Mental Health and Physical Activity 18 (2020) 100319

Contents lists available at ScienceDirect

Mental Health and Physical Activity

journal homepage: www.elsevier.com/locate/menpa

Associations of physical activity with positive mental health: A population- T

based study
Nina Tamminena,∗, Jaakko Reinikainenb, Kaija Appelqvist-Schmidlechnerc, Katja Borodulinb,d,
Tomi Mäki-Opase, Pia Solina
a
WHO Collaborating Centre for Mental Health Promotion, Prevention and Policy, Mental Health Unit, Finnish Institute for Health and Welfare, Helsinki, Finland
b
Public Health Evaluation and Projection Unit, Finnish Institute for Health and Welfare, Helsinki, Finland
c
Mental Health Unit, Finnish Institute for Health and Welfare, Helsinki, Finland
d
The Age Institute, Helsinki, Finland
e
University of Eastern Finland (UEF), Department of Social Sciences, Kuopio, Finland

A R T I C LE I N FO A B S T R A C T

Keywords: Introduction: The relationship between physical activity (PA) and positive mental health (PMH) has been studied
Physical activity domains mainly by either concentrating on total PA or focusing on leisure time PA. This study investigated whether total
Positive mental health PA and PA domains of leisure time, commuting and occupational PA and screen time sitting at home were
Warwick-Edinburgh mental well-being scale associated with PMH.
Population study
Methods: The study used a national population study, the FinHealth 2017 Study that included the Warwick-
Edinburgh Mental Well-being Scale (WEMWBS) as a measurement for PMH. Of all (n = 10305, of whom 10247
were eligible) invited, 58.1% (5952) participated in health examination and were given a questionnaire in-
cluding the WEMWBS. A total of 5337 (52.1% of the eligible sample, 89.7% of the health examination parti-
cipants) returned the questionnaire. The WEMWBS scale was adequately completed by 5090 participants (55.6%
women, mean age 55.5 years). Low, moderate and high level of PMH categories were formed. Leisure time,
occupational and commuting PA domains were assessed separately and together as a total PA index. Also, screen
time sitting at home was measured. Binary logistic regression models were utilized to estimate the odds ratios for
having low PMH compared with moderate PMH and separately for having high PMH compared with moderate
PMH. The models were estimated both with and without adjustment for potential confounders.
Results: Physical inactivity (total PA) was associated with lower levels of PMH. Some of the PA domains were
strongly associated with low PMH; leisure time physical inactivity and long screen time sitting at home were
strongly related to higher odds of low PMH. Commuting PA was associated with high PMH.
Conclusions: The study highlights the importance of PA with relation to PMH among adults. Especially, physical
inactivity was strongly associated with low PMH. As low levels of PMH can potentially present a risk for mental
ill-health, this study suggests it is important to increase PA among the least active population.

1. Introduction illness (WHO, 2018). The term is often used interchangeably with the
term mental well-being. PMH is understood to comprise two main
The World Health Organization (WHO) has defined positive mental elements: feeling good (hedonic perspective) and functioning well
health as “a state of well-being in which the individual realizes his or (eudaimonic perspective). The hedonic perspective focuses on the
her own abilities, can cope with the normal stresses of life, can work subjective experience of happiness and life satisfaction, and the eu-
productively and fruitfully, and is able to make a contribution to his or daimonic perspective on psychological functioning and self-realisation
her community” (WHO, 2018, p. 1). Positive mental health (PMH) is (Ryan & Deci, 2001; Stewart-Brown, 2015). PMH includes aspects such
thus based on the assumption that mental health is something positive as self-esteem, optimism, a sense of mastery and coherence, satisfying
and a resource for well-being, and more than just the absence of mental personal relationships and resilience, that is, the ability to cope with

∗
Corresponding author. Finnish Institute for Health and Welfare, P.O. Box 30, 00271, Helsinki, Finland.
E-mail addresses: nina.tamminen@thl.fi (N. Tamminen), jaakko.reinikainen@thl.fi (J. Reinikainen),
kaija.appelqvist-schmidlechner@thl.fi (K. Appelqvist-Schmidlechner), katja.borodulin@ikainstituutti.fi (K. Borodulin), tomi.maki-opas@uef.fi (T. Mäki-Opas),
pia.solin@thl.fi (P. Solin).

https://doi.org/10.1016/j.mhpa.2020.100319
Received 10 September 2019; Received in revised form 19 January 2020; Accepted 20 January 2020
Available online 23 January 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/).
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

