Physical activity protects from incident anxiety: A meta‐analysis of prospective cohort studies
Background: Prospective cohorts have suggested that physical activity (PA) can decrease the risk of incident anxiety. However, no meta‐analysis has been conducted. Aims: To examine the prospective relationship between PA and incident anxiety and explore potential moderators. Methods: Searches were conducted on major databases from inception to October 10, 2018 for prospective studies (at least 1 year of follow‐up) that calculated the odds ratio (OR) of incident anxiety in people with high PA against people with low PA. Methodological quality was assessed using the Newcastle‐Ottawa Scale (NOS). A random‐effects meta‐analysis was conducted and heterogeneity was explored using subgroup and meta‐regression analysis. Results: Across 14 cohorts of 13 unique prospective studies (N = 75,831, median males = 50.1%) followed for 357,424 person‐years, people with high self‐reported PA (versus low PA) were at reduced odds of developing anxiety (adjusted odds ratio [AOR] = 0.74; 95% confidence level [95% CI] = 0.62, 0.88; crude OR = 0.80; 95% CI = 0.69, 0.92). High self‐reported PA was protective against the emergence of agoraphobia (AOR = 0.42; 95% CI = 0.18, 0.98) and posttraumatic stress disorder (AOR = 0.57; 95% CI = 0.39, 0.85). The protective effects for anxiety were evident in Asia (AOR = 0.31; 95% CI = 0.10, 0.96) and Europe (AOR = 0.82; 95% CI = 0.69, 0.97); for children/adolescents (AOR = 0.52; 95% CI = 0.29, 0.90) and adults (AOR = 0.81; 95% CI = 0.69, 0.95). Results remained robust when adjusting for confounding factors. Overall study quality was moderate to high (mean NOS = 6.7 out of 9). Conclusion: Evidence supports the notion that self‐reported PA can confer protection against the emergence of anxiety regardless of demographic factors. In particular, higher PA levels protects from agoraphobia and posttraumatic disorder.
Keywords: agoraphobia; anxiety; exercise; incidence; meta‐analysis; panic; physical activity; posttraumatic stress disorder; protection
1 INTRODUCTION
Anxiety disorders are common and burdensome across the world (Baxter, Vos, Scott, Norman, et al., [6]; Baxter, Vos, Scott, Ferrari, & Whiteford, [5]). The point prevalence varies across world regions, ranging between 2.1% (range, 1.8–2.5%) in East Asia up to 6.1% (range, 5.1–7.4%) in North Africa/Middle East (Baxter, Vos, Scott, Ferrari, et al., [5]; Baxter, Vos, Scott, Norman, et al., [6]). The prevalence of anxiety symptoms is even higher, at approximately 11% worldwide (Stubbs, Koyanagi et al., [55]). Globally, anxiety disorders are the sixth leading cause of disability regarding years of life lived with disability across all ages, and fourth in adults (18–49 years; Meier et al., [38]). In addition, anxiety disorders are associated with a 39% and 146% increased risk of premature mortality from natural causes and unnatural causes, respectively (Meier et al., [38]).
People with anxiety disorders are at higher risk of cardiometabolic diseases, such as diabetes and acute cardiac events (Edmondson & von Kanel, [22]; Smith, Deschenes, & Schmitz, [51]). This is potentially due to shared etiological biological factors between anxiety and cardiovascular disorders (e.g., increased inflammation and oxidative stress; Belem da Silva et al., [10]), but also due to modifiable risk behaviors like lower physical activity (PA) levels and increased sedentary behavior (Stubbs, Koyanagi et al., [55]; Vancampfort, Stubbs, Herring, Hallgren, & Koyanagi, [61]). Although several previous cross‐sectional studies suggest that people with anxiety disorders or higher levels of anxiety symptoms are more likely to engage in lower levels of PA (22.9% versus 16.6%; Dillon, McMahon, O'Regan, & Perry, [19]; Stubbs, Koyanagi et al., [55]) and spend more time in sedentary activities when compared with people without anxiety (Dillon et al., [19]; Teychenne, Costigan, & Parker, [58]; Vancampfort et al., [60], [61]), PA is not always inversely associated with anxiety symptoms (Nguyen et al., [42]).
Randomized controlled trials have demonstrated that exercise, the structured and planned form of PA, has anxiolytic effects in people with and without anxiety disorders (Herring, Lindheimer, & O'Connor, [27]; Herring, O'Connor, & Dishman, [28]; Stubbs, Vancampfort et al., [56], [60]). Previous reviews have also demonstrated that a single bout of exercise can help alleviate anxiety symptoms in people with anxiety disorders (Strohle, Ströhle, Graetz, & Scheel, [53]). Moreover, previous systematic and nonsystematic reviews (Martinsen, [35]; Teychenne et al., [58]) have suggested that higher levels of PA are linked to a decreased risk of future anxiety disorders. No study, however, performed a meta‐analysis to estimate the magnitude of the effect. Whilst helpful, the lack of meta‐analysis has prevented wider quantitative inferences to be made about this relationship and reduced the ability to investigate potential sources of heterogeneity. Thus, it remains unclear if and to what degree PA confers protection against the emergence of anxiety.
Considering the aforementioned gaps in the literature, our aim was to systematically review and meta‐analyze prospective cohort studies examining the role of PA in reducing the risk of incident anxiety. We also aimed to explore potential moderators, including age at baseline, geographical location, sex, length of follow‐up, study quality, number of covariates used in the model, study sample size, and total person‐years.
2 METHODS
This review was conducted following the Meta‐analysis Of Observational Studies in Epidemiology (Stroup et al., [54]) guidelines and the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (Moher, Liberati, Tetzlaff, & Altman, [39]) statement. To eliminate researcher bias, the search strategy, inclusion criteria, and data extraction, as well as the overall and preplanned subgroup analyses, strictly adhered to those adopted in a previous systematic review of PA as a protective factor for the incident of depression (Schuch et al., [47]). The protocol was not registered in PROSPERO but is available upon request.
2.1 Search procedure
Four researchers (A. H., F. S., S. H., and P. Z.) searched PubMed, Embase, Web of Science, and SPORTDiscus from database inception to October 10, 2018 for eligible studies. Keywords included a combination of terms related to PA, anxiety, anxiety disorders, and longitudinal design studies, adapted for each database. Full details of the searches and terms used are displayed in Materials S1. Manual searches of the reference lists from recovered articles and other systematic reviews investigating the association between PA, sedentary behavior or fitness and anxiety disorders were conducted (Teychenne et al., [58]).
2.2 Inclusion and exclusion criteria
Articles were eligible if they met the following criteria: (a) evaluated participants, of any age, free from anxiety at baseline; (b) measured PA with a self‐report questionnaire such as the International Physical Activity Questionnaire (Craig et al., [17]), single or multiple questions of exercise, sports or PA participation, or any objective PA measures (e.g., pedometers and accelerometers). PA definition used in the present study was: Any bodily movement produced by skeletal muscles and which requires energy expenditure (Caspersen, Powell, & Christenson, [16]). Only evaluations of high versus low PA, using any criterion, were eligible; (c) Used a prospective cohort study design with a follow‐up period of 1 year or longer. Prospective cohort studies with less than 1 year follow‐up were not included since it may not be considered a sufficient period for risk and protective factors to exert a meaningful influence on mental health symptoms (Cairns, Yap, Pilkington, & Jorm, [15]); (d) evaluated incident (new cases from baseline to follow‐up) anxiety as the outcome, namely increased anxiety symptoms identified via established cut‐offs of anxiety screening instruments (e.g., Hospital Anxiety and Depression Scale and Beck Anxiety Scale; Beck, Ward, Mendelson, Mock, & Erbaugh, [8]); or anxiety disorders, diagnosed using structured or semi‐structured diagnostic interviews (e.g., instruments using Diagnostic and Statistical Manual (DSM) for Mental Disorders [American Psychiatric Association, [1]] or International Classification of Disease criteria [World Health Organization, [62]]). For the present meta‐analysis, we included posttraumatic stress disorder (PTSD) in the analysis as PTSD was considered an anxiety disorder until the DSM‐IV; (e) Report an adjusted or nonadjusted odds ratio (OR) and 95% confidence intervals (95% CIs) or the raw numbers of exposed and nonexposed participants who developed anxiety at follow‐up in a way that allows calculation of OR. In instances where data were not available or additional information was needed, we contacted the corresponding authors up to three times (or until the data were provided, whichever came first) over a 4‐week period to request the data to enable inclusion in the present analysis (see Acknowledgments).
