Mental Health and Physical Activity 19 (2020) 100351
Available online 8 August 2020
1755-2966/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Associations of exercise frequency and cardiorespiratory fitness with
symptoms of depression and anxiety - a cross-sectional study of
36,595 adults
Mats Hallgren
a,*, Aaron Kandola
b, Brendon Stubbs
c, Thi-Thuy-Dung Nguyen
d, Peter Wallin
e,
Gunnar Andersson
e, Elin Ekblom-Bak
faEpidemiology of Psychiatric Conditions, Substance Use and Social Environment (EPiCSS), Department of Public Health Sciences, Karolinska Institutet, Stockholm 171 77, Sweden
bDivision of Psychiatry, University College London, London, United Kingdom
cDepartment of Psychological Medicine, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, De Crespigny Park, Box SE5 8AF; South London and Maudsley NHS Foundation Trust, London, United Kingdom
dMedical Epidemiology and Biostatistics (MEB), Karolinska Institutet, Stockholm, 171 77, Sweden eResearch Department, HPI Health Profile Institute, Danderyd, Sweden
fÅstrand Laboratory of Work Physiology, The Swedish School of Sport and Health Sciences, Stockholm, Sweden
A R T I C L E I N F O Keywords: Exercise Cardiorespiratory fitness Physical activity Depression Anxiety A B S T R A C T
Introduction: Regular exercise has been associated with better mental health, but relationships of cardiorespi-ratory fitness (CRF) with mental health symptoms are less certain. We conducted a large cross-sectional study to examine associations of exercise frequency and objectively measured CRF with symptoms of depression and anxiety.
Methods: Data originates from the Swedish Health Profile Assessment (HPA) database, a general health assess-ment offered to all employees working for companies or organizations connected to occupational and health services. Participants reported the frequency (last 30 days) of structured exercise and completed a 6-min VO2max test on a cycle ergometer to determine CRF. Data on relevant covariates including leisure-time sedentary behavior were also collected. The study outcome was self-reported frequent (often/very often) symptoms of depression and anxiety.
Results: Complete data from 36,595 participants were analyzed (41% female, mean age = 42 years). In fully- adjusted models, compared to those in the lowest exercise frequency category (never/sometimes), those exer-cising 1–2 times/week and ≥3 times/week had 0.75 (95% CI = 0.68–0.82) and 0.72 (95% CI = 0.65–0.79) lower odds of reporting frequent symptoms of depression/anxiety, respectively. For CRF, crude models indicated a dose-response with the odds of frequent depression/anxiety symptoms reducing with higher CRF levels (low versus medium and high, respectively). No associations were found after adjustment for leisure-time sedentary behavior.
Conclusions: Exercising at least 1–2 times per week is associated with lower odds of depression/anxiety symp-toms. Leisure-time sedentary behavior attenuates relationships of CRF with depression/anxiety sympsymp-toms.
1. Introduction
Depression and anxiety are prevalent psychiatric conditions associ-ated with marked socio-occupational impairment and somatic co- morbidity (Correll et al., 2017), including a high prevalence of cardio-vascular disease (Batelaan, Seldenrijk, Bot, van Balkom, & Penninx,
2016). Mood disorders are often preceded by a ‘prodromal’ period, characterized by increasingly frequent symptoms of sadness, lowered mood, and heightened anxiety (Karsten et al., 2011). Universal pre-vention strategies commonly target these sub-threshold symptoms, with the aim of curtailing transitions to major depression or an anxiety dis-order (Beyond Blue, 2020).
* Corresponding author. Karolinska Institutet, Solnav¨agen 1e, level 6, 113 65, Solna, Sweden. E-mail address: mats.hallgren@ki.se (M. Hallgren).
Contents lists available at ScienceDirect
Mental Health and Physical Activity
journal homepage: www.elsevier.com/locate/menpa
https://doi.org/10.1016/j.mhpa.2020.100351
Physical activity and its sub-set, exercise, are widely promoted as effective prevention strategies for depression and anxiety (Hallgren et al., 2017; McDowell, Gordon, Andrews, MacDonncha, & Herring, 2018). In a review of prospective studies (n = 49; 266,939 participants)
Schuch et al. (2018) found that, compared to those with low levels of physical activity, those with high levels had lower odds of developing depression (AOR = 0.83, 95% CI = 0.79–0.88). Comparable effect sizes were seen across different age groups and geographic regions. Beneficial relationships have also been reported in prospective studies of anxiety symptoms and disorders (McDowell, Dishman, Gordon, & Herring, 2019). A recent Mendelian randomization study (Choi et al., 2019), and meta-analysis of randomized controlled trials (Gordon et al., 2018;
Gordon, McDowell, Lyons, & Herring, 2017), suggest causal relation-ships between exercise and mental health symptoms.
Despite this evidence, there is active discussion in the literature concerning the relative importance of exercise ‘dose’ and the role of cardiorespiratory fitness (CRF) in these relationships. Exercise dose re-fers to the combined intensity, duration and frequency of exercise. Of these three components, the frequency of exercise is particularly important in helping novice exercisers to establish and maintain new lifestyle routines (Lachman, Lipsitz, Lubben, Castaneda-Sceppa, & Jette, 2018). Regular exercise could have stronger neurobiological effects than infrequent exercise (Dunn & Jewell, 2010; Dunn, Trivedi, Kampert, Clark, & Chambliss, 2005). From a behavioral perspective, regular ex-ercise might also have stronger reinforcing properties, leading to better adherence and ultimately greater psychological benefits (Hallgren, Moss, & Gastin, 2010). While several experimental studies have assessed the effects of exercise dose or exercise intensity on depressive symptoms (Dunn et al., 2005; Helgadottir, Hallgren, Ekblom, & Forsell, 2016), observational studies of associations between exercise frequency and symptoms of depression and anxiety are scarce (McDowell et al., 2019;
Schuch et al., 2018). These studies could inform mental health promo-tion strategies, and potentially enhance the effectiveness of exercise prescriptions for those presenting with symptoms of mood disorders.
