies for qualitative review.
We found 11 randomized, clinical trials that fulfilled our inclusion criteriaReference Blumenthal, Babyak and Moore52–Reference Veale, Le Fevre and Pantelis62 and 2 non-randomized trials.Reference de la Cerda, Cervelló, Cocca and Viciana63, Reference Deslandes, Moraes and Alves64 The main characteristics of the included studies are shown in Table 1.
Table 1 Main characteristics of included studies
BDI: Beck Depression Inventory; BRMS: Bech-Rafaelsen Melancholy Scale; CES-D: Centre for Epidemiologic Studies Depression; CIS: Clinical Interview Schedule; HAM-D: Hamilton Rating Scale for Depression; RCT: randomized controlled trial; SF 36: Short Form Health Survey-36; WHOQOL-Bref: World Health Organization Quality of Life-Bref
*Percentage of depressed patients in the sample. **Percentage of participants undergoing antidepressant therapy.
Of the studies, 2 were conducted in the USA,Reference Blumenthal, Babyak and Moore52, Reference Matthews, Hsu and Walkup57 2 in the UK,Reference Mather, Rodriguez and Guthrie56, Reference Veale, Le Fevre and Pantelis62 2 in Brazil,Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61, Reference Deslandes, Moraes and Alves64 and 1 each in Germany,Reference Knubben, Reischies and Adli53 Norway,Reference Martinsen, Medhus and Sandvik54 Denmark,Reference Martiny, Refsgaard and Lund55 Portugal,Reference Mota-Pereira, Silverio and Carvalho58 Italy,Reference Pilu, Sorba and Hardoy59 Iran,Reference Rashidi, Rashidi, Rouzbahani and Rezaei60 and Spain.Reference de la Cerda, Cervelló, Cocca and Viciana63 All of the included papers but the Iranian oneReference Rashidi, Rashidi, Rouzbahani and Rezaei60 were published in English.
Samples
The number of participants in the studies ranged widely from 20 to 424, with a total number of 1131 and a mean per study of 87. Considering only the studies in which the whole sample was taking antidepressants, the number of participants notably decreased (total: n = 581, mean per study n = 51).
Three of the trials recruited voluntary participants through flyers, media advertisements, and letters sent to local physicians and mental health facilities.Reference Blumenthal, Babyak and Moore52, Reference Martiny, Refsgaard and Lund55, Reference Mather, Rodriguez and Guthrie56 Four trials focused on depressed inpatients recruited in psychiatric wards,Reference Knubben, Reischies and Adli53, Reference Martinsen, Medhus and Sandvik54, Reference Rashidi, Rashidi, Rouzbahani and Rezaei60, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 and 4 on depressed outpatients.Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59, Reference de la Cerda, Cervelló, Cocca and Viciana63, Reference Deslandes, Moraes and Alves64 Matthews etalReference Matthews, Hsu and Walkup57 recruited participants through telephone screens and direct mails.
Two trials also admitted bipolar patients in the depressive phase.Reference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55 Three trials were focused on treatment-resistant depressed patientsReference Mather, Rodriguez and Guthrie56, Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59 and 3 on participants with severe depression.Reference Knubben, Reischies and Adli53, Reference Martinsen, Medhus and Sandvik54, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 The remaining trials were carried out on patients diagnosed with mild to moderate depression.
Eight trials achieved psychiatric diagnosis of depression through Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnostic criteria for MDD,Reference Blumenthal, Babyak and Moore52–Reference Martiny, Refsgaard and Lund55, Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61, Reference Deslandes, Moraes and Alves64 and 2 trials used International Classification of Diseases (ICD) criteria.Reference Mather, Rodriguez and Guthrie56, Reference de la Cerda, Cervelló, Cocca and Viciana63 Matthews etalReference Matthews, Hsu and Walkup57 assessed participants’ depressive symptoms with the Centre for Epidemiologic Studies Depression (CES-D), and Veale etalReference Veale, Le Fevre and Pantelis62 used the clinical interview schedule (CIS). Rashidi etalReference Rashidi, Rashidi, Rouzbahani and Rezaei60 did not specify diagnostic criteria.
In 10 trials, the whole sample was formed by participants who were undergoing antidepressant therapy.Reference Blumenthal, Babyak and Moore52, Reference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55, Reference Mather, Rodriguez and Guthrie56, Reference Mota-Pereira, Silverio and Carvalho58–Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61, Reference de la Cerda, Cervelló, Cocca and Viciana63, Reference Deslandes, Moraes and Alves64 The percentages of those on antidepressant drug treatment in the remaining trials ranged from around 24% to over 50% (mean 33.43%).
