OUP user menu

Educational inequalities in general and mental health: differential contribution of physical activity, smoking, alcohol consumption and diet

Nanna Kurtze, Terje A. Eikemo, Carlijn B. M. Kamphuis
DOI: http://dx.doi.org/10.1093/eurpub/cks055 223-229 First published online: 10 May 2012

Abstract

Background: Behavioural, material and psychosocial risk factors may explain educational inequalities in general health. To what extent these risk factors have similar or different contributions to educational inequalities in mental health is unknown. Methods: Data were derived from the Norwegian Survey of Level of Living from 2005, comprising 5791 respondents aged ≥25 years. The study objectives were addressed by means of a series of logistic regression analyses in which we examined: (i) educational inequalities in self-reported general and mental health; (ii) the associations between behavioural, material and psychosocial risk factors and general and mental health, controlled for sex, age and education; and (iii) the contribution of risk factors to the observed health gradients. Results: The lower educated were more likely to be in poor health [odds ratio (OR): 3.46 (95% confidence interval, CI: 2.84–4.21)] and to be in poor mental health [OR: 1.41 (95% CI: 1.12–1.78)] than the highest educated. The joint contribution of behavioural, material and psychosocial risk factors explained all the variations of mental health inequalities, whereas these were able to explain ∼40% of the inequalities in general health. Both behavioural and material risk factors contributed substantially to the explanation of general and mental health inequalities, whereas the psychosocial risk factor (i.e. having close persons to communicate with) only seemed to make a larger difference for the explanation of mental health inequalities. Conclusion: Policies and interventions to reduce health inequalities should have a broad focus. Combined strategies should be applied to improve physical activity, decrease smoking and improve material and psychosocial conditions among lower educated groups, to achieve the true potential of reducing inequalities in both general and mental health.

Introduction

Many studies have shown gradients in physical ill health and mortality by socio-economic position (SEP).1 These gradients have also been found for major mental illness and for common mental disorders including depression.2,3 The question arises, as there are similar gradients in both physical and mental morbidity, could there be factors that contribute to the gradients in both physical and mental health? This is important information for intervention development, as interventions to reduce inequalities in health could potentially have the largest effect when targeted towards those risk factors that affect both general and mental health. Behavioural risk factors, such as physical inactivity, smoking and poor diet, are known for their important contribution to socio-economic inequalities in general health outcomes,4 in combination with material and psychosocial circumstances.57 For socio-economic inequalities in mental health, on the other hand, less is known about the contribution of behavioural risk factors.8 Physical activity (PA) and exercise seem to have beneficial effects across several mental health outcomes9; however, their contribution to inequalities in mental health is less consistent.8 We know that smoking is an independent risk factor for psychological distress, which is also unequally distributed according to SEP.8 Smoking has been linked with depression, which is also the case for non-daily fruit and vegetable consumption.10 Overall, little is known about the association between lifestyle and self-rated mental and physical health.1114

In the following, we will draw extra attention towards PA among the lifestyle factors. While smoking, alcohol consumption and diet can be regarded as traditional lifestyle factors, PA is a risk factor that is more distinctively associated with the modern Western lifestyle because technological advances in television, video, the internet and gaming increasingly encourage sedentary lifestyles. Also, while most activity occurs during work, chores or transport in low-income countries, most activity occurs during leisure time in high-income countries. Physical inactivity is estimated to cause ∼21–25% of the breast and colon cancer burden, 27% of the diabetes and ∼30% of the ischaemic heart disease burden, and improvements in PA levels may even prevent up to 6% of the annual deaths.15 PA may become one of the most important risk factors in the next generation. From a public health perspective, it is therefore important that PA is promoted among all social groups to fulfil true health potential.