Regular exercise promotes higher CRF and both factors are shown to independently have beneficial effects on somatic health (Myers et al., 2015). The extent to which these related factors also confer independent mental health benefits is unclear. A small number of observational studies have examined associations of CRF with depression and anxiety symptoms, but their findings have been inconsistent. In one systematic review and meta-analysis (Papasavvas, Bonow, Alhashemi, & Mickle-wright, 2016) four cross-sectional studies were identified, with sample sizes of 37, 53, 131 and 2428, respectively. Two of the included studies found an inverse association between higher CRF and depression (Rice, Katzel, & Waldstein, 2010; Valtonen et al., 2009), while two studies showed no association (Raso, Natale, Duarte, Greve, & Shephard, 2012;
Stewart et al., 2003). Recently, Kandola, Ashdown-Franks, Hendrikse, Sabiston, and Stubbs (2019) systematically evaluated the relationship between CRF and the incidence of common mental health disorders in prospective cohort studies. Compared to high CRF, low and medium CRF were associated with 47% and 23% greater risk of these disorders, respectively. However, only four eligible studies were included and heterogeneity was substantial in the meta-analysis. Moreover, one study used an estimated CRF algorithm (Shigdel, Stubbs, Sui, & Ernstsen, 2019), and two studies involved the same participant sample (Aberg et al., 2012; Nyberg et al., 2018), highlighting the need for more research using objective measures of CRF. If changes in CRF are asso-ciated with mental health symptoms, then CRF might be an important biomarker to identify at-risk groups or a benchmark for exercise-based treatments.
Several factors are shown to be associated with both exercise and depressive symptoms, including body mass, smoking, and pain (Craft, Perna, Freund, & Culpepper, 2008). Recently, we have also shown that leisure-time sedentary behaviors (e.g. TV-viewing) are inversely asso-ciated with depressive symptoms after adjustment for physical activity levels (Hallgren et al., 2018). This suggests that sedentary behavior is an
important co-variate in studies of relationships between exercise and mental health.
Further studies measuring associations of both exercise frequency and CRF with symptoms of depression and anxiety are needed to determine their relative importance in the prevention of common mental health symptoms. We examined cross-sectional relationships of exercise frequency and objectively measured CRF (assessed as maximal oxygen consumption, VO2max) with self-reported symptoms of
depres-sion and anxiety. As these symptoms are more prevalent among females (McLean, Asnaani, Litz, & Hofmann, 2011; Parker & Brotchie, 2010), and sex differences in CRF have been reported previously (Ekblom-Bak et al., 2019), a secondary aim was to explore possible sex differences in these relationships.
2. Materials and methods 2.1. Study population
Data originate from the Swedish Health Profile Assessment (HPA) database, managed by the HPI Health Profile Institute (Stockholm, Sweden: www.hpihealth.se/). HPA includes a one-page questionnaire about lifestyle and health experiences during the last 30 days, mea-surement of anthropometrics and blood pressure, estimation of VO2max
from a submaximal fitness test on a cycle ergometer, and a brief dialogue with a HPA ‘coach’ (with qualifications in allied health science) to promote wellbeing. The HPA is offered nationally to employees working for companies or organizations connected to occupational (e.g. employee wellness, ergonomics and occupational therapy) or health related services. Participation is voluntary and free-of-charge. Although HPA has been running since 1976, we based our analyses on data collected from January 2017 (when questions on relevant co-variates were first introduced) to June 2019. The total initial sample comprised 50,264 participants before removal of missing co-variate data (described below). The final fully adjusted sample included 18,429 unique cases. The original study complies with the guidelines of the Declaration of Helsinki. The Research Ethics Vetting Board in Stockholm approved the original study and all subjects gave informed consent. Informed consent was obtained from participants after the procedure was fully explained.
2.2. Study outcome: frequent symptoms of depression and anxiety
The primary outcome was assessed with a single question: ‘During the past 30 days…I experience worry, depressed mood, or anxiety...’ with five response alternatives; Never, Rarely, Sometimes, Often, Very often. As previously reported (Hallgren, Nguyen, et al., 2020b), the last two categories were merged and coded as ‘1’ indicating frequent symptoms of depression/anxiety. The remaining responses were merged and coded as ‘0’ (i.e. does not have frequent symptoms of depression/anxiety).
2.3. Exposures
Exercise frequency: This was assessed with the question: ‘During the
past 30 days…I exercise/train…’ with eight response alternatives; Never, Sometimes, 1 time/week, 2 times/week, 3 times/week, 4 times/ week, 5 times/week, ≥6 times/week. Based on the distribution of these scores, which included low frequencies in the highest two categories, these were re-coded as: Never/sometimes, 1–2 times/week, and ≥3 times/week.
Cardiorespiratory fitness: This was assessed as VO2max, expressed as
ml/min/kg, estimated from heart rate response after completing the Åstrand 6-min submaximal exercise test on a Monarch cycle ergometer (Åstrand, 1960). VO2max test scores were divided into three groups (low,
medium and high) based on the inter-quartile range. In previous vali-dation studies on adult population samples, we have shown small and
non-significant mean differences between estimated VO2max using the
Åstrand test and directly measured VO2max during treadmill running
(Bjorkman, Ekblom-Bak, Ekblom, & Ekblom, 2016).
2.4. Covariates
Based on previous evidence of associations of exercise and CRF with depression (Hallgren, Nguyen, et al., 2020b; Zhai, Zhang, & Zhang, 2014) and anxiety (Teychenne, Costigan, & Parker, 2015), the variables listed below were included in the statistical models. Furthermore, in light of recent studies showing robust inverse associations between leisure-time sedentary behavior and depression (Huanga, Gana, Wanga, Caoa, & Lu, 2019), this variable was also included to explore possible attenuating effects.
Age and sex: These were self-reported and included as continuous
and categorical variables, respectively.
Education level: This was not directly assessed in the HPA survey,
but derived by converting occupation codes from the Swedish Standard Classification of Occupation (2012) into three education levels: Primary school, Secondary school, University/vocational. A detailed description of the conversion method is presented elsewhere (Ekblom-Bak, Ekblom, Andersson, Wallin, & Ekblom, 2018).
Body mass index: Weight was assessed with a calibrated scale in
lightweight clothing to the nearest 0.5 kg. Height was measured to the nearest 0.5 cm using a wall-mounted stadiometer. With these two var-iables, body mass index (BMI; kg/m2) was calculated, then categorised
according to the WHO’s classification for adults; normal weight (<25), overweight (25-<30), and obese (≥30).
