Scottish Health Survey - topic report: The Glasgow Effect

4. Adverse Health Behaviours

SUMMARY

Overweight

• Women in Greater Glasgow and Clyde were no more likely to be overweight (including obese) than women in the rest of Scotland, before or after adjusting for age.
• After adjusting for age, SIMD, socio-economic and behavioural variables, men in Greater Glasgow and Clyde were significantly less likely to be overweight (including obese).
• However additional adjustment for biological variables removed the effect, specifically when the model adjusted for equivalised income, marital status, smoking status, cholesterol and C-reactive protein.

Obese

• Men in Greater Glasgow and Clyde were no more likely to be obese than men in the rest of Scotland, before or after adjusting for age.
• However after adjusting for age and SIMD, women in Greater Glasgow and Clyde were less likely to be obese than those in the rest of Scotland, although this difference was explained by further adjusting for equivalised income, a socio-economic variable.

Binge drinking

• Residents in Greater Glasgow and Clyde had a significantly higher prevalence of binge drinking than the rest of Scotland, but this was fully explained by adjusting for age and sex.

Drinking over the recommended weekly alcohol limit

• There was no significant difference in odds of drinking over the recommended weekly alcohol limit between Greater Glasgow and Clyde and the rest of Scotland, either before or after adjusting for age and sex.

Potential problem drinking

• Residents of Greater Glasgow and Clyde had significantly higher odds of potential problem drinking than the rest of Scotland, but this was fully explained by adjusting for age and sex.

Current smoking

• Residents of Greater Glasgow and Clyde had significantly higher odds of being a current smoker than the rest of Scotland, but this was fully explained by adjusting for age and sex.

Heavy smoking

• Residence in Greater Glasgow and Clyde was associated with higher odds of heavy smoking after adjusting for age and sex; however further adjustment for SIMD fully explained this effect.

Fruit and vegetable consumption

• Residents of Greater Glasgow and Clyde had significantly higher odds of consuming less than 2 portions of fruit and vegetables per day than the rest of Scotland, but this was fully explained by adjusting for age and sex.

4.1 Introduction

This section describes the outcome variables which were used to investigate the existence of a 'Glasgow Effect' for adverse health behaviours.

Overweight/obesity

Body Mass Index ( BMI) was used to classify individuals as overweight and obese. BMI is defined as weight (kg) divided by height squared (m ²), and therefore uses the participants' heights and weights which were measured by the interviewer as part of the main interview.

Participants were classified as overweight if their BMI was greater than or equal to 25kg/m ², and obese if their BMI was greater than or equal to 30kg/m ².

Alcohol

In the Scottish Health Survey data on alcohol consumption was self-reported, and self-reported data often produces lower estimates of alcohol consumption than alcohol sales data would suggest. However these data can still be used to compare relative values between groups.

Weekly alcohol consumption was estimated by first asking participants aged 16 and over how often during the last 12 months they had consumed the following six types of drinks:

• normal beer, lager, cider and shandy
• strong beer, lager and cider
• sherry and martini
• spirits and liqueurs
• wine
• alcoholic soft drinks ("alcopops").

The average number of times per week each type of drink had been drunk was estimated from this question. They were then asked how much of each drink they had usually consumed on each occasion. This data was then converted into units; for more details see the Scottish Health Survey 2009 main report ¹².

Women are advised not to consume more than 14 units per week, with 21 units the recommended limit for men: participants who drink above these levels, therefore, are considered to drink more than the recommended weekly alcohol limit.

Daily consumption was measured by asking participants aged 16 and over about their alcohol consumption on their heaviest drinking day from the week preceding the interview. They were asked how much they had consumed on that day of the six types of drinks mentioned above, and from this an estimate of units was calculated. Binge drinking has been defined as the consumption of more than 6 units on one occasion for women, and more than 8 units for men; participants who drank above these cut-offs are considered to be binge drinkers.

Participants self-completed the CAGE questionnaire which highlights indicators of potential problem drinking. A positive answer to 2 or more of the questions was taken as an indicator of potential problem drinking.

