Abstract

Background

The extent to which the poor health profile of Glasgow, the city with the highest mortality rates in the UK, can be explained solely by socio-economic factors is unclear. This paper additionally considers behavioural and biological factors as explanations of excess risk.

Methods

Scottish Health Survey data for 2008–09 were analysed using logistic regression models to compare the odds of physical and mental health outcomes, as well as adverse health behaviours, for residents of the Greater Glasgow and Clyde (GGC) conurbation compared with the rest of Scotland.

Results

After adjustment for age and sex, significant differences were observed among Glasgow residents for most mental and physical health outcomes, but not for most adverse health behaviours. Adjustment for area and individual-level socio-economic characteristics explained the differences for all outcomes except anxiety, psychological ill-health, heart attack and men being overweight. After additional adjustment for behavioural and biological characteristics, significantly higher odds of anxiety and heart attack remained for residents of the Glasgow area.

Conclusions

Adjusting for area- and individual-level socio-economic conditions explained the excess risk associated with residents of GGC for most (16 out of 18) outcomes; however, significant excess risks for two outcomes remained: anxiety and heart attack. Additional explanations are, therefore, required.

Introduction

The poor health profile of Scotland relative to other comparatively wealthy nations is well known: the country has the highest mortality rates (particularly among adults of working age) and lowest overall life expectancy in western Europe.1,2 This is driven to a large degree by particularly high mortality in Glasgow and the West Central Scotland conurbation.3,4 Traditional explanations for this phenomenon have focused on material deprivation and related socio-economic conditions, principally in comparison to England and Wales.5,6 In recent years, however, an increasing amount of evidence has suggested that factors over and above socio-economic deprivation may be contributing to this situation.7–11 For example, in 2001 all-cause mortality in Scotland was shown to be 8% higher than that in England and Wales even once all differences in area-based deprivation had been accounted for,8 while premature mortality in Glasgow has been shown to be 30% higher than those in the identically deprived cities of Liverpool and Manchester.12 These unknown drivers of poor health have speculatively been referred to as the ‘Scottish Effect’ and, in the case of the country's largest city, the ‘Glasgow Effect’. Importantly, however, the essential question of what explains poor health over and above the effects of socio-economic deprivation is also relevant to, and has been shown to be relevant to, other parts of the UK.13,14

Such an ‘effect’ has also been shown for Glasgow relative to the rest of Scotland for both morbidity and mortality.10,15 The former analyses, however, were limited by the area-based deprivation measures that were available at the time of the study and by the range of risk factor variables that were included in the analyses. In this study, therefore, we seek to overcome these weaknesses by employing more sophisticated, spatially sensitive, measures of deprivation, as well as a much broader range of socio-economic, behavioural and biological variables to establish whether differences in health between Glasgow and Scotland can be fully explained in these terms, or whether there is evidence of any such unexplained ‘excess’ poor health.

Methods

Data

Data are from the 2008 and 2009 Scottish Health Surveys,16,17 cross-sectional studies of Scotland's population, commissioned by the Scottish Government Health Directorates. Based on face-to-face interviews, the surveys collect a wide range of demographic, economic, behavioural and biological variables. Sub-samples from both the 2008 and 2009 surveys were selected for a nurse visit to collect biological measurements [e.g. blood pressure; lung function; urine sample-based measures of sodium, potassium and creatinine; cotinine (from a saliva sample); waist and hip measurement and demispan], and some of these participants agreed to provide a blood sample (from which measures such as high-density lipoprotein (HDL) and total cholesterol, fibrinogen, glycated haemoglobin and C-reactive protein could be analysed). Between the two data sets, data were available for 13 996 adults (aged 16+), of whom 3242 were adults resident in the Greater Glasgow and Clyde (GGC) area. [Note: this relates to the National Health Service (NHS) Board area of Greater Glasgow and Clyde.) The nurse and blood sub-samples were available for 414 and 322 respondents in GGC, respectively; the equivalent figures for the rest of Scotland were 1823 and 1466. More information on the sample design and data collection is available from the 2008 and 2009 Scottish Health Survey reports.16,17

Three groups of outcomes were considered: general and mental health (e.g. poor self-assessed health, anxiety and depression); physical health (e.g. history of heart attack or stroke) and adverse health behaviours (e.g. smoking, alcohol and diet). The full list of variables with definitions is presented in Table 1. Data on anxiety and depression were only collected in the nurse sub-sample. Analyses of the outcomes ‘overweight’ and ‘obese’ were carried out separately for men and women (this was done due to the different relationships between weight and SEP for men and women). Exposure data are listed in Table 2.