Smoking: This was assessed by asking participants how often during
the past 30 days they smoked tobacco (cigarettes), with the response alternatives; ≥20/day, 11–19/day, 1–10/day, Occasionally, Never. The first three responses were merged as ‘daily smokers’ (versus occasional, and non-smoker).
Use of pain medication: This was assessed with the question ‘During
the past 30 days…I use pain medicines…’ with five response alternatives (Very often, Often, Sometimes, Rarely, Never). This question was included as a continuous variable.
Leisure-time sedentary behavior: This was assessed with the
ques-tion ‘During the past 30 days…I sit still during my leisure-time…’ with five response alternatives: Almost always, 75% of the time, 50% of the time, 25% of the time, Almost never. Due to the small proportion of responses to the lowest and highest categories, these were merged into: high (≥75% of the time), medium (50% of the time), and low (≤25% of the time).
2.5. Statistical analyses
Baseline characteristics of the fully adjusted sample were calculated using descriptive statistics (mean, standard deviation (SD), percentage %). To assess associations of exercise and CRF with frequent symptoms of depression/anxiety, data were analyzed in two steps. First, separate logistic regression models were run to assess the crude, partly-adjusted (Model 2: age, gender, smoking, pain medication, BMI), and fully- adjusted (Model 3: model 2 plus leisure-time sedentary behavior) asso-ciations of each exposure variable (i.e. exercise frequency and CRF) with the primary outcome. These are shown as odds ratios (ORs) with 95% confidence intervals (CIs) and corresponding p-values. To assess a po-tential dose-response relationship between exercise frequency and symptoms of depression and anxiety, ‘Never/sometimes’ was used as the reference category and compared to the remaining two categories; 1–2 times/week, and ≥3 times/week. Similarly, the ‘Low’ CRF group was used as the reference category and compared to ‘Medium’ and ‘High’. Recognizing that both CRF levels and the prevalence of depression/ anxiety symptoms may differ by sex (Ekblom-Bak et al., 2019; Parker & Brotchie, 2010) we included a sex interaction term in the partly (Model 2) and fully-adjusted analyses (Model 3). In a second analytic step,
fully-adjusted models were run separately within each of the three ex-ercise frequency and CRF categories, respectively. This enabled associ-ations of exercise frequency with symptoms of depression/anxiety to be assessed within low, medium, and high CRF levels. Analyses were per-formed using SPSS version 25, and R version 3.5.2.
3. Results 3.1. Missing data
Of the initial sample consisting of 50,264 people, missing data on relevant co-variates ranged from 0.1% (Age) to 36% (CRF). For the Education variable, 52% of cases were missing. This occurred due to the process (described above) of converting occupational codes into edu-cation levels. As some occupations could not be coded, the conversion process was not always possible. We therefore conducted two sets of analyses, (1) primary analyses without the Education variable included, shown below (n = 36,595), and (2) sensitivity analyses with the Edu-cation variable included (n = 18,429). Results from sensitivity analyses are shown in Supplementary Table 1 (exercise frequency) and 2 (CRF), respectively. Differences between participants with and without the Education variable are shown in Supplementary Table 3.
3.2. Participant characteristics
Characteristics of the study sample are shown in Table 1. Participants were predominantly middle-aged (mean age = 41.7 years, SD = 11.5)
Table 1
Characteristics of the study participants (n = 36,595). Characteristic
Gender (Female), n (%) 15,077 (41.2)
Age mean (SD); [range] 41.7 (11.5); [16–80] Education, n (%) Primary school 1427 (3.9) Secondary school 18,187 (49.7) University 16,980 (46.4) Smoking status, n (%) Non-smoker 30,703 (83.9) Occasional smoker 3220 (8.8) Daily smoker 2671 (7.3)
Body Mass Index (BMI), n (%)
Normal weight 16,174 (44.2) Overweight 13,832 (37.8) Obese 6587 (18.0) Pain medication, n (%) Never 7831 (21.4) Rarely 19,139 (52.3) Sometimes 7209 (19.7) Often 1610 (4.4) Very often 805 (2.2)
Leisure-time sedentary behavior, n (%)
Low (≤25% of the time) 19,431 (53.1) Medium (50% of the time) 12,552 (34.3) High (≥75% of the time) 4611 (12.6) Exercise frequency
Never/Sometimes 11,820 (32.3)
1-2 times/week 10,905 (29.8)
≥3 times/week 13,869 (37.9)
VO2max ml/min/kg: mean (SD), [range] 35.8 (10.0), [15.0–77.9] Cardiorespiratory fitness (VO2max) categories
Low (14.9–28.3 ml/min/kg) 9148 (25.0) Medium (28.4–41.7 ml/min/kg) 18,297 (50.0) High (41.8–77.9 ml/min/kg) 9148 (25.0) Frequent symptoms of depression/anxiety
No (Never/Rarely/Sometimes) 33,594 (91.8)
and there were more males (58.8%). Approximately half of the sample was university educated, and the majority were non-smokers. In total, 8.2% of the sample reported experiencing frequent (‘Often/Very often’) symptoms of depression/anxiety (11.6% males; 5.8% males). Thirty-two percent of the participants exercised ‘Never/Sometimes’, 29.8% exer-cised 1–2 times/week, and 37.9% exerexer-cised ≥3 times/week. Cardiore-spiratory fitness scores were normally distributed and comparable to other population based surveys of Swedish adults (Ekblom, Engstrom, & Ekblom, 2007).CRF values did not differ significantly between males and females at any level (mean score, low, medium, or high).
3.3. Associations of exercise (last 30 days) with frequent symptoms of depression and anxiety
In crude models, a dose-response relationship was found. Compared to those in the lowest exercise frequency category (Never/sometimes), those exercising 1–2 times/week and ≥3 times/week had 0.67 (95% CI =0.62–0.74) and 0.56 (95% CI = 0.52–0.62) lower odds of reporting frequent symptoms of depression/anxiety, respectively. In partly- adjusted models, these results remained stable. In fully adjusted models including leisure-time sedentary behavior, the strength of these associations was attenuated slightly. Compared to those who never/ sometimes exercised, those exercising 1–2 times/week and ≥3 times/ week had 0.75 (95% CI = 0.68–0.82) and 0.72 (95% CI = 0.65–0.79) lower odds of reporting frequent symptoms of depression/anxiety, respectively (see Table 2). When analyses were run separately within each of the three levels of CRF (low, medium and high), the results were comparable to the pooled analyses (Fig. 1). The mean weekly frequency of exercise (total and within each of the three exercise categories) did not differ significantly between males and females. The sex interaction term was not significant in any models (p > 0.32).