Intervention groups
Nine studies used an aerobic exercise intervention,Reference Knubben, Reischies and Adli53, Reference Martinsen, Medhus and Sandvik54, Reference Mota-Pereira, Silverio and Carvalho58–Reference Deslandes, Moraes and Alves64 1 used anaerobic exercise,Reference Mather, Rodriguez and Guthrie56 and 1 had an intervention based on mixed aerobic/anaerobic exercise.Reference Matthews, Hsu and Walkup57 One study had 2 active treatments, with supervised aerobic exercise and supervised aerobic exercise plus sertraline.Reference Blumenthal, Babyak and Moore52 Another study had 2 intervention groups, which compared exercise plus duloxetine to wake/light therapy (chronotherapy) plus duloxetine.Reference Martiny, Refsgaard and Lund55
Control groups
Among the trials in which the whole sample underwent treatment with antidepressants, 4 studies compared the intervention with a control group undergoing only antidepressantsReference Blumenthal, Babyak and Moore52, Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61; 2 used stretching, flexibility, or relaxation plus antidepressantsReference Knubben, Reischies and Adli53, Reference Rashidi, Rashidi, Rouzbahani and Rezaei60; 1 used chronotherapy plus antidepressantsReference Martiny, Refsgaard and Lund55; and 1 used health educational talks plus antidepressants control group.Reference Mather, Rodriguez and Guthrie56 Among those studies with a subsample undergoing antidepressant drug treatment, 1 had an occupational therapy control group,Reference Martinsen, Medhus and Sandvik54 1 used health talks and stretching,Reference Matthews, Hsu and Walkup57 and one had a no-treatment control group.Reference Veale, Le Fevre and Pantelis62
Duration
The duration of the trials varied from 10 days to 1 year; the mean duration was 19.3 weeks. A 34-week follow-up assessment was performed in 1 study.Reference Mather, Rodriguez and Guthrie56
Quality assessment
Treatment allocation was adequately conceived in 4 studies out of 13.Reference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55–Reference Matthews, Hsu and Walkup57 Intention to treat analysis was performed in 4 studies out of 13.Reference Blumenthal, Babyak and Moore52, Reference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55, Reference Mather, Rodriguez and Guthrie56 Due to the obvious nature of group exercising, none of the studies had a double-blind design; 10 studies assessed the main outcome using blinded assessors.Reference Blumenthal, Babyak and Moore52, Reference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55–Reference Veale, Le Fevre and Pantelis62
The main outcome was a significant reduction compared to baseline of Hamilton rating scale for Depression-17 items (HAMD-17) score in 4 studies,Reference Martiny, Refsgaard and Lund55, Reference Mather, Rodriguez and Guthrie56, Reference Pilu, Sorba and Hardoy59, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 of BDI score in 3 studies,Reference Martinsen, Medhus and Sandvik54, Reference Rashidi, Rashidi, Rouzbahani and Rezaei60, Reference de la Cerda, Cervelló, Cocca and Viciana63 and of the CES-D score in 1 study.Reference Matthews, Hsu and Walkup57 In 5 studies, the assessment of participants was performed both with an observer-administered and a self-administered questionnaire.Reference Blumenthal, Babyak and Moore52, Reference Knubben, Reischies and Adli53, Reference Mota-Pereira, Silverio and Carvalho58, Reference Veale, Le Fevre and Pantelis62, Reference Deslandes, Moraes and Alves64 Two studies evaluated quality of life by World Health Organization Quality of Life-Bref (WHOQOL-Bref)Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 or short form health survey-36 items (SF-36).Reference Deslandes, Moraes and Alves64 For 2 trials,Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59 further papers were availableReference Carta, Hardoy and Pilu30, Reference Mota-Pereira, Carvalho and Silverio65 that examined quality of life by WHOQOL-BriefReference Carta, Hardoy and Pilu30 and WHOQOL-Brief and SF-36Reference Mota-Pereira, Carvalho and Silverio65 of the same trials.
Further information retrieved by the authors
We asked some authors to clarify some methodological concerns that we had regarding their trials. Pilu etalReference Pilu, Sorba and Hardoy59 and Rashidi etalReference Rashidi, Rashidi, Rouzbahani and Rezaei60 were asked about blinding, intention to treat analysis, and drop-out rates; Rashidi etal were also asked about the statistical difference between the treatment and control groups. Matthews etalReference Matthews, Hsu and Walkup57 were asked about the separate data concerning only patients undergoing antidepressants.