adversities and face stressors (Huppert, 2009; Lehtinen, 2008; Vaillant, We had the novel possibility to examine various PA domains in a large
2012). national population study, and to look at whether the possible asso-
PMH is currently receiving increased attention in research, policy ciations with PMH were different depending on the specific PA domain.
making and clinical practice (EU Joint Action, 2016; Forsman et al., Furthermore, we explored the potential associations in relation to dif-
2015). It has been recognized as a key resource for the health and well- ferent levels of PMH. Earlier studies have suggested that associations
being of individuals, families and societies (Huppert, 2009; Jenkins (e.g. between socioeconomic factors and PMH) with low level of PMH
et al., 2008; Wahlbeck, 2011, 2015). Research has shown PMH to be follow a different pattern than associations with high level of PMH (Ng
associated with a reduced risk of mortality, better physical health, Fat, Mindell, Boniface, & Stewart-Brown, 2016; Stewart-Brown,
better academic achievement and social functioning, and with reduced Samaraweera, Taggart, Kandala, & Stranges, 2015; Stranges,
risk for mental illness (Keyes, 2005; Keyes & Simoes, 2012; Suldo & Samaraweera, Taggart, Kandala, & Stewart-Brown, 2014). Differences
Shaffer, 2008). There is also growing evidence on the economic benefits between predictors of the low end of the mental well-being scale with
of promoting PMH. Actions targeted especially at children and ado- the high end of the mental well-being scale have also been found with
lescents and some work-place interventions have shown to be cost-ef- health behaviours such as diet, smoking and alcohol consumption
fective (Zechmeister, Kilian, McDaid, & the MHEEN group, 2008; (Stewart-Brown et al., 2015). Thus, the aim of the current study was to
Knapp, McDaid, & Parsonage, 2011; McDaid, Park, & Wahlbeck, 2019). investigate whether total PA and PA domains such as leisure time PA,
Improved PMH at work-place, for example, can help employees to stay commuting PA, occupational PA and screen time sitting at home were
at work (less sickness and absenteeism) and work to their full produc- associated with PMH using a large national population study, the Fin-
tive potential (less presenteeism, i.e. lost productivity while at work). Health 2017 Study (Borodulin & Sääksjärvi, 2019). We examined the
Interventions promoting PMH can also generate significant savings in relationship between PA and levels of low, moderate and high PMH.
public health expenditure such as reductions in health and social care This is the first study, to our knowledge, to investigate associations
costs (Knapp, McDaid, & Parsonage, 2011; Zechmeister et al., 2008). between different domains of PA with different levels of PMH in a large
The association between physical activity (PA) and mental health is population study.
well-documented providing strong evidence on mental health benefits
of PA (Biddle, 2016; Chekroud et al., 2018; White et al., 2017). How- 2. Methods
ever, the focus of previous research has been on mental health problems
(Bennie, Teychenne, De Cocker, & Biddle, 2019; Chekroud et al., 2018; 2.1. Study population
Suetani et al., 2017) rather than on the perspective of PMH. There is,
nonetheless, growing evidence of the potential benefits of PA in in- The FinHealth 2017 Study is a nationally representative population-
creasing PMH, especially related to leisure time PA (Cerin, Leslie, based health examination study aiming to collect up-to-date informa-
Sugiyama, & Owen, 2009; Mason, Curl, & Kearns, 2016; Richards et al., tion on the health and well-being of adults residing in Finland, and on
2015; Zhang & Chen, 2019). A report on the 2012/13 North West the factors influencing their health and well-being (Borodulin &
Mental Wellbeing Survey (Jones et al., 2013) showed that respondents Sääksjärvi, 2019). In 2017, 10 305 persons (of whom 10 247 were
who met the PA target (at least 30 min of moderate-intensity activity on eligible) over 18 years of age living in Finland were invited by two-
5–7 days per week) had significantly higher PMH as compared to those stage clustered random sampling to participate in a health examination.
who did not meet the target. A study on a community –wide gamifi- Participation rate in the health examination was 58.1% (5952). The
cation-based PA intervention (Harris, 2018) found that an increase in participants underwent clinical measurements and filled in ques-
PA following the intervention was positively correlated with PMH. The tionnaires. Those who participated in the health examination received a
study also revealed that increases in PMH were significantly greater for questionnaire that measured PMH with the Warwick-Edinburgh Mental
the least active prior to the intervention. Further, clear associations Well-being Scale (WEMWBS) (Tennant et al., 2007). A total of 5337
were found between various domains of PA and PMH measure in a participants (52.1% of the eligible sample, 89.7% of the health ex-
study conducted in deprived neighbourhoods in Glasgow (Mason et al., amination participants) returned the questionnaire. The scale was
2016). The researchers found that people engaging in PA for pleasure, adequately completed by 5090 participants.
through active leisure and family activities and in active travel, gen-
erally had better PMH. Commuting PA has shown mixed associations 2.2. Positive mental health measure
with PMH (Humphreys, Goodman, & Ogilvie, 2013; Mytton, Panter, &
Ogilvie, 2016). However, several studies have established a positive The WEMWBS was used to measure the level of PMH (i.e. mental
relationship between commuting PA and PMH (Martin, Goryakin, & well-being) of the survey respondents. WEMWBS is a validated, popular
Suhrcke, 2014; Mason et al., 2016; Mytton et al., 2016; White et al., measure that covers both subjective well-being (hedonic perspective)
2017). Majority of the studies (e.g. Harris, 2018; Jones et al., 2013; and psychological functioning (eudaimonic perspective) addressing
Mason et al., 2016) investigating the relationship between PA and PMH aspects of PMH (Tennant et al., 2007). The English version of the
have been cross-sectional and thus cannot reliably distinguish causality WEMWBS has shown good validity and high levels of internal con-
from association. sistency (Cronbach's alpha 0.89/0.91) and reliability (test-retest relia-
While increasing number of studies have established associations of bility 0.83). The scale is suitable for use in measuring mental well-being
leisure time PA and, to some extent, commuting PA with PMH, less is at a population level due to its strong psychometric performance and
known about the relationships between PA domains such as occupa- lack of ceiling effects. (Stranges et al., 2014; Taggart, Stewart-Brown, &
tional PA and sedentary behaviour (e.g. screen time sitting) and PMH. Parkinson, 2016; Tennant et al., 2007). The scale is consisted of 14
Occupational PA has been found to be unrelated or negatively asso- positively worded items, each item answered on a five-point Likert
ciated with mental well-being (Cerin et al., 2009). Conversely, White scale. The item scores are summed to produce a total score ranging
and partners (2017) found occupational PA positively associated with between 14 and 70, with higher scores representing higher levels of
mental ill-health. Knowledge about the relationship of sedentary be- PMH (Taggart et al., 2016). Finnish, Swedish and English versions of
haviour with mental well-being seems to be non-existent. Some mixing the scale were used in the FinHealth 2017 Study.
results have been shown regarding sedentary behaviour and mental ill- Three PMH categories were formed by classifying the WEMWBS
health (Hamer, Stamatakis, & Mishra, 2010; Teychenne et al., 2014, scores: low level of PMH if the score was less than one standard de-
2017, 2020). viation (SD) below the mean (score < 44), high level of PMH if the
The relationship between PA and PMH, accordingly, seems to vary score was more than one SD above the mean (score > 61), and
between different PA domains (Cerin et al., 2009; White et al., 2017). moderate PMH for the rest (Taggart et al., 2016).

