Journal of Public Health | Vol. 37, No. 2, pp. 258 – 268 | doi:10.1093/pubmed/fdu035 | Advance Access Publication June 8, 2014
No mean city: adolescent health and risk behaviours in a UK urban setting Kate A. Levin1, David Walsh2, Gerry McCartney3 1
NHSGGC, Public Health Directorate, West House, Gartnavel Royal Hospital, Glasgow G12 0HX Glasgow Centre for Population Health, Glasgow G2 4NE, UK Public Health Science Directorate, NHS Health Scotland, Glasgow G2 4DL, UK Address correspondence to Kate Levin, E-mail: [email protected] 2 3
Background The adult population of Glasgow has worse health than in the rest of Scotland, only partially explained by deprivation. Little is known about the health of young Glaswegians. Methods The 2010 Health Behaviour in School-aged Children survey data were analysed using multilevel modelling to compare outcomes in Glasgow relative to the rest of Scotland. Results Glasgow adolescents had similar or better self-reported health on some measures—e.g. adjusting for age and sex, OR for ‘very happy’ was 0.93 (95% CI ¼ (0.75, 1.14))—and the beta coefficient for positive GHQ-12 was 2.79 (0.72, 4.85) compared with the rest of Scotland. However, many health aspects were worse in Glasgow especially for eating and sedentary behaviour, subjective health and aggression, e.g. the OR for ‘daily consumption of vegetables’ was 0.59 (0.46, 0.77), of reporting ‘excellent health’ was 0.66 (0.50, 0.87); headaches was 1.40 (1.09, 1.80); however drinking alcohol in the past week was lower (OR 0.71 (0.50, 0.99)) and smoking, similar. Adjustment for family affluence and school type marginally attenuated the association with Glasgow. Conclusions The worse health experienced by Glasgow adults is only partially seen among young people in Glasgow; however, these are seen at the youngest ages in the study. Keywords places, public health, young people, financial crisis
Introduction The poor health of people in Scotland is well documented, being labelled the ‘Sick Man of Europe’ on account of having the lowest, and most slowly improving, life expectancy in Western Europe.1,2 Scotland also has the largest inequalities in premature mortality within the UK3 and Western Europe.4,5 In particular, the city of Glasgow has the worst health in Scotland with life expectancy at birth of .6 years below the UK average for men (71.6 years, compared with a UK average of 78.2 years), and .4 years below the average for women (78.0 years, compared with a UK average of 82.3 years).6 Although traditional explanations have focussed on material deprivation and deindustrialization, a considerable amount of recent research has highlighted ‘excess’ levels of poor health ‘over and above that’ attributable to differences in socioeconomic characteristics and post-industrial decline.
258
This has been shown nationally (Scotland compared with the rest of GB), regionally (West Central Scotland compared with similar post-industrial European regions), at a city level (Glasgow relative to Liverpool and Manchester) and measured in terms of both area deprivation and individual socioeconomic status (SES).7 – 10 The ‘excess’ has been shown to be ubiquitous in Scotland, but most concentrated in and around the city of Glasgow. Indeed, an excess has also been shown for Glasgow itself compared with the rest of Scotland.11,12 This excess mortality has been shown to be most pronounced among those of working age, with childhood mortality
Kate A. Levin, Senior Researcher David Walsh, Public Health Programme Manager, GCPH Gerry McCartney, Head of Public Health Observatory, NHS Health Scotland
# The Author 2014. Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved. For permissions, please e-mail: [email protected].