Table 1

Outcome variables

Outcome variables How defined 
General and mental health 
 Poor self-assessed health Self-reported ‘bad’ or ‘very bad’ 
 Anxiety ≥2 on the anxiety scale of the revised clinical interview schedule 
 Depression ≥2 on the depression scale of the revised clinical interview schedule 
 Psychological ill-health ≥4 on GHQ-12 
 Low mental well-being <1 SD below the mean WEMWBSa score 
Physical health 
 Heart attack Self-reported doctor diagnosed 
 CVD Self-reported doctor diagnosed 
 COPD Self-reported doctor diagnosed 
 Stroke Self-reported doctor diagnosed 
 Limiting long-standing illness Self-reported 
Adverse health behaviours 
 Overweight BMI ≥25 kg/m2 
 Obese BMI ≥30 kg/m2 
 Binge drinking Men: >8 units in a day, women: >6 units in a day 
 Drinking over the recommended  weekly limit Men: >21 units in a week, women: >14 units in a week 
 Potential problem drinking ≥2 on the CAGE questionnaire 
 Current cigarette smoking Self-reported 
 Heavy cigarette smoking ≥ 20 cigarettes a day 
 Poor fruit and vegetable consumption <2 portions per day 
Outcome variables How defined 
General and mental health 
 Poor self-assessed health Self-reported ‘bad’ or ‘very bad’ 
 Anxiety ≥2 on the anxiety scale of the revised clinical interview schedule 
 Depression ≥2 on the depression scale of the revised clinical interview schedule 
 Psychological ill-health ≥4 on GHQ-12 
 Low mental well-being <1 SD below the mean WEMWBSa score 
Physical health 
 Heart attack Self-reported doctor diagnosed 
 CVD Self-reported doctor diagnosed 
 COPD Self-reported doctor diagnosed 
 Stroke Self-reported doctor diagnosed 
 Limiting long-standing illness Self-reported 
Adverse health behaviours 
 Overweight BMI ≥25 kg/m2 
 Obese BMI ≥30 kg/m2 
 Binge drinking Men: >8 units in a day, women: >6 units in a day 
 Drinking over the recommended  weekly limit Men: >21 units in a week, women: >14 units in a week 
 Potential problem drinking ≥2 on the CAGE questionnaire 
 Current cigarette smoking Self-reported 
 Heavy cigarette smoking ≥ 20 cigarettes a day 
 Poor fruit and vegetable consumption <2 portions per day 

aThe Warwick-Edinburgh Mental Well-being Scale (WEMWBS), a questionnaire scale used for scale for assessing positive mental health.

Table 2

Exposure variables

Exposure variables 
Socioeconomic 
 SIMD 
 Income-related benefits 
 NS-SECa 
 Economic activity 
 Educational qualifications 
 Housing tenure 
 Marital status 
Behavioural 
 Smoking status 
 Binge drinking 
 Drinking over weekly limit 
 Abstaining from alcohol 
 Potential problem drinking 
 Physical activity 
 Fruit and vegetable consumption 
Biological 
 BMI 
 Waist–hip ratio 
 Blood pressure 
 Lung function 
 Total cholesterol 
 HDL cholesterol 
 Fibrinogen 
 C-reactive protein 
Exposure variables 
Socioeconomic 
 SIMD 
 Income-related benefits 
 NS-SECa 
 Economic activity 
 Educational qualifications 
 Housing tenure 
 Marital status 
Behavioural 
 Smoking status 
 Binge drinking 
 Drinking over weekly limit 
 Abstaining from alcohol 
 Potential problem drinking 
 Physical activity 
 Fruit and vegetable consumption 
Biological 
 BMI 
 Waist–hip ratio 
 Blood pressure 
 Lung function 
 Total cholesterol 
 HDL cholesterol 
 Fibrinogen 
 C-reactive protein 

aNational Statistics Socio-economic Classification (NS-SEC categorized as managerial and professional occupations, intermediate occupations, small employers and own account workers, lower supervisory and technical occupations and semi-routine occupations, as well as a category for people for whom the NS-SEC is not applicable, such as full-time students).

Note that exposure data on relationships and social mobility were included in some of the models. The rationale for their inclusion, and the results, are described elsewhere,18 and are not considered further in this paper.

Analyses

First, the prevalence of each outcome in GGC was compared with the prevalence in the rest of Scotland, with chi-square tests for significance carried out. A series of logistic regression models were then run, initially adjusting for age, sex and residence in GGC, then additionally adjusting for area-level deprivation, individual-level socio-economic factors, behavioural factors and biological factors (Table 2). Backward selection was carried out after each group of variables had been added, until all the variables in the model were significant at the 5% level. As the sample size was greatly reduced for the nurse variables, and reduced again for the blood variables, if all the blood variables were dropped from the model then the model was re-run excluding the blood variables, and similarly with the nurse variables, thereby enabling a larger sample to be used. McFadden's pseudo R2 values were compared for each model, and the model with the highest McFadden's pseudo R2 provided the best fit to the data.