In the PA literature, it is debated that regular moderate activity has the most positive health effects,16,17 whereas others argue that vigorous PA is most important for physical health.18 Some have found that being moderately active is enough to have a life-prolonging effect, whereas others show that vigorous activity has an extra health-enhancing effect, also among those being moderately active.19,20 Other findings indicate that all activities beyond the sedentary level are beneficial for mental health.17 We will add to this debate by investigating whether moderate and vigorous activities have independent contributions to inequalities in general and mental health.

Besides lifestyle factors, we also know that material and psychosocial factors contribute to the explanation of socio-economic inequalities in general health and mental health, respectively.5,7,21,22 As lifestyle factors are partly influenced by material and psychosocial factors,23,24 these should be taken into account when investigating the role of lifestyle for socio-economic inequalities in general and mental health.22

Therefore, the goals of this article are 3-fold: (i) to explore socio-economic gradients in general and mental health; (ii) to assess associations of behavioural (moderate and vigorous PA, smoking, nutrition and alcohol consumption), material (household income) and psychosocial risk factors (social support) with SEP, and with general and mental health; and (iii) to investigate the independent and joint contributions of lifestyle factors, material factors and psychosocial factors to health inequalities (figure 1).

Methods

The study population selected for this study consists of respondents from a population-based cross-sectional study from Statistics Norway, the Survey of Level of Living 2005 (the Health study).25 The data and extensive documentation are freely available for downloading at the Norwegian Social Sciences Data Services (NSD) website (www.nsd.uib.no). Available data from 6732 individuals were included (16–79 years) in our study. After excluding those aged ≤24 years (because many of them will not have completed their education at these ages) and listwise deletion of included variables, the analytic sample consisted of 5764 respondents.

Health outcomes

We used two subjective indicators of health: general health and mental health. Self-reported general health was constructed from a variable asking: ‘How is your health in general?’ Eligible responses were ‘very good’, ‘good’, ‘fair’, ‘bad’ and ‘very bad’. We dichotomised the variable into ‘less than good’ health (‘fair’, ‘bad’ and ‘very bad’) vs. ‘very good or good health’. The 25-item Hopkin’s Symptom Check List (HSCL)26 was used to assess anxiety and depression. Respondents were defined to have presence of depressive symptoms if they had a value of 1.55 or above, which is in line with previous research.27

Independent variables

We applied four indicators of behavioural lifestyle factors. Two variables describing the frequency of PA were applied. First, moderate PA was based on a question asking on how many days in the last week the respondent had carried out moderate PA for at least 10 min consecutively. Moderate activities refer to activities that make you breathe somewhat faster than normal. Examples were easy exercises, swimming, walking, biking at normal speed or light lifting. Second, vigorous PA was based on a question asking on how many days during the last 7 days had the respondent carried out exhausting PA activities for at least 10 min consecutively. Examples were heavy lifting, heavy domestic work, fast jogging or cycling or ball playing. For both variables, we compare the non-physically active with those who were active at least 1 day in the last week. Poor nutrition was measured according to the daily intake of fruits and vegetables, respectively. Two variables were collapsed in the order that we compare the health outcomes of those who stated that they neither ate fruits nor vegetables on a daily basis, with others. Alcohol consumption was measured according to any kind of alcohol consumption 4–7 times weekly in the last 12 months. Finally, we operationalised smoking according to daily smokers (including former daily smokers) and non-smokers (including occasional smokers).

Household income (after tax) was applied as material factor, whereas social support, measured by the number of persons being close to the respondent, was used as psychosocial factor.

Educational level was measured as the highest level of education the person had completed. Using the OECD guidelines,28 the data were classified into three categories. Sex and age were included as demographic control variables in all analyses. A descriptive overview of all variables is presented in table 1.