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
A B S T R AC T
N O M EA N C I TY
Methods
†
† † †
†
Study design
Data from the 2010 Scottish Health Behaviour in School-aged Children (HBSC) survey were used. Full details of the sampling, recruitment and data collection are described elsewhere.17 Briefly, representative samples of children in school years Primary 7 (P7), Secondary 2 (S2) and Secondary 4 (S4) (ages 11, 13 and 15 years, respectively) were recruited using stratified (at education authority level) random sampling. The Glasgow authority sample was large enough to be representative (comprising 8 primary and 17 secondary schools), enabling comparison with the rest of Scotland. Response rate of schools across Scotland was 74%, and of pupils 89%. The questionnaire was completed anonymously by children in class under teacher supervision. The research protocol was approved by University of Edinburgh ethics committee. Data
Outcome variables Data were collected on eating habits, physical activity and sedentary behaviour, weight control behaviour and body image, puberty, tooth-brushing, mental well-being, subjective health and medicine use, substance use, sexual health, bullying and fighting, injuries and aspirations. A full description of each outcome, including questionnaire items and responses, can be found in the 2010 HBSC National report.17 Items not described in the report have relevant references attached below. The following outcome variables were examined under each of these topics: † Eating habits—nine items: daily consumption of fruit, vegetables, sweets, crisps, chips, sugary drinks and diet soft
†
† †
† †
drinks; family meals eaten on 4 days per week; breakfast consumed daily on weekdays. Physical activity and sedentary behaviour—10 items: at least 1 h of moderate physical activity daily; vigorous physical activity outside school daily; watch TV for .2/4 h a day; play computer games for .2 h a day; use of computer for something other than computer games for .2 h; sedentary for .2 h a day; number of days per week when TV/computers/computer games are used for 2 h or more. Weight control behaviour and body image—three items: currently on a diet; report too fat; report good looks. Tooth-brushing—one item: brush teeth at least twice a day. Mental well-being—eight items: very happy; always happy; always confident; never left out; life satisfaction score; health related quality of life (HRQoL), using the KIDSCREEN-10 score;18 GHQ-1219 measure of social functioning (‘positive scores’) and GHQ-12 measure of anxiety and depression (‘negative scores’). Subjective health and medicine use—14 items: excellent health; headaches/stomach ache/nerves/difficulty sleeping/ back ache/feeling low/bad temper/dizzy occurring more than once a week in the last 6 months; multiple health complaints (MHC), defined as more than one health complaint occurring more than once a week; taken medicine for headaches/stomach ache/nerves/sleeping difficulties in the last month. Substance use—six items: smoke tobacco; weekly smoking; daily smoking; drink alcohol at least once a week; been drunk at least twice; used cannabis three or more times in the last 12 months. Sexual health—one item: has had sexual intercourse. Bullying and fighting—three items: been bullied twice in the last two months; bullied in the last 2 months; been in a physical fight three times or more in the last 12 months. Injuries—one item: injured at least once in the last 12 months requiring medical attention. Aspirations—one item: plan to go onto further/higher education.
Explanatory variables Age was included as a continuous variable in the analyses. Glasgow City education authority school pupils were defined as being from Glasgow. The Family Affluence Scale (FAS), a measure of material wealth, was calculated using responses to questions about having own bedroom, the number of cars and computers in the home and the number of family holidays in the previous 12 months.20 The four items were combined using categorical principal component analysis, a method recommended by Batista-Foguet et al.21 to produce tertiles of low, medium and high family affluence.
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
either no worse (in the case of national and regional comparisons) or better (in the case of Glasgow compared with Liverpool and Manchester).13 Nevertheless, it is likely that poor health in mid-life reflects experiences and influences from earlier in the life course. Recent research has, however, shown little difference in the home environment of children aged 3–7 years living in the Glasgow conurbation relative to those living in similar areas elsewhere.14 The health and health behaviour of adolescents in Glasgow compared with the rest of Scotland is only partially understood15 and is likely to be an important determinant of future adult health.16 The overall objectives of the proposed study are to: (i) assess whether there is evidence of ‘excess’ levels of poor health and risk behaviour in adolescence in Glasgow compared with the rest of Scotland, (ii) identify differences independent of material deprivation and (iii) determine whether these emerge during adolescence.