Table 3

Prevalence of health conditions and health behaviours in GGC and the rest of Scotland

  Prevalence (%)
 
Significant difference 
GGC Rest of Scotland 
General and mental health    
 Self-assessed health 
 Anxiety 14 
 Depression 13 
 Psychological ill-health 18 14 
 Mental well-being 17 14 
Physical health 
 Heart attack 
 CVD 16 15 
 COPD 
 Stroke 
 Limiting long-standing illness 28 25 
Adverse health behaviours 
 Overweight men 63 70 
 Overweight women 60 62 
 Obese men 25 27 
 Obese women 25 28 
 Binge drinking 23 21 
 Drinking over recommended  weekly limit 23 24 
 Drinking over recommended  daily limit 12 10 
 Current cigarette smoking 27 25 
 Heavy cigarette smoking 
 <2 portions fruit/veg per day 33 30 
  Prevalence (%)
 
Significant difference 
GGC Rest of Scotland 
General and mental health    
 Self-assessed health 
 Anxiety 14 
 Depression 13 
 Psychological ill-health 18 14 
 Mental well-being 17 14 
Physical health 
 Heart attack 
 CVD 16 15 
 COPD 
 Stroke 
 Limiting long-standing illness 28 25 
Adverse health behaviours 
 Overweight men 63 70 
 Overweight women 60 62 
 Obese men 25 27 
 Obese women 25 28 
 Binge drinking 23 21 
 Drinking over recommended  weekly limit 23 24 
 Drinking over recommended  daily limit 12 10 
 Current cigarette smoking 27 25 
 Heavy cigarette smoking 
 <2 portions fruit/veg per day 33 30 

*denotes significance (p < 0.05) from chi-square tests.

Analyses were carried out using SPSS 16.0.

Further details of all methodological issues are reported elsewhere.18

Results

Prevalence overview

Table 3 presents the prevalence of each outcome in GGC and the rest of Scotland. There are significantly higher levels of each general and mental health outcome in GGC, with the prevalence close to double those in the rest of Scotland for levels of moderate-to-severe anxiety and depression, and over twice for poor self-assessed health. For physical health, although the prevalence is higher in GGC for all the outcomes, it is only significantly higher for heart attack, limiting long-standing illness and chronic obstructive pulmonary disease (COPD). For adverse health behaviours, GGC had lower levels of overweight and obesity for both men and women, significantly so for overweight men and obese women. The prevalence was also lower for drinking over the recommended weekly alcohol limit, although this difference was not significant. However, residents of GGC had significantly higher levels of binge drinking, current smoking, heavy smoking and consuming less than two portions of fruit and vegetables a day compared with the rest of the country.

Regression models

Table 4 presents the odds ratio (OR) for each outcome in GGC compared with the rest of Scotland, both unadjusted and adjusted for age and sex. For reasons of clarity and space, Table 5 presents ORs for those outcomes where the effect of residence in GGC was not fully attenuated by adjustment for age and sex, and shows the effects of adjustment for area-based deprivation, individual-level socio-economic status, behavioural factors and biological markers. All these results are discussed in more detail below.

Table 4

ORs for residents of Greater Glasgow and Clyde compared with the rest of Scotland for health conditions and health behaviours

  OR for GGC (95% CI)
 
Unadjusted Adjusted for age and sex 
General and mental health   
 Self-assessed health 1.44 (1.22, 1.70) 1.56 (1.32, 1.83) 
 Anxiety 2.11 (1.49, 3.00) 2.20 (1.54, 3.15) 
 Depression 1.85 (1.22, 2.80) 1.92 (1.27, 2.89) 
 Psychological ill-health 1.39 (1.22, 1.60) 1.38 (1.20, 1.58) 
 Mental well-being 1.24 (1.08, 1.43) 1.25 (1.08, 1.44) 
Physical health 
 Heart attacka 1.26 (1.00, 1.60) 1.44 (1.13, 1.82) 
 CVD 1.09 (0.96, 1.25) 1.20 (1.05, 1.37) 
 COPD 1.32 (1.04, 1.66) 1.44 (1.14, 1.81) 
 Stroke 1.11 (0.87, 1.44) 1.24 (0.96, 1.60) 
 Limiting long-standing illness 1.21 (1.08, 1.37) 1.33 (1.17, 1.51) 
Adverse health behaviours 
 Overweight men 0.75 (0.62, 0.89) 0.78 (0.65, 0.94) 
 Overweight women 0.89 (0.77, 1.03) 0.94 (0.81, 1.09) 
 Obese men 0.92 (0.77, 1.10) 0.98 (0.82, 1.17) 
 Obese women 0.86 (0.73, 1.02) 0.89 (0.76, 1.05) 
 Binge drinking 1.22 (1.02, 1.46) 1.16 (0.96, 1.39) 
 Drinking over recommended weekly limit 0.93 (0.81, 1.08) 0.91 (0.79, 1.05) 
 Drinking over recommended daily limit 1.13 (0.98, 1.29) 1.07 (0.94, 1.23) 
 Current cigarette smoking 1.11 (0.97, 1.27) 1.09 (0.95, 1.25) 
 Heavy cigarette smoking 1.18 (0.99, 1.41) 1.21 (1.01, 1.44) 
 <2 portions fruit/veg per day 1.15 (1.01, 1.30) 1.13 (1.00, 1.28) 
  OR for GGC (95% CI)
 