2
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

2.3. Different domains of PA and total PA index adjusting variables, which were thought to be possible confounders,
were fitted separately. Age and household income per consumption unit
Three different domains of PA were assessed. These included leisure were treated as continuous variables and total PA index, leisure time
time PA, occupational PA and commuting PA (Borodulin et al., 2016). PA, screen time sitting at home, occupational PA, commuting PA, sex,
In addition, we measured screen time sitting at home (Borodulin & marital status, educational level, employment status and self-rated
Sääksjärvi, 2019). Leisure time PA was assessed as: “How much do you health as categorical variables in logistic regressions. Model 1 was ad-
exercise and stress yourself physically in your leisure time?” The an- justed for age and sex, Model 2 additionally for marital status, educa-
swer options captured both the amount of time and the level of in- tional level, household income per consumption unit, employment
tensity of the PA and were dichotomized into inactive and active. The status and self-perceived health. Model 3 was adjusted for all the
response of not moving much and not straining physically was cate- variables. However, Model 3 did not include the total PA index to avoid
gorized as inactive; other three answer categories as active. Occupa- the problem of multicollinearity, as it included all the other PA vari-
tional PA was assessed as: “How demanding is your work physically?” ables from which the total PA was derived.
From the four available response options, four categories of inactive, The stratified sampling design was taken into account, and non-
low, medium and high were established. Commuting PA was measured participation was adjusted by weighting utilizing the R package survey
with a question: “On your way to work or school, how many minutes do (Lumley, 2004). The weights were formed using sociodemographic
you travel on foot, by bicycle or similar? Add up the journeys to and variables and information about hospitalizations obtained from na-
from work/school.” The response options ranged from not being at tional administrative registers (Borodulin & Sääksjärvi, 2019). For each
work to more than 60 min per day and were further dichotomized into model, individuals with missing values in variables under consideration
less than 30 min daily and 30 min or more daily. Screen time sitting at were removed. All the analyses were carried out using the R statistical
home was measured as: “How many hours on average do you sit in a software version 3.5.1 (R Core Team, 2018).
weekday at home in front of screen?” Responses were given in hours
and minutes and were dichotomized into 3 h at most and more than 3 h. 3. Results
The PA questions used in this study have shown good criterion validity
against morbidity and mortality (Grimby et al., 2015; Hu et al., 2007) Table 1 shows the characteristics of the study participants for dif-
and the leisure time PA question a moderate correlation against ac- ferent PA and sociodemographic variables by PMH categories. Ac-
celerometer counts among the working age population (Fagt et al., cording to these unadjusted numbers, it seemed that all the variables
2011). were associated with PMH except commuting PA and sex. Higher levels
The total PA index was created using the leisure time, commuting of PA were usually related to higher PMH and inactivity to lower PMH.
and occupational PA questions (Borodulin et al., 2016). Four categories Low level of PMH was associated with young age, being single, low
were created: inactivity (inactivity in all PA domains), low (inactivity in education, low household income, unemployment and poor self-per-
leisure time PA; low activity in at least one of commuting PA or oc- ceived health. Old age, being divorced or widowed, high education,
cupational PA), medium (moderate in leisure time or occupational PA high household income, retirement and good self-perceived health were
or 15–29 min in commuting PA and no high activity in any of the do- associated with high level of PMH.
mains) and high (high activity in at least one of leisure time, occupa- The adjusted associations of PA with low and high PMH are pre-
tional or commuting PA). Full description of the PA questions and the sented in Table 2. The upper part of the table shows the odds for having
total PA index used in the study is presented in Supplementary File 1. low PMH and the lower part shows the odds for having high PMH.
Separate models were fitted to adjust for different sets of socio-
2.4. Sociodemographic and personal characteristics demographic and PA variables. Based on the total PA index, having
medium or lower level (inactive) of activity compared to having high
Background variables used in the analyses included age, sex, marital level of activity increased the odds of having low level of PMH. Inactive
status (three categories: married/cohabiting, divorced/widow and persons were 1.47 times more likely to have low PMH than highly ac-
single), educational level (low = primary school, middle = vocational tive persons. Leisure time physical inactivity (OR = 1.46) and high
school or high school, high = college or university level), household screen time sitting at home (OR = 1.51) were strongly related to higher
income per consumption unit (total household income divided by the odds of low PMH, whereas occupational and commuting PA were not
square root of household size) (OECD, 2019), employment status (four found to associate with low PMH at all.
categories: in employment or studying, unemployed, retired and other) PA variables were not so strongly related to high level of PMH as to
and self-rated health (four categories: good, rather good, moderate and low level of PMH (Table 2). Only commuting PA associated with high
poor/rather poor [the poorest class was combined with rather poor due PMH. Those who reported physically active commuting 30 min or more
to very small frequencies]). The selection of covariates and confounders daily had 2.05 times higher odds of having high PMH than those with
was based on previous studies from similar data sets, which focused on less commuting PA. We also tested whether PA variables associated
demographic and socioeconomic correlates of health and well-being with PMH differently for men and women, but no evidence of inter-
(Borodulin & Sääksjärvi, 2019; Stranges et al., 2014; Lundqvist & Mäki- action effects was found.
Opas, 2016).
4. Discussion
2.5. Statistical methods
4.1. Main findings and study implications
Descriptive analyses using χ2 tests were carried out to test for the
associations of PA and sociodemographic variables with PMH levels. This cross-sectional study examined the relationship between PA
We built models where the outcome variable was categorized PMH. and PMH in the FinHealth 2017 Study, a large nationally representative
Binary logistic regression models were utilized to estimate the odds sample of the Finnish adult population (Koponen, Borodulin, Lundqvist,
ratios for having low PMH compared with moderate PMH and sepa- Sääksjärvi, & Koskinen, 2018). We had the opportunity to investigate
rately for high PMH compared with moderate PMH. The decision to the associations of both total PA levels and different domains of PA with
model WEMWBS as a categorical variable rather than continuous was PMH using the WEMWBS as a measure of PMH (Tennant et al., 2007).
based on the different associations at the low and high end of the In addition to looking at PA domains such as leisure time PA, com-
spectrum found in previous studies (Ng et al., 2016; Stewart-Brown muting PA and occupational PA, we included sedentary behaviour in
et al., 2015; Stranges et al., 2014). Models with different numbers of the form of screen time sitting at home in our analysis. Moreover, we