Excluded were: (a) Reviews, commentaries, editorials, or any other study without primary data; (b) conference or congress presentations without clear information about the design, methods, or the outcomes; (c) studies written in languages other than Dutch, English, French, German, Portuguese, or Spanish.
Studies of the same epidemiological cohort were included in different analyses when they reported the results with different metrics (i.e., adjusted and nonadjusted OR). When two or more studies reported data of the same cohort, we selected the most recently published. Studies reporting subsamples of cohorts were excluded.
2.3 Study selection
In the first stage of study selection, four authors (F. S., S. H., A. H., and P. Z.) independently screened titles and abstracts of all articles retrieved from the search in duplicate. Afterward, the full‐text of potentially eligible references were reviewed in detail by the same investigators. Disagreements were resolved through discussion until consensus was achieved.
2.4 Outcomes
The primary outcome was the adjusted odds ratio (AOR) for incident anxiety or anxiety disorders and 95% CI.
2.5 Data extraction
One author (F. S.) extracted and a second (S. H.) cross‐checked the data including geographical location, name of cohort, number of participants included at baseline, age at baseline, % of females, PA assessment (instrument or questions used, what aspects of PA were considered by the measure to define PA levels [e.g., frequency, intensity, time, type, energetic amount expended, steps, or other]), anxiety assessment (e.g., instrument and cut‐off used, diagnostic criteria, medical records), follow‐up period, OR and 95% CI, and covariates. Data from figures were extracted using the GetData Graph Digitizer (version 2.26).
2.6 Study quality
The Newcastle‐Ottawa Scale (NOS) was used to assess the methodological quality of studies by two authors (F. S. and J. M.). The NOS scale evaluates the risk of bias of prospective studies with three domains: (a) selection of participants, with four items assessing (a.1) representativeness of the exposed cohort, (a.2) same derivation between source of exposed and nonexposed participants, (a.3) ascertainment of the exposure, and (a.4) demonstration that the outcome of interest was not present at the baseline, (b) comparability, with one item regarding comparability of cohorts on the basis of the design of the analysis; studies where the OR were calculated on the basis of the raw number of participants provided from the original papers or did not present adjusted analyses received zero points for comparability, and (c) outcomes, with three items regarding (c.1) adequate assessment of the outcome of interest, (c.2) adequate duration of follow‐up, and (c.3) adequacy of follow‐up. A study can be awarded a maximum of one point for each numbered item within the selection (a), and outcome (c) categories and a maximum of two stars can be given for comparability. The maximum score according to the NOS scale is 9 (highest quality). Scores of 0–3, 4–6, and 7–9 were assigned for the low, moderate, and high quality of studies, respectively (Wu et al., [63]). In case of disagreement, a consensus was reached through a discussion.
2.7 Meta‐analysis
A random‐effects meta‐analysis was conducted investigating the relationship between baseline PA and incident anxiety. First, we pooled data across all studies comparing incident anxiety (henceforth overall analysis) in highest PA levels group (the group of greater frequency, intensity, volume, energetic expenditure or other, from each study, as defined by the authors) vs the lowest PA level (reference group). Analysis of AOR and crude OR were conducted separately. For the AOR we pooled the estimates using the model with the greatest number of covariates presented by the authors. Second, subgroup analyses were conducted investigating the protective effects of PA on different anxiety classifications (anxiety symptoms only, any diagnosed anxiety disorders, and specific anxiety disorders, whenever data were available), continents, age groups of the sample at baseline (e.g., children/adolescents [<18 years], adults [18‐65 years] or elderly [>65 years]), PA assessments (e.g., comparison of high and low PA‐based on intensity, frequency, volume [time spent in PA, or a composition of two or more variables including metabolic equivalents [METS] units), and adjustments for potential confounders in AOR analysis (age and sex, body mass index [BMI] and smoking, age and sex and either BMI or smoking, and age and sex and both BMI and smoking). Third, we evaluated potential moderators through meta‐regression for AOR. The moderators tested were, length of follow‐up, person‐years, the total number of participants at baseline, study quality according to the NOS overall score, and the individual NOS scores for the selection of participants, outcome, and comparability, and the number of covariates included in the model. The covariates tested were chosen based on the previous studies to evaluate whether studies with greater sample size, length of follow‐up or greater person‐years were more powered to detect stronger effects, and to understand if studies with lesser bias risk or more adjusted for potential covariates are more likely to find significant or larger effects (Schuch et al., [47]; Thompson & Higgins, [59]). Fourth, we evaluated the publication bias using the Begg and Mazumdar ([9]) and Egger tests (Egger, Smith, Schneider, & Minder, [23]) and corrected through the Duval and Tweedie trim and fill procedure (Duval & Tweedie, [20]). To maximize statistical power, studies pooling participants with incident anxiety disorders along with incident depressive disorders were included in the main analysis. However, a sensitivity analysis excluding those papers was performed to evaluate whether they influenced the obtained results. The Q and I2 statistic were used to assess and to quantify the heterogeneity, respectively. Scores of <25%, 25–50%, and >50% indicated low, moderate, and high heterogeneity, respectively (Higgins, Thompson, Deeks, & Altman, [29]). Finally, the fail‐safe number of negative studies that would be required to nullify (i.e., make p > .05) the effect size was calculated (Rosenthal, [45]). All analyses were performed using the Comprehensive Meta‐Analysis software (version 3).
3 RESULTS
3.1 Search results
The initial search identified 9,966 titles. After the removal of duplicates, 7,382 abstracts were considered. At the full‐text review stage, 93 studies were assessed for eligibility and 80 were subsequently excluded (see Figure S1 for the flowchart and Material S2 for a list of excluded articles). One study was found using the reference lists of previously published studies. In total, 14 cohorts of 13 unique prospective studies were included in this review (Baumeister et al., [4]; Beard, Heathcote, Brooks, Earnest, & Kelly, [7]; Cynthia et al., [18]; Da Silva et al., [50]; Jonsdottir, Rodjer, Hadzibajramovic, Borjesson, & Ahlborg, [32]; Kang et al., [33]; McDowell et al., [36]; McDowell, Gordon, Andrews, MacDonncha, & Herring, [37]; Muller, Ganeshamoorthy, & Myers, [41]; Pasco et al., [43]; Sanchez‐Villegas et al., [46]; Strohle et al., [52]; Ten Have, de Graaf, & Monshouwer, [26]; Zainal & Newman, [65]).
3.2 Studies and participants characteristics
Across the 14 cohorts of 13 unique prospective studies, 75,831 individuals were included, with nearly equal sex distribution (median 50.1% males; interquartile range [IQ] = 43.72–65.37), followed up for a median of 3.5 years (IQ = 2.0–6.5). The total person‐years was 357,424. Eleven cohorts provided data for AOR (n = 69,037) and six cohorts provided for OR (n = 22,902). No study used an objective measure to evaluate PA. Across the 14 cohorts, 10 studies evaluated anxiety disorders using structured or semi‐structured diagnostic instruments or self‐reported physician diagnosis of anxiety disorders, and four used cut‐offs of anxiety screening instruments. A description of the included studies can be found in Table. Detailed information on PA classification within studies can be seen at Table S1.