Unadjusted Adjusted for age and sex 
General and mental health   
 Self-assessed health 1.44 (1.22, 1.70) 1.56 (1.32, 1.83) 
 Anxiety 2.11 (1.49, 3.00) 2.20 (1.54, 3.15) 
 Depression 1.85 (1.22, 2.80) 1.92 (1.27, 2.89) 
 Psychological ill-health 1.39 (1.22, 1.60) 1.38 (1.20, 1.58) 
 Mental well-being 1.24 (1.08, 1.43) 1.25 (1.08, 1.44) 
Physical health 
 Heart attacka 1.26 (1.00, 1.60) 1.44 (1.13, 1.82) 
 CVD 1.09 (0.96, 1.25) 1.20 (1.05, 1.37) 
 COPD 1.32 (1.04, 1.66) 1.44 (1.14, 1.81) 
 Stroke 1.11 (0.87, 1.44) 1.24 (0.96, 1.60) 
 Limiting long-standing illness 1.21 (1.08, 1.37) 1.33 (1.17, 1.51) 
Adverse health behaviours 
 Overweight men 0.75 (0.62, 0.89) 0.78 (0.65, 0.94) 
 Overweight women 0.89 (0.77, 1.03) 0.94 (0.81, 1.09) 
 Obese men 0.92 (0.77, 1.10) 0.98 (0.82, 1.17) 
 Obese women 0.86 (0.73, 1.02) 0.89 (0.76, 1.05) 
 Binge drinking 1.22 (1.02, 1.46) 1.16 (0.96, 1.39) 
 Drinking over recommended weekly limit 0.93 (0.81, 1.08) 0.91 (0.79, 1.05) 
 Drinking over recommended daily limit 1.13 (0.98, 1.29) 1.07 (0.94, 1.23) 
 Current cigarette smoking 1.11 (0.97, 1.27) 1.09 (0.95, 1.25) 
 Heavy cigarette smoking 1.18 (0.99, 1.41) 1.21 (1.01, 1.44) 
 <2 portions fruit/veg per day 1.15 (1.01, 1.30) 1.13 (1.00, 1.28) 

aUsing different age variables: 16–54, 55–64, 75+.

Table 5

ORs for residents of GGC compared with the rest of Scotland for health conditions and health behaviours (where the effect of residence in GGC was not fully attenuated by adjustment for age and sex), showing effects of adjustment for area-based deprivation, individual-level SES, behavioural factors and biological markers

