Scottish Health Survey - topic report: physical activity

5. Results - Multivariate Analysis (Logistic Regression) of Factors Associated with Physical Activity Levels

5.1 Factors associated with meeting the new physical activity guidelines

Table 2 provides a list of the factors from the bivariate analysis which were entered into the logistic regression model and an indication of whether the characteristics retained significant association with meeting physical activity recommendations in final models. Results are presented for all respondents and by sex.

Table 2: Factors associated with achieving recommendations for physical activity (new guidelines), 2012

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	√	√	√
Marital Status	ns	ns	ns
Socioeconomic
Equivalised Income quintiles	ns	ns	ns
Deprivation - SIMD Quintiles	ns	ns	ns
Economic Activity Status (EAS)	√	√	√
Highest Education Qualification	ns	ns	ns
Health and Lifestyle
Self-Assessed health	√	√	√
Longstanding Illness	√	√	ns
Cigarette smoking	√	ns	√
Mental wellbeing	√	√	√
life satisfaction	ns	ns	ns
BMI	√	√	√

√ = significant in final model; ns = non-significant; N/A = not applicable

For all adults, demographic and health and lifestyle factors were of primary importance associated with meeting the 2012 physical activity guidelines, whereas socioeconomic factors appeared overall less influential, with the exception of economic activity status. The factors remaining significantly associated in the logistic regression models were sex, age, economic activity status, self-assessed health, disability, cigarette smoking, mental wellbeing and BMI.

When analysis was conducted separately for men and women, the range of factors was similar except that disability was a factor for men but not women and cigarette smoking was a factor for women but not men.

Marital status, equivalised income, deprivation, educational attainment and life satisfaction were not significant in any of the models.

Sex - Women were found to be less likely to achieve recommended levels of physical activity than men. In 2012, the odds for men to achieve the physical activity guideline in comparison to women were 1.47 (CI 1.24 - 1.73).

Age - In comparison to those aged 75 or older, all the younger age categories were significantly more likely to achieve recommended physical activity levels with the greatest difference for those aged 16-24 years (OR 4.07, CI 2.20 - 7.54).

A decrease in physical activity observed with increasing age appears more marked in men than in women, with a particularly large difference between men aged 16-24 years and those aged 75 or older (OR 12.81, CI 4.81 - 34.12), however, the large confidence intervals suggest there is a large degree of variation within the age category 16-24, such that this apparent marked effect should be viewed with caution.

The bivariate analyses indicated a steady decline with increasing age in the proportion meeting physical activity guidelines, seen mainly in men. Logistic regression revealed a more complex relationship, once other factors had been controlled for. In women, likelihood of meeting the guidelines increased to age 44 and then declined to age 64. There was a slight increase again at age 65-74. In men, there was a large decrease in the likelihood of meeting the guidelines between ages 16-24 and 25-34, an increase to age 44 and then a steady decline to age 74.

Economic activity status - In comparison to those in paid work, those who were either unable to work, looking for work or retired were all less likely to meet the physical activity recommendations. Those who were unable to work had the lowest odds and were least likely to be active at the recommended level (OR 0.22, CI 0.13 - 0.36). There was no significant difference between those in paid work and those either in full-time education or looking after family/home or those who responded 'other'.

When the analysis was conducted separately for men and women, some differences were apparent. Men in full-time education were less likely to achieve recommended physical activity levels compared to those in paid work. No such difference was observed in women. Women who are retired were less likely to achieve physical activity recommendations compared to women in paid work, whereas no such effect was observed for men.

Self-assessed health - As health status improved so too did the likelihood of achieving the recommended physical activity level. Those who reported very good/good health had higher odds of achieving the recommended level of physical activity than those who rated their health as bad/very bad (OR 2.68, CI 1.81 - 3.95). Those with fair health were also more likely to be active than those with poor health (OR 1.82, CI 1.25 - 2.66).

When the analysis was conducted separately for men and women, the relationship was more marked for women than men and followed a similar pattern to that for all adults, with increasing odds of meeting physical activity guidelines with increasing self-rated health. Women with good health had odds of 3.01 (CI 1.84 - 5.22) compared to women with poor health. For men, there was still a higher likelihood of meeting physical activity recommendations for those with good health compared to bad health (OR 2.34, CI 1.26 - 4.32), however, there was no difference between those with fair health and good health.