View this table:
Table 1

Descriptive statistics of applied variables (N = 5764)

VariablesDescriptionLow valueHigh valueN (%) for discrete variables
Poor mental health and poor general health (dependent variables)
    Mental healthBased on HSCL. Poor mental health = 1 if the answers from 25 questionsa have a mean larger than 1.55 (categories were 1 = not afflicted, 2 = slightly afflicted, 3 = pretty afflicted and 4 = very much afflicted)01605 (10.5)
    General healthVery good and good health (0) vs. less than good health (1)011174 (20.4)
Demographical variables and educational attainment
    Age, mean (SE)Age at the time of the interview259849.70 (15.76)
    WomenSex, 0 = men, 1 = women012877 (49.9)
    EducationPrimary1336 (23.2)
Secondary2573 (44.6)
Tertiary1855 (32.2)
No information90 (1.6)
Behavioural
    Smoking‘Do you smoke or have you ever smoked on a daily basis?’Daily (or former daily) smoker012740 (47.5)
Has never smoked (reference)013024 (52.5)
    Poor nutritionDoes not eat vegetables or fruit/berries on a daily basis011286 (22.3)
Physical activity
    No moderate0 (vs. 1–7) days of moderate physical activity last week011988 (34.5)
    No vigorous0 (vs. 1–7) days of hard physical activity last week012768 (48.0)
    AlcoholConsumed any kind of alcohol 2–7 times weekly the last 12 months01795 (23.3)
Material
    IncomeHousehold income after tax
        First income quartileLowest011441 (25.0)
        Second income quartile011441 (25.0)
        Third income quartile011441 (25.0)
        Fourth income quartileHighest011441 (25.0)
Psychosocial
    Social supportHow many persons are close to you
        0–2 persons011317 (22.8)
        3–5 persons012539 (44.0)
        ≥5 persons011908 (33.1)
  • a: Headache/shivering/faintness or dizziness/nervousness, inner unrest/sudden fear for no reason/constantly afraid or worried/heartbeat/feeling of tenseness, jittery/attack of anxiety or panic/restless to the extent that it is difficult to sit quietly/lacking energy, everything elapses slower than usual/blaming yourself/cry easily/suicidal/poor appetite/trouble sleeping/feeling of hopelessness regarding the future/dejected, sad/feeling of loneliness/lack of sexual appetite and interest/feeling of being tricked into a trap or captured/worried a lot or restless/no interest for anything/feeling that everything is a strain/feeling of being useless?

Statistical procedure

All analyses were performed using logistic regressions with 95% confidence intervals (CIs) applying poor general health and poor mental health as outcome variables, respectively. The independent and joint contributions of the various risk factors to educational inequalities in health were calculated using a stepwise exclusion approach. This approach requires three sets of regression analyses. First, an empty Model A with only educational attainment (adjusted for sex and age). Furthermore, regression models in which only one risk factor category is excluded (B) were needed to estimate the reduced effect of education that is attributable to each risk factor. Finally, a full Model C was run with educational attainment and all risk factor variables adjusted for sex and age (figure 1). The joint effect of all risk factors combined was calculated as AC/(A – 1) × 100. The independent effect of each risk factor was derived from the join effect model in this way: [(AC)/(A – 1)] × 100 – [(AB)/(A – 1)] × 100.

Figure 1

Conceptual model relating SEP and risk factors to general health and mental health

Results

The presentation of the results will follow the three research questions, and we will therefore start by describing the health gradients in Norway. As shown in the first model of tables 2 and 3, respectively, educational gradients in both poor general and mental health were observed. Lower educated were more likely to be in poor general health [odds ratio (OR): 3.46 (95% CI: 2.84–4.21)] and to be in poor mental health [OR: 1.41 (95% CI: 1.12–1.78)] than the highest educated.