3
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

Table 1
Sample characteristics and distributions of participants’ positive mental health by physical activity and sociodemographic variables.
Total n (5090) Positive mental health, N (%) p-valuea Item non-response, Nc

Low Moderate High

721 (14) 3660 (72) 709 (14)

Total PA index 6
Inactive 140 (25) 361 (64) 63 (11) < 0.001
Low 67 (22) 205 (69) 26 (9)
Medium 313 (14) 1656 (73) 315 (14)
High 201 (10) 1433 (74) 304 (16)

Leisure time PA 46
Inactive 293 (24) 807 (66) 123 (10) < 0.001
Active 422 (11) 2815 (74) 584 (15)

Screen time sitting at home a day 120

3 h at most 391 (12) 2483 (74) 490 (15) < 0.001
More than 3 h 313 (19) 1086 (68) 207 (13)

Occupational PAb 101

Inactive 149 (12) 916 (75) 152 (12) < 0.001
Low 65 (10) 488 (76) 90 (14)
Medium 116 (18) 461 (71) 73 (11)
High 18 (8) 170 (80) 24 (11)

Commuting PAb 75
Less than 30 min daily 305 (13) 1790 (75) 285 (12) 0.130
30 min or more daily 43 (12) 250 (72) 55 (16)

Age 0
18–29 58 (18) 235 (74) 24 (8) < 0.001
30–49 216 (14) 1155 (76) 152 (10)
50–69 306 (14) 1519 (71) 323 (15)
70–100 141 (13) 751 (68) 210 (19)

Sex 0
Men 329 (15) 1617 (72) 314 (14) 0.772
Women 392 (14) 2043 (72) 395 (14)

Marital status 19
Married/cohabiting 428 (12) 2670 (74) 516 (14) < 0.001
Divorced/widow 150 (18) 553 (66) 139 (17)
Single 138 (22) 424 (69) 53 (9)

Educational level 72
Low 256 (16) 1110 (71) 208 (13) 0.017
Middle 228 (14) 1207 (73) 224 (14)
High 221 (12) 1297 (73) 267 (15)

Household income per consumption unit 171

1st quintile 185 (21) 610 (68) 103 (11) < 0.001
2nd quintile 164 (17) 686 (72) 106 (11)
3rd quintile 155 (13) 833 (72) 163 (14)
4th quintile 95 (10) 696 (76) 126 (14)
5th quintile 91 (9) 725 (73) 181 (18)

Employment status 10
In employment or studying 348 (13) 2047 (75) 340 (12) < 0.001
Unemployed 84 (30) 174 (62) 23 (8)
Retired 251 (13) 1290 (69) 329 (18)
Other 37 (19) 140 (72) 17 (9)

Self-rated health 14
Good 64 (5) 972 (73) 301 (23) < 0.001
Rather good 231 (11) 1528 (75) 270 (13)
Moderate 275 (20) 971 (71) 114 (8)
Poor/rather poor 146 (42) 183 (52) 21 (6)

Note. N = number, PA = physical activity.

a
p-values from χ2 tests used to determine possible associations between PA/sociodemographic variables and PMH categories.
b
Restricted to those who are employed or studying.
c
Of those who have responded to the WEMWBS scale.

4
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

Table 2
Odds ratios and 95% confidence intervals separately for low level of positive mental health (PMH) compared with the moderate level of PMH and for high level of
PMH compared with the moderate level of PMH.
Model 1 (age and sex Model 2 (sociodemo-graphic Model 3 (fully p-value for interaction
adjusted), n, OR (95% CI) factors adjusteda), n, OR (95% adjustedb), n, OR (95% with sexd
CI) CI)

Low PMH compared with Total PA index n = 4376 n = 4172 0.210

moderate PMH Inactive 3.20 (2.35, 4.37) 1.47 (1.02, 2.11)
Low 2.06 (1.47, 2.89) 1.66 (1.15, 2.40)
Medium 1.54 (1.22, 1.96) 1.33 (1.05, 1.68)
High ref. ref.

Leisure time PA n = 4337 n = 4138 n = 4033 0.742

Inactive 2.39 (1.95, 2.94) 1.57 (1.26, 1.95) 1.46 (1.18, 1.82)
Active ref. ref. ref.

Screen time sitting at n = 4273 n = 4115 n = 4033 0.756

home a day
More than 3h 1.98 (1.60, 2.45) 1.66 (1.31, 2.11) 1.51 (1.19, 1.91)
3h at most ref. ref. ref.

Occupational PAc n = 2383 n = 2327 n = 2292 0.307

Inactive 1.01 (0.51, 1.99) 1.50 (0.73, 3.10) 1.36 (0.67, 2.77)
Low 0.90 (0.43, 1.88) 1.33 (0.61, 2.93) 1.29 (0.60, 2.77)
Medium 1.53 (0.77, 3.03) 1.92 (0.94, 3.92) 1.85 (0.91, 3.74)
High ref. ref. ref.

Commuting PAc n = 2388 n = 2332 n = 2292 0.786

Less than 30 min daily 0.94 (0.62, 1.42) 0.94 (0.56, 1.58) 1.00 (0.57, 1.76)
30 min or more daily ref. ref. ref.