1 Description of studies
| Study | Acronym | n | Follow‐up (years) | Persons‐years | Country | National representative | % males | Age group at baseline | Anxiety/stress disorder definition | PA measure | PA parameters |
|---|
| Baumeister et al. (4) | Study of Health in Pomerania (SHIP) | 1,952 | 4.5 | 8,784 | Germany | No | 49.7 | Adults | Munich‐CIDI | Baecke questionnaire | Frequency |
| Beard et al. (7) | Northern Rivers Mental Health Study (NoRMHS) | 968 | 2 | 1,936 | Australia | No | 43.3 | Adults | CIDI OR DIS | PA questionnaire (not specified) | Volume |
| Da Silva et al. (50) | The Whitehall II study | 9,309 | 8 | 74,472 | UK | No | 68.5 | Adults | GHQ (five items for anxiety scores ≥8) | Two questions on time and intensity of PA | METS/composed |
| Jonsdottir et al. (32) | ‐ | 2,818 | 2 | 5,636 | Sweden | No | 13 | Adults | HADS‐A > 10 | Adapted Santin and Grimby scale | Intensity |
| Kang et al. (33) | ‐ | 745 | 2 | 1,490 | Korea | No | 48.5 | Older adults | GMS‐B3 together with the AGECAT | Active versus inactive lifestyle "estimated by measuring work and leisure activities" | ? |
| LeardMann et al. (34) | The Millennium Cohort Study | 30,218 | 3 | 90,654 | US | No | 77.7 | Adults | PCL‐C ≥ 50 and DSM‐IV criteria for PTSD | Questions on average number of min/day and day/week in strength training, vigorous physical activity, and moderate or light physical activity. | METS/composed |
| McDowell, Dishman et al. (36) | The Irish Longitudinal Study on Ageing (TILDA) | 3,165 | 2 | 6,330 | Ireland | No | ? | Older adults | HADS‐A ≥ 8 | IPAQ | METS/composed |
| McDowell, Dishman et al. (36) | The Irish Longitudinal Study on Ageing (TILDA) | 3,236 | 2 | 6,472 | Ireland | No | 43.8 | Older adults | CIDI‐SF | IPAQ | METS/composed |
| Muller et al. (41) | Veterans Exercise Testing Study (VETS) | 5,826 | 9.6 | 55,929 | US | No | 96.4 | Adults | PTSD (ICD‐9‐CM 309.81) as diagnosed and recorded electronically by the patient's primary care physician | One question on engagement in moderate to vigorous PA, at least three times per week | ? |
| Pasco et al. (43) | Geelong Osteoporosis Study (GOS) | 547 | 4.1 | 2,242 | Australia | No | 56 | Older adults | SCID‐I/NP | PASE | METS/composed |
| Sanchez‐Villegas et al. (46) | The SUN study (Seguimiento University of Nav arra) | 10,381 | 6 | 62,286 | Spain | No | ? | Adults | Self‐report of physician diagnosis of anxiety | Questionnaire assessing time spent in 17 PA per week | METS/composed |
| Strohle et al. (52) | Early Developmental Stages of Psychopathology (EDSP) | 2,501 | 4 | 9,832 | Germany | Yes | 50.9 | Children/adolescents | CIDI | Four questions on PA frequency | Frequency |
| Ten Have et al. (26) | Netherlands Mental Health Survey and Incidence Study (NEMESIS) | 4,796 | 3 | 14,388 | The Netherlands | Yes | 50.6 | Adults | CIDI | Single question on hours per week of exercise | Volume |
| Zainal & Newman (65) | Midlife Development in the United States (MIDUS) | 2,605 | 9 | 23,445 | US | Yes | 43.7 | Adults | CIDI‐SF | Unclear | ? |
1 Abbreviations: AGECAT, Automated Geriatric Examination for Computer‐Assisted Taxonomy algorithm; CIDI, World Mental Health Composite International Diagnostic Interview; DIS, Diagnostic Interview Schedule; DSM, Diagnostic and Statistical Manual for Mental Disorders; GHQ, General Health Questionnaire; GMS‐B3, Geriatric Mental State Schedule; HADS‐A, Hospital Anxiety and Depression Scale; ICD, International Classification of Disease; IPAQ, International Physical Activity Questionnaire; METS, metabolic equivalents; PA, physical activity; PCL‐C, PTSD Checklist‐Civilian Version; PASE, Physical Activity Scale for the Elderly; SCID‐I/NP, Structured Clinical Interview for DSM‐IV Axis I Disorders, nonpatient version; UK, United Kingdom; US, United States.
- 2 a Adjusted odds ratio (OR).
- 3 b Data sent by the author.
- 4 c Crude OR.
- 5 d Data were calculated using raw numbers.
3.3 Study quality
The mean (standard deviation [SD]) NOS score was 6.71 (0.9), which is suggestive for a moderate to high methodological quality and a low risk of bias of the included studies. More detailed information about the quality assessment of the studies included in this meta‐analysis can be seen in Table S2.
3.4 PA and incident anxiety
3.4.1 Highest versus lowest PA
Higher self‐reported PA levels were associated with decreased incident anxiety when compared with lower PA levels in adjusted (AOR = 0.74, 95% CI = 0.62, 0.88, p = .001, I2 = 23.96, Q = 13.15, p = .21, N = 11, n = 69,037) (Figure) and crude OR analyses (OR = 0.80, 95% CI = 0.69, 0.92, p = 0.002, I2 = 0.00, Q = 3.78, p = .57, N = 7, n = 22,902) (Figure S2). Publication bias was evident for the AOR analysis in the Egger's (intercept = −1.61, p = .01) but not in the Begg and Mazumdar (τ = −0.21, p = .35) test. Both Egger's (intercept = −1.29, p = .05) and the Begg and Mazumdar (τ = −0.66, p = .06) are suggestive of publication bias for crude OR analysis. The adjusted values, according to the Duval and Tweedie trim and fill technique remained significant: AOR = 0.86 (95% CI = 0.69, 0.99) and OR = 0.83 (95% CI = 0.72, 0.95). The number of studies with negative results required to nullify the effects of PA on incident anxiety was 40 and 13, in AOR and OR analyses, respectively.
3.4.2 Subgroup analyses
Subgroup analyses were performed to explore the effects of self‐reported PA on incident anxiety symptoms only, on anxiety disorders, excluding anxiety symptoms, on specific disorders (PTSD, obsessive–compulsive disorder [OCD], generalized anxiety disorder [GAD], social phobia, panic, agoraphobia) and, exploring potential differences across different geographical regions, age groups, PA assessments, completers of the public health recommended dose of PA, and the use of potential confounders. Subgroup analyses were performed both for adjusted and crude analysis. The results are briefly summarized below and the detailed results of subgroups analyses can be seen in Table. The detailed list of studies included in each analysis can be seen in Table S3 for AOR and Table S4 for OR.