Smoking

Smoking status was self-reported. Information about smoking status in adults aged 16 and 17 was collected via a self-completion questionnaire, whereas for adults aged 20 and over this information was collected as part of the main interview. For adults aged 18 and 19 the data were collected either by means of the self-completion questionnaire, or at the main interview (this was at the interviewer's discretion).

Participants were defined as current smokers if they reported being a current cigarette smoker, and participants were classified as heavy smokers if they smoked 20 or more cigarettes a day.

Fruit and vegetable consumption.

Participants were asked about the portions of fruit and vegetables they had consumed in the 24 hours preceding the interview. Portion sizes are defined in detail in the Scottish Health Survey 2009 main report. As fruit and vegetable consumption is being used here as a measure of poor diet, two cut-off points of no portions and less than two portions of fruit and vegetables were used.

4.2 Weight

The analyses for weight were carried out separately for each sex, due to the different patterns often observed in men and women.

4.2.1 Overweight

61% of adult women were overweight (including obese). The prevalence of overweight increased with age until age 55-64, then decreased. The lowest prevalence was among 16-24 year olds (39%), with the highest prevalence among 55-64 year olds (75%). There was also significant variation between SIMD quintiles although there wasn't a clear linear pattern, with the highest prevalence found in the middle quintile (66%) and the lowest prevalence in the least deprived quintile (54%). There was no significant difference in overweight prevalence between women in Greater Glasgow and Clyde and the rest of Scotland, nor a difference in odds of overweight when adjusting for age.

68% of adult men were overweight (including obese). The prevalence of overweight increased with age from 35% of 16 - 24 year olds to 83% of 55 - 64 year olds, then decreased. There was also significant variation between SIMD quintiles, with the highest prevalence found in the middle quintile (72%) and the lowest prevalence in the most deprived quintile (65%). Male residents of Greater Glasgow and Clyde had a significantly lower prevalence of overweight than the rest of Scotland (63% vs. 70%).

The model development process for men can be found in Appendix 5, along with McFadden's pseudo R ²s for the different models. In view of these, the best fitting model was chosen, and the results for that model using the full data available are described here.

The model which best fit the data for overweight men started with all the socio-economic, behavioural and biological variables, except HDL cholesterol as only 34 men had low HDL. After backward selection the model contained: equivalised income, marital status, smoking status, cholesterol level and C-reactive protein. As blood analytes are included in this model, the sample size was 654.

Men in the lowest equivalised income quintile had less than a third of the odds of being overweight than men in the top quintile (odds ratio of 0.32). Single men, men who were married/in a civil partnership and separated and men who were widowed were significantly less likely to be overweight compared to men who were married/in a civil partnership and living together (respective odds ratios of 0.37, 0.23 and 0.27). Moderate smokers had significantly lower odds of being overweight than men who had never smoked or who were ex-occasional smokers (odds ratio of 0.40). Men with high cholesterol (=5mmol/l) were two-thirds more likely to be overweight (odds ratio of 1.68), and the odds of overweight increased with increasing C-reactive protein quintile up to the fourth quintile (odds ratio of 6.46), then decreased slightly (odds ratio of 3.88).

4.2.2 Obesity

26% of adult men were obese. The prevalence of obesity increased with age until age 55-64, then decreased. The lowest prevalence was among 16-24 year olds (9%), and the highest prevalence among 55-64 year olds (38%). Although the prevalence increased with increasing deprivation from 24% in the least deprived quintile to 29% in the most deprived quintile, this difference was not significant. There was no difference in obesity prevalence between men in Greater Glasgow and Clyde and the rest of Scotland, nor was there a difference in odds of obesity when adjusting for age.

28% of adult women were obese. The prevalence of obesity increased with age from 17% of 16 - 24 year olds to 35% of 65 - 74 year olds. There was also significant variation between SIMD quintiles, with prevalence increasing with increasing deprivation from 20% in the least deprived quintile to 32% in the most deprived quintile. Female residents of Greater Glasgow and Clyde had a significantly lower prevalence of obesity than the rest of Scotland (25% vs. 28%).