Disability - Having a long term limiting illness (LLI) had a negative association with achieving physical activity recommendations (OR 0.72, CI 0.57 - 0.89). There was no difference between those with a long term illness that was not limiting and those without an LLI. When the analysis was conducted separately for men and women, the relationship remained only for men (OR 0.68, CI 0.49 - 0.96).

Cigarette smoking - Compared to those who have never smoked, a current cigarette smoker was found to be significantly less likely to achieve recommended levels of physical activity (OR 0.71, CI 0.57 - 0.87). No difference was found between those who were previous smokers and those who had never smoked. Separate analysis by sex revealed that cigarette smoking was not a significant factor predicting physical activity levels for men but was for women (OR 0.67, CI 0.50 - 0.89). Women who have never smoked were more likely to achieve physical activity recommendations compared to current smokers.

Mental wellbeing - Lower than average mental wellbeing was significantly associated with a lower likelihood of achieving physical activity recommendations (OR 0.6, CI 0.46 - 0.79). This was true for both men and women.

BMI - Obesity, i.e. a BMI of 30 or more, had a negative impact on likelihood of achieving recommended levels of physical activity. Those with a BMI of less than 30 were significantly more likely to achieve the recommendations (OR 1.48, CI 1.24 - 1.77). This relationship was true for both men and women, but more marked in men (Men: OR 1.63, CI 1.24 - 2.15. Women: OR 1.42, CI 1.11 - 1.81).

Chi-square results in the logistic regression combined with odd ratios were used to identify key individual factors and the category of each relevant variable most likely to be associated with meeting the physical activity guidelines. Using this information, Figures 18 and 19 illustrate the 'types' of individuals (in this study) most likely to meet the recommended level of moderate to vigorous physical activity in 2012.

Figure 18: Characteristics most strongly associated with men achieving the recommended physical activity levels in 2012:

Figure 19: Characteristics most strongly associated with women achieving the recommended physical activity levels in 2012:

5.2 Comparison between meeting the old physical activity guidelines and the new guidelines

One possible impact of the removal of the stipulation that activity should be carried out regularly rather than a total given for the week, is that it affects the distribution of factors associated with meeting the physical activity guidelines in some way. To investigate this, logistic regression of the factors associated with meeting the old guidelines (30 minutes moderate or vigorous physical activity (MVPA) on at least 5 days per week) was carried out (see Table 3) using data from the 2011 SHeS, and results compared to those for the current guidelines (150 minutes MVPA per week) in 2012.

For all adults, the factors remaining significantly associated with meeting the old physical activity guidelines were age, sex, economic activity status, self-assessed health, disability and BMI. A similar pattern of relationships was found also when the analysis was conducted separately for men and women. Figures 20 and 21 illustrate the 'types' of individuals (in this study) most likely to meet the recommended level of moderate to vigorous physical activity in 2011. Further details on the results of the analysis on the 2011 SHeS data can be found in Table 20, Annex B.

Table 3: Factors associated with achieving recommendations for physical activity (old guidelines), 2011

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	√	√	√
Marital Status	ns	ns	ns
Socioeconomic
Equivalised Income quintiles	ns	ns	ns
Deprivation - SIMD Quintiles	ns	ns	ns
Economic Activity Status (EAS)	√	√	√
Highest Education Qualification	ns	ns	ns
Health and Lifestyle
Self-Assessed health	√	√	√
Longstanding Illness	√	√	√
Cigarette smoking	ns	ns	ns
Mental wellbeing	ns	ns	ns
life satisfaction	ns	ns	ns
BMI	√	√	√

√ = significant in final model; ns = non-significant: N/A = not applicable

Figure 20: Characteristics most strongly associated with men achieving the recommended physical activity levels in 2011:

Figure 21: Characteristics most strongly associated with women achieving the recommended physical activity levels in 2011:

Overall, very little difference was found in the demographic, socioeconomic and health and lifestyle patterning of those who met the new compared to the old guidelines. The main associations in both years were found to be demographic and health and lifestyle with no significant relationships with income, deprivation or educational status.