 n Unadjusted Age and sex SIMD (area deprivation) Individual-level socio-economic variables Behavioural variables Biological variables 
General and mental health 
 Self-assessed health 13988 1.44 (1.22, 1.90) 1.56 (1.32, 1.83) 1.20 (1.03, 1.40) n/s   
 Anxiety 1887 2.11 (1.41, 3.15) 2.11 (1.42, 3.14) 1.95 (1.31, 2.90) 1.93 (1.28, 2.90) 1.92 (1.28, 2.88) 1.92 (1.28, 2.88)a 
 Depression 2220 1.85 (1.22, 2.80) 1.92 (1.27, 2.89) 1.75 (1.16, 2.63) n/s   
 Psychological ill-health 10900 1.40 (1.21, 1.62) 1.38 (1.19, 1.60) 1.26 (1.08, 1.46) 1.20 (1.02, 1.40) 1.20 (1.03, 1.40) n/s 
 Mental well-being 12667 1.24 (1.08, 1.43) 1.25 (1.08, 1.44) n/s    
Physical Health 
 Heart attackb 11230 1.30 (0.99, 1.71) 1.46 (1.12, 1.91) 1.35 (1.05, 1.75) 1.36 (1.05, 1.78) 1.37 (1.05, 1.81) 1.44 (1.10, 1.90)c 
 CVD 13984 1.09 (0.96, 1.25) 1.20 (1.05, 1.37) n/s    
 COPD 13995 1.32 (1.04, 1.66) 1.44 (1.14, 1.81) n/s    
 Limiting long-standing  illness 10392 1.29 (1.12, 1.47) 1.40 (1.21, 1.61) 1.24 (1.08, 1.42) n/s   
Adverse health behaviours 
 Overweight men 4656 0.75 (0.62, 0.91) 0.78 (0.64, 0.94) 0.80 (0.65, 0.97) 0.81 (0.67, 0.99) 0.79 (0.65, 0.96) n/s 
 Heavy cigarette smoking 13918 1.18 (0.99, 1.41) 1.21 (1.01, 1.44) n/s    
 n Unadjusted Age and sex SIMD (area deprivation) Individual-level socio-economic variables Behavioural variables Biological variables 
General and mental health 
 Self-assessed health 13988 1.44 (1.22, 1.90) 1.56 (1.32, 1.83) 1.20 (1.03, 1.40) n/s   
 Anxiety 1887 2.11 (1.41, 3.15) 2.11 (1.42, 3.14) 1.95 (1.31, 2.90) 1.93 (1.28, 2.90) 1.92 (1.28, 2.88) 1.92 (1.28, 2.88)a 
 Depression 2220 1.85 (1.22, 2.80) 1.92 (1.27, 2.89) 1.75 (1.16, 2.63) n/s   
 Psychological ill-health 10900 1.40 (1.21, 1.62) 1.38 (1.19, 1.60) 1.26 (1.08, 1.46) 1.20 (1.02, 1.40) 1.20 (1.03, 1.40) n/s 
 Mental well-being 12667 1.24 (1.08, 1.43) 1.25 (1.08, 1.44) n/s    
Physical Health 
 Heart attackb 11230 1.30 (0.99, 1.71) 1.46 (1.12, 1.91) 1.35 (1.05, 1.75) 1.36 (1.05, 1.78) 1.37 (1.05, 1.81) 1.44 (1.10, 1.90)c 
 CVD 13984 1.09 (0.96, 1.25) 1.20 (1.05, 1.37) n/s    
 COPD 13995 1.32 (1.04, 1.66) 1.44 (1.14, 1.81) n/s    
 Limiting long-standing  illness 10392 1.29 (1.12, 1.47) 1.40 (1.21, 1.61) 1.24 (1.08, 1.42) n/s   
Adverse health behaviours 
 Overweight men 4656 0.75 (0.62, 0.91) 0.78 (0.64, 0.94) 0.80 (0.65, 0.97) 0.81 (0.67, 0.99) 0.79 (0.65, 0.96) n/s 
 Heavy cigarette smoking 13918 1.18 (0.99, 1.41) 1.21 (1.01, 1.44) n/s    

n/s, not significant

aModel providing the best fit after adding biological variables did not include any biological variables.

bUsing different age variables: 16–54, 55–64, 75+.

cBMI was the only biological variable adjusted for in this model (as only one participant who had blood data available had experienced a doctor-diagnosed heart attack).

General and mental health

Residents of GGC had significantly higher odds for each of the general and mental health outcomes, both unadjusted and when adjusted for age and sex (Table 4). The odds were still significantly higher after adjustment for area-based deprivation [Scottish Index of Multiple Deprivation (SIMD)] for all outcomes except the Warwick-Edinburgh Mental Well-being Scale (WEMWBS; mental well-being). For the other outcomes, the ORs for GGC were reduced, but not fully attenuated. Further adjustment for the individual-level socio-economic variables fully attenuated the risk for depression and self-assessed health, but significant differences remained for psychological ill-health [measured by the General Health Questionnaire (GHQ); OR = 1.20 (95% confidence interval CI: 1.02, 1.40)] and anxiety [OR = 1.93 (95% CI: 1.28, 2.90); Table 5]. This was also true after further adjustment for behavioural variables [for example: OR for anxiety 1.92 (95% CI: 1.28, 2.88)]. After further adjustment (based on the smaller sample size) for biological variables [body mass index (BMI) and the nurse variables,18 Appendix 3], the risk of poor psychological health for residents of GGC was fully attenuated. However, this was not the case for the anxiety measure: after adjustment for all exposure variables, residents of GGC were still nearly twice as likely to report high levels of anxiety in the model providing the best fit [OR = 1.92 (95% CI: 1.28, 2.88)].

Physical health

After adjusting for age and sex, stroke was the only physical health outcome for which there was no significant difference between residents of GGC and the rest of Scotland. The excess risk of cardiovascular disease (CVD) (OR = 1.20, 95% CI: 1.05, 1.37) and COPD (OR = 1.44, 95% CI: 1.14, 1.81) after adjusting for age and sex were fully attenuated when adjusted for SIMD, showing that area-level deprivation explained the observed difference in levels. Residents of GGC also had an increased risk of having a limiting long-standing illness compared with the rest of Scotland after adjusting for age and sex (OR = 1.40, 95% CI: 1.21, 1.61), which was partially attenuated by adjusting for SIMD, but not fully (OR = 1.24, 95% CI: 1.08, 1.42), with residents of GGC still having odds 24% higher than residents of the rest of Scotland. However, further adjustment for individual-level socio-economic position (SEP) measures in addition to the area-level SIMD fully attenuated the excess risk of limiting long-standing illness.