High PMH compared with Total PA index n = 4363 n = 4171 0.341

moderate PMH Inactive ref. ref.
Low 0.93 (0.51, 1.70) 1.15 (0.63, 2.10)
Medium 1.32 (0.90, 1.95) 1.25 (0.83, 1.88)
High 1.80 (1.27, 2.55) 1.34 (0.88, 2.05)

Leisure time PA n = 4329 n = 4141 n = 4033 0.614

Inactive ref. ref. ref.
Active 1.48 (1.16, 1.89) 1.11 (0.84, 1.45) 1.04 (0.78, 1.37)

Screen time sitting at n = 4266 n = 4116 n = 4033 0.767

home a day
More than 3h ref. ref. ref.
3h at most 1.36 (1.12, 1.65) 1.22 (0.96, 1.56) 1.20 (0.94, 1.53)

Occupational PAc n = 2374 n = 2325 n = 2295 0.142

Inactive ref. ref. ref.
Low 1.14 (0.80, 1.60) 1.20 (0.85, 1.69) 1.20 (0.85, 1.71)
Medium 1.13 (0.68, 1.87) 1.10 (0.73, 1.66) 1.09 (0.73, 1.61)
High 0.73 (0.47, 1.12) 0.88 (0.50, 1.56) 0.91 (0.53, 1.57)

Commuting PAc n = 2380 n = 2331 n = 2295 0.148

Less than 30 min daily ref. ref. ref.
30 min or more daily 1.90 (1.12, 3.24) 2.06 (1.24, 3.42) 2.05 (1.25, 3.39)

Note. OR = odds ratio, CI = confidence interval, PMH = positive mental health, PA = physical activity.
a
Adjusted for age, sex, marital status, educational level, household income per consumption unit, employment status and self-perceived health.
b
Fully adjusted models are adjusted for all the sociodemographic variables and leisure time PA, screen time sitting at home, occupational PA and commuting PA.
c
Restricted to those who are employed or studying.
d
Interactions for total PA index tested in Model 2 and for other PA variables in Model 3.

aimed to explore these associations in relation to different levels of associated with low level of PMH. Our data showed that leisure time
PMH as earlier studies have suggested that the associations may follow physical inactivity and long screen time sitting at home were strongly
a different pattern (Stranges et al., 2014; Stewart-Brown et al., 2015; Ng related to higher odds of low PMH. Sedentary behaviour such as long
Fat et al., 2016). Our findings give further support to the evidence that screen time sitting at home in leisure time was thus connected to having
the relationship between PA, total and domain-specific, and PMH differ low PMH. This is a novel result as the relationship between sedentary
at the low end of the mental well-being scale from the high end of the behaviour and PMH has been an unexplored research area. These
mental well-being scale. We found that when looking at total PA, findings could suggest that an increase in leisure time PA and a re-
physical inactivity was associated with lower levels of PMH. This result, duction of screen time sitting at home might have a positive influence
thus, seems to support the earlier notions of the potential benefits of PA on the level of PMH. Leisure time PA is most likely based on one's own
in increasing PMH (e.g. Cerin et al., 2009; Harris, 2018; Mason et al., preferences and enjoyment (Teychenne et al., 2020; White et al., 2018)
2016; Zhang & Chen, 2019). which can therefore be seen more easily affecting mental well-being.
We also found that some of the studied PA domains were strongly Furthermore, leisure time PA can provide possibilities for