Some differences were found between specific health and lifestyle factors:

• Disability - Those with a longstanding illness were less likely to meet the physical activity recommendations in both 2011 and 2012, however, in 2011 this also applied to both men and women, whereas in 2012 disability was a factor only for men.
• Cigarette smoking - In 2011, there was no relationship between the cigarette smoking status of a person and meeting physical activity recommendations, however, in 2012 cigarette smoking status was found to be related to all adults and women's likelihood of achieving the recommended level of physical activity.
• Mental wellbeing - In 2011, there was no relationship between mental wellbeing and meeting physical activity recommendations, however, in 2012 mental wellbeing was found to be related to both men and women's likelihood of achieving the recommended level of physical activity.

5.3 Factors associated with participation in different domains of physical activity

This sections presents the results from logistic regression of the factors associated specifically with participation across five different domains of physical activity in 2012: sport and exercise, walking, housework, manual work/gardening/DIY and activity at work (occupational physical activity).

Sport and exercise

Demographic, socioeconomic and health and lifestyle factors were all associated with whether a person was likely to report participation in sport and exercise or not, see Table 4 below and Table 21 in Appendix B.

Table 4: Significant factors associated with sport and exercise participation

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	√	√	√
Marital Status	ns	ns	ns
Socioeconomic
Equivalised Income quintiles	√	ns	ns
Deprivation - SIMD Quintiles	ns	√	ns
Economic Activity Status (EAS)	ns	ns	ns
Highest Education Qualification	√	√	√
Health and Lifestyle
Self-Assessed health	√	√	√
Longstanding Illness	ns	ns	ns
Cigarette smoking	√	ns	√
Mental wellbeing	√	√	√
life satisfaction	√	√	ns
BMI	√	√	√

Men were found to be significantly more likely to participate in sport and exercise than women (OR 1.49, CI 1.26 - 1.75).

Increasing age had a negative relationship with sport and exercise participation, with those aged 16-24 found to be substantially more likely to participate in sport and exercise than those aged 75 or over (OR 7.63, CI 4.09 - 14.24). This relationship was even more marked in men than for women, however, as noted previously this should be interpreted with caution due to large confidence intervals indicating wide variation within the youngest age category.

Those in the highest equivalised income quintile were significantly more likely to participate in sport and exercise compared to those in the lowest quintile (OR 1.60, CI 1.17 - 2.17). There was no difference between other income quintiles. This was not a significant factor when analyses were conducted separately for men and women. This may be due to a small variation relative to the bands of equivalised income included in the analysis.

Better self-rated health was positively associated with participation in sport and exercise. Those reporting better health were significantly more likely to participate compared to those reporting poor health (OR 2.41, CI 1.60 - 3.65). A similar pattern was found when analyses were conducted separately for men and women.

There was a positive association between increasing level of educational attainment and participation in sport and exercise. Those with a degree or higher were more likely to participate in sport and exercise than those with no qualifications (OR 2.67, CI 2.02 - 3.52). A similar pattern was found when analyses were conducted separately for men and women.

A lower BMI was found to be positively associated with participation in sport and exercise. This was found to be the case for both men and women. Adults with a BMI less than 30 were more likely to participate in sport and exercise than those classed as obese with a BMI of 30 or more (OR 1.53, CI 1.29 - 1.83).

A lower mental wellbeing score, as assessed by the WEMWBS measure, was associated with lower levels of participation in sport and exercise (OR 0.73, CI 0.55 - 0.96). This was the case for both men and women. In a similar vein, those with a lower life satisfaction score were less likely to report participation in sport and exercise (OR 0.82, CI 0.69 - 0.98). When analyses were conducted separately for men and women, life satisfaction remained a significant factor only for men.

For women only, cigarette smoking had a relationship with sport and exercise participation. Those who were current smokers were less likely to report participation in sport and exercise compared to those who had never smoked (OR 0.52, CI 0.39 - 0.70). Level of area deprivation was a significant factor but for men only, however it is not possible to state the direction of association.