Residents of GGC had significantly higher odds of having had a doctor diagnosed heart attack after adjusting for age and sex (OR = 1.46, 95% CI: 1.12, 1.91). This was partially attenuated by adjusting for area-level deprivation via the SIMD (OR = 1.35, 95% CI: 1.05, 1.75), showing that some of the difference could be explained by area-level deprivation. However, further adjustment for individual-level SEP variables did not attenuate the excess risk [OR = 1.36 (95% CI: 1.05, 1.78)], nor did additional adjustment for behavioural variables [OR = 1.37 (95% CI: 1.05, 1.81)]. As there was a low prevalence of doctor-diagnosed heart attacks, it was not possible to include nurse or blood covariates in the model. Therefore, the only biological variable added to the model was BMI, which resulted in a slight increase in the OR (1.44 (95% CI: 1.10, 1.90).

Adverse health behaviours

After adjusting for age and sex, there were only significant differences between residents of GGC and the rest of Scotland in the odds of being a heavy smoker (OR = 1.21, 95% CI: 1.01, 1.44), and being overweight for men (OR = 0.78, 95% CI: 0.64, 0.94). Adjusting for SIMD fully attenuated the excess odds of being a heavy smoker, but the odds of male residents of GGC being overweight remained lower than the rest of Scotland (OR = 0.80, 95% CI: 0.65, 0.97). The same was true when first socio-economic (OR = 0.81, 95% CI: 0.67, 0.99) and then behavioural variables (OR = 0.79, 95% CI: 0.65, 0.96) were added to the analysis. However, when biological variables were added to the model, greatly reducing the sample size, there was no significant difference in male overweight between residents of GGC and the rest of Scotland. However, with this reduced sample there was no significant difference after adjusting for SIMD, implying that the reduced sample were not representative of the initial sample.

Discussion

Main findings of this study

Using the 2008 and 2009 Scottish Health Surveys, this study sought to examine the extent to which residence in Glasgow (and its wider conurbation) was significantly and independently associated with the risk of a range of adverse health behaviours, and both mental and physical health outcomes, and the degree to which whether any association was explained by area-level deprivation, and individual socio-economic, behavioural and biological factors. It is notable that for the majority of outcomes analysed (16 out of 18), differences between the Glasgow conurbation and the rest of Scotland were fully explained by a combination of these factors. This is not surprising for two reasons: first, on account of the wide range of explanatory variables included in the models; and second, because poor health has been shown to be significantly higher in all parts of Scotland compared with the rest of Great Britain (whilst controlling for the effects of socio-economic deprivation), with the difference merely more pronounced in the Glasgow area. Given the ubiquitous level of poor health seen across the country, therefore, it would be surprising if any significant differences in health outcomes remained after adjustment for such a broad range of explanatory factors.

That said, a significantly higher risk for residents of GGC was observed for two health outcomes after adjusting for all variables: the first is 42% increased odds of having experienced a doctor-diagnosed heart attack. As the analyses are based on survey data, this outcome obviously only relates to survivors of heart attacks. However, other administrative data show higher rates of hospital admission and mortality for ischaemic heart disease (IHD) among residents of GGC, suggesting this is unlikely to be a solely survivor-specific issue.19

This finding echoes results of analyses by Mitchell et al.,9 who used health survey data to analyse differences in self-reported IHD between Scottish and English populations: a 50% higher risk of IHD was found among Scottish respondents compared with their English equivalents, even after controlling for socio-economic and other risk factors.

The second significant difference which remained was for anxiety (defined as moderate to severe anxiety), for which residents of GGC had close to twice the odds [OR = 1.92 (95% CI: 1.28, 2.88)]. It is important to note that the relationships may not be causative; due to the cross-sectional design of the study it is not possible to investigate causation. Nonetheless this finding of adverse mental health in Glasgow compared with elsewhere in Scotland which cannot be accounted for by socio-economic and other characteristics of the population is echoed by other research. Gray,15 for example, showed this to be the case in relation to poor psychological morbidity (high GHQ12 scores), a similar finding to this study when based on the larger sample size models (i.e. excluding the biological variables). However, despite this, and the similar approaches of the two studies, there are a number of differences between the results of Gray's study and this one. (for example in relation to general health and binge drinking. There were significant differences shown between the two populations after adjusting for SEP in the earlier study, but the difference was fully attenuated in our own analyses) There are a number of reasons why this might be the case: first, different years of survey data have been used, and the profile of the sample may have changed over time. Secondly, the geographical definition of the Glasgow area is different: our study is based on the current NHS Board area of Greater Glasgow and Clyde, which has a larger population size than the previous, smaller, NHS Board area that was the focus of Gray's analyses (NHS Greater Glasgow and Clyde came into being in 2006. It replaced the previous NHS Greater Glasgow health board area, which was merged with part of the old NHS Argyll and Clyde area. It therefore covers the old Greater Glasgow area, but additionally covers areas such as Inverclyde and Renfrewshire. The population size has increased from c. 870 000 to c. 1.2 million. The Scottish Health Survey sample was stratified by NHS Board and therefore analysis at this level is more representative than sub-NHS Board geographies, thus the new NHS Board area was chosen as the preferred geography for the analyses); thirdly, we have used a more sophisticated, spatially sensitive measure of area-based deprivation,20 based on very small areas of around 750 people (whereas the previous analyses were constrained by use of postcode sectors which have an average population size of 5000–6000). Finally, we have employed a much larger selection of individual socio-economic, behavioural and biological variables in the models, resulting in a much richer and comprehensive analytical approach.