In the initial logistic regression model for women containing age and residence, residents in Greater Glasgow and Clyde did not have a significantly different odds of obesity from the rest of Scotland, however residence became significant with further adjustment for SIMD (odds ratio of 0.83, meaning female residents of Greater Glasgow and Clyde had 17% lower odds of obesity). SIMD was also significant, with increasing odds of obesity for increasing deprivation. Further adjustment for socio-economic variables removed this effect, with the final model containing age, SIMD and equivalised income. This shows that any difference in the level of obesity between women in Greater Glasgow and Clyde and the rest of Scotland were explained by socio-economic effects.

4.3 Alcohol Consumption

4.3.1 Binge drinking

22% of Scottish adults binge drink, with a significantly higher prevalence in men than women (26% vs. 17%). There was a significant difference by age group, with the prevalence decreasing with increasing age, from 35% of 16-24 year olds to 1% of those aged 75 and over. There was no significant difference in prevalence between the SIMD quintiles, yet the prevalence in Greater Glasgow and Clyde was significantly higher than the rest of Scotland (23% vs. 21%).

In the initial logistic regression model containing only age, sex and residence, residence in Greater Glasgow and Clyde had an odds ratio of 1.07, which was not significant. This shows that the difference in prevalence of binge drinking between Greater Glasgow and Clyde and the rest of Scotland has been explained by the differing age and sex distributions.

4.3.2 Drinking over the recommended weekly alcohol limit

24% of adults consumed more than the recommended weekly alcohol limit, with a significantly higher prevalence for men than women (29% vs. 19%). There was a significant difference by age group, with the prevalence generally decreasing with increasing age, from 34% of 16-24 year olds to 7% of those age 75+, but with a prevalence of 27% for 45-54 year olds, which was the second highest prevalence. There was a significant difference in prevalence by SIMD quintile, with the highest prevalence for those in the least deprived quintile (27%). The level of drinking over the recommended weekly alcohol limit was very similar in the three most deprived quintiles, with a range in prevalence of 21% - 23%. There was no significant difference in prevalence between Greater Glasgow and Clyde and the rest of Scotland, nor any difference in odds of drinking over the weekly alcohol limit after controlling for age and sex.

4.3.3 Potential problem drinking

10% of adults had a score of 2 or more on the CAGE questionnaire, indicating potential problem drinking, with a significantly higher prevalence among men than women (13% vs. 8%). There was also a significant difference by age, with prevalence decreasing as age increases, from 18% of 16-24 year olds to 2% of those aged 75 and over. The prevalence was highest in the most deprived SIMD quintile (13%), and decreased with decreasing deprivation to 8% of the least deprived SIMD quintile. The prevalence in Greater Glasgow and Clyde was significantly higher than the rest of Scotland (12% vs. 10%).

In the initial logistic regression model containing only age, sex and residence, residence in Greater Glasgow and Clyde had an odds ratio of 1.16, which was not significant. This shows that the difference in prevalence of potential problem drinking between Greater Glasgow and Clyde and the rest of Scotland was explained by the differing age and sex distributions.

4.4 Smoking

4.4.1 Current smoking status

25% of adults were current cigarette smokers, with no significant difference between men and women. There was a significant difference by age, with similar rates for those aged 16 to 54 (range: 28% - 31%), and decreasing rates with increasing age from age 55. Current smoking rates were almost three times higher in the most deprived SIMD quintile compared to the least deprived SIMD quintile (39% vs. 14%). Residents of Greater Glasgow and Clyde had a significantly higher current smoking prevalence than the rest of Scotland (27% vs. 25%).

In the initial logistic regression model containing only age, sex and residence, residence in Greater Glasgow and Clyde was not significant. Age was the only variable which was a significant predictor of smoking status in the model, implying the difference in prevalence of current cigarette smokers between Greater Glasgow and Clyde and the rest of Scotland was explained by the differing age distributions.