View this table:
Table 2

Logistic regressions (95% CI) of the effect of lifestyle factors on poor general health for people aged ≥25 years (N = 5764)

VariablesOR (95% CI) of poor general health
Model 1Model 2aModel 2bModel 3Model 4Model 5
Education
    Primary3.46 (2.84–4.21)2.57 (2.10–3.15)2.79 (2.28–3.41)2.77 (2.26–3.39)2.51 (2.04–3.08)2.43 (1.98–2.99)
    Secondary1.96 (1.63–2.36)1.65 (1.36–1.99)1.75 (1.45–2.11)1.73 (1.43–2.08)1.63 (1.35–1.97)1.60 (1.33–1.94)
    Tertiary111111
Women1.04 (0.91–1.19)1.00 (0.96–1.26)1.05 (0.91–1.20)1.02 (0.881.17)0.98 (0.85–1.13)1.00 (0.87–1.15)
Age1.03 (1.02–1.03)1.03 (1.02–1.03)1.02 (1.02–1.03)1.02 (1.02–1.03)1.02 (1.02–1.02)1.02 (1.01–1.02)
Behaviour
    No moderate PA1.34 (1.16–1.54)1.34 (1.16–1.55)1.32 (1.14–1.52)
    No vigorous PA1.97 (1.70–2.29)1.95 (1.67–2.27)1.93 (1.66–2.25)
    Poor nutrition1.18 (1.00–1.39)1.15 (0.98–1.36)1.12 (0.95–1.33)1.10 (0.93–1.31)
    Daily smoking1.49 (1.30–1.72)1.46 (1.27–1.68)1.45 (1.26–1.67)1.45 (1.26–1.66)
Material
    Income quartile 12.01 (1.62–2.50)2.08 (1.68–2.57)1.89 (1.52–2.35)1.85 (1.49–2.31)
    Income quartile 21.80 (1.45–2.23)1.84 (1.48–2.27)1.77 (1.43–2.20)1.75 (1.41–2.17)
    Income quartile 31.39 (1.11–1.73)1.39 (1.11–1.73)1.38 (1.10–1.72)1.37 (1.10–1.71)
    Income quartile 41111
Psychosocial
    0–2 close persons1.44 (1.20–1.73)1.47 (1.23–1.76)1.38 (1.15–1.66)1.35 (1.12–1.62)
    3–5 close persons1.02 (0.86–1.20)1.02 (0.87–1.20)1.00 (0.84–1.17)1.00 (0.85–1.18)
    ≥5 close persons1111
Joint and independent contributions of lifestyle factors to the explanation of educational health inequalitiesPhysical activityBehaviourMaterialPsychosocialAll combined
P: 5.7% (moderate: 1.7%; vigorous: 2.9%)P: 14.7%P: 13.9%P: 3.3%P: 41.9%
S: 4.2% (moderate PA: 0.1%; vigorous PA: 4.2%)S: 14.6%S: 12.5%S: 2.1%S: 36.5%
  • Calculations: joint contributions of live style factors on educational health inequalities: [100 × (Model 1 – Model 5)/(Model 1 – 1)].

  • Independent contributions of live style factors on educational health inequalities:

  • Physical activity: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 2a)/(Model 1 – 1)].

  • Behaviour: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 2b)/(Model 1 – 1)].

  • Material: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 3)/(Model 1 – 1).

  • Psychosocial: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 4)/(Model 1 – 1)].

  • P, primary vs. tertiary education; S, secondary vs. tertiary education.

  • Significant results are boldfaced.

View this table:
Table 3

Logistic regressions (95% CI) of the effect of lifestyle factors on poor mental health for people aged ≥25 years (N = 5764)