Trials in Which the Entire Sample Was Undergoing Antidepressant Treatment
Exercise plus antidepressants versus antidepressants alone
Among the 4 randomized controlled trials (RCTs) that compared add-on aerobic exercise with antidepressants with antidepressants alone,Reference Blumenthal, Babyak and Moore52, Reference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 3 showed a significant difference between treatment and control group outcomeReference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 and were focused on particular samples; 2 examined the efficacy of combined therapy (antidepressants plus aerobic exercise) in participants affected by treatment-resistant depressionReference Mota-Pereira, Silverio and Carvalho58, Reference Pilu, Sorba and Hardoy59; and 1 utilized the same intervention on patients with severe depression.Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 Two non-randomized trials also reported positive results.Reference de la Cerda, Cervelló, Cocca and Viciana63, Reference Deslandes, Moraes and Alves64
Mota-Pereira etalReference Mota-Pereira, Silverio and Carvalho58 carried out a trial on 33 patients (males and females aged 18–60 years) who were affected by treatment-resistant MDD and who were taking appropriate combined therapy, in doses considered appropriate, for more than 9 months and less than 15 months. Participants were randomized to the treatment (5 weekly sessions of aerobic exercise, 1 supervised and 4 unsupervised, consisting of 30–45 min/day walks, as an adjunction to the usual antidepressant treatment) or a control group (usual antidepressant treatment alone) for 12 weeks. At the end of the trial, the treatment group showed lower HAMD-17 and Beck Depression Inventory (BDI) rates compared to the control group. While in the control group, none of the participants showed response (defined as a decrease from baseline to endpoint of ≥50% on the HAMD-17 total score) or remission (defined as an endpoint HAMD-17 total score ≤7), in the treatment group, there was a 21% rate of response and a 26% rate of remission, although these data were not significant. Another paper that analyzed data on QoL of the same trialReference Mota-Pereira, Carvalho and Silverio65 reported positive findings on the physical domain of SF-36 and on the social domain of WHOQOL-Brief.
Pilu etalReference Pilu, Sorba and Hardoy59 performed a randomized trial on a small sample of 30 female patients affected by MDD, aged 40–60 years, who were undergoing but not responding to antidepressant therapy. Participants were randomized in a 2:1 fashion either to the treatment (2 weekly supervised aerobic exercise sessions of 1 hour each plus their usual antidepressant) or the control group (antidepressant alone) for 8 months. Only the treatment group showed significant reduction in HAMD-17 scores. Another paper that analyzed data from the same trialReference Carta, Hardoy and Pilu30 found that exercise also improved the perceived physical domain of quality of life: the score of WHO-QOL-Brief scale in the physical domain significantly improved in the exercise plus antidepressants group, as compared to the control group. The perceived QoL in the other domains did not change during the treatment in either group.
Schuch etalReference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 published the preliminary results of a randomized controlled trial of aerobic exercise added to antidepressants, compared to antidepressants alone, on 26 inpatients affected by severe depression as diagnosed using the mini international neuropsychiatric interview (MINI) (according to DSM-IV criteria). Patients in the exercise group exercised to 16 kcal/kg/week during 3 sessions/week (ie, “public health dose”) in association with conventional treatment; in order to improve compliance, patients could choose the exercise (stationary bicycle, a treadmill, or an elliptical machine) and the intensity, with the sole objective to complete the total calories estimated. Both the exercise and control groups achieved statistically significant improvements in HAMD-17 at 2 weeks and at discharge compared to baseline mean scores. Between groups, there was a difference favorable to the exercise group at discharge. Similarly, both groups showed improvements in physical and psychological domains of the WHO-QOL Brief at the second week of hospitalization, with significant differences favoring the exercise group at discharge on the psychological domain and as well as a trend on the physical domain.
Blumenthal etalReference Blumenthal, Babyak and Moore52 conducted a 16-week RCT on 156 male and female participants with mild to moderate MDD, aged ≥50 years, who were assigned to a program of aerobic exercise, antidepressant (sertraline), or combined exercise and medication. Bipolar disorder and current use of antidepressants at baseline were exclusion criteria. Depression assessment was done with HAMD-17 and BDI. Treatments did not differ in improving depression; nevertheless, some differences were found depending on the severity of depression at baseline, ie, the medication group exhibited a more rapid initial response. Interestingly, participants with mild depression at baseline responded more quickly to the combination treatment than did moderately to severely depressed participants.
De la Cerda etalReference de la Cerda, Cervelló, Cocca and Viciana63 performed a non-randomized, controlled study in which a sample of 82 female depressed patients (ICD criteria), aged between 20 and 64 years (mean age 32.4 years), were divided into 2 age-balanced groups: 1 group receiving fluoxetine 20 mg/day and aerobic exercise, 1 group receiving ongoing fluoxetine 20 mg/day alone, for a duration of 8 weeks. The aerobic training program consisted of 3 sessions per week, of increasing duration from 45 to 60 minutes, of aerobic gymnastics, fun dance, and walking. Assessment was performed by BDI, and International Classification of Diseases, 10th revision (ICD-10) was used to confirm diagnosis both at baseline and at the end-point of the trial. While the majority of patients in the pharmacotherapy/exercise group had decreased BDI and ICD-10 severity ratings from moderate to minimal or no depression, participants in the pharmacotherapy group decreased from moderate to mild depression. Interestingly, the authors noted that the women in the sample belonged to middle–low income social class, and were generally suffering from loneliness and isolation; thus it could be that exercising in a group setting functioned as an opportunity for socializing and leisure time. An analogous experience was experienced by participants in the trial by Pilu etal,Reference Pilu, Sorba and Hardoy59 who also belonged to a low-income social class (personal communication, 2013).