3.4. Associations of cardiorespiratory fitness with frequent symptoms of depression and anxiety
In crude models, compared to those with low CRF, those with me-dium and high CRF had 0.88 (95% CI = 0.81–0.97) and 0.86 (95% CI = 0.77–0.95) lower odds of reporting frequent symptoms of depression/ anxiety, respectively. In partly-adjusted models, compared to those with low CRF, those with high CRF had 0.83 (95% CI = 0.73–0.95) lower odds of depression/anxiety symptoms. In fully adjusted models that included leisure-time sedentary behavior, there was no association (Table 3). When analyses were run separately within each of the three exercise frequency categories, the results were comparable to the pooled analyses (Fig. 2). Mean CRF scores (total and within each of the three CRF categories) did not differ significantly between males and females. The sex interaction term was not significant in any models (p > 0.38). Results from sensitivity analyses with the Education variable included were almost identical to those shown here (Supplementary Tables 2 and 3).
4. Discussion
In this large cross-sectional study involving employed adults, we found that participation in at least 1–2 exercise sessions per week was
associated with lower odds of reporting frequent symptoms of depres-sion and anxiety. After adjustment for potential confounders, including leisure-time sedentary behavior (Hallgren, Nguyen, et al., 2020a), no associations were found between objectively measured CRF levels and frequent symptoms of depression/anxiety. Sex did not appear to mod-erate these relationships.
To optimize the effectiveness of exercise-based prevention and treatment strategies for mental health, information is needed regarding the optimal frequency of exercise, and the relative importance of CRF. To our knowledge, no previous studies have assessed relationships of
both exercise frequency and CRF with symptoms of depression and
anxiety in the same population. Our findings suggest that exercise fre-quency may have a stronger association with common mental health symptoms than CRF, where the association was nullified by leisure-time sedentary behavior. While exercise frequency and CRF were positively correlated (r = 0.32), they appear to have somewhat independent re-lationships with mental health symptoms; a finding also reported in studies of the relationship between physical activity, CRF, and somatic health (Myers et al., 2015). A related observation is that some forms of structured exercise (e.g. resistance training) are shown to be effective treatments for mood disorders and related symptoms (Gordon et al., 2017), yet may not significantly improve CRF levels. Taken together, this suggests that high CRF may not be necessary to prevent common mental health symptoms. Instead, regular participation in a preferred form of structured exercise may be of greater relative importance.
Another possible interpretation is that the benefits of CRF on mental health symptoms only become apparent when comparing those with very low CRF to those with moderate CRF. Studies of the relationship between CRF and somatic health suggest that most of the benefits are gained through modest improvements in CRF among those with very low baseline fitness levels (Blair et al., 1996). All participants in this study were employed, and employment status is linked to higher average CRF (Van Domelen et al., 2011). Thus, it is possible that our study did not include enough participants with low CRF levels to observe associations with depression and anxiety symptoms.
Our findings regarding CRF differ from those reported in a longitu-dinal Swedish study. To determine whether CRF at 18 years of age was associated with future risk of being diagnosed with an affective disorder,
Aberg et al. (2012) prospectively followed 1,117,292 adolescents with no history of mental illness for 3–40 years. In fully adjusted models, low CRF (compared to medium and high) was associated with increased risk for clinician-diagnosed depression. Similarly, a recent cross-sectional study (n = 26,615; mean age = 55.7 years) found that, compared to those with low non-exercise estimated (e) CRF levels, those with me-dium and high eCRF had 21% and 26% lower odds of depression, respectively (Shigdel et al., 2019). In prospective analyses (n = 14,020; mean age = 52.2 years) the odds were 22% and 19% lower for the same comparisons. These disparate findings could be attributable to several factors, including differences in study design, participant mean age, measurement of the primary outcome (clinical depression versus symptoms of depression/anxiety), and adjustment for different co-variates in the statistical models. Unlike previous studies, we included leisure-time sedentary behavior in the fully-adjusted models, as it is shown to be inversely associated with symptoms of depression (Hallgren, Dunstan, & Owen, 2020b; Hallgren, Nguyen, et al., 2020a).
Table 2
Associations of exercise frequency (last 30 days) with frequent symptoms of depression/anxiety (n = 36,595).
Exercise frequency Crude Model Model 2 Model 3
OR 95% CI p OR 95% CI p OR 95% CI p
Never/Sometimes 1 - - 1 - - 1 - -
1-2 times/week 0.67 0.62–0.74 <0.001 0.69 0.63–0.76 <0.001 0.75 0.68–0.82 <0.001 ≥3 times/week 0.56 0.52–0.62 <0.001 0.59 0.53–0.64 <0.001 0.72 0.65–0.79 <0.001 Model 2 adjusted for: age, gender, smoking status, BMI, use of pain medication, exercise-gender interaction. Model 3 adjusted for Model 2 co-variates plus leisure-time sedentary behavior. OR= Odds Ratio; CI = Confidence Interval.
Inclusion of this co-variate may have reduced residual confounding. Our finding that ‘regular’ exercise (defined here as ≥1–2 times/ week) is associated with lower odds of depression/anxiety symptoms, compared to exercising ‘never/sometimes’, is broadly consistent with previous observational studies. Prospective data from the British Whitehall II study (n = 9,309, 8 year follow-up) demonstrated that, compared to non-regular exercise, regular moderate intensity exercise lasting at least 2.5 h/week was associated with lower odds of future depression and anxiety symptoms (Da Silva et al., 2012). Although ex-ercise duration was not assessed in this study, it is reasonable to assume that for most people, exercise lasting ‘2.5 h/week or more’ implies a frequency of at least 1–2 times/week. However, the Whitehall II study
results contrast with a recent prospective study involving 33,908 ‘healthy’ adults followed over 11 years, where just 30 min/week of moderate exercise was associated with lower odds of developing depression (Harvey et al., 2017). Longer average weekly durations of exercise (lasting 1–2 h, 3–4 h, and 5–6 h, respectively) did not confer additional protective effects.