It was notable that the differences for the majority of health outcomes between GGC and Scotland were explained by either area or individual-level socio-economic factors. This underlines the importance of improving both the area- and individual-level socio-economic circumstances of those experiencing poverty and deprivation. However, the fact that residents of GGC are associated with such higher risks of aspects of poor mental (anxiety) and physical (heart disease) health, even after controlling for all socio-economic, behavioural and biological differences, clearly warrants further research. At present, we can only speculate as to the likely causes. An assessment of the latter in relation to Scotland's excess mortality compared with elsewhere in the UK (the ‘Scottish Effect’) has recently been undertaken, with the authors highlighting historical, political and economic factors (resulting in levels of hopelessness and community disruption) as potentially important.11 However, as that report also points out, the explanation will clearly be multi-factorial, rather than relating to one single cause, and specific research focusing on the aspects of physical and mental health in Glasgow that have been highlighted in this report may shed additional light.

Limitations of this study

There are some limitations to the study, such as the use of so many self-reported measures; however, as this was the same for residents of GGC and the rest of Scotland, the results are still comparable between the two regions. That said, measures of self-assessed health have been shown to be influenced by demographic, socio-economic and cultural factors,21–23 and it is possible measures such as ‘feelings of anxiety’ may be influenced by such factors.

Complete case analyses were carried out, although weights were used to ensure the participating sample represented Scotland, this does not account for item non-response within participants, which occasionally led to different results when the same model was run using the nurse/blood sample and the complete sample.

The number of participants who had suffered strokes or heart attacks was small, and this meant certain biological relationships could not be investigated.

In addition, this is a cross-sectional data set, so it is not possible to draw conclusions regarding the direction of effects.

Finally, there is a danger that in analysing such a large number of outcomes, some spurious findings may emerge. However, for both outcomes where an increased risk for Glasgow residents was found, the size of the increased odds (42 and 92%) was such that this seems unlikely in this case.

However, the many advantages of the study outweigh the disadvantages, such as the large sample size and breadth of information available in the Scottish Health Surveys, allowing these analyses to be carried out, extending previous work which only investigated the extent of socio-economic variables in explaining differences between Glasgow and the rest of Scotland. An additional advantage is the use of a more spatially specific variable to examine the effect of area-level deprivation.

What is already known on this topic

Scotland has the poorest health of any UK country (and indeed of any Western European nation), driven by high levels of morbidity and mortality in its largest city, Glasgow, and the surrounding conurbation. Socio-economic circumstances neither fully explain Scotland's poor health profile compared with elsewhere in the UK, nor Glasgow's poor health status compared with the rest of Scotland. Previous analyses of deprivation and health between Glasgow and the rest of the country were limited by the area-based deprivation measures and range of risk factors used.

What this study adds

Individual economic status and area-based deprivation explain the majority, but not all, of the differences between the Glasgow area and the rest of Scotland for key measures of health. Other differences were explained by behavioural and biological factors. However, there remained an increased risk of anxiety and doctor-diagnosed heart attacks even after the additional adjustment for behavioural and biological risk factors in addition to socio-economic conditions. Further explanations for this phenomenon are required.

Conclusions

This study has sought to improve on previous analyses to examine the extent to which the adverse health profile of Glasgow, compared with the rest of Scotland, can be explained by differences in socio-economic, behavioural and biological risk factors. It showed that combined area and individual-level socio-economic circumstances explained the differences found between GGC and the rest of Scotland for the vast majority (16 out of 18) of outcomes investigated. However, four outcomes remained where differences were not explained by socio-economic factors: anxiety, doctor diagnosed heart attack, high GHQ scores and being overweight. Of these, the latter two were explained by differences in behavioural and biological factors. However, there remained an unexplained excess in relation to prevalence of anxiety and doctor diagnosed heart attack, with higher, unexplained, levels found in GGC that warrant further investigation.

Funding

This work was supported by the Scottish Government.