Deslandes etalReference Deslandes, Moraes and Alves64 published the results of a quasi-experimental study that aimed to identify changes in depressive symptoms, quality of life, and cortical asymmetry produced by aerobic activity on a small sample of 20 patients (70% females, 71 ± 3 years) with a diagnosis of major depressive disorder according to DSM-IV criteria. Participants were divided into an exercise group (pharmacological treatment plus aerobic training) and a control group (pharmacological treatment alone). Assessment was done by BDI, HAMD-17, Montgomery-Asberg Depression Rating Scale (MADRS) for depression, SF-36 for the QoL, and electroencephalographic measurements (frontal and parietal alpha asymmetry) before treatment and after 1 year of treatment. After 1 year, the exercise group showed a statistically significant decrease of depressive symptoms compared to the control group, both according to the depression questionnaires and in the subscale “role-physical” of SF-36. The control group showed a decrease in cortical activity on the right hemisphere (increase of alpha power), which was not observed in the exercise group. Since the sample was composed of elderly subjects, it is possible that this last result was due to the physical activity counteracting the participants’ mental decline.
Exercise plus antidepressants versus stretching/flexibility/relaxation plus antidepressants
The trial carried out by Knubben etalReference Knubben, Reischies and Adli53 focused on the short-term (10 days) benefit of physical activity in a small sample of 38 depressed inpatients who were undergoing antidepressant therapy. The study assessed the patients with both an observer-administered questionnaire, the Beck Rafaelsen Melancholy Scale (BRMS), and a self-administered one, the Center for Epidemiologic Studies Depression scale (CES-D). Patients were randomly assigned to an exercise (daily walking on a treadmill) or placebo (low-intensity stretching and relaxation exercises) group. The exercise group had a substantially greater reduction in depression scores (BRMS) than the control group. The subjective evaluation of depression (CES-D scores) at the end of the study was significantly lower in the exercise group than the control group.
Rashidi etalReference Rush, Golden and Hall69 performed a trial on 35 depressed female inpatients who were randomly allocated to a treatment group (which played volleyball, badminton, or dodgeball in teams of 4–5 every day) or a control group (only encouraged to perform warm up and stretching exercises) for 2 months. At the end of the 60-day period, BDI was filled out by physicians based on the statements of patients. Both individual and group exercises had positive effects on their mental states; nevertheless, the mental health of the treatment group improved considerably more than the control group, but the p-value was not declared in the paper, and the authors did not respond to our request for further information.
Exercise plus antidepressants versus chronotherapy plus antidepressants
Martiny etalReference Martiny, Refsgaard and Lund55 carried out a trial that aimed to establish the effectiveness of wake and light therapy (ie, chronotherapy), as an adjunction to antidepressants, in inducing a rapid and sustained remission in MDD. In this trial, chronotherapy was compared to exercise plus antidepressants, which was assumed to be an active comparator condition to balance treatment expectations. The sample was formed by 75 patients, the majority suffering from recurrent, unipolar, long-standing melancholic depression. Around 84% were receiving antidepressant treatment at inclusion. A small percentage of patients was affected by bipolar depression (3 patients with bipolar I disorder and 3 patients with bipolar II disorder in each group), and 63% of participants were classified as having treatment-resistant depression. Participants were randomly assigned to a 9-week chronotherapeutic intervention using wake therapy, bright light therapy, and sleep time stabilization (n = 37) or a 9-week intervention using daily exercise (n = 38). During a 1-week run-in phase, all patients began treatment with duloxetine 60 mg/day; all other antidepressants were discontinued, while other psychopharmacologic drugs were maintained, including mood stabilizers. The first phase was followed by a 1-week intervention phase in which patients in the wake therapy group underwent 3 wake therapies in combination with daily morning light therapy and sleep time stabilization and patients in the exercise group began daily exercise. This phase was followed by a 7-week continuation phase with daily chronotherapy or daily exercise. Assessment was performed with HAMD-17. The type of exercise (aerobic or anaerobic) was not defined in the paper; it was performed both indoor and outdoor, supervised by a physiotherapist, with a daily duration of 30 minutes, and, interestingly, it could be flexible. In fact, it was intensified in 44.8% of patients (n = 17) and reduced in 2.6% (n = 1), with a mean exercise duration of 63.0 minutes/day. While both treatments were well tolerated, and both had a clinically relevant antidepressant response, participants assigned to chronotherapy had an immediate and clinically significantly better response (defined as a reduction from baseline score on HAMD-17 of ≥50%) and remission (defined as a score<8 on HAMD-17) compared to the exercise group: at the end-point, response was obtained in 71.4% of wake therapy patients versus 47.3% of exercise patients, and remission was achieved in 45.6% of wake therapy patients versus 23.1% of exercise patients.