5
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

improvements in self-esteem and mastery of skills or goals (Paluska & promote and support active lifestyles are necessary. Actions such as
Schwenk, 2000; Teychenne et al., 2020), both components of PMH. creating affordable leisure time PA opportunities for all and building PA
Regarding sedentary behavior and its association with PMH, more re- promoting living and commuting environments are a recommended
search is needed before we can make any interpretations or implica- way to increase the activity levels of the population. In addition,
tions of the relationship between the two. community level measures such as easily accessible community PA
Interestingly, we did not find association between occupational PA groups and paid/volunteer PA guides could promote PA levels espe-
and PMH thus adding to the scarce and somewhat controversial evi- cially among those who are physically least active.
dence of their relationship (Cerin et al., 2009; White et al., 2017). This
finding highlights the need for more extensive investigations of occu-
pational PA in order to determine when it may beneficial to mental 4.2. Study strengths and limitations
well-being. Occupational PA is likely to be viewed as a compulsory task
rather than voluntary activity and not providing opportunities for en- A major strength of this study is that it relies on a large, nationally
joyment, mastery or improved self-esteem, thereby not improving representative sample of the population, which provides sufficient
mental well-being. statistical power to distinguish associations between PA variables and
Another interesting finding in our study is the positive association PMH. Further, a well-established and validated measure, the WEMWBS,
between commuting PA and high PMH. In our study, commuting PA was used in the survey to measure respondents’ PMH which benefits
included travel on foot, by bicycle or similar 30 min or more per day to this study. The large sample size together with the use of a recognized
and from work or school. Relatively few studies have explored these measurement enables comparisons with findings of other studies in the
associations showing to some extent conflicting results (Humphreys same research area. In addition, the scope of the FinHealth 2017 Study
et al., 2013; Mytton et al., 2016). However, our observation mirrors allowed us to examine various domains of PA in addition to total PA
those that have found a positive relationship between commuting PA levels. We were able to improve the representativeness further by using
and mental well-being (Martin et al., 2014; Mason et al., 2016; Mytton versatile information from national administrative registers to calculate
et al., 2016; White et al., 2017). Our study was cross-sectional and survey weights that were used to adjust for non-participation.
therefore does not allow for an understanding of the temporal asso- Nevertheless, the study has some limitations. First, due to the cross-
ciations between commuting PA and PMH. Nonetheless, perhaps sectional nature of the study, no causal directions could be established
walking or cycling to school or work creates feelings of enjoyment and from the observed relationships. Second, the survey respondents self-
mastery similar to those argued with leisure time PA. This could imply reported their PA levels, which may have led to overestimation of their
that policy actions such as city planning to develop and provide safe activity levels. Objectively measured PA levels tend to be lower than
pedestrian and cycle paths to encourage active commuting, and active subjective assessments of actual activity levels (Choi et al., 2019; Craig,
leisure time for that matter, may be beneficial to PMH. Incentives at Mindell, & Hirani, 2009). However, the PA measures used in the survey
workplaces such as showers and storage facilities to promote and en- are widely used in large population-based studies (Strath et al., 2013)
courage active commuting including walking and cycling to work could and have shown good psychometric properties (Fagt et al., 2011;
also contribute to better mental well-being of employees. Grimby et al., 2015; Hu et al., 2007). Third, the Finnish and the
A note on the possible direction of the association between PA and Swedish versions of the WEMWBS used in the survey have not yet been
PMH is needed here. As our study was cross-sectional in design, we validated. Nevertheless, the short version of the scale, SWEMWBS, has
were not able to determine the causality or direction of the relationship been validated in Swedish (Haver, Akerjordet, Caputi, Furunes, &
between PA and PMH. Little potential evidence exists on the causal Magee, 2015). Further, the WEMWBS was translated from English into
direction between PA and PMH (e.g. Harris, 2018) due to cross-sec- Finnish, and then independently back-translated to ensure that the
tional nature of the studies (e.g. Mason et al., 2016; Richards et al., content and meaning remained unchanged (language equivalence)
2015). There is, however, some evidence of bidirectional relationship (Beaton, Bombardier, Guillemin, & Bosi-Ferraz, 2000). The Swedish
between PA and mental ill-health, and a stronger suggestion of a po- version of the WEMWBS was developed similarly. Finally, although the
tential causal relationship between PA and a reduced risk for depression study population consisted of a nationally representative sample of
(Choi et al., 2019; Pinto Pereira, Geoffroy, & Power, 2014; Schuch adults over 18 years of age, care must be taken to generalise the study
et al., 2018) the latter proposing a possible protective role of PA against results to younger population or to populations of different nature.
depression. Interestingly, some studies have speculated that associa-
tions with low PMH mirror those of mental ill-health (Stewart-Brown
et al., 2015; Stranges et al., 2014; Taggart et al., 2016). This suggests 5. Conclusion
that at least with relation to low PMH, PA could provide protection
against low PMH. Further, if the relationship between PA and PMH This study highlights the importance of PA with relation to PMH
would potentially be bidirectional, such as suggested between PA and among adults. Especially physical inactivity was strongly associated
mental ill-health, low PMH could possibly result in lower levels of PA. with low PMH. As low levels of PMH can potentially present a risk for
Associations with high levels of PMH, on the other hand, have found to mental ill-health, it seems vital to promote and increase PA among the
follow a different pattern to those of low PMH (Stewart-Brown et al., least active population. However, as the causal direction between PA
2015). The mechanisms and explanations of possible causalities be- and PMH remains unsolved, future experimental research designs are
tween PA and PMH lay beyond this current study, thus future inquiries needed to investigate the relationship further, including the direction of
and experimental research designs are needed. causation between these two domains. In addition, more information
It is also worth to bear in mind, that there are many factors that may and deeper understanding, for example using qualitative methods, on
influence PMH, PA being potentially one of them. Our study found the relationship and potential mechanisms between PA and PMH is
higher levels of PMH generally among those respondents, who were warranted.
older, higher educated, had high household income and perceived their
health as good. It seems, therefore, that the relationship between PA
and PMH is a complex one. We may need to look at other, potentially Funding
mediating or moderating factors such as socioeconomical factors, and
what kind of role they play in the relationship between PA and PMH. This study received research funding from Signe and Ane
Nevertheless, our findings further strengthen the evidence of the Gyllenberg Foundation. TMO was funded by the Academy of Finland,
positive relationship between PA and PMH. Population strategies to Strategic Research Council (#303615).