2 Subgroup analysis exploring the effects of PA on incident anxiety in different continents, physical activity assessment unity, presence of diagnosed depression and age
| Analysis | | | Meta‐analysis | Heterogeneity | |
|---|
| Studies with adjusted odds ratio (AOR) | Number of cohorts (arms) | Number participants | AOR | 95% CI | p value | I2 | Q | p value | Classic fail‐safe N |
|---|
| Overall | 11 | 69,037 | 0.748 | 0.629 | 0.889 | .001 | 23.96 | 13.15 | 0.21 | 40 |
| Diagnosis | | | | | | | | | | |
| Any disorder | 8 | 56,236 | 0.748 | 0.608 | 0.921 | .006 | 35.83 | 9.35 | .18 | 17 |
| Anxiety symptoms | 4 | 16,037 | 0.711 | 0.483 | 1.046 | .083 | 13.77 | 3.47 | .32 | 1 |
| Agoraphobia | 2 | 7,297 | 0.426 | 0.185 | 0.988 | .047 | 0.00 | 0.14 | .70 | n/a |
| Panic | 2 | 7,317 | 0.417 | 0.141 | 1.222 | .111 | 0.00 | 0.76 | .38 | n/a |
| OCD | 2 | 7,333 | 1.002 | 0.213 | 4.714 | .998 | 0.00 | 0.12 | .73 | n/a |
| PTSD | 2 | 32,753 | 0.579 | 0.393 | 0.855 | .006 | 0.00 | 0.12 | .72 | n/a |
| Social phobia | 2 | 7,220 | 0.790 | 0.411 | 1.506 | .473 | 0.00 | 0.79 | .37 | n/a |
| Specific phobia | 2 | 7,076 | 0.653 | 0.401 | 1.048 | .074 | 0.00 | 0.14 | .70 | n/a |
| GAD | 4 | 13,164 | 0.768 | 0.488 | 1.209 | .255 | 0.00 | 1.18 | .75 | 0 |
| Continent | | | | | | | | | | |
| Asia | 1 | 745 | 0.310 | 0.100 | 0.963 | .043 | 0.00 | 0.00 | .00 | n/a |
| North America | 2 | 32,823 | 0.695 | 0.468 | 1.031 | .071 | 30.12 | 1.43 | .23 | n/a |
| Europe | 7 | 34,922 | 0.821 | 0.692 | 0.976 | .025 | 11.36 | 6.77 | .34 | 8 |
| Oceania | 1 | 547 | 0.555 | 0.233 | 1.301 | .174 | 0 | 0 | .00 | n/a |
| Age at baseline | | | | | | | | | | |
| Adults | 7 | 62,079 | 0.815 | 0.695 | 0.957 | .013 | 12.18 | 6.83 | .33 | 11 |
| Children/adolescents | 1 | 2,051 | 0.520 | 0.299 | 0.901 | .021 | 0.00 | 0.00 | .00 | n/a |
| Older | 3 | 4,457 | 0.593 | 0.324 | 1.083 | .089 | 20.21 | 2.50 | .28 | n/a |
| PA assessment | | | | | | | | | | |
| Frequency | 2 | 4,453 | 0.751 | 0.412 | 1.370 | .351 | 76.87 | 4.32 | .03 | n/a |
| Intensity | 2 | 33,036 | 0.580 | 0.405 | 0.830 | .003 | 0.00 | 0.03 | .85 | n/a |
| METS/composed | 4 | 23,402 | 0.772 | 0.594 | 1.000 | .050 | 0.00 | 1.03 | .79 | 0 |
| Volume | 1 | 4,796 | 0.710 | 0.374 | 1.347 | .295 | 0.00 | 0.00 | .00 | n/a |
| 150 min mod/vig PA per week | 3 | 42,692 | 0.717 | 0.542 | 0.949 | .020 | 0.00 | 1.85 | .70 | n/a |
| Adjustment | | | | | | | | | | |
| Age and sex | 10 | 68,292 | 0.779 | 0.667 | 0.910 | .002 | 13.75 | 10.43 | .31 | 27 |
| BMI | 5 | 36,188 | 0.680 | 0.541 | 0.851 | .001 | 0.00 | 2.14 | .71 | 10 |
| Smoking | 5 | 45,191 | 0.810 | 0.639 | 1.028 | .083 | 32.60 | 5.93 | .20 | 2 |
| Age and sex + (BMI or smoking) | 7 | 51,686 | 0.795 | 0.656 | 0.963 | .019 | 20.96 | 7.59 | .27 | 9 |
| Age and sex + BMI + smoking | 5 | 45,191 | 0.810 | 0.639 | 1.028 | .083 | 32.60 | 5.93 | .20 | 2 |
| Studies with crude odds ratio (OR) | | | OR | 95% CI | p value | I2 | Q | | |
| Overall | 6 | 22,902 | 0.804 | 0.699 | 0.924 | .002 | 0.00 | 3.798 | .57 | 13 |
| Diagnosis | | | | | | | | | | |
| Any disorder | 3 | 10,030 | 0.676 | 0.361 | 1.269 | .223 | 38.76 | 3.26 | .19 | n/a |
| Anxiety symptoms | 4 | 16,037 | 0.741 | 0.597 | 0.920 | .007 | 0.00 | 0.86 | .75 | 4 |
| Continent | | | | | | | | | | |
| Asia | 1 | 745 | 0.650 | 0.320 | 0.920 | .234 | 0.00 | 0.00 | .00 | n/a |
| North America | 1 | 5,826 | 0.910 | 0.707 | 1.171 | .464 | 0.00 | 0.00 | .00 | n/a |
| Europe | 3 | 15,292 | 0.751 | 0.599 | 0.943 | .014 | 0.00 | 0.72 | .60 | 2 |
| Oceania | 1 | 968 | 0.470 | 0.119 | 1.858 | .282 | 0.00 | 0.00 | .00 | n/a |
| Age at baseline | | | | | | | | | | |
| Adults | 4 | 18,921 | 0.850 | 0.712 | 1.014 | .072 | 0.00 | 2.63 | .45 | 1 |
| Older | 2 | 3,910 | 0.735 | 0.587 | 0.921 | .007 | 0.00 | 0.18 | .22 | n/a |
| PA assessment | | | | | | | | | | |
| Intensity | 1 | 2,818 | 0.550 | 0.242 | 1.248 | .153 | 0.00 | 0.00 | .00 | n/a |
| METS/composed | 3 | 18,300 | 0.833 | 0.719 | 0.965 | .015 | 0.00 | 1.37 | .50 | n/a |
| Volume | 1 | 968 | 0.470 | 0.094 | 2.358 | .359 | 0.00 | 0.00 | .00 | n/a |
| 150 min mod/vig PA per week | 2 | 12,474 | 0.771 | 0.609 | 0.977 | .031 | 0.00 | 0.11 | .64 | n/a |
- 6 Abbreviations: AOR, Adjusted odds ratio; BMI, Body mass index; 95% CI, 95% confidence interval; n/a, not available; GAD, generalized anxiety disorder; METS/composed, metabolic equivalents; OCD, obsessive–compulsive disorder; OR, odds ratio; PA, physical activity; PTSD, posttraumatic stress disorder.
- 7 Bold values indicates p < 0.05.
Analysis of adjusted data reported in studies
Higher self‐reported PA levels decreased the risk of incident anxiety disorders (AOR = 0.62, 95% CI = 0.62, 0.88). PA had significant protective effects against the development of agoraphobia (AOR = 0.42, 95% CI = 0.18, 0.98) and PTSD (AOR = 0.57, 95% CI = 0.39, 0.88), with no significant effects among the other analyzed anxiety disorders, including phobias, panic, and GAD. The protective effects were significant across two continents: Asia (AOR = 0.31, 95% CI = 0.10, 0.96), and Europe (AOR = 0.82, 95% CI = 0.69, 0.97); for children/adolescents (AOR = 0.52, 95% CI = 0.29, 0.90) and adults (AOR = 0.81, 95% CI = 0.69, 0.95). Instruments assessing PA according to the different intensities (AOR = 0.58, 95% CI = 0.40, 0.83), or METS (AOR = 0.77, 95% CI = 0.59, 1.00) found that higher intensity or energetic expenditure during PA were significantly associated with reduced incident anxiety. Also, completing the recommendation of 150 min of moderate/vigorous PA was associated with a lower risk of incident anxiety (AOR = 0.71, 95% CI = 0.54, 0.94). Analyses adjusting for age and sex, BMI, or combining age, sex, with BMI or smoking were significant.