VariablesOR (95% CI) of poor mental health
Model 1Model 2aModel 2bModel 3Model 4Model 5
Education
    Primary1.41 (1.12–1.78)1.05 (0.82–1.34)1.16 (0.911.48)1.09 (0.861.39)1.09 (0.851.39)1.01 (0.791.30)
    Secondary1.11 (0.901.36)0.93 (0.761.16)1.00 (0.811.23)0.95 (0.771.17)0.96 (0.781.19)0.92 (0.741.14)
    Tertiary111111
Women1.72 (1.452.05)1.84 (1.54–2.21)1.80 (1.51–2.14)1.81 (1.51–2.17)1.70 (1.42–2.03)1.79 (1.49–2.14)
Age0.99 (0.980.99)0.99 (0.980.99)0.99 (0.980.99)0.99 (0.980.99)0.99 (0.980.99)0.99 (0.980.99)
Behaviour
    No moderate PA1.13 (0.941.36)1.16 (0.961.39)1.13 (0.941.36)
    No vigorous PA1.34 (1.12–1.62)1.35 (1.12–1.62)1.32 (1.10–1.60)
    Poor nutrition1.10 (0.901.36)1.11 (0.901.37)1.09 (0.891.35)1.07 (0.871.32)
    Daily smoking1.58 (1.32–1.88)1.56 (1.31–1.86)1.57 (1.31–1.87)1.55 (1.30–1.85)
Material
    Income quartile 11.57 (1.22–2.02)1.63 (1.27–2.09)1.57 (1.22–2.03)1.53 (1.19–1.98)
    Income quartile 21.27 (0.981.65)1.31 (1.01–1.69)1.30 (1.01–1.69)1.27 (0.981.64)
    Income quartile 31.11 (0.851.43)1.12 (0.861.45)1.12 (0.871.46)1.11 (0.851.43)
    Income quartile 4111
Psychosocial
    0–2 close persons1.90 (1.50–2.42)1.90 (1.49–2.41)1.86 (1.47–2.37)
    3–5 close persons1.38 (1.12–1.70)1.38 (1.12–1.70)1.38 (1.12–1.70)
    ≥5 close persons111
Joint and independent contributions of lifestyle factors to the explanation of educational health inequalitiesPhysical activityBehaviourMaterialPsychosocialAll combined
P: 9.8% (Moderate: 2.5%; Vigorous: 4.9%)P: 36.6%P: 19.6%P: 19.6%P: 97.6%
S: 100.0% (Moderate PA: 100%; Vigorous PA: 100%)S: 100.0%S:100.0%S:100.0%S: 100.0%
  • Calculations: joint contributions of live style factors on educational health inequalities: [100 × (Model 1 – Model 5)/(Model 1 – 1)].

  • Independent contributions of lifestyle factors on educational health inequalities:

  • Physical activity: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 2a)/(Model 1 – 1)].

  • Behaviour: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 2b)/(Model 1 – 1)].

  • Material: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 3)/(Model 1 – 1).

  • Psychosocial: [100 × (Model 1 – Model 5)/(Model 1 – 1)] – [100 × (Model 1 – Model 4)/(Model 1 – 1)].

  • P, primary vs. tertiary education; S, secondary vs. tertiary education.

  • Significant results are boldfaced.

Associations of lifestyle, material and psychosocial factors with education and with health

Supplementary Web table A presents prevalences and OR of poor general and mental health for each risk factor according to educational groups. The prevalence rates show that the occurrence of unhealthy behaviour was more frequent in the lower educational groups, with one main exception. The consumption of alcohol was more frequent among the higher educated (5.4%) when compared with the mid (2.8%) and lower (1.2%) educated. It appeared that lack of PA (44.2% and 59.3% for no moderate and no vigorous PA, respectively) and smoking (55.3%) had the highest prevalences of all risk factors for the lower educated. The risk factor-specific ORs (adjusted for age, sex and education) reflect these differences, which were generally larger for poor general health when compared with poor mental health.

Among the behavioural risk factors, the ORs of ‘no moderate PA’ were 1.67 (95% CI: 1.46–1.92) for poor general health and 1.31 (95% CI: 1.10–1.56) for poor mental health. These were slightly larger for ‘no vigorous PA’, which were OR: 2.22 (95% CI: 1.92–2.57) for poor general health and OR: 1.47 (95% CI: 1.23–1.76) for poor mental health. Furthermore, no intake of fruit and vegetables was associated with smaller OR: 1.28 (95% CI: 1.09–1.51) for poor general health and OR: 1.19 (95% CI: 0.97–1.46) for poor mental health. The ORs of smoking and alcohol consumption were, on the other hand, larger for poor mental health when compared with poor general health. Alcohol consumption demonstrates a rather strong association with poorer mental health but an inverse prevalence gradient by education. For this reason, alcohol consumption could not contribute to the explanation of educational inequalities and was therefore left out of the explanatory models.