6
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

Declaration of competing interest Knapp, M., McDaid, D., & Parsonage, M. (Eds.). (2011). Mental health promotion and
mental illness prevention: The economic case. London: Department of Health.
Koponen, P., Borodulin, K., Lundqvist, A., Sääksjärvi, K., & Koskinen, S. (Eds.). (2018).
The authors declare that they have no conflicts of interest. Health, functional capacity and welfare in Finland – FinHealth 2017 studyHelsinki:
National Institute for Health and Welfare (THL).
Appendix A. Supplementary data Lehtinen, V. (2008). Building up good mental health. Guidelines based on existing
knowledge. Helsinki: Stakes.
Lumley, T. (2004). Analysis of complex survey samples. Journal of Statistical Software,
Supplementary data to this article can be found online at https:// 9(1), 1–19.
doi.org/10.1016/j.mhpa.2020.100319. Lundqvist, A., & Mäki-Opas, T. (Eds.). (2016). Health 2011 Survey – Methods. Report 8/
2016. Helsinki: National Institute for Health and Welfare.
Martin, A., Goryakin, Y., & Suhrcke, M. (2014). Does active commuting improve psy-
References chological wellbeing? Longitudinal evidence from eighteen waves of the British
household panel survey. Preventive Medicine, 69, 296–303.
Mason, P., Curl, A., & Kearns, A. (2016). Domains and levels of physical activity are
Beaton, D. E., Bombardier, C., Guillemin, F., & Bosi-Ferraz, M. (2000). Guidelines for the
linked to adult mental health and wellbeing in deprived neighbourhoods: A cross-
process of cross-cultural adaptation of self-report measures. Spine, 25(24),
sectional study. Mental Health and Physical Activity, 11, 19–28.
3186–3191.
McDaid, D., Park, A. L., & Wahlbeck, K. (2019). The economic case for the prevention of
Bennie, J. A., Teychenne, M. J., De Cocker, K., & Biddle, S. J. H. (2019). Associations
mental illness. Annual Review of Public Health, 40. https://doi.org/10.1146/annurev-
between aerobic and muscle-strengthening exercise with depressive symptom se-
publhealth-040617-013629.
verity among 17,839 U.S. adults. Preventive Medicine, 121, 121–127. https://doi.org/
Mytton, O. T., Panter, J., & Ogilvie, D. (2016). Longitudinal associations of active com-
10.1016/j.ypmed.2019.02.022.
muting with wellbeing and sickness absence. Preventive Medicine, 84, 19–26.
Biddle, S. (2016). Physical activity and mental health: Evidence is growing. World
Ng Fat, L., Mindell, J., Boniface, S., & Stewart-Brown, S. (2016). Evaluating and estab-
Psychiatry: Official Journal of the World Psychiatric Association (WPA), 15(2), 176–177.
lishing national norms for the short Warwick-Edinburgh Mental Well-being Scale
https://doi.org/10.1002/wps.20331.
(SWEMWBS) using the health survey for England. Quality of Life Research, 26(5),
Borodulin, K., Harald, K., Jousilahti, P., Laatikainen, T., Männistö, S., & Vartiainen, E.
1129–1144.
(2016). Time trends in physical activity from 1982 to 2012 in Finland. Scandinavian
OECD (2019). Under pressure: The squeezed middle class. Paris: OECD Publishinghttps://
Journal of Medicine & Science in Sports, 26(1), 93–100. https://doi.org/10.1111/sms.
doi.org/10.1787/689afed1-en.
12401.
Paluska, S. A., & Schwenk, T. L. (2000). Physical activity and mental health. Sports
Borodulin, K., & Sääksjärvi, K. (Eds.). (2019). FinHealth 2017 Study – Methods. Report 17/
Medicine, 29, 167–180. https://doi.org/10.2165/00007256-200029030-00003.
2019.. Helsinki, Finland: Finnish Institute for Health and Welfare.
Pinto Pereira, S. M., Geoffroy, M., & Power, C. (2014). Depressive symptoms and physical
Cerin, E., Leslie, E., Sugiyama, T., & Owen, N. (2009). Associations of multiple physical
activity during 3 decades in adult life: Bidirectional associations in a prospective
activity domains with mental well-being. Mental Health and Physical Activity, 2,
cohort study. JAMA Psychiatry, 71(12), 1373–1380. https://doi.org/10.1001/
55–64.
jamapsychiatry.2014.1240.
Chekroud, S. R., Gueorguieva, R., Zheutlin, A. B., Paulus, M., Krumholz, H. M., Krystal, J.
R Core Team (2018). R: A language and environment for statistical computing. Vienna,
H., et al. (2018). Association between physical exercise and mental health in 1.2
Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-
million individuals in the USA between 2011 and 2015: A cross-sectional study. The
project.org/.
Lancet Psychiatry, 5(9), 739–746. https://doi.org/10.1016/S2215-0366(18)30227-X.
Richards, J., Jiang, X., Kelly, P., Chau, J., Bauman, A., & Ding, D. (2015). Don't worry, be
Choi, K. W., Chen, C., Stein, M. B., Klimentidis, Y. C., Wang, M. J., Koenen, K. C., et al.
happy: Cross-sectional associations between physical activity and happiness in 15
(2019). Assessment of bidirectional relationships between physical activity and de-
European countries. BMC Public Health, 15, 53. https://doi.org/10.1186/s12889-015-
pression among adults: A 2-sample mendelian randomization study. JAMA Psychiatry,
1391-4.
76(4), 399–408. https://doi.org/10.1001/jamapsychiatry.2018.4175.
Ryan, R., & Deci, E. (2001). On happiness and human potentials: A review of research on
Craig, R., Mindell, J., & Hirani, V. (Eds.). (2009). Health survey for England 2008: Volume 1
hedonic and eudaimonic well-being. Annual Review of Psychology, 52, 141–166.
- physical activity and fitness. National Centre for Social Research and UCL Medical
Schuch, F. B., Vancampfort, D., Firth, J., Rosenbaum, S., Ward, P. B., Silva, E. S., et al.
School. Leeds: NHS Information Centre for Health and Social Care.
(2018). Physical activity and incident depression: A meta-analysis of prospective
EU Joint Action on Mental Health and Wellbeing (2016). European framework for action on
cohort studies. American Journal of Psychiatry, 175, 631–648. https://doi.org/10.
mental health and wellbeing, final conference - Brussels, 21-22 January.
1176/appi.ajp.2018.17111194 PMID: 29690792.
Fagt, S., Andersen, L. F., Anderssen, S. A., Becker, W., Borodulin, K., Fogelholm, M., et al.
Stewart-Brown, S., Samaraweera, P. C., Taggart, F., Kandala, N. B., & Stranges, S. (2015).
(2011). Nordic Monitoring of diet, physical activity and overweight. Validation of in-
Socioeconomic gradients and mental health: Implications for public health. The
dicators. Nordic Council of Ministers.
British Journal of Psychiatry, 206, 461–465. https://doi.org/10.1192/bjp.bp.114.
Forsman, A. K., Wahlbeck, K., Aarø, L. E., Alonso, J., Barry, M. M., Brunn, M., et al.
147280.
(2015). Research priorities for public mental health in Europe: Recommendations of
Stranges, S., Samaraweera, P. C., Taggart, F., Kandala, N. B., & Stewart-Brown, S. (2014).
the ROAMER project. The European Journal of Public Health, 25, 249–254.
Major health-related behaviours and mental well-being in the general population:
Grimby, G., Börjesson, M., Jonsdottir, I. H., Schnohr, P., Thelle, D. S., & Saltin, B. (2015).
The health survey for England. BMJ Open, 4, e005878. https://doi.org/10.1136/
The "Saltin-Grimby physical activity level scale" and its application to health re-
bmjopen-2014-005878.
search. Scandinavian Journal of Medicine & Science in Sports, 25(Suppl 4), 119–125.
Strath, S. J., Kaminsky, L. A., Ainsworth, B. E., Ekelund, U., Freedson, P. S., Gary, R. A.,
https://doi.org/10.1111/sms.12611.
et al. (2013). Guide to the assessment of physical activity: Clinical and research ap-
Hamer, M., Stamatakis, E., & Mishra, G. D. (2010). Television- and screen-based activity
plications. A scientific statement from the American Heart Association. Circulation,
and mental well-being in adults. American Journal of Preventive Medicine, 38(4),
128(20), 2259–2279. https://doi.org/10.1161/01.cir.0000435708.67487.da.
375–380. https://doi.org/10.1016/j.amepre.2009.12.030.
Suetani, S., Mamun, A., Williams, G. M., Najman, J. M., McGrath, J. J., & Scott, J. G.
Harris, M. A. (2018). The relationship between physical inactivity and mental wellbeing:
(2017). Longitudinal association between physical activity engagement during ado-
Findings from a gamification-based community-wide physical activity intervention.
lescence and mental health outcomes in young adults: A 21-year birth cohort study.
Health Psychology Open. https://doi.org/10.1177/2055102917753853 January-
Journal of Psychiatric Research, 94, 116–123. https://doi.org/10.1016/j.jpsychires.
June, 1–8.
2017.06.013.
Haver, A., Akerjordet, K., Caputi, P., Furunes, T., & Magee, C. (2015). Measuring mental
Suldo, S., & Shaffer, E. (2008). Looking beyond psychopathology: The dual-factor model
well-being: A validation of the short Warwick–Edinburgh mental well-being scale in
of mental health in youth. School Psychology Review, 37, 52–68.
Norwegian and Swedish. Scandinavian Journal of Public Health, 43(7), 721–727.
Taggart, F., Stewart-Brown, S., & Parkinson, J. (2016). Warwick-Edinburgh mental well-
https://doi.org/10.1177/1403494815588862.
being scale (WEMWBS). User guide (version 2). Edinburgh: NHS Health Scotland,
Hu, G., Jousilahti, P., Borodulin, K., Barengo, N., Lakka, T. A., Nissinen, A., et al. (2007).
Warwick Medical School, University of Warwick.
Occupational, commuting, and leisure-time physical activity in relation to coronary
Tennant, R., Hiller, L., Fishwick, R., Platt, S., Joseph, S., Weich, S., et al. (2007). The
heart disease among middle-aged Finnish men and women. Atherosclerosis, 194(2),
Warwick-Edinburgh mental well-being scale (WEMWBS): Development and UK va-
490–497.
lidation. Health and Quality of Life Outcomes, 5, 63. https://doi.org/10.1186/1477-
Humphreys, D. K., Goodman, A., & Ogilvie, D. (2013). Associations between active
7525-5-63.
commuting and physical and mental wellbeing. Preventive Medicine, 57, 135–139.
Teychenne, M., Abbott, G., Ball, K., & Salmon, J. (2014). Prospective associations be-
Huppert, F. A. (2009). Psychological well-being: Evidence regarding its causes and con-
tween sedentary behaviour and risk of depression in socio-economically dis-
sequences. Applied Psychology: Health and Well-Being, 1, 137–164.
advantaged women. Preventive Medicine, 65, 82–86. https://doi.org/10.1016/j.
Jenkins, R., Meltzer, H., Jones, P., Brugha, T., Bebbington, P., Farrell, M., et al. (2008).
ypmed.2014.04.025.
Mental health: Future challenges. Foresight, 104-08-Fo/on. London, UK: The
Teychenne, M., Abbott, G., Lamb, K. E., Rosenbaum, S., & Ball, K. (2017). Is the link
Government Office for Science.
between movement and mental health a two-way street? Prospective associations
Jones, A., Perkins, C., Stansfield, J., Mason, J., O'Keefe, M., McHale, P., et al. (2013).
between physical activity, sedentary behaviour and depressive symptoms among
North West Mental wellbeing survey 2012/13. London: Public Health England,
women living in socioeconomically disadvantaged neighbourhoods. Preventive
Liverpool John Moores University.
Medicine, 102, 72–78. https://doi.org/10.1016/j.ypmed.2017.07.005.
Keyes, C. (2005). Chronic physical disease and aging: Is mental health a potential pro-
Teychenne, M., White, R. L., Richards, J., Schuch, F. B., Rosenbaum, S., & Bennie, J. A.
tective factor? Ageing International, 33, 88–104.
(2020). Do we need physical activity guidelines for mental health: What does the
Keyes, C., & Simoes, E. (2012). To flourish or not. Positive mental health and all-cause
evidence tell us? Mental Health and Physical Activity, 18, 100315. https://doi.org/10.
mortality. American Journal of Public Health, 102, 2164–2172.
1016/j.mhpa.2019.100315.