References

1
ScotPHO
Scotland and European Health for All (HfA) Database 2007
2007
ScotPHO
 
http://www.scotpho.org.uk/scotlandhfadb (October 2011, date last accessed)
2
Whyte
B
Scottish mortality in a European context, 1950–2000. An analysis of comparative mortality trends
2007
ScotPHO
 
3
Hanlon
P
Walsh
D
Whyte
B
Let Glasgow flourish
2006
Glasgow Centre for Population Health
 
4
Leyland
AH
Dundas
R
McLoone
P
, et al.  . 
Inequalities in mortality in Scotland 1981–2001
2007
MRC Social and Public Health Sciences Unit
 
5
Carstairs
V
Morris
R
Deprivation—explaining differences in mortality between Scotland and England and Wales
BMJ
 , 
1989
, vol. 
299
 
6704
(pg. 
886
-
9
)
6
Carstairs
V
Morris
R
Deprivation and Health in Scotland
Health Bull (Edinb)
 , 
1990
, vol. 
48
 
4
(pg. 
162
-
75
)
7
Bloor
M
Gannon
M
Hay
G
, et al.  . 
Contribution of problem drug users’ deaths to excess mortality in Scotland: secondary analysis of cohort study
BMJ
 , 
2008
, vol. 
337
 pg. 
a478
 
8
Hanlon
P
Lawder
RS
Buchanan
D
, et al.  . 
Why is mortality higher in Scotland than in England and Wales? Decreasing influence of socioeconomic deprivation between 1981 and 2001 supports the existence of a ‘Scottish Effect
J Public Health
 , 
2005
, vol. 
27
 
2
(pg. 
199
-
204
)
9
Mitchell
R
Fowkes
G
Blane
D
, et al.  . 
High rates of ischaemic heart disease in Scotland are not explained by conventional risk factors
J Epidemiol Community Health
 , 
2005
, vol. 
59
 
7
(pg. 
565
-
7
)
10
Walsh
D
Bendel
N
Jones
R
, et al.  . 
It's not ‘just deprivation’: why do equally deprived UK cities experience different health outcomes?
Public Health
 , 
2010
, vol. 
124
 
9
(pg. 
487
-
95
)
11
McCartney
G
Collins
C
Walsh
D
, et al.  . 
Accounting for Scotland's excess mortality: towards a synthesis
2011
Glasgow Centre for Population Health
 
12
Walsh
D
Bendel
N
Jones
R
, et al.  . 
Investigating a ‘Glasgow Effect’: why do equally deprived UK cities experience different health outcomes?
2010
Glasgow Centre for Population Health
 
13
Phillimore
PR
Morris
D
Discrepant legacies: premature mortality in two industrial towns
Soc Sci Med
 , 
1991
, vol. 
33
 
2
(pg. 
139
-
52
)
14
Whynes
DK
Deprivation and self-reported health: are there ‘Scottish effects’ in England and Wales?
Journal of Public Health
 , 
2009
, vol. 
31
 
1
(pg. 
147
-
53
)
15
Gray
L
Comparisons of health-related behaviours and health measures between Glasgow and the rest of Scotland
2007
Glasgow Centre for Population Health
 
16
Scottish Government
The 2008 Scottish Health Survey, Volume 1. Main report
2009
Edinburgh
Scottish Government
 
www.scotland.gov.uk/Publications/2009/09/28102003/0 (August 2011, date last accessed)
17
Scottish Government
The 2009 Scottish Health Survey, Volume 1. Main report
2010
Edinburgh
Scottish Government
 
www.scotland.gov.uk/Publications/2010/09/27093010/0 (August 2011, date last accessed)
18
Landy
R
Walsh
D
Ramsay
J
The Scottish Health Survey: topic report: the Glasgow effect
2010
Edinburgh
Scottish Government
 
19
ISD Scotland
Heart disease statistics update
2011
ISD Scotland
 
20
Scottish Government
Scottish index of multiple deprivation 2009. Technical report
2009
Edinburgh
Scottish Government
 
http://www.scotland.gov.uk/Topics/Statistics/SIMD (October 2011, date last accessed)
21
OECD
Self-reported health and disability
2010
OECD Social Issues/Migration/Health
(pg. 
34
-
5
Volume 2010, Number 21, December (2)
22
Young
H
Grundy
E
O'Reilly
D
, et al.  . 
Self-rated health and mortality in the UK: results from the first comparative analysis of the England and Wales, Scotland, and Northern Ireland longitudinal studies
Popul Trends
 , 
2010
, vol. 
139
 (pg. 
11
-
36
)
23
Mitchell
R
Commentary: the decline of death—how do we measure and interpret changes in self-reported health across cultures and time?
Int J Epidemiol
 , 
2005
, vol. 
34
 
2
(pg. 
306
-
8
)