The material risk factor of income demonstrated a powerful and almost linear social gradient for both health outcomes. This gradient also showed that income was a more important determinant for poor general health than for poor mental health within all income quartiles. When compared with the highest income quartile, the OR of reporting poor general health was 2.14 (95% CI: 1.73–2.65) in the lowest quartile, OR: 1.87 (95% CI: 1.52–2.32) in the second and OR: 1.40 (95% CI: 1.12–1.75) in third. The same numbers were OR: 1.67 (95% CI: 1.30–2.14), OR: 1.34 (95% CI: 1.04–1.73) and OR: 1.13 (95% CI: 0.87–1.46) for poor mental health.

As for the psychosocial risk factor, having close persons to communicate with seems more important for poor mental health when compared with poor general health. Having only 0–2 persons to rely on, gave an OR: 1.53 (95% CI: 1.28–1.82) for poor general health and OR: 1.99 (95% CI: 1.57–2.52) for poor mental health.

Explanatory models

In the fully adjusted models of tables 2 and 3, the educational gradients for general and mental health were reduced by 41.9% [to OR: 2.43 (95% CI: 1.98–2.99)] and 97.6% [to OR: 1.01 (95% CI: 0.79–1.30)], respectively. The behavioural, material and psychosocial contribution to the explanation of these inequalities varied between the two health outcomes.

All behavioural risk factors together explained 14.7% of the inequalities in general health between lower and higher educated. Income explained approximately the same amount (13.9%), whereas social support only explained 3.3%. Both moderate and vigorous PA made an independent contribution to educational inequalities in general health also when other behavioural factors, and material and psychosocial factors were taken into account (Model 2a, table 2). The joint effect of both PA measures amounted to 5.7% (primary vs. tertiary) and 4.2% (secondary vs. tertiary) of the educational differences in poor general health.

With respect to mental health inequalities, behavioural factors explained 36.6%, whereas material and psychosocial risk factors explained 19.6% each. PA explained more of the differences in mental health when compared with general health (Model 2a, table 3). The joint effect of both PA measures amounted to 9.8% (primary vs. tertiary) and 100% (secondary vs. tertiary) of the educational differences in mental health. For the latter case, it should be mentioned that the observed OR reduction was from 1.11 to 0.99.

Discussion

This study showed educational differences in both general and mental health in Norway, with a larger gradient for general than for mental health. Behavioural, material and psychosocial risk factors contributed differently to the explanation of inequalities in general and mental health. The joint contribution of these risk factors explained all the variation of mental health inequalities, whereas these were only able to explain ∼40% of the inequalities in general health. Furthermore, both behavioural and material risk factors contributed substantially to the explanation of general and mental health inequalities, whereas the psychosocial risk factor (i.e. having close persons to communicate with) only seemed to make a larger difference for the explanation of mental health inequalities. Finally, both vigorous and moderate PA seemed to have independent associations with both health outcomes, in which vigorous PA is probably the most important (although the real health potential likely lies within the combination to these two types of activities). Before we further discuss the main results, we will address some important limitations of this study.

Limitations

The first limitation concerns the cross-sectional design of the study. All studies with a similar design are subject to some well-known methodological problems. For example, we cannot draw inferences regarding the direction of causation in the observed relationships. PA may not only have positive health effects but also it may be easier for people with good health to participate in vigorous physical activities, such as sporting activities, than for people with poorer health. This effect may be stronger for general health than for mental health (e.g. poor physical health may very well pose stronger limits to exercise than poor mental health). This is important, as this would imply that any overestimation of effects due to selectivity would be stronger for general than for mental health. Furthermore, smoking, alcohol consumption and poor nutrition may be predictors of poor mental health, but a reverse association is also likely, i.e. having poor mental health may result in an unhealthy lifestyle, for instance, when smoking and alcohol consumption are applied to cope with feelings of anxiety and depression. Therefore, as opposed to a longitudinal design, it is hard to speculate about the effects in a prospective manner.