259
260
J O U RN A L O F P U B LI C H E A LT H
Statistical analysis
Preliminary analyses presented frequencies for each binary variable and means for each continuous variable, for Glasgow and the rest of Scotland, weighting by school grade and type. Logistic multilevel regression models were fitted for each of the binary outcome variables, using reweighted iterative generalized least squares (RIGLS) in MLwiN 2.27.23 Linear multilevel regression models in MLwiN 2.27 were fitted for continuous outcomes such as life satisfaction and KIDSCREEN scores, also using RIGLS estimation. Wald tests were carried out to identify the precision of parameter estimates and associations with Glasgow. The models had three levels: education authority, school and individual child and were fitted for boys and girls separately, adjusting for age and geography (Glasgow/rest of Scotland). The education authority level was insignificant and was therefore removed. Family Affluence Scale was then added and the models re-run. Interaction terms between age and Glasgow were included to examine whether an association
between living in Glasgow and health emerged during adolescence. Additional analyses were carried out adjusting for school type (state/independent), to reduce the possibility of bias, although these are not presented. These showed only very slight differences for some outcome measures and did not change the conclusions of the study. A sensitivity analysis with independent school children in Glasgow schools coded as living outside Glasgow is available from the authors on request.
Results There was a greater proportion of S4 pupils, and particularly girls, and a smaller proportion of S2 pupils in the Glasgow City sample relative to the rest of Scotland (Table 1). A greater proportion of respondents from Glasgow were of low FAS, and a smaller proportion were of high FAS (under a chi-squared test, P , 0.01). After weighting by school grade and school type, there was a greater proportion of Glasgow pupils eating more unhealthy foods (sweets, chips, crisps and sugary drinks) and fewer eating fruit and vegetables. There was also more television viewing and computer use among Glasgow pupils (Tables 1 and 2). Excellent health was more prevalent among girls living outside Glasgow: 17.5% compared with 10.5% (Table 1) and Glaswegian boys were more likely to report having headaches (16.6% compared with 11.2%) and dizziness (12.5% compared with 8.5%). When the data were modelled, adjusting for age and sex (Tables 3 and 4), 8 of the 9 eating behaviour items, 5 of the 8 sedentary behaviour items and 4 of the 14 subjective health items were significant across the pooled gender data set, indicating a greater prevalence of poor health and unhealthy living in Glasgow. In particular, the odds of drinking sugary drinks were over two times those of young people living in the rest of Scotland. Furthermore, the odds of sharing a family meal on four or more days a week were lower among Glaswegian youth. Odds of being in a physical fight were also greater (1.40) among those from Glasgow, whereas Glaswegian girls were less likely to plan to go onto further or higher education. However, the odds of weekly alcohol consumption were lower among those living in Glasgow. In addition, Glaswegian girls reported watching 2 h or more of TV on one more day a week than those living outside Glasgow and played computer games for .2 h on 0.73 days more per week (Table 4). Girls from Glasgow had an average GHQ-12 negative score of 2.45 greater than elsewhere (indicating illness); however, young people from Glasgow also had a GHQ-12 positive score of 2.79 greater than those from elsewhere. Further adjustment for family affluence for the most part reduced the odds ratios, but only marginally (Tables 3 and 4). Odds of girls’ daily consumption of fruit and chips and of
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
Data set representativeness Respondents with missing data for age or family affluence were excluded (33 cases, 1 from Glasgow and 32 from the rest of Scotland with missing age; 148, 12 from Glasgow and 136 from elsewhere with missing FAS), leaving a sample of 6593, of which 633 were from Glasgow. Where outcome data were missing, participants were only excluded from relevant analyses. The proportion of missing cases per variable was generally ,5% with the majority of ,2% for both Glasgow and the rest of Scotland, with no discernible pattern difference between Glasgow and the rest of Scotland. The only exception was for the survey item on sexual activity, which had a higher proportion of missing data in Glasgow (18% compared with 6% in the rest of Scotland). Model procedures adjusting only for age described below were re-run including the 148 coded as ‘missing’ FAS data, with results suggesting no obvious bias arising from the omission of these cases. Comparison with school roll statistics22 showed that there was over-representation of children attending independent schools in P7 in the sample. Primary schools in Glasgow in 2010 had on average 39% of pupils receiving free school meals (FSM) (with a range of 2 –76%), whereas secondary schools had 32% with FSM (2– 51%). In the study sample, these figures were 32% (with a range of 3 –65%) and 27% (2 –44%), a slightly lower proportion of low SES children than that of the population of Glasgow. For the rest of Scotland sample, the average proportion of children receiving FSM (20% in primary and 12% in secondary) was in line with population figures although the range also suggested a higher SES bias in the sample.