Regular exercise influences a range of biological and psychosocial processes implicated in the pathophysiology of mood disorders. Un-derlying mechanisms include neurogenesis, monoamine synthesis, reduced systemic inflammation and oxidative stress, and regulation of the endocrine system (Kandola, Ashdown-Franks, Hendrikse, et al., 2019). Emerging research suggests that exercise could also improve the
Fig. 1. Associations of exercise frequency (last 30 days) with frequent symptoms of depression/anxiety measured within three strata of cardiorespiratory fitness (n =36,595).
Table 3
Associations of cardiorespiratory fitness (VO2max) with self-reported frequent symptoms of depression/anxiety (n = 36,595).
Fitness level Crude Model Model 2 Model 3
OR 95% CI p OR 95% CI p OR 95% CI p
Low 1 - - 1 - - 1 - -
Medium 0.88 0.81–0.97 0.011 0.91 0.81–1.00 0.063 0.99 0.89–1.11 0.975
High 0.86 0.77–0.95 0.005 0.83 0.73–0.95 0.012 1.01 0.88–1.16 0.827
Model 2 adjusted for: age, gender, smoking status, BMI, use of pain medication, exercise-gender interaction. Model 3 adjusted for Model 2 co-variates plus leisure-time sedentary behavior. Cardiorespiratory fitness levels correspond to the following values: Low = 14.9–28.3 ml/kg/min; Medium = 28.4–41.7 ml/kg/min; High = 41.8–77.9 ml/kg/min. OR= Odds Ratio; CI = Confidence Interval.
Fig. 2. Associations of cardiorespiratory fitness (VO2max) with frequent symptoms of depression/anxiety measured within three strata of exercise frequency (n =36,595).
diversity and composition of the gut microbiota, which in turn can have beneficial mental health effects (Clapp et al., 2017; Monda et al., 2017). From a psychological perspective, regular exercise is shown to improve self-esteem, self-efficacy, and may engender greater social support (Awick et al., 2017; McAuley, Mihalko, & Bane, 1997) A review of the purported mechanisms, including a conceptual framework to guide future research, is presented elsewhere (Kandola, Ashdown-Franks, Hendrikse, et al., 2019).
Strengths of the current study include the large participant sample, objective measurements of CRF, and the assessment of important co- variates (e.g. pain medication, sedentary behavior) excluded from many previous studies. Several limitations are acknowledged, include the cross-sectional study design, which precludes assertions about the direction of these associations. The single-item outcome measure is a weakness, and should not be equated with a diagnosis of depression or anxiety; this should be determined using a structured, validated ques-tionnaire or interview. While validated, multiple-item quesques-tionnaires are preferred, related studies involving employed adults have shown that single-item measures of physical activity are robust predictors of health outcomes (Neupane, Karstad, Hallman, Rugulies, & Holtermann, 2019), and two recent studies using the Health Profile Institute data that we employed have demonstrated the predictive validity of these single-item questions (Ekblom-Bak et al., 2018; Eriksson et al., 2019). Development of the survey and the rationale for item selection is described in a published report (Andersson & Wallin, 2019). Symptoms of depression and anxiety were assessed together, rather than sepa-rately, which is not optimal. While these symptoms often develop simultaneously, we recognize that depression and anxiety are distinct clinical disorders with unique etiologies. Sedentary behavior during leisure-time was included as a co-variate, but information about the type of sedentary behavior was not assessed. This is relevant as evidence suggests that prolonged sedentary behaviors which are mentally ‘pas-sive’ (e.g. TV-viewing) are detrimentally related to depression, while mentally-active behaviors (e.g. reading) could have protective mental health effects (Hallgren, Nguyen, et al., 2020). As noted, exercise dose is comprised of three elements, and we have measured only one
component (i.e. frequency). While the average frequency of exercise tends to increase with longer average durations of training, making them potentially comparable in this sense, exercise intensity could have as-sociations with depression/anxiety symptoms that are largely indepen-dent of exercise frequency and duration. Finally, it is recognized that BMI is a relatively poor indicator of the percent of body fat. Waist and hip circumference are more accurate indicators of overweight and obesity (Adab, Pallan & Wincup, 2018).
In conclusion, our findings suggest that participation in regular (≥1–2 times/week) exercise is associated with lower odds of frequent symptoms of depression/anxiety. The association between CRF and these symptoms was substantially attenuated when leisure-time seden-tary behavior was included as a co-variate. Taking into account the limitations noted above, the results suggest that regular exercise may be sufficient to reduce the risk of depression/anxiety symptoms at the population level. However, as poor mental health is consistently asso-ciated with somatic co-morbidity (Correll et al., 2017), it may be advisable to promote exercise routines that improve CRF in those with identified low levels. Prospective studies of these relationships are needed. Studies examining associations of exercise intensity, duration and frequency with common mental health symptoms are also warranted.
Author contributions
MH conceived the study and wrote the first draft; TTDN and MH conducted the statistical analyses. All co-authors contributed substan-tive comments that were incorporated into the manuscript. PW manages the database on which the study is based.
Funding
MH is supported by research grants from the Swedish Council for Health Working Life and Welfare (FORTE) and the Ekhaga Foundation, Sweden.
Supplementary Table 1
Associations of exercise frequency (last 30 days) with frequent symptoms of depression/anxiety (n = 18,429)
Exercise frequency Crude Model Model 2 Model 3
OR 95% CI p OR 95% CI p OR 95% CI p
Never/Sometimes 1 - - 1 - - 1 - -
1-2 times/week 0.66 0.61–0.71 <0.001 0.66 0.58–0.76 <0.001 0.73 0.64–0.81 <0.001 ≥3 times/week 0.56 0.52–0.60 <0.001 0.58 0.51–0.66 <0.001 0.74 0.66–0.83 <0.001 Model 2 adjusted for: age, gender, smoking status, BMI, use of pain medication, exercise-gender interaction. Model 3 adjusted for Model 2 co-variates plus leisure-time sedentary behavior. OR= Odds Ratio; CI = Confidence Interval.