Analysis of crude data reported in studies
In nonadjusted analyses, higher levels of self‐reported PA conferred protective effects on the emergence of anxiety symptoms (OR = 0.74, 95% CI = 0.59, 0.92). There were no studies presenting crude ORs for specific anxiety disorders. Effects were only significant for studies in Europe (OR = 0.75, 95% CI = 0.59, 0.94) and those that included older adults at baseline (OR = 0.73, 95% CI = 0.58, 0.92). Studies that established PA using questions on measures based in METS or combined intensity, frequency or volume of PA found significant effects (OR = 0.83, 95% CI = 0.71, 0.96). Performing at least 150 min of moderate or vigorous PA per week conferred protective effects (OR = 0.77, 95% CI = 0.60, 0.97).
3.5 Sensitivity analysis
One study (Pasco et al., [43]) pooled data for the incidence of depression and anxiety disorders together and one study used self‐report of physician diagnosis of anxiety as the outcome measure (Sanchez‐Villegas et al., [46]). Sensitivity analyses removing the study that pooled depression and anxiety together (AOR = 0.75, 95% CI = 0.63, 0.90), or the study that used self‐report of physician diagnosis of anxiety as the outcome measure did not change the findings (AOR = 0.73, 95% CI = 0.60, 0.90), or removing both (AOR = 0.74, 95% CI = 0.60, 0.91) have not changed the protective effects of PA on incident anxiety.
3.6 Meta‐regression
We explored the potential moderating role of sample size at baseline, the length of follow‐up, individual study person‐years, the % of males, the number of covariates used in each study for adjusted analyses and the study quality according to the NOS scale. All variables tested were used as continuous variables. No significant moderators were identified via meta‐regression (Table).
3 Univariate meta‐regression of moderators of the effects of PA on incident anxiety for overall analyses at a time
| Moderator | Number cohorts | β | 95% CI | p value | R2 |
|---|
| Studies presenting AOR | | | | | | |
| Sample size | 11 | <−0.001 | <−0.001 | <0.001 | .23 | 0.45 |
| Length of follow‐up | 11 | 0.048 | −0.043 | 0.139 | .30 | 0.00 |
| Person‐years | 11 | <0.001 | <−0.001 | <0.001 | .62 | 0.00 |
| Number of covariates | 9 | −0.010 | −0.054 | 0.032 | .62 | 0.11 |
| % of dropout | 8 | <−0.001 | <−0.026 | 0.020 | .80 | 0.00 |
| Study quality | 11 | −0.100 | −0.522 | 0.321 | .64 | 0.00 |
| Study quality (selection of participants) | 11 | 0.091 | −0.385 | 0.567 | .95 | 0.00 |
| Study quality (comparability) | 11 | 0.410 | −0.203 | 1.023 | .19 | 0.41 |
| Study quality (outcome) | 11 | −0.568 | −1.084 | 0.052 | .06 | 1.00 |
8 Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; PA, physical activity.
4 DISCUSSION
To the best of our knowledge, this is the first meta‐analysis demonstrating that higher levels of self‐reported PA are associated with a decreased risk of both anxiety symptoms and anxiety disorders in people free of anxiety at baseline. This protective effect of self‐reported PA was evident across Asia and Europe, children/adolescents and adults, regardless of how self‐reported PA was captured and after the adjustment for publication bias and other confounders such as age, sex, smoking, and BMI. Specifically, our data suggest that people with higher levels of self‐reported PA have a 27% and 21% lower risk for developing anxiety when compared to people with lower levels of self‐reported PA, even after adjusting for potential confounders, in AOR and OR analyses, respectively.
The protective effects of self‐reported PA on anxiety development are in line with previous cross‐sectional studies showing that lower PA levels are associated with current anxiety symptoms (da Silva, Schuch, Costa, Hirakata, & Manfro, [49]), and that exercise training has anxiolytic effects in people with and without anxiety disorders (Herring et al., [28]; Herring et al., [27]; Stubbs, Koyanagi et al., [55]). In the present study, we have found initial evidence that PA has potential protective effects on incident‐specific anxiety disorders, such as agoraphobia and PTSD. No protective effects on another diagnosis, such as panic, OCD, social phobia, specific phobias, or GAD were found. Previous cohorts have found the existence of disorder‐specific risk factors (Blanco et al., [12]). For instance, the early parental loss is a protective factor for panic (OR = 0.9, 95% CI = 0.7–1.0), but a risk factor for PTSD (OR = 1.2, 95% CI = 1.0–1.4) (Blanco et al., [12]). Although we have found effects only for PTSD and agoraphobia, due to the small number of studies included in subgroup analysis, only two studies provided data for incident agoraphobia (Strohle et al., [52]; Ten Have et al., [26]), PTSD (LeardMann et al., [34]; Strohle et al., [52]), panic Strohle et al., [52]; Ten Have et al., [26]), OCD (Strohle et al., [52]; Ten Have et al., [26]), social phobia (Strohle et al., [52]; Ten Have et al., [26]), and specific phobias (Strohle et al., [52]; Ten Have et al., [26]), and four for GAD (McDowell, Dishman et al., [36]; Strohle et al., [52]; Ten Have et al., [26]; Zainal & Newman, [65]), it is precocious to determine whether these effects are common to all anxiety disorders or are disorder‐specific and more research is needed to adequately address this point.
The potential mechanisms that may underlie the protective effects of self‐reported PA on incident anxiety are unclear. From a biological perspective, self‐reported PA influence some of the core biological processes thought to be involved in the onset of anxiety; for instance, habitual self‐reported PA may promote neuroregeneration, as shown by increased levels of neuroregeneration markers, such as brain‐derived neurotrophic factor (Szuhany, Bugatti, & Otto, [57]) or the balance between inflammatory/anti‐inflammatory (Gleeson et al., [24]) and oxidative/antioxidative markers (Bogdanis et al., [13]; Schuch et al., [48]). Research in healthy adults has shown that higher levels of PA can reduce interleukin 6 and this may be the mechanism through which it exerts its protective effect (Jankord & Jemiolo, [31]). From a psychological perspective, self‐reported PA can reduce anxiety sensitivity (Asmundson et al., [3]), which is the main focus in the genesis and maintenance of panic and other anxiety disorders (Paulus, Gallagher, Bartlett, Tran, & Vujanovic, [44]). Also, engaging in PA, particularly structured exercise, may directly improve psychological factors such as increasing self‐efficacy regarding the ability to exert control over potential threats (Anderson & Shivakumar, [2]), thereby reducing the risk of developing anxiety. The role of self‐reported PA in anxiety may also not necessarily be causal, but pleiotropic, with common genetic factors underlying both propensity to exercise and mental health problems (De Moor, Beem, Stubbe, Boomsma, & De Geus, [40]).