7
N. Tamminen, et al. Mental Health and Physical Activity 18 (2020) 100319

Vaillant, G. E. (2012). Positive mental health: Is there a cross-cultural definition? World An observational study of the moderating roles of autonomous and controlled mo-
Psychiatry, 11(2), 93–99. https://doi.org/10.1016/j.wpsyc.2012.05.006. tivation. International Journal of Behavioral Nutrition and Physical Activity, 15, 85.
Wahlbeck, K. (2011). European mental health policy should target everybody. The https://doi.org/10.1186/s12966-018-0722-0.
European Journal of Public Health, 21(5), 551–553. https://doi.org/10.1093/eurpub/ World Health Organization (2018). Mental health: Strengthening our response. Fact sheet.
ckr122. Retrieved from https://www.who.int/en/news-room/fact-sheets/detail/mental-
Wahlbeck, K. (2015). Public mental health: The time is ripe for translation of evidence health-strengthening-our-response.
into practice. World Psychiatry, 14, 36–42. Zechmeister, I., Kilian, R., McDaid, D., & the MHEEN group (2008). Is it worth investing
White, R. L., Babic, M. J., Parker, P. D., Lubans, D. R., Astell-Burt, T., & Lonsdale, C. in mental health promotion and prevention of mental illness? A systematic review of
(2017). Domain-specific physical activity and mental health: A meta-analysis. the evidence from economic evaluations. BMC Public Health, 8, 20. https://doi.org/
American Journal of Preventive Medicine, 52(5), 653–666. https://doi.org/10.1016/j. 10.1186/1471-2458-8-20.
amepre.2016.12.008. Zhang, Z., & Chen, W. (2019). A systematic review of the relationship between physical
White, R. L., Parker, P. D., Lubans, D. R., MacMillan, F., Olson, R., Astell-Burt, T., et al. activity and happiness. Journal of Happiness Studies, 20, 1305. https://doi.org/10.
(2018). Domain-specific physical activity and affective wellbeing among adolescents: 1007/s10902-018-9976-0.