4.4.2 Heavy smokers

8% of adults were heavy smokers, with a significantly higher prevalence among men than women (9% vs. 7%). The relationship of heavy smoking with age was shaped like a negative quadratic, with the prevalence increasing from 16 - 24 (prevalence 3%) until age 45 - 54 (12%), then decreasing again, with a prevalence of 3% for those aged 75 and over. There was also a strong relationship with SIMD, with prevalence in the most deprived quintile more than 4 times higher than the least deprived quintile (14% vs. 3%). The prevalence was significantly higher in Greater Glasgow and Clyde than in the rest of Scotland (9% vs. 8%).

In the initial logistic regression model containing only age, sex and residence, residence in Greater Glasgow and Clyde had an odds ratio of 1.21, indicating residents of Greater Glasgow and Clyde had a 21% increased risk of being a heavy smoker compared to the rest of Scotland. However once the model adjusted for SIMD, the effect of residence was fully explained. SIMD was highly significant in the model, with increasing odds of being a heavy smoker associated with increasing levels of deprivation; the odds of being a heavy smoker for someone in the most deprived quintile were more than five times the odds for someone in the least deprived quintile, after adjusting for age and sex.

4.5 Fruit and vegetable consumption

31% of adults consumed less than 2 portions of fruit and vegetables per day, with a significantly higher rate for men than women (34% vs. 28%). There was also a significant difference by age, with decreasing rates corresponding to increasing age, from 42% of 16-24 year olds to 25% - 27% of those aged 55 and over. A significant difference was also found by SIMD quintile, with those living in the most deprived quintile twice as likely to consume less than 2 portions of fruit and vegetables per day than those in the least deprived quintile (42% vs. 22%). There was a small but significant difference for residents of Greater Glasgow and Clyde (33%) compared to the rest of Scotland (30%).

9% of adults consumed no portions of fruit and vegetables per day, with similar patterns found for the relationship between consuming no portions of fruit and vegetables per day and age, sex and SIMD quintile as were found for consuming less than two portions of fruit and vegetables per day; however there was not a significant difference in the proportion who consumed no portions of fruit and vegetables per day in Greater Glasgow and Clyde and the rest of Scotland. When controlling for age and sex there was still no significant difference between residents of Greater Glasgow and Clyde and the rest of Scotland.

When considering the outcome 'consuming less than two portions of fruit and vegetables per day', the initial logistic regression model containing only age, sex and residence found that residence in Greater Glasgow and Clyde had an odds ratio of 1.13, of borderline significance (p=0.058). However once the model also adjusted for SIMD quintiles the odds ratio dropped to 1.01, completely removing the relationship. SIMD was highly significant, with increasing odds for increasing deprivation.

4.6 Assessing the impact of the socio-economic variables individually

For the adverse health behaviours there were a range of socio-economic variables which resulted in the model which provided the best fit to the data (as they had the highest McFadden's pseudo R ²) when only one socio-economic variable was included. For all the alcohol-related outcomes, the socio-economic variable which produced the model which best fit the data was equivalised income, but different variables provided the best fitting model for the other adverse health behaviours.

4.7 Conclusions and Discussion

The overall result of this chapter is that after adjusting for age and sex, almost all of the adverse health behaviours were found not to be significantly different in Greater Glasgow and Clyde from the rest of the country. The exceptions to this were heavy smoking, where the effect was explained by SIMD, obesity for women, where the effect was explained by socio-economic variables, and overweight for men, where the effect was explained by additional adjustment for socio-economic, behavioural and biological variables.