Economic activity status had no significant relationship with participation in sport and exercise. This contrasts with the findings regarding the range of factors associated with meeting physical activity guidelines, where economic activity status was the only socioeconomic factor to remain significant in the final model.

Disability/illness and marital status were also found not to be associated with sport and exercise participation.

Walking (at least 10 minutes)

There were some notable differences between the factors associated with walking and those for meeting physical activity guidelines and participating in sport and exercise, see Table 5. In contrast to most outcomes for physical activity, women were found to be significantly more likely to participate in walking than men. Age has been consistently associated with levels of physical activity, yet it was not an influential factor for participation in walking, except in women. Once other factors were controlled for in the models, the likelihood of participating in walking tended to increase with age but was only statistically significant for the three age groups above age 45 years, see Table 22 in Appendix B.

Table 5: Significant factors associated with walking participation

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	ns	ns	√
Marital Status	√	√	√
Socioeconomic
Equivalised Income quintiles	ns	ns	ns
Deprivation - SIMD Quintiles	ns	ns	ns
Economic Activity Status (EAS)	√	√	ns
Highest Education Qualification	√	√	ns
Health and Lifestyle
Self-Assessed health	√	ns	√
Longstanding Illness	√	ns	√
Cigarette smoking	√	√	ns
Mental wellbeing	ns	ns	ns
life satisfaction	ns	ns	√
BMI	ns	ns	ns

Socioeconomic factors were not relevant for women but economic activity status and highest educational qualification did demonstrate an association for men. Health factors were more relevant for women than for men.

Men - economic activity status, educational qualification, cigarette smoking and marital status were all related to participation in walking. Men who reported being unable to work were less likely to report participation in walking (OR 0.22, CI 0.09 - 0.53).

Women - self-assessed health, age, marital status, illness and life satisfaction were all found to be related to participation in walking in women. Those aged between 45 to 74 were all far more likely to report participation in walking compared to those aged 75 or more. Women with a longstanding illness were less likely to participate in walking than those without (OR 0.44, CI 0.24 - 0.83). Those women with a lower level of life satisfaction were less likely to participate in walking than those with a life satisfaction score of 8 or more (OR 0.58, CI 0.36 - 0.96).

Housework

Sex, age, income, health, smoking and BMI were all associated with participation in heavy housework, see Table 6. Overall, age, sex and health were the key factors here with little influence exerted by socioeconomic factors. Men were less likely than women to participate in housework (OR 0.14, CI 0.10 - 0.21). There was a general trend for participation in housework to decrease with age with the exception of the youngest category (16-24) who were not significantly different to those aged 75 or older. Those with better health were more likely to participate in housework than those rating their health as poor or very poor.

Table 6: Significant factors associated with heavy housework participation

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	√	√	√
Marital Status	ns	ns	√
Socioeconomic
Equivalised Income quintiles	√	ns	ns
Deprivation - SIMD Quintiles	ns	ns	√
Economic Activity Status (EAS)	ns	√	ns
Highest Education Qualification	ns	ns	√
Health and Lifestyle
Self-Assessed health	√	√	√
Longstanding Illness	ns	ns	ns
Cigarette smoking	√	ns	√
Mental wellbeing	ns	ns	ns
life satisfaction	ns	ns	ns
BMI	√	√	ns

Age and self-assessed health remained significant when analyses were conducted separately for both men and women. Mental wellbeing, long standing illness and life satisfaction were not associated in any of the housework related analyses. Other associated factors were quite different between the sexes.

Men - In addition to age and self-assessed health, BMI and economic activity status were associated with whether men participated in heavy housework. Those with a BMI less than 30 were more likely to participate in heavy housework than those who were classified as obese. Those who were retired were more likely to participate in heavy housework than those in paid work.

Women - In addition to age and self-assessed health, cigarette smoking, deprivation, educational attainment and marital status were associated with whether women participated in heavy housework, however, considerable variation within categories makes interpretation of the impact of the additional factors here difficult.

Manual work/gardening/DIY

Overall, demographic and socioeconomic factors were more influential on this domain with little influence from health and lifestyle factors, except self-assessed health, see Table 7.