Second, the study is subject to the problem of common method variance as both the independent and dependent variables are based on self-reports. Reports of longstanding illness or serious disease are considered less subjective than self-rated health measures, and evidence suggests that the former represents a source of reliable and valid data on health status.29,30 Third, our findings may have been influenced by the veracity of the respondents. Sensitive questionnaire items, such as those related to smoking and alcohol are potentially subject to biased reporting. However, self-reported smoking habits have been found to be reliable in Norwegian men and women by an anonymous designed study to foster truthful reporting.31 Finally, the limitation of recalled PA on questionnaires has been well documented.32

Discussion of the results

Our observed gradients in general and mental health in Norway were in line with previous studies.8,33 Moreover, the steepness of the educational gradient in mental health (i.e. presence of depressive symptoms) was similar to the gradient found elsewhere.2 Other studies have also found behavioural, material and psychosocial risk factors important for the explanation of inequalities in mortality,47 whereas the evidence for inequalities in general health and particularly for mental health is much more limited.8

Our analyses have revealed strong associations between PA and health from a nationally representative sample. The observed effect of PA on mental and general health was surprisingly stable (and significant) also after controlling for SEP, social support and lifestyle factors. Pisinger et al.34 found higher PA at baseline to be associated with better self-reported health and improved PA to be associated with improved physical health at a 5-year follow-up. Another study15 reported that cross-sectional associations were mainly observed for physical components, whereas longitudinal associations were observed for mental components.

Besides PA, other lifestyle factors, and in particular smoking, also made a significant contribution to inequalities in both mental and general health. The underlying mechanism between smoking and mental health is, unlike that of PA, probably not of a causal but more likely, of a selective nature: those experiencing mental health problems are more likely to engage in risk behaviours, such as smoking and heavy alcohol consumption. These behaviours could also be applied for reasons of coping with mental difficulties or difficult life circumstances in general.

Furthermore, income and social support were important explanatory factors of both general and mental health inequalities as well, which is in line with previous research.7,8,35 This underlines the complexity of addressing inequalities in physical and mental health, as, ideally, all these behavioural, material and psychosocial risk factors need to be addressed by policies or interventions simultaneously to decrease health inequalities.

Alcohol consumption was more prevalent among high compared with low socio-economic groups. This is contrary to the findings of a Dutch study,36 but in line with the study from Marmot37 suggesting that moderate drinking is more common among those of higher socio-economic status. It is important to be aware of the inverse gradient of alcohol consumption when analysing socio-economic inequalities in mental health in Norway, as an inclusion of alcohol consumption would have led to an underestimation of socio-economic inequalities in mental health.

Conclusion

Policies and interventions aiming at improving the average mental and general health status of populations should have a broad focus. Combined strategies should be applied to improve PA, decrease smoking and improve material and psychosocial conditions among lower educated groups, to achieve the true potential of reducing inequalities in both general and mental health.

Supplementary Data

Supplementary Data are available at Eurpub online.

Funding

Friluftslivets fellesorganisasjon (FRIFO) supported this study.

Conflicts of interest: None declared.

Key points

  • Behavioural and material risk factors were important for the explanation of socio-economic inequalities in both general and mental health, whereas psychosocial risk factors seemed to matter most for the explanation of inequalities in mental health.

  • Both vigorous and moderate PA seemed to have independent associations with both health outcomes, in which vigorous PA probably was the most important.

  • Interventions aiming at reducing educational inequalities in mental ill health and physical illness simultaneously should be targeted towards behavioural and material risk factors.

References

View Abstract