Supplementary Table 2
Associations of cardiorespiratory fitness (VO2max) with self-reported frequent symptoms of depression/anxiety (n = 18,429)
Fitness level Crude Model Model 2 Model 3
OR 95% CI P OR 95% CI p OR 95% CI p
Low 1 - - 1 - - 1 - -
Medium 0.89 0.81–0.98 0.017 0.91 0.80–1.01 0.067 1.06 0.90–1.23 0.489
High 0.83 0.73–0.93 0.002 0.82 0.72–0.93 0.013 1.01 0.81–1.24 0.942
Model 2 adjusted for: age, gender, smoking status, BMI, use of pain medication, exercise-gender interaction. Model 3 adjusted for Model 2 co-variates plus leisure-time sedentary behavior. Cardiorespiratory fitness levels correspond to the following values: Low = 14.9–28.3 ml/kg/min; Medium = 28.4–41.7 ml/kg/min; High = 41.8–77.9 ml/kg/min. OR= Odds Ratio; CI = Confidence Interval.
Supplementary Table 3
Characteristics of participants with and without the Education variable Characteristic Missing Education n =
26,139 (52%) With Education N = 24,125 (48%) Female*, n (%) 10,343 (39.6) 10,263 (42.5) Age mean (SD); [range] 42.44 (11.62); [16–80] 42.56 (12.03);
[16–80] Smoking*, n (%)
Non-smoker 21,942 (84.7) 19,944 (83.0) Occasional smoker 2268 (8.8) 2127 (8.9)
Daily smoker 1694 (6.5) 1947 (8.1)
Body Mass Index (BMI)*, n (%)
Normal weight 10,598 (45.5) 10,229 (42.9) Overweight 8737 (37.5) 9076 (38.1) Obese 3955 (17.0) 4518 (19.0) Pain medication*, n (%) Very often 481 (1.9) 588 (2.4) Often 1055 (4.1) 1148 (4.8) Sometimes 5010 (19.3) 4830 (20.1) Rarely 13,838 (53.4) 12,262 (51.0) Never 5509 (21.3) 5197 (21.6)
Leisure-time sedentary behavior* n (%) High (Always + 75% of the
time) 3535 (13.6) 2746 (11.4)
Medium (50% of the time) 9307 (35.9) 7805 (32.5) Low (Never + 25% of the
time) 13,069 (50.4) 13,467 (56.1)
Exercise frequency
Never/Sometimes 8129 (31.4) 7986 (33.3) 1-2 times/week 7901 (30.5) 6979 (29.1) At least 3 times/week 9849 (38.1) 9034 (37.6) VO2 max ml/kg/min mean
(SD), [range] 36.12 (10.09), [14.95–77.78] 35.67 (10.03), [14.95–77.94] Fitness (categories of VO2 max)
Low (14.9–28.3 ml/kg/min) 3520 (19.3) 3828 (20.7) Medium (28.4–41.7 ml/kg/
min) 10,968 (60.1) 11,078 (59.9)
High (41.8–77.9 ml/kg/
min) 3768 (20.6) 3576 (19.3)
Frequent symptoms of depression/anxiety No (Never/Rarely/
Sometimes) 23,563 (91.0) 22,024 (91.7) Yes (Often/Very often) 2335 (9.0) 1996 (8.3) *Significantly different p < 0.01.
Kandola et al., 2019
Declaration of competing interest
PW and GA are employed at Health Profile Institute (HPI), Danderyd, Sweden. This study is based on data from the HPI database, managed by HPI, which is also responsible for the development and standardization of methods used and education of the HPA coaches since the start of the administration of HPAs. GA and PW were not involved in the analyses of data presented in this paper.
Acknowledgements
None
References
Aberg, M. A. I., Waern, M., Nyberg, J., Pedersen, N. L., Bergh, Y., Aberg, N. D., et al. (2012). Cardiovascular fitness in males at age 18 and risk of serious depression in adulthood: Swedish prospective population-based study. British Journal of Psychiatry, 201(5), 352–359. https://doi.org/10.1192/bjp.bp.111.103416.
Adab, P., Pallan, M., & Whincup, P. H. (2018). Is BMI the best measure of obesity? It works for most people most of the time. BMJ, 360, k1274. https://doi.org/10.1136/ bmj.k1274.
Andersson, G., & Wallin, P. (2019). H¨ALSOPROFILBED¨OMNING™ – tv¨arvetenskap, evidens och bepr¨ovad erfarenhe (health profile Assessment: Interdisciplinary evidence and proven experience). Sweden: Health Profile Institute (HPI) Danderyd.
Åstrand, I. (1960). Aerobic work capacity in men and women with special eference to age. Acta Physiologica Scandivavica Suppl, 49(169), 1–92.
Awick, E. A., Ehlers, D., Fanning, J., Phillips, S. M., W´ojcicki, T., Mackenzie, M. J., et al. (2017). Effects of a home-based DVD-delivered physical activity program on self- esteem in older adults: Results from a randomized controlled trial. Psychosomatic Medecine, 79(1), 71–80. https://doi.org/10.1097/PSY.0000000000000358. Batelaan, N. M., Seldenrijk, A., Bot, M., van Balkom, A. J. L. M., & Penninx, B. W. J. H.
(2016). Anxiety and new onset of cardiovascular disease: Critical review and meta- analysis. British Journal of Psychiatry, 208(3), 223–231. https://doi.org/10.1192/bjp. bp.114.156554.
Beyond Blue. (2020). Depression facts: Signs and symptoms. Beyond Blue Australia. Retrieved 15th March 2020 from: https://www.beyondblue.org.au/the-facts/d epression/signs-and-symptoms.
Bjorkman, F., Ekblom-Bak, E., Ekblom, O., & Ekblom, B. (2016). Validity of the revised Ekblom Bak cycle ergometer test in adults. European Journal of Applied Physiology, 116(9), 1627–1638. https://doi.org/10.1007/s00421-016-3412-0.
Blair, S. N., Kampert, J. B., Kohl, H. W., Barlow, C. E., Macera, C. A., Paffenbarger, R. S., et al. (1996). Influences of cardiorespiratory fitness and other precursors on cardiovascular disease and all-cause mortality in men and women. Journal of the American Medical Association, 276(3), 205–210. https://doi.org/10.1001/ jama.276.3.205.