Exercise plus antidepressants versus health educational talks plus antidepressants
Mather etalReference Mather, Rodriguez and Guthrie56 published a RCT on the effectiveness of exercise as an adjunct to antidepressant therapy in a sample of 86 depressed patients aged 53 years or older (median age: exercise group 63 years, control group 65 years). Participants were randomly assigned to a treatment group (a twice-weekly exercise class, which contained elements of endurance, muscle strengthening, and stretching) or a control group (twice-weekly health education talks) for 10 weeks, and were assessed with HAMD-17 on 3 occasions: baseline, 10 weeks, and 34 weeks. Because the study focused on a particular population (ie, a group who had failed to respond to initial treatment), the general convention in trials of antidepressant therapy to use a ≥ 50% reduction in HAMD-17 score as the definition of a response was modified by the authors, who assumed that a ≥30% reduction in HAMD-17 score associated with participation in exercise would be of clinical interest. At 10 weeks, the exercise group achieved a higher response frequency compared to the control group.
Trials in Which a Part of the Sample Underwent Antidepressant Treatment
The earliest trial, to our knowledge, that compared add-on exercise with another treatment was the one performed in 1985 by Martinsen etal.Reference Martinsen, Medhus and Sandvik54 Forty-three inpatients of both genders, aged 17–60 years (mean age 40 years), who were affected by MDD were randomly assigned to an intervention group (1 hour of supervised aerobic exercise 3 times/week for 9 weeks, at 50–70% of maximum aerobic capacity) or to a control group (occupational therapy). Nine patients of 24 (37.5%) in the intervention group and 14 of 19 (73.7%) in the control group were also taking Tricyclic Antidepressants (TCAs). Depression was assessed with the BDI. Mean reductions in BDI scores were significantly larger in the exercise group compared to the control group.
Later, in 1991, Veale etalReference Veale, Le Fevre and Pantelis62 carried out a trial that showed the combined results of 2 studies. The first trial aimed to determine whether aerobic exercise, added to the standard treatment of patients suffering from depression, produced additional benefit compared with no extra intervention; the second trial aimed to investigate whether the therapeutic component of the exercise program was due to the patients’ improvement in aerobic fitness level. In that second study, patients were allocated to either an aerobic exercise group or to a low-intensity exercise group. Because of the selection criteria of the present review, we consider only the findings of the first study. Eighty-three depressed outpatients undergoing standard treatment (ie, medications, psychotherapy, or social interventions) were randomly assigned in a 3:2 ratio to an intervention group (3 sessions per week of aerobic fitness for 12 weeks) or to a control group (no extra treatment). Depression assessment used the Clinical Interview Schedule (CIS) and BDI. There was a randomization bias, so participants in the control group had higher depression rates according to the BDI than those in the intervention group (p < 0.05); nevertheless, the percentages of those taking antidepressants at baseline between the groups did not significantly differ (45% in the intervention and 34% in the control group). A significant difference emerged in favor of the exercising group as assessed by CIS mean rates at the end of the trial, while BDI mean rates did not significantly differ. The percentages of participants taking antidepressant therapy at the end of the trial were 44% in the treatment group and 24% in the control group.
Matthews etalReference Matthews, Hsu and Walkup57 carried out a post-hoc analysis of data from the Lifestyle Interventions and Independence for Elders Pilot (LIFE-P) study, a 12-month single-blind RCT that compared a moderate intensity physical activity intervention (PA), consisting of a combination of aerobic, strength, flexibility, and balance exercises in 3 phases (individualized, center-based, and finally home-based exercise program, with group-based counseling sessions), with the successful aging (SA) control, a series of sessions on health topics relevant to older adults followed by 5–10 minutes of stretching. The participants were 424 aged volunteers (mean age 76.77 years). Of the sample, 23.8% had depressive symptoms, and 15.8% had elevated depressive symptom scores (CES-D ≥14). The percentages of participants taking antidepressant medications were significantly higher in those with high depressive (42.1% in the PA intervention group, and 55.2% in the SA control group) versus low depressive symptoms (18.9% in the PA intervention group, and 19.8% in the SA control group), but were comparable between the 2 intervention arms (23% in the PA group, 24.6% in the SA group). There was no significant improvement in CES-D score over time as a result of participation in either intervention group. No significant changes in CES-D scores were found in association with either intervention when examined in participants with high and low depressive symptoms over the duration of the trial.