The present study is, however, not free from limitations. First, evidence of publication bias was encountered both for adjusted and crude analysis. Nonetheless, the analyses remained significant after adjusting for publication bias, trimming three studies in each analysis: AOR = 0.80 (95% CI = 0.66–0.98) and OR = 0.82 (95% CI = 0.70–0.95). The publication bias found is at best modest and have not changed the overall effect size significantly. Second, we have included only 11 studies that provided effects adjusting for relevant covariates and six studies that provided crude OR and the results should be read with this limitation in consideration. The small number of studies also affects the subgroup analyses for many of the potential moderators, that should be considered in light of a potential lack of statistical power to drawn definitive conclusions. Similarly, given the few numbers of studies that could be included, we were likely underpowered to detect significant linear moderators of the relationship between PA on anxiety. Third, the included studies assessed self‐reported PA levels in a time frame of days or weeks. Thus, it is not clear whether being engaged in higher levels of self‐reported PA for longer periods implies greater protection in comparison to shorter periods, nor whether increasing or decreasing PA over the years would alter the protective effects. Fourth, no studies applied objective measures of PA, which makes the present analyses more vulnerable to recall biases, potentially resulting in over or even under estimations when compared with objective measures (Dyrstad, Hansen, Holme, & Anderssen, [21]). Also, only five of the included studies used validated PA measures and PA characterization differed across instruments. Fifth, studies included in this meta‐analysis did not specifically exclude people with incident depression from analysis, which is highly comorbid with anxiety and also associated with low PA. Previous meta‐analytic evidence demonstrates that PA is protective against depression, with a slightly higher AOR than observed here for anxiety (AOR = 0.83, 95% CI = 0.79, 0.88; I2 = 0.00, N = 49) (Schuch et al., [47]). Although it is possible that the effect of PA on anxiety is in part mediated through the effect of PA on depression, previous research that effects of exercise on anxiety are equal to and independent of effects of exercise on depression (Hiles, Lamers, Milaneschi, & Penninx, [30]). Sixth, it is unclear whether the use of psychotropic medications, particularly the off‐label use influence this relationship (Bernard & Carayol, [11]). Last, two studies identified in our screening presented their results in risk ratio or hazard ratio (Brunes, Gudmundsdottir, & Augestad, [14]; Yu, Ter Riet, Puhan, & Frei, [64]), and one using a referent group different than the lower or the higher self‐reported PA levels (Harvey et al., [25]), could not be included in the meta‐analysis. Of these, two studies, are in accordance with the findings of the present meta‐analysis (Brunes et al., [14]; Yu et al., [64]), and one not (Harvey et al., [25]).
In light of these limitations, we propose that: (a) Further well‐designed prospective studies using nationally representative samples and validated and objective PA measures are warranted; (b) the effects of different PA types and "dosages" on subsequent risk for anxiety and anxiety disorders should be investigated; (c) considering the burden of disease and the global impact of mental illness, further studies should evaluate the cost‐effectiveness of PA in the prevention of anxiety.
Anxiety is a highly prevalent disorder and one of the leading causes of global burden and our data suggests that higher levels of PA may confer protection against new cases in the general population. Given the established health benefits of PA more broadly, our data add further evidence to the growing calls for the necessity to promote PA at the population‐level. This goal is facilitated by multiple layers of intervention from individual‐level healthcare, workplace, and education‐based health behavior change or PA programs, to broad‐scale national policy change including urban planning to create PA‐promoting, accessible and clean environments. Increasing PA should be considered a priority on preventing mental and physical chronic conditions and reducing overall mortality.
5 CONCLUSION
Higher levels of self‐reported PA are associated with lower odds of developing future anxiety and anxiety disorders. Our data further emphasize the importance of policies targeting increased PA levels to reduce incident anxiety.
ACKNOWLEDGMENTS
The authors thank the following researchers for providing the additional information and/or data for this study: Ingborn Jonsdotirr and Sebastian Baumeister. B. S. is supported by Health Education England and the National Institute for Health Research HEE NIHR ICA Program Clinical Lectureship (ICA‐CL‐2017‐03‐001). B. S. is part supported by the Maudsley Charity and the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South London at King's College Hospital NHS Foundation Trust. The views expressed in this article are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. S. H. salary for research unrelated to this study is supported by an AstraZeneca grant paid to her employer (University of Newcastle). J. F. is supported by a Blackmores Institute Fellowship.
CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.
AUTHOR CONTRIBUTIONS
F. S. and B. S. designed the protocol of the study. F. S., A. H., S. H., and P. Z. performed the searches within the this found. F. S. and S. H. extracted the data from the studies. F. S. and J. M. evaluated the quality of the studies. All authors have read and contributed with the interpretation of the results, and with the elaboration of the text.
GRAPH: Supporting information
GRAPH: Supporting information
GRAPH: Supporting information
[
REFERENCES
1
American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders (5th ed ed.). Arlington, TX : American Psychiatric Association.
2
Anderson, E., & Shivakumar, G. (2013). Effects of exercise and physical activity on anxiety. Frontiers in Psychiatry, 4, 27. https://doi.org/10.3389/fpsyt.2013.00027
3
Asmundson, G. J. G., Fetzner, M. G., DeBoer, L. B., Powers, M. B., Otto, M. W., & Smits, J. A. J. (2013). Let's get physical: A contemporary review of the anxiolytic effects of exercise for anxiety and its disorders. Depression and Anxiety, 30 (4), 362 – 373. https://doi.org/10.1002/da.22043
4
Baumeister, S. E., Leitzmann, M. F., Bahls, M., Dörr, M., Schmid, D., Schomerus, G., ... Grabe, H. J. (2017). Associations of leisure‐time and occupational physical activity and cardiorespiratory fitness with incident and recurrent major depressive disorder, depressive symptoms, and incident anxiety in a general population. Journal of Clinical Psychiatry, 78 (1), e41 – e47. https://doi.org/10.4088/JCP.15m10474
5
Baxter, A. J., Vos, T., Scott, K. M., Ferrari, A. J., & Whiteford, H. A. (2014). The global burden of anxiety disorders in 2010. Psychological Medicine, 44 (11), 2363 – 2374. https://doi.org/10.1017/s0033291713003243
6
Baxter, A. J., Vos, T., Scott, K. M., Norman, R. E., Flaxman, A. D., Blore, J., & Whiteford, H. A. (2014). The regional distribution of anxiety disorders: Implications for the Global Burden of Disease Study, 2010. International Journal of Methods in Psychiatric Research, 23 (4), 422 – 438.
7
Beard, J. R., Heathcote, K., Brooks, R., Earnest, A., & Kelly, B. (2007). Predictors of mental disorders and their outcome in a community based cohort. Social Psychiatry and Psychiatric Epidemiology, 42 (8), 623 – 630. https://doi.org/10.1007/s00127‐007‐0182‐3
8
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561 – 571.
9
Begg, C. B., & Mazumdar, M. (1994). Operating characteristics of a rank correlation test for publication bias. Biometrics, 50 (4), 1088 – 1101.
Belem da Silva, C. T., Costa, M. A., Bortoluzzi, A., Pfaffenseller, B., Vedana, F., Kapczinski, F., & Manfro, G. G. (2017). Cytokine levels in panic disorder: Evidence for a dose‐response relationship. Psychosomatic Medicine, 79 (2), 126 – 132. https://doi.org/10.1097/psy.0000000000000384
Bernard, P., & Carayol, M. (2015). A commentary on the importance of controlling for medication use within trials on the effects of exercise on depression and anxiety. Mental Health and Physical Activity, 9, 10 – 15. https://doi.org/10.1016/j.mhpa.2015.07.001
Blanco, C., Rubio, J., Wall, M., Wang, S., Jiu, C. J., & Kendler, K. S. (2014). Risk factors for anxiety disorders: Common and specific effects in a national sample. Depression and Anxiety, 31 (9), 756 – 764. https://doi.org/10.1002/da.22247
Bogdanis, G. C., Stavrinou, P., Fatouros, I. G., Philippou, A., Chatzinikolaou, A., Draganidis, D., ... Maridaki, M. (2013). Short‐term high‐intensity interval exercise training attenuates oxidative stress responses and improves antioxidant status in healthy humans. Food and Chemical Toxicology, 61, 171 – 177. S0278‐6915(13)00352‐9 [pii]. https://doi.org/10.1016/j.fct.2013.05.046
Brunes, A., Gudmundsdottir, S. L., & Augestad, L. B. (2015). Gender‐specific associations between leisure‐time physical activity and symptoms of anxiety: The HUNT study. Social Psychiatry and Psychiatric Epidemiology, 50 (3), 419 – 427. https://doi.org/10.1007/s00127‐014‐0915‐z
Cairns, K. E., Yap, M. B. H., Pilkington, P. D., & Jorm, A. F. (2014). Risk and protective factors for depression that adolescents can modify: A systematic review and meta‐analysis of longitudinal studies. Journal of Affective Disorders, 169, 61 – 75. https://doi.org/10.1016/j.jad.2014.08.006
Caspersen, C. J., Powell, K. E., & Christenson, G. M. (1985). Physical activity, exercise, and physical fitness: Definitions and distinctions for health‐related research. Public Health Reports, 100 (2), 126 – 31.