Choi, K. W., Chen, C. Y., Stein, M. B., Klimentidis, Y. C., Wang, M. J., Koenen, K. C., et al. (2019). Assessment of bidirectional relationships between physical activity and depression among adults A 2-sample mendelian randomization study. JAMA Psychiatry, 76(4), 399–408. https://doi.org/10.1001/jamapsychiatry.2018.4175. Clapp, M., Aurora, N., Herrera, L., Bhatia, M., Wilen, E., & Wakefield, S. (2017). Gut
microbiota’s effect on mental health: The gut-brain axis. Clinical Practice, 7(4), 987.
https://doi.org/10.4081/cp.2017.987.
Correll, C. U., Solmi, M., Veronese, N., Bortolato, B., Rosson, S., Santonastaso, P., et al. (2017). Prevalence, incidence and mortality from cardiovascular disease in patients with pooled and specific severe mental illness: A large-scale meta-analysis of 3,211,768 patients and 113,383,368 controls. World Psychiatry, 16(2), 163–180.
https://doi.org/10.1002/wps.20420.
Craft, L. L., Perna, F. A., Freund, K. M., & Culpepper, L. (2008). Psychosocial correlates of exercise in women with self-reported depressive symptoms. Journal of Physical Activity and Health, 5(3), 469–480. https://doi.org/10.1123/jpah.5.3.469. Da Silva, M. A., Singh-Manoux, A., Brunner, E. J., Kaffashian, S., Shipley, M. J.,
Kivimaki, M., et al. (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. Dunn, A. L., & Jewell, J. S. (2010). The effect of exercise on mental health. Current Sports
Medicine Reports, 9(4), 202–207. https://doi.org/10.1249/Jsr.0b013e3181e7d9af. Dunn, A. L., Trivedi, M. H., Kampert, J. B., Clark, C. G., & Chambliss, H. O. (2005).
Exercise treatment for depression: Efficacy and dose response. American Journal of Preventive Medicine, 28(1), 1–8. https://doi.org/10.1016/j.amepre.2004.09.003. Ekblom-Bak, E., Ekblom, O., Andersson, G., Wallin, P., & Ekblom, B. (2018). Physical
education and leisure-time physical activity in youth are both important for adulthood activity, physical performance, and health. Journal of Physical Activity and Health, 15(9), 661–670. https://doi.org/10.1123/jpah.2017-0083.
Ekblom-Bak, E., Ekblom, O., Andersson, G., Wallin, P., Soderling, J., Hemmingsson, E., et al. (2019). Decline in cardiorespiratory fitness in the Swedish working force between 1995 and 2017. Scandinavian Journal of Medicine & Science in Sports, 29(2), 232–239. https://doi.org/10.1111/sms.13328.
Ekblom, B., Engstrom, L. M., & Ekblom, O. (2007). Secular trends of physical fitness in Swedish adults. Scandinavian Journal of Medicine & Science in Sports, 17(3), 267–273.
https://doi.org/10.1111/j.1600-0838.2006.00531.x.
Eriksson, J. S., Ekblom, B., Kallings, L. V., Hemmingsson, E., Andersson, G., Wallin, P., et al. (2019). Active commuting in Swedish workers between 1998 and 2015-trends, characteristics, and cardiovascular disease risk. Scandinavian Journal of Medicine & Science in Sports, 30(2), 370–379. https://doi.org/10.1111/sms.13581.
Gordon, B. R., McDowell, C. P., Hallgren, M., Meyer, J. D., Lyons, M., & Herring, M. P. (2018). Association of efficacy of resistance exercise training with depressive symptoms meta-analysis and meta-regression analysis of randomized clinical trials. JAMA Psychiatry, 75(6), 566–576. https://doi.org/10.1001/
jamapsychiatry.2018.0572.
Gordon, B. R., McDowell, C. P., Lyons, M., & Herring, M. P. (2017). The effects of resistance exercise training on anxiety: A meta-analysis and meta-regression analysis of randomized controlled trials. Sports Medicine, 47(12), 2521–2532. https://doi. org/10.1007/s40279-017-0769-0.
Hallgren, M., Dunstan, D. W., & Owen, N. (2020a). Passive versus mentally active sedentary behaviors and depression. Exercise and Sport Sciences Reviews, 48(1), 20–27. https://doi.org/10.1249/JES.0000000000000211.
Hallgren, M. A., Moss, N. D., & Gastin, P. (2010). Regular exercise participation mediates the affective response to acute bouts of vigorous exercise. Journal of Sports Science and Medicine, 9(4), 629–637. https://www.ncbi.nlm.nih.gov/pubmed/24149790. Hallgren, M., Nguyen, T. T., Owen, N., Vancampfort, D., Dunstan, D. W., Wallin, P., et al.
(2020b). Associations of sedentary behavior in leisure and occupational contexts with symptoms of depression and anxiety. Preventive Medicine, 133. https://doi.org/ 10.1016/j.ypmed.2020.106021.
Hallgren, M., Stubbs, B., Vancampfort, D., Lundin, A., Jaakallio, P., & Forsell, Y. (2017). Treatment guidelines for depression: Greater emphasis on physical activity is needed. European Psychiatry, 40, 1–3. https://doi.org/10.1016/j.
eurpsy.2016.08.011.
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 Jouornal of Psychiatry, 175(1), 28–35. https://doi.org/10.1176/appi. ajp.2017.16111223.
Helgadottir, B., Hallgren, M., Ekblom, O., & Forsell, Y. (2016). Training fast or slow? Exercise for depression: A randomized controlled trial. Preventive Medecine, 91, 123–131. https://doi.org/10.1016/j.ypmed.2016.08.011.
Huanga, Y., Gana, Y., Wanga, C., Caoa, S., & Lu, X. (2019). Sedentary behavior and risk of depression: A meta-analysis of prospective studies. Translational Psychiatry, 10(1), 26. https://doi.org/10.1038/s41398-020-0715-z.
Kandola, A., Ashdown-Franks, G., Hendrikse, J., Sabiston, C. M., & Stubbs, B. (2019). Physical activity and depression: Towards understanding the antidepressant mechanisms of physical activity. Neuroscience & Biobehavioral Reviews, 107, 525–539. https://doi.org/10.1016/j.neubiorev.2019.09.040.