N O M EA N C I TY
261
Table 1 Prevalence of adolescent health and risk behaviours for boys and girls in Glasgow and the rest of Scotland Boys
Girls
Glasgow
Rest of Scotland
Glasgow
Rest of Scotland
N
300
2931
333
3029
Grade: P7
102 (31.1%)
912 (31.1%)
90 (27.0%)
Grade: S2
82 (27.3%)
946 (32.3%)
80 (24.0%)
933 (30.8%)
Grade: S4
116 (38.7%)
1073 (36.6%)
163 (48.9%)
1148 (37.9%)
FAS: Low FAS
Demographics 948 (31.3%)
920 (31.4%)
167 (50.2%)
936 (30.9%)
87 (29.0%)
960 (32.8%)
98 (29.4%)
1011 (33.4%)
FAS: High FAS
68 (22.7%)
1051 (35.9%)
68 (20.4%)
1082 (35.7%)
% (SE)a
Outcome measures
P-valueb
% (SE)a
P-valueb
Eating habits Daily consumption of fruit
30.6 (4.5)
33.9 (1.2)
0.469
31.2 (3.1)
40.4 (1.2)
0.006
Daily consumption of vegetables
24.1 (3.1)
34.4 (1.1)
0.002
28.8 (2.7)
40.5 (1.1)
No mean city: adolescent health and risk behaviours in a UK urban setting Kate A. Levin1, David Walsh2, Gerry McCartney3 1
NHSGGC, Public Health Directorate, West House, Gartnavel Royal Hospital, Glasgow G12 0HX Glasgow Centre for Population Health, Glasgow G2 4NE, UK Public Health Science Directorate, NHS Health Scotland, Glasgow G2 4DL, UK Address correspondence to Kate Levin, E-mail: [email protected] 2 3
Background The adult population of Glasgow has worse health than in the rest of Scotland, only partially explained by deprivation. Little is known about the health of young Glaswegians. Methods The 2010 Health Behaviour in School-aged Children survey data were analysed using multilevel modelling to compare outcomes in Glasgow relative to the rest of Scotland. Results Glasgow adolescents had similar or better self-reported health on some measures—e.g. adjusting for age and sex, OR for ‘very happy’ was 0.93 (95% CI ¼ (0.75, 1.14))—and the beta coefficient for positive GHQ-12 was 2.79 (0.72, 4.85) compared with the rest of Scotland. However, many health aspects were worse in Glasgow especially for eating and sedentary behaviour, subjective health and aggression, e.g. the OR for ‘daily consumption of vegetables’ was 0.59 (0.46, 0.77), of reporting ‘excellent health’ was 0.66 (0.50, 0.87); headaches was 1.40 (1.09, 1.80); however drinking alcohol in the past week was lower (OR 0.71 (0.50, 0.99)) and smoking, similar. Adjustment for family affluence and school type marginally attenuated the association with Glasgow. Conclusions The worse health experienced by Glasgow adults is only partially seen among young people in Glasgow; however, these are seen at the youngest ages in the study. Keywords places, public health, young people, financial crisis
Introduction The poor health of people in Scotland is well documented, being labelled the ‘Sick Man of Europe’ on account of having the lowest, and most slowly improving, life expectancy in Western Europe.1,2 Scotland also has the largest inequalities in premature mortality within the UK3 and Western Europe.4,5 In particular, the city of Glasgow has the worst health in Scotland with life expectancy at birth of .6 years below the UK average for men (71.6 years, compared with a UK average of 78.2 years), and .4 years below the average for women (78.0 years, compared with a UK average of 82.3 years).6 Although traditional explanations have focussed on material deprivation and deindustrialization, a considerable amount of recent research has highlighted ‘excess’ levels of poor health ‘over and above that’ attributable to differences in socioeconomic characteristics and post-industrial decline.