Craig, C. L., Marshall, A. L., Sjostrom, M., Bauman, A. E., Booth, M. L., Ainsworth, B. E., ... Oja, P. (2003). International physical activity questionnaire: 12‐country reliability and validity. Medicine and Science in Sports and Exercise, 35 (8), 1381 – 1395. https://doi.org/10.1249/01.MSS.0000078924.61453.FB
Cynthia, A. L., Molly, L. K., Besa, S., Alyson, J. L., Edward, J. B., Timothy, S. W., ... Smith, T. C. Millennium Cohort Study Team (2011). Prospectively assessed posttraumatic stress disorder and associated physical activity. Public Health Reports, 126 (3), 371 – 383. https://doi.org/10.1177/003335491112600311
Dillon, C. B., McMahon, E., O'Regan, G., & Perry, I. J. (2018). Associations between physical behaviour patterns and levels of depressive symptoms, anxiety and well‐being in middle‐aged adults: A cross‐sectional study using isotemporal substitution models. BMJ Open, 8 (1), e018978. https://doi.org/10.1136/bmjopen‐2017‐018978
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel‐plot‐based method of testing and adjusting for publication bias in meta‐analysis. Biometrics, 56 (2), 455 – 463.
Dyrstad, S. M., Hansen, B. H., Holme, I. M., & Anderssen, S. A. (2014). Comparison of self‐reported versus accelerometer‐measured physical activity. Medicine and Science in Sports and Exercise, 46 (1), 99 – 106. https://doi.org/10.1249/MSS.0b013e3182a0595f
Edmondson, D., & von Kanel, R. (2017). Post‐traumatic stress disorder and cardiovascular disease. Lancet Psychiatry, 4 (4), 320 – 329. https://doi.org/10.1016/s2215‐0366(16)30377‐7
Egger, M., Smith, G. D., Schneider, M., & Minder, C. (1997). Bias in meta‐analysis detected by a simple, graphical test. BMJ, 315 (7109), 629 – 634. https://doi.org/10.1136/bmj.315.7109.629
Gleeson, M., Bishop, N. C., Stensel, D. J., Lindley, M. R., Mastana, S. S., & Nimmo, M. A. (2011). The anti‐inflammatory effects of exercise: Mechanisms and implications for the prevention and treatment of disease. Nature Reviews Immunology, 11 (9), 607 – 615.
Harvey, S. B., Overland, S., Hatch, S. L., Wessely, S., Mykletun, A., & Hotopf, M. (2017). Exercise and the prevention of depression: Results of the HUNT Cohort Study. American Journal of Psychiatry, 175, 28 – 36. appiajp201716111223. https://doi.org/10.1176/appi.ajp.2017.16111223
Ten Have, M., de Graaf, R., & Monshouwer, K. (2011). Physical exercise in adults and mental health status findings from the Netherlands mental health survey and incidence study (NEMESIS). Journal of Psychosomatic Research, 71 (5), 342 – 348. https://doi.org/10.1016/j.jpsychores.2011.04.001
Herring, M. P., Lindheimer, J. B., & O'Connor, P. J. (2014). The effects of exercise training on anxiety. American Journal of Lifestyle Medicine, 8 (6), 388 – 403. https://doi.org/10.1177/1559827613508542
Herring, M. P., O'Connor, P. J., & Dishman, R. K. (2010). The effect of exercise training on anxiety symptoms among patients: A systematic review. Archives of Internal Medicine, 170 (4), 321 – 331. https://doi.org/10.1001/archinternmed.2009.530
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in meta‐analyses. BMJ (Clinical Research Ed.), 327 (7414), 557 – 560.
Hiles, S. A., Lamers, F., Milaneschi, Y., & Penninx, B. W. J. H. (2017). Sit, step, sweat: Longitudinal associations between physical activity patterns, anxiety and depression. Psychological Medicine, 47 (8), 1466 – 1477. https://doi.org/10.1017/S0033291716003548
Jankord, R., & Jemiolo, B. (2004). Influence of physical activity on serum IL‐6 and IL‐10 levels in healthy older men. Medicine and Science in Sports and Exercise, 36 (6), 960 – 964.
Jonsdottir, I. H., Rodjer, L., Hadzibajramovic, E., Borjesson, M., & Ahlborg, G., Jr. (2010). A prospective study of leisure‐time physical activity and mental health in Swedish health care workers and social insurance officers. Preventive Medicine, 51 (5), 373 – 377. https://doi.org/10.1016/j.ypmed.2010.07.019
Kang, H.‐J., Bae, K.‐Y., Kim, S.‐W., Shin, I.‐S., Yoon, J. ‐S., & Kim, J.‐M. (2015). Anxiety symptoms in Korean elderly individuals: A two‐year longitudinal community study. International Psychogeriatrics, 28 (3), 423 – 433. https://doi.org/10.1017/S1041610215001301
LeardMann, C. A., Kelton, M. L., Smith, B., Littman, A. J., Boyko, E. J., Wells, T. S., ... Millennium Cohort Study Team (2011). Prospectively assessed posttraumatic stress disorder and associated physical activity. Public Health Reports, 126 (3), 371 – 383.
Martinsen, E. W. (2008). Physical activity in the prevention and treatment of anxiety and depression. Nordic Journal of Psychiatry, 62 (sup47), 25 – 29. https://doi.org/10.1080/08039480802315640
McDowell, C. P., Dishman, R. K., Vancampfort, D., Hallgren, M., Stubbs, B., MacDonncha, C., & Herring, M. P. (2018). Physical activity and generalized anxiety disorder: Results from The Irish Longitudinal Study on Ageing (TILDA). International Journal of Epidemiology, 47, dyy141 – 1453. dyy141. https://doi.org/10.1093/ije/dyy141
McDowell, C. P., Gordon, B. R., Andrews, K. L., MacDonncha, C., & Herring, M. P. (2018). Associations of physical activity with anxiety symptoms and status: Results from The Irish longitudinal study on ageing. Epidemiology and Psychiatric Sciences, 1 – 10. https://doi.org/10.1017/s204579601800001x
Meier, S. M., Mattheisen, M., Mors, O., Mortensen, P. B., Laursen, T. M., & Penninx, B. W. (2016). Increased mortality among people with anxiety disorders: Total population study. British Journal of Psychiatry, 209 (3), 216 – 221. https://doi.org/10.1192/bjp.bp.115.171975
Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Preferred reporting items for systematic reviews and meta‐analyses: The PRISMA statement. PLoS Medicine, 6 (7), e1000097. https://doi.org/10.1371/journal.pmed.1000097
De Moor, M. H., Beem, A. L., Stubbe, J. H., Boomsma, D. I., & De Geus, E. J. (2006). Regular exercise, anxiety, depression and personality: A population‐based study. Preventive Medicine, 42 (4), 273 – 279. doi:S0091‐7435(05)00233‐1 [pii]. https://doi.org/10.1016/j.ypmed.2005.12.002
Muller, J., Ganeshamoorthy, S., & Myers, J. (2017). Risk factors associated with posttraumatic stress disorder in US veterans: A cohort study. PLoS One, 12 (7), e0181647. https://doi.org/10.1371/journal.pone.0181647
Nguyen, H. Q., Fan, V. S., Herting, J., Lee, J., Fu, M., Chen, Z., ... Adams, S. G. (2013). Patients with COPD with higher levels of anxiety are more physically active. Chest, 144 (1), 145 – 151. https://doi.org/10.1378/chest.12‐1873
Pasco, J. A., Williams, L. J., Jacka, F. N., Henry, M. J., Coulson, C. E., Brennan, S. L., ... Berk, M. (2011). Habitual physical activity and the risk for depressive and anxiety disorders among older men and women. International Psychogeriatrics, 23 (2), 292 – 298. https://doi.org/10.1017/s1041610210001833
Paulus, D. J., Gallagher, M. W., Bartlett, B. A., Tran, J., & Vujanovic, A. A. (2018). The unique and interactive effects of anxiety sensitivity and emotion dysregulation in relation to posttraumatic stress, depressive, and anxiety symptoms among trauma‐exposed firefighters. Comprehensive Psychiatry, 84, 54 – 61. https://doi.org/10.1016/j.comppsych.2018.03.012
Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psychological Bulletin, 86 (3), 638 – 641.