Discussion
Considered overall, the studies included in the present review showed a strong effectiveness of exercise combined with antidepressants. Nine of the trials showed a statistically significant superiority in reducing depression scores in favor of combined interventions,Reference Knubben, Reischies and Adli53, Reference Martinsen, Medhus and Sandvik54, Reference Mather, Rodriguez and Guthrie56, Reference Mota-Pereira, Silverio and Carvalho58–Reference Rashidi, Rashidi, Rouzbahani and Rezaei60, Reference Veale, Le Fevre and Pantelis62–Reference Deslandes, Moraes and Alves64 and, among the other 4 trials, 3Reference Blumenthal, Babyak and Moore52, Reference Martiny, Refsgaard and Lund55, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 reported good results for both the intervention and control groups. Whenever we considered only trials with the whole sample in treatment with antidepressants, the findings appear to be significant.
Moreover, 4 of the included studiesReference Carta, Hardoy and Pilu30, Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61, Reference Deslandes, Moraes and Alves64, Reference Mota-Pereira, Carvalho and Silverio65 found an improvement in quality of life, and corroborated the results of some previous studies on the efficacy of exercise as a sole treatment for depression.Reference Singh, Stavrinos and Scarbek20, Reference Brenes, Williamson and Messier66, Reference Singh, Clements and Fiatarone67
Also, certain secondary considerations in some trials should be highlighted regarding the compliance to the add-on strategy. Drop-out rates in the selected studies varied between 0 and 30%, with 7 trials in which there were no drop outs at allReference Knubben, Reischies and Adli53, Reference Martiny, Refsgaard and Lund55, Reference Mather, Rodriguez and Guthrie56, Reference Pilu, Sorba and Hardoy59–Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61, Reference de la Cerda, Cervelló, Cocca and Viciana63 and another with a drop out percentage under 10%.Reference Mota-Pereira, Silverio and Carvalho58 While the remaining studies’ withdrawal rates did not differ from those of trials with antidepressants treatment,Reference Rush, Golden and Hall68, Reference Rush, Golden and Hall69 this is a significant fact that can be explained by the high acceptability and, in some case, flexibility, of physical activity intervention.
Some authors also reported the number of patients who did not agree to participate in the proposed trial: Schuch etal,Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 while reporting no drop-outs, highlighted that 14 of the 40 invited patients “were not interested” in participating in the study, Rashidi etalReference Rashidi, Rashidi, Rouzbahani and Rezaei60 excluded 4 patients in the treatment group who were not interested in participating in group sports, and Martinsen etalReference Martinsen, Medhus and Sandvik54 reported 6 patients who dropped out of the study at the beginning (4 in the intervention group and 2 in the control group), with “withdrawals not related to treatment.” Nevertheless, all these authors carried out the analyses with the remaining participants. Mather etalReference Mather, Rodriguez and Guthrie56 reported that initial scrutiny of the data revealed that 1 patient in the exercise group out of 87 initially scrutinized participants had not met the eligibility criteria for the study, so this individual's results were removed from the analysis. Furthermore, Matthews etalReference Matthews, Hsu and Walkup57 reported that adherence did not differ between participants with higher depressive symptoms compared with those with lower depressive symptoms, and Mota-Pereira etalReference Mota-Pereira, Silverio and Carvalho58 even highlighted that participants in the exercise group significantly differed from those in the control group, as they showed greater depression severity (higher HAMD-17 and BDI total scores) and worse functioning (lower global assessment of functioning (GAF) and higher clinical global impression - severity scale (CGI-S)). Martiny etalReference Martiny, Refsgaard and Lund55 reported that recruitment was stopped at 75 patients due to time constraints and funding limits. Thus, no authors but but Schuch etalReference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 reported significant rates of selected patients who refused to participate in the trials, implying that exercise in adjunction to antidepressants is an acceptable treatment for the majority of the patients.