258
This has been shown nationally (Scotland compared with the rest of GB), regionally (West Central Scotland compared with similar post-industrial European regions), at a city level (Glasgow relative to Liverpool and Manchester) and measured in terms of both area deprivation and individual socioeconomic status (SES).7 – 10 The ‘excess’ has been shown to be ubiquitous in Scotland, but most concentrated in and around the city of Glasgow. Indeed, an excess has also been shown for Glasgow itself compared with the rest of Scotland.11,12 This excess mortality has been shown to be most pronounced among those of working age, with childhood mortality
Kate A. Levin, Senior Researcher David Walsh, Public Health Programme Manager, GCPH Gerry McCartney, Head of Public Health Observatory, NHS Health Scotland
# The Author 2014. Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved. For permissions, please e-mail: [email protected].
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
A B S T R AC T
N O M EA N C I TY
Methods
†
† † †
†
Study design
Data from the 2010 Scottish Health Behaviour in School-aged Children (HBSC) survey were used. Full details of the sampling, recruitment and data collection are described elsewhere.17 Briefly, representative samples of children in school years Primary 7 (P7), Secondary 2 (S2) and Secondary 4 (S4) (ages 11, 13 and 15 years, respectively) were recruited using stratified (at education authority level) random sampling. The Glasgow authority sample was large enough to be representative (comprising 8 primary and 17 secondary schools), enabling comparison with the rest of Scotland. Response rate of schools across Scotland was 74%, and of pupils 89%. The questionnaire was completed anonymously by children in class under teacher supervision. The research protocol was approved by University of Edinburgh ethics committee. Data
Outcome variables Data were collected on eating habits, physical activity and sedentary behaviour, weight control behaviour and body image, puberty, tooth-brushing, mental well-being, subjective health and medicine use, substance use, sexual health, bullying and fighting, injuries and aspirations. A full description of each outcome, including questionnaire items and responses, can be found in the 2010 HBSC National report.17 Items not described in the report have relevant references attached below. The following outcome variables were examined under each of these topics: † Eating habits—nine items: daily consumption of fruit, vegetables, sweets, crisps, chips, sugary drinks and diet soft
†
† †
† †
drinks; family meals eaten on 4 days per week; breakfast consumed daily on weekdays. Physical activity and sedentary behaviour—10 items: at least 1 h of moderate physical activity daily; vigorous physical activity outside school daily; watch TV for .2/4 h a day; play computer games for .2 h a day; use of computer for something other than computer games for .2 h; sedentary for .2 h a day; number of days per week when TV/computers/computer games are used for 2 h or more. Weight control behaviour and body image—three items: currently on a diet; report too fat; report good looks. Tooth-brushing—one item: brush teeth at least twice a day. Mental well-being—eight items: very happy; always happy; always confident; never left out; life satisfaction score; health related quality of life (HRQoL), using the KIDSCREEN-10 score;18 GHQ-1219 measure of social functioning (‘positive scores’) and GHQ-12 measure of anxiety and depression (‘negative scores’). Subjective health and medicine use—14 items: excellent health; headaches/stomach ache/nerves/difficulty sleeping/ back ache/feeling low/bad temper/dizzy occurring more than once a week in the last 6 months; multiple health complaints (MHC), defined as more than one health complaint occurring more than once a week; taken medicine for headaches/stomach ache/nerves/sleeping difficulties in the last month. Substance use—six items: smoke tobacco; weekly smoking; daily smoking; drink alcohol at least once a week; been drunk at least twice; used cannabis three or more times in the last 12 months. Sexual health—one item: has had sexual intercourse. Bullying and fighting—three items: been bullied twice in the last two months; bullied in the last 2 months; been in a physical fight three times or more in the last 12 months. Injuries—one item: injured at least once in the last 12 months requiring medical attention. Aspirations—one item: plan to go onto further/higher education.