The analyses in this report found no significant difference in current smoking levels between Greater Glasgow and Clyde and the rest of Scotland after adjusting for age and sex. This is different to the results found in a study which investigated smoking status with respect to the Glasgow Effect ²⁴ using data from the 1995, 1998 and 2003 Scottish Health Surveys. Using a multilevel analysis they found that before adjustment residents of Greater Glasgow were more likely to smoke; however after adjusting for socio-economic variables this association was no longer significant. The same data were used in a study by Gray ¹⁰ which compared health related behaviours and health measures between Glasgow and the rest of Scotland, and found that adjusting for age and year of survey did not remove the effect, but adjusting for socio-economic variables did, without using a multilevel approach. There are various possible explanations for the difference between the results in this report and those using the earlier Scottish Health Surveys. One is that the smoking levels have changed differently in Greater Glasgow and Clyde compared to the rest of Scotland over time; another allows for the different division of Scotland into health boards between the earlier and later Scottish Health Surveys, as the earlier study only uses Greater Glasgow, whereas this study investigates the health board of Greater Glasgow and Clyde.

This report found that there was a significantly higher level of binge drinking in Greater Glasgow and Clyde, but that adjusting for age and sex removed this effect, without the need to adjust for any socio-economic variables; however Gray ¹⁰ found that residence in Greater Glasgow was associated with increased levels of binge drinking for men, which was not fully attenuated by adjusting for socio-economic variables. This report found no difference in prevalence in drinking over the weekly alcohol limit between Greater Glasgow and Clyde and the rest of Scotland. This differs to results reported by Gray, where residence in Greater Glasgow was associated with exceeding the recommended weekly alcohol limit for men, with the effect not attenuated at all by adjusting for socio-economic factors.

The prevalence of both binge drinking and drinking over the recommended weekly alcohol limit has decreased slowly since 2003 for men and women ¹². Gray reported a prevalence of 35% for binge drinking among men in Greater Glasgow, the highest of all regions, whereas in Greater Glasgow and Clyde this value was 26% for 2008 and 2009 combined. From these analyses it is not possible to know whether this is due to the use of the larger health board area in the analysis, which would lower the binge drinking levels, or whether levels of binge drinking have declined faster in Greater Glasgow than in the rest of Scotland. It is also important to note that unit conversion factors for alcohol were revised in 2008, which may have contributed to the difference in results.

The analyses in this report using the 2008 and 2009 Scottish Health Surveys found no difference in prevalence of consuming no portions of fruit and vegetables per day between Greater Glasgow and Clyde and the rest of Scotland, but found that a difference did exist in prevalence for consuming less than two portions of fruit and vegetables per day. However, this difference was of borderline significance when adjusting for age and sex, and was removed completely by further adjusting for SIMD. One study by Gray and Leyland has previously investigated fruit and vegetable consumption with respect to the Glasgow Effect ²⁵ using data from the 1995 and 1998 Scottish Health Surveys, but looked at consumption of 5 or more portions of fruit and vegetables per day, rather than using fruit and vegetable consumption as an indicator of poor diet as was done in this report. Gray and Leyland found there was no difference in the proportion consuming 5 or more portions of fruit or vegetables per day for either men or women between residents of the Greater Glasgow area and the rest of Scotland. In a separate report Gray ¹⁰ used the 2003 Scottish Health Survey data and found no difference in consumption of five or more portions of fruit or vegetable per day for women when comparing Greater Glasgow with the rest of Scotland. For men there was no significant difference when adjusting for age, but when socio-economic factors were also adjusted for, residence in Greater Glasgow became significant, with men in Greater Glasgow more likely to eat at least 5 portions of fruit and vegetables per day.

An interesting relationship exists for obesity among women in Greater Glasgow and Clyde; there was a significantly lower prevalence of obesity in women in Greater Glasgow and Clyde, which was removed by adjusting for age and sex, but then became significant again when adjusting for SIMD, with an odds ratio of 0.83, showing less risk of obesity than women the rest of Scotland. The same results were found by Gray using the 1995, 1998 and 2003 Scottish Health Survey data ¹⁰; for women there was no significant difference before adjusting for socio-economic factors, but after adjustment residence in Greater Glasgow was associated with lower likelihoods of being obese compared to the rest of Scotland. No significant difference was found between obesity in men in Greater Glasgow and the rest of Scotland for men, either before or after adjusting for socio-economic factors.