Significantly, some authors proposed physical activity interventions with characteristics of flexibility and pleasantness: de la Cerda etalReference de la Cerda, Cervelló, Cocca and Viciana63 and Rashidi etalReference Rashidi, Rashidi, Rouzbahani and Rezaei60 utilized varied interventions (ie, different activities at each of 3 weekly sessions), Martiny etalReference Martiny, Refsgaard and Lund55 made sure that patients could choose the intensity of exercise, and Schuch etalReference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 varied the type of activity. Recently, a trial carried out by Callaghan etalReference Callaghan, Khalil, Morres and Carter70 affirmed the superiority of preferred, rather than prescribed, intensity exercise in reducing depression severity and improving QoL. Moreover, Mather etal,Reference Mather, Rodriguez and Guthrie56 Pilu etal,Reference Pilu, Sorba and Hardoy59 Rashidi etal,Reference Rashidi, Rashidi, Rouzbahani and Rezaei60 Veale etal,Reference Veale, Le Fevre and Pantelis62 and de la Cerda etalReference de la Cerda, Cervelló, Cocca and Viciana63 had participants who exercised in groups. Thus, exercise provides an opportunity to have leisure time and social contacts, which are experiences generally lacking in depressed people and which are not directly addressed by pharmacotherapy.Reference de la Cerda, Cervelló, Cocca and Viciana63
One of the major limits of the included studies is the exercise-related placebo effect. Martinsen etal,Reference Martinsen, Medhus and Sandvik54 Pilu etal,Reference Pilu, Sorba and Hardoy59 de la Cerda etal,Reference de la Cerda, Cervelló, Cocca and Viciana63 and Veale etalReference Veale, Le Fevre and Pantelis62 underlined that therapeutically effective components in their trials could be due to the beneficial effect of socialization due to exercise, rather than a therapeutic component of exercise per se. To minimize this issue, some trials had control groups with “placebo treatments” with similar components of socialization, without the active form of treatment related to physical activity. Stretching, flexibility, and relaxationReference Knubben, Reischies and Adli53, Reference Matthews, Hsu and Walkup57, Reference Rashidi, Rashidi, Rouzbahani and Rezaei60 seem to be an acceptable compromise to address this issue; nevertheless, some studies chose control groups based on occupational therapyReference Martinsen, Medhus and Sandvik54 or health educational talksReference Mather, Rodriguez and Guthrie56, Reference Matthews, Hsu and Walkup57 that could have introduced uncontrolled variables in the analysis. Furthermore, Martiny etalReference Martiny, Refsgaard and Lund55 compared an unproven treatment (chronotherapy) to exercise.
On the other hand, researchers often tend to disregard that depression sometimes spontaneously improves, and such an improvement is not totally due to placebo response, as the estimated effect size for the change in depression scores during wait-list control was 0.505, which represents an average improvement of 4 points on the Hamilton Rating Scale for Depression.Reference Rutherford, Mori, Sneed, Pimontel and Roose71 Moreover, placebo response rates of antidepressant therapies tend to increase from ≥30% and ≥40%, respectively, for monotherapy and adjunctive trials.Reference Iovieno and Papakostas72 Still, the gold standard for testing the efficacy of a new treatment for MDD continues to be antidepressants.
Among the studies that were included in the present review, a variety of methodological weaknesses could devalue the relevance of results. Above all, allocation was adequately conceived in only 4 studies, and intention to treat analysis also was only conducted in 4 studies. Samples were small among studies that showed the largest effect of treatment, and it could be argued that the effect would be smaller in larger samples.
Nevertheless, we have focused on a question that reflects very well a natural clinical setting, but which may be difficult to transfer appropriately to the setting of a clinical trial: how effective is exercise in addition to antidepressants? Interestingly, all of the included trials that focused on patients suffering from treatment-resistant depressionReference Knubben, Reischies and Adli53, Reference Mather, Rodriguez and Guthrie56, Reference Mota-Pereira, Silverio and Carvalho58–Reference Schuch, Vasconcelos-Moreno, Borowsky and Fleck61 showed a larger improvement in depression scores in the exercise-plus-antidepressants intervention groups compared to control groups, thus confirming previous findings of a possible role of exercise as an adjuvant treatment for patients with severe MDD.Reference Mota-Pereira, Silverio, Carvalho, Ramos and Ribeiro73 While exercise has a high cost-effectivenessReference Chalder, Wiles and Campbell74 compared to a first-line antidepressant drug treatment, and it may be not always available, it might be more useful as a step-2 treatment for antidepressant medication nonresponders.