Sanchez‐Villegas, A., Ara, I., Guillén‐Grima, F., Bes‐Rastrollo, M., Javier, J., Varo, C., & Martínez‐González, M. A. (2008). Physical activity, sedentary index, and mental disorders in the SUN Cohort Study. Medicine & Science in Sports & Exercise, 40 (5), 827 – 834.
Schuch, F., Vancampfort, D., Firth, J., Rosenbaum, S., Ward, P. B., Silva, E., ... Stubbs, B. (2018). Physical activity and incident depression: A meta‐analysis of prospective cohort studies. The American Journal of Psychiatry, 175 (7), 631 – 648.
Schuch, F. B., Vasconcelos‐Moreno, M. P., Borowsky, C., Zimmermann, A. B., Wollenhaupt‐Aguiar, B., Ferrari, P., & de Almeida Fleck, M. P. (2014). The effects of exercise on oxidative stress (TBARS) and BDNF in severely depressed inpatients. European Archives of Psychiatry and Clinical Neuroscience, 264 (7), 605 – 613. https://doi.org/10.1007/s00406‐014‐0489‐5
da Silva, C. T. B., Schuch, F., Costa, M., Hirakata, V., & Manfro, G. G. (2014). Somatic, but not cognitive, symptoms of anxiety predict lower levels of physical activity in panic disorder patients. Journal of Affective Disorders, 164, 63 – 68.
Da Silva, M. A., Singh‐Manoux, A., Brunner, E. J., Kaffashian, S., Shipley, M. J., Kivimaki, M., & Nabi, H. (2012). Bidirectional association between physical activity and symptoms of anxiety and depression: The Whitehall II study. European Journal of Epidemiology, 27 (7), 537 – 546. https://doi.org/10.1007/s10654‐012‐9692‐8
Smith, K. J., Deschenes, S. S., & Schmitz, N. (2018). Investigating the longitudinal association between diabetes and anxiety: A systematic review and meta‐analysis. Diabetic Medicine, 35, 677 – 693. https://doi.org/10.1111/dme.13606
Strohle, A., Hofler, M., Pfister, H., Muller, A. G., Hoyer, J., Wittchen, H. U., & Lieb, R. (2007). Physical activity and prevalence and incidence of mental disorders in adolescents and young adults. Psychological Medicine, 37 (11), 1657 – 1666. https://doi.org/10.1017/s003329170700089x
Strohle, A., Graetz, B., Scheel, M., Wittmann, A., Feller, C., Heinz, A., & Dimeo, F. (2010). The acute antipanic and anxiolytic activity of aerobic exercise in patients with panic disorder and healthy control subjects. Journal of Psychiatric Research, 43, 1013 – 1017.
Stroup, D. F., Berlin, J. A., Morton, S. C., Olkin, I., Williamson, G. D., Rennie, D., ... Thacker, S. B. (2000). Meta‐analysis of observational studies in epidemiology: A proposal for reporting. Meta‐analysis Of Observational Studies in Epidemiology (MOOSE) group. Journal of the American Medical Association, 283 (15), 2008 – 2012.
Stubbs, B., Koyanagi, A., Hallgren, M., Firth, J., Richards, J., Schuch, F., ... Vancampfort, D. (2017). Physical activity and anxiety: A perspective from the World Health Survey. Journal of Affective Disorders, 208, 545 – 552. https://doi.org/10.1016/j.jad.2016.10.028
Stubbs, B., Vancampfort, D., Rosenbaum, S., Firth, J., Cosco, T., Veronese, N., ... Schuch, F. (2017). An examination of the anxiolytic effects of exercise for people with anxiety and stress‐related disorders: A meta‐analysis. Psychiatry Research, 249, 102 – 108.
Szuhany, K. L., Bugatti, M., & Otto, M. W. (2015). A meta‐analytic review of the effects of exercise on brain‐derived neurotrophic factor. Journal of Psychiatric Research, 60 (0), 56 – 64. https://doi.org/10.1016/j.jpsychires.2014.10.003
Teychenne, M., Costigan, S. A., & Parker, K. (2015). The association between sedentary behaviour and risk of anxiety: A systematic review. BMC Public Health, 15, 513. https://doi.org/10.1186/s12889‐015‐1843‐x
Thompson, S. G., & Higgins, J. (2002). How should meta‐regression analyses be undertaken and interpreted? Statistics in Medicine, 21 (11), 1559 – 1573.
Vancampfort, D., Richards, J., Stubbs, B., Akello, G., Gbiri, C. A., Ward, P. B., & Rosenbaum, S. (2016). Physical activity in people with posttraumatic stress disorder: A systematic review of correlates. Journal of Physical Activity and Health, 13 (8), 910 – 918. https://doi.org/10.1123/jpah.2015‐0436
Vancampfort, D., Stubbs, B., Herring, M. P., Hallgren, M., & Koyanagi, A. (2018). Sedentary behavior and anxiety: Association and influential factors among 42,469 community‐dwelling adults in six low‐ and middle‐income countries. General Hospital Psychiatry, 50, 26 – 32. https://doi.org/10.1016/j.genhosppsych.2017.09.006
World Health Organization (1993). The ICD‐10 Classification of Mental and Behavioural Disorders – Diagnostic Criteria for Research. Geneva, Switzerland : World Health Organization.
Wu, W., Tong, Y., Zhao, Q., Yu, G., Wei, X., & Lu, Q. (2015). Coffee consumption and bladder cancer: A meta‐analysis of observational studies. Scientific Reports, 5, 9051.
Yu, T., Ter Riet, G., Puhan, M. A., & Frei, A. (2017). Physical activity and risk of comorbidities in patients with chronic obstructive pulmonary disease: A cohort study. NPJ Primary Care Respiratory Medicine, 27 (1) https://doi.org/10.1038/s41533‐017‐0034‐x
Zainal, N. H., & Newman, M. G. (2018). Executive function and other cognitive deficits are distal risk factors of generalized anxiety disorder 9 years later. Psychological Medicine, 48 (12), 2045 – 2053. https://doi.org/10.1017/s0033291717003579
]
By Felipe B. Schuch; Brendon Stubbs; Jacob Meyer; Andreas Heissel; Philipp Zech; Davy Vancampfort; Simon Rosenbaum; Jeroen Deenik; Joseph Firth; Philip B. Ward; Andre F. Carvalho and Sarah A. Hiles
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author