Kandola, A., Ashdown-Franks, G., Stubbs, B., Osborn, D. P. J., & Hayes, J. F. (2019). The association between cardiorespiratory fitness and the incidence of common mental health disorders: A systematic review and meta-analysis. Journal of Affective Disorders, 257, 748–757. https://doi.org/10.1016/j.jad.2019.07.088.
Karsten, J., Hartman, C. A., Smit, J. H., Zitman, F. G., Beekman, A. T. F., Cuijpers, P., et al. (2011). Psychiatric history and subthreshold symptoms as predictors of the occurrence of depressive or anxiety disorder within 2 years. British Journal of Psychiatry, 198(3), 206–212. https://doi.org/10.1192/bjp.bp.110.080572. Lachman, M. E., Lipsitz, L., Lubben, J., Castaneda-Sceppa, C., & Jette, A. M. (2018).
When adults don’t exercise: Behavioral strategies to increase physical activity in sedentary middle-aged and older adults. Innovation in Aging, 2(1). https://doi.org/ 10.1093/geroni/igy007.
McAuley, E., Mihalko, S. L., & Bane, S. M. (1997). Exercise and self-esteem in middle- aged adults multidimensional relationships and physical fitness and self-efficacy influences. Journal of Behavioral Medecine, 20(1), 67–83. https://doi.org/10.1023/a: 1025591214100.
McDowell, C. P., Dishman, R. K., Gordon, B. R., & Herring, M. P. (2019). Physical activity and anxiety: A systematic review and meta-analysis of prospective cohort studies. American Journal of Preventive Medicine, 57(4), 545–556. https://doi.org/10.1016/j. amepre.2019.05.012.
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.
McLean, C. P., Asnaani, A., Litz, B. T., & Hofmann, S. G. (2011). Gender differences in anxiety disorders: Prevalence, course of illness, comorbidity and burden of illness. Journal of Psychiatric Research, 45(8), 1027–1035. https://doi.org/10.1016/j. jpsychires.2011.03.006.
Monda, V., Villano, I., Messina, A., Valenzano, A., Esposito, T., Moscatelli, F., et al. (2017). Exercise modifies the gut microbiota with positive health effects. Oxidative medicine and cellular longevity. https://doi.org/10.1155/2017/3831972. Artn 3831972.
Myers, J., McAuley, P., Lavie, C. J., Despres, J. P., Arena, R., & Kokkinos, P. (2015). Physical activity and cardiorespiratory fitness as major markers of cardiovascular risk: Their independent and interwoven importance to health status. Progress in Cardiovascular Diseases, 57(4), 306–314. https://doi.org/10.1016/j.
pcad.2014.09.011.
Neupane, S., Karstad, K., Hallman, D. M., Rugulies, R., & Holtermann, A. (2019). Objectively Measured versus self-reported occupational physical activity and multisite musculoskeletal pain: a prospective follow-up study at 20 nursing homes in Denmark. International Archives of Occupational and Environmental Health. https:// doi.org/10.1007/s00420-019-01495-4.
Nyberg, J., Henriksson, M., Aberg, M. A. I., Rosengren, A., Soderberg, M., Aberg, N. D., et al. (2018). Cardiovascular fitness in late adolescent males and later risk of serious non-affective mental disorders: A prospective, population-based study. Psychological Medicine, 48(3), 416–425. https://doi.org/10.1017/S0033291717001763. Papasavvas, T., Bonow, R. O., Alhashemi, M., & Micklewright, D. (2016). Depression
symptom severity and cardiorespiratory fitness in healthy and depressed adults: A systematic review and meta-analysis. Sports Medicine, 46(2), 219–230. https://doi. org/10.1007/s40279-015-0409-5.
Parker, G., & Brotchie, H. (2010). Gender differences in depression. International Review of Psychiatry, 22(5), 429–436. https://doi.org/10.3109/09540261.2010.492391. Raso, V., Natale, V. M., Duarte, A. J., Greve, J. M., & Shephard, R. J. (2012).
Immunological parameters in elderly women: Correlations with aerobic power, muscle strength and mood state. Brain, Behavior, and Immunity, 26(4), 597–606.
https://doi.org/10.1016/j.bbi.2012.01.012.
Rice, M. C., Katzel, L. I., & Waldstein, S. R. (2010). Sex-specific associations of depressive symptoms and cardiovascular risk factors in older adults. Aging & Mental Health, 14 (4), 405–410. https://doi.org/10.1080/13607860903586185. Pii 921986295. Schuch, F. B., Vancampfort, D., Firth, J., Rosenberg, S., Ward, P. B., Silva, E. S., et al.
(2018). Physical activity and incident depression: A meta-analysis of prospective cohort studies. American Journal of Psychiatry, 75(7), 631–648. https://doi.org/ 10.1176/appi.ajp.2018.17111194.
Shigdel, R., Stubbs, B., Sui, X. M., & Ernstsen, L. (2019). Cross-sectional and longitudinal association of non-exercise estimated cardiorespiratory fitness with depression and anxiety in the general population: The HUNT study. Journal of Affective Disorders, 252, 122–129. https://doi.org/10.1016/j.jad.2019.04.016.
Stewart, K. J., Turner, K. L., Bacher, A. C., DeRegis, J. R., Sung, J., Tayback, M., et al. (2003). Are fitness, activity, and fatness associated with health-related quality of life and mood in older persons? Journal of Cardiopulminary Rehabilitation, 23(2), 115–121. https://doi.org/10.1097/00008483-200303000-00009.
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.
Valtonen, M., Laaksonen, D. E., Laukkanen, J., Tolmunen, T., Rauramaa, R., Viinamaki, H., et al. (2009). Leisure-time physical activity, cardiorespiratory fitness and feelings of hopelessness in men. BMC Public Health, 9, 204. https://doi.org/ 10.1186/1471-2458-9-204.
Van Domelen, D. R., Koster, A., Caserotti, P., Brychta, R. J., Chen, K. Y., McClain, J. J., et al. (2011). Employment and physical activity in the U.S. American Journal of Preventive Medicine, 41(2), 136–145. https://doi.org/10.1016/j.
amepre.2011.03.019.
Zhai, L., Zhang, Y., & Zhang, D. (2014). Sedentary behaviour and the risk of depression: A meta-analysis. British Journal of Sports Medicine. https://doi.org/10.1136/bjsports- 2014-093613.