A growing number of studies, both performed on animal models of depression and on depressed humans, has focused on other possible molecular targets for antidepressant treatment, besides monoamines. This has led researchers to postulate the “neurotrophic hypothesis of depression.”Reference Neto, Borges, Torres-Sanchez, Mico and Berrocoso75 According to this hypothesis, several alterations in the levels of neurotrophins, particularly of brain-derived neurotrophic factor (BDNF), might produce the structural and neurochemical changes that underlie depression. In particular, the limbic structures, most notably the hippocampus, have been shown to be affected by neuronal atrophyReference Mendez-David, Hen, Gardier and David76, Reference Tanti and Belzung77 in depressed patients. Both pharmacological and nonpharmacological interventions for depression have been shown to produce changes in the levels of neurotrophins. BDNF increases have been reported to follow the administration of antidepressant drugs,Reference De Foubert, Carney and Robinson78–Reference Nibuya, Morinobu and Duman82 which suggests that BDNF expression may mediate the action of antidepressants. Moreover, an increase in BDNF has also been detected with electroconvulsive therapy (ECT)Reference Nibuya, Morinobu and Duman82, Reference Altar, Whitehead, Chen, Wortwein and Madsen83 and exercise.Reference Russo-Neustadt, Alejandre, Garcia, Ivy and Chen80, Reference Neeper, Gomez-Pinilla, Choi and Cotman84–Reference Arunrut, Alejandre, Chen, Cha and Russo-Neustadt87 Interestingly, exercise in adjunction with the antidepressant reboxetine has been shown to quickly increase BDNF levels, whereas reboxetine alone did not.Reference Russo-Neustadt, Alejandre, Garcia, Ivy and Chen80 A similar add-on strategy with tranylcypromine was shown to increase BDNF and have a more favorable effect on a rat model of depressive behavior than did an antidepressant alone.Reference Russo-Neustadt, Ha, Ramirez and Kesslak88 Notably, this effect was demonstrated both in young and in aged rats, confirming the possibility of adult neurogenesis.Reference Garza, Ha, Garcia, Chen and Russo-Neustadt89
It has been recently hypothesized that the increase in serotonin levels induced by antidepressant drugs, specifically SSRIs, may not improve depression per se, but rather enhances neuroplasticity through the modulation of BDNF mRNA, consequently rending the individual more susceptible to the influence of the environment.Reference Branchi, Santarelli and Capoccia90 According to this interesting hypothesis, researchers showed that in an animal model of chronic depression, animals that were chronically treated with fluoxetine did not display the expected increase in BDNF and even worsened in a stressful environment. Thus, the authors suggested that the effects of antidepressants medications could be enhanced by training patients to cope with harsh environments, for instance through cognitive behavioral therapy. We believe that there is enough evidence to recommend exercise in adjunction to antidepressants to manage treatment-resistant depression.
A number of reviews and meta-analyses on the efficacy of exercise on depression, which were published in the last 30 years, have testified to the researchers’ interest in this subject; nevertheless, while generally positive, findings have continued to be somewhat inconclusive and debated, and the literature has been criticized because a variety of methodological limitations.Reference Brosse, Sheets, Lett and Blumenthal39, Reference Daley42, Reference Greer and Trivedi44, Reference Lawlor and Hopker46, Reference Mead, Morley and Campbell47, Reference Rethorst, Wipfli and Landers49, Reference Rimer, Dwan and Lawlor50, Reference Josefsson, Lindwall and Archer91 Furthermore, direct comparisons between studies were often difficult due to wide varieties in assessment or diagnosis of depression, level of severity of the disorder, size of the sample, type, frequency and duration of the intervention, and assessment of outcome.Reference Blake31
Few of the meta-analyses and systematic reviews carried out have considered studies of physical activity as additions to antidepressant medications,Reference Craft and Perna41, Reference Mura and Carta51, Reference Mota-Pereira, Silverio, Carvalho, Ramos and Ribeiro73, Reference Silveira, Moraes and Oliveira92, Reference Danielsson, Noras, Waern and Carlsson93 and, to the best of our knowledge, no meta-analysis or systematic review has focused specifically on this topic. Craft and PernaReference Craft and Perna41 converted the overall effect sizes of –0.72, –0.94, and –1.1 of 3 meta-analysesReference Lawlor and Hopker46, Reference Craft and Landers94, Reference North, McCullagh and Tran95 into binomial effect sizes, reflecting an increase in success rate due to exercise of 67%, 71%, and 74%, respectively. The authors pointed out that clinical guidelines consider a 50% reduction in symptoms a treatment response, and argued that, considering clinical criteria more relevant than statistical significance, there is strong evidence to advocate the use of exercise as a potentially powerful adjunct to existing treatment (pharmacotherapies and psychological therapies).
Further high-quality research could establish the efficacy of exercise as an adjunction to antidepressants to treat MDD, and our results should encourage efforts in such a promising direction.
Conclusions
This is the first review to have focused on exercise as an add-on strategy in the treatment of MDD. Our findings corroborate some previous observations that were based on few studies and which were difficult to generalize.Reference Craft and Perna41, Reference Mura and Carta51, Reference Mota-Pereira, Silverio, Carvalho, Ramos and Ribeiro73, Reference Silveira, Moraes and Oliveira92, Reference Danielsson, Noras, Waern and Carlsson93 Given the results of the present article, it seems that exercise might be an effective strategy to enhance the antidepressant effect of medication treatments. Moreover, we hypothesize that the main role of exercise on treatment-resistant depression is in inducing neurogenesis by increasing BDNF expression, as was demonstrated by several recent studies.
Further analyses and higher quality studies are needed; nevertheless, as we have focused on a particular intervention (exercise in adjunction to antidepressants) that perhaps more closely reflects daily clinical practice rather than the structured context of a clinical trial, we have shown that this strategy could appropriately and safety translate into a real context, in particular if exercise has both flexible and acceptable features.
Disclosures
The authors do not have an affiliation with or financial interest in any organization that might pose a conflict of interest.