Table 7: Significant factors associated with manual work/gardening/DIY participation

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	√	N/A	N/A
Age	√	√	√
Marital Status	√	√	ns
Socioeconomic
Equivalised Income quintiles	√	√	ns
Deprivation - SIMD Quintiles	√	√	√
Economic Activity Status (EAS)	√	√	ns
Highest Education Qualification	√	√	√
Health and Lifestyle
Self-Assessed health	√	√	√
Longstanding Illness	ns	√	ns
Cigarette smoking	ns	ns	ns
Mental wellbeing	ns	ns	ns
life satisfaction	ns	ns	ns
BMI	ns	ns	ns

Sex, age, income, marital status, deprivation, economic activity status, education and self-assessed health were all associated with participation in manual work/gardening/DIY. Age, self-assessed health, deprivation and educational attainment were also all associated with participation in manual work/gardening/DIY when analyses were conducted separately for men and women.

Men were significantly more likely to participate in manual work/gardening/DIY than women (OR 2.30, CI 1.96 - 2.69). Those aged 35 to 74 were all more likely to participate in this domain compared to those aged 75 or older and those in the youngest age category were less likely to, although this disguises some differences by age between men and women. Overall, the higher the socioeconomic indicator the higher the likelihood of participating in manual work/gardening/DIY. Decreasing level of deprivation was generally associated with increased participation, more so for the 3^rd and 4^th quintile with a slight decrease in participation in quintile 5. In a similar fashion, all four higher equivalised income quintiles were associated with greater participation but there was a decreasing trend from income quintile 4 to the highest earners.

Smoking, mental wellbeing, life satisfaction and BMI were not associated factors in any of the analyses for manual work/gardening/DIY. Several differences were found to exist by gender.

Men - For men, income, marital status, economic activity status (EAS) and disability status were associated with participation in manual work/gardening/DIY in addition to age, self-assessed health, deprivation and educational attainment.

Women - For women, there were no further significant factors in addition to age, self-assessed health, deprivation and educational attainment. With respect to age, those aged 16-34 were less likely to participate in manual work/gardening/DIY than those aged 75 or older and those aged 35 to 74 were no different to the oldest age group.

Occupational physical activity

Age and socioeconomic factors income, deprivation and highest qualification level were all strongly associated with who was physically active at work. Health and lifestyle factors had very little influence, see Table 8.

Table 8: Significant factors associated with occupational physical activity participation (respondents in paid work only)

FACTOR	ALL	MEN	WOMEN
Demographic
Sex	ns	ns	ns
Age	√	√	√
Marital Status	ns	ns	ns
Socioeconomic
Equivalised Income quintiles	√	√	√
Deprivation - SIMD Quintiles	√	ns	√
Economic Activity Status (EAS)	n/a	n/a	n/a
Highest Education Qualification	√	√	√
Health and Lifestyle
Self-Assessed health	ns	ns	ns
Longstanding Illness	ns	ns	ns
Cigarette smoking	ns	ns	√
Mental wellbeing	ns	ns	ns
life satisfaction	ns	ns	ns
BMI	ns	√	ns

It is notable that this is the only physical activity outcome in this study where sex was found to be non-significant.

The results indicate that those in the youngest age category are more likely to be active in their paid job than those age 75+. In general, participation in occupational physical activity increased with decreasing indicators of socioeconomic status. Those in the highest equivalised income quintile (£39520+) were less likely to be active at work than those earning less than £10672 (OR 0.53 CI 0.32 - 0.87). Adults in the most deprived quintile were more likely to be active in their paid job than adults in the other deprivation quintile groups. The higher the qualification, the less likely the respondent was of being active at work. This contrasts with the finding for participation in sport and exercise and manual work/gardening/DIY where participation tended to increase with increase in the indicators of socioeconomic status.

When analyses were conducted separately for men and women, all but deprivation remained significantly associated in both.

Men - Age, income and highest qualification level remained strongly associated and in addition a relationship was found to BMI. Although ORs were non-significant, the suggestion was that those with a BMI less than 30 were more likely to participate in occupational physical activity.

Women - Age, income, deprivation and highest qualification level were all strongly associated with who was physically active at work. In addition, an association was found between smoking behaviour and occupational physical activity. Current smokers were more likely to be active in their paid work than those who have never smoked.