Explanatory variables Age was included as a continuous variable in the analyses. Glasgow City education authority school pupils were defined as being from Glasgow. The Family Affluence Scale (FAS), a measure of material wealth, was calculated using responses to questions about having own bedroom, the number of cars and computers in the home and the number of family holidays in the previous 12 months.20 The four items were combined using categorical principal component analysis, a method recommended by Batista-Foguet et al.21 to produce tertiles of low, medium and high family affluence.
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
either no worse (in the case of national and regional comparisons) or better (in the case of Glasgow compared with Liverpool and Manchester).13 Nevertheless, it is likely that poor health in mid-life reflects experiences and influences from earlier in the life course. Recent research has, however, shown little difference in the home environment of children aged 3–7 years living in the Glasgow conurbation relative to those living in similar areas elsewhere.14 The health and health behaviour of adolescents in Glasgow compared with the rest of Scotland is only partially understood15 and is likely to be an important determinant of future adult health.16 The overall objectives of the proposed study are to: (i) assess whether there is evidence of ‘excess’ levels of poor health and risk behaviour in adolescence in Glasgow compared with the rest of Scotland, (ii) identify differences independent of material deprivation and (iii) determine whether these emerge during adolescence.
259
260
J O U RN A L O F P U B LI C H E A LT H
Statistical analysis
Preliminary analyses presented frequencies for each binary variable and means for each continuous variable, for Glasgow and the rest of Scotland, weighting by school grade and type. Logistic multilevel regression models were fitted for each of the binary outcome variables, using reweighted iterative generalized least squares (RIGLS) in MLwiN 2.27.23 Linear multilevel regression models in MLwiN 2.27 were fitted for continuous outcomes such as life satisfaction and KIDSCREEN scores, also using RIGLS estimation. Wald tests were carried out to identify the precision of parameter estimates and associations with Glasgow. The models had three levels: education authority, school and individual child and were fitted for boys and girls separately, adjusting for age and geography (Glasgow/rest of Scotland). The education authority level was insignificant and was therefore removed. Family Affluence Scale was then added and the models re-run. Interaction terms between age and Glasgow were included to examine whether an association
between living in Glasgow and health emerged during adolescence. Additional analyses were carried out adjusting for school type (state/independent), to reduce the possibility of bias, although these are not presented. These showed only very slight differences for some outcome measures and did not change the conclusions of the study. A sensitivity analysis with independent school children in Glasgow schools coded as living outside Glasgow is available from the authors on request.
Results There was a greater proportion of S4 pupils, and particularly girls, and a smaller proportion of S2 pupils in the Glasgow City sample relative to the rest of Scotland (Table 1). A greater proportion of respondents from Glasgow were of low FAS, and a smaller proportion were of high FAS (under a chi-squared test, P , 0.01). After weighting by school grade and school type, there was a greater proportion of Glasgow pupils eating more unhealthy foods (sweets, chips, crisps and sugary drinks) and fewer eating fruit and vegetables. There was also more television viewing and computer use among Glasgow pupils (Tables 1 and 2). Excellent health was more prevalent among girls living outside Glasgow: 17.5% compared with 10.5% (Table 1) and Glaswegian boys were more likely to report having headaches (16.6% compared with 11.2%) and dizziness (12.5% compared with 8.5%). When the data were modelled, adjusting for age and sex (Tables 3 and 4), 8 of the 9 eating behaviour items, 5 of the 8 sedentary behaviour items and 4 of the 14 subjective health items were significant across the pooled gender data set, indicating a greater prevalence of poor health and unhealthy living in Glasgow. In particular, the odds of drinking sugary drinks were over two times those of young people living in the rest of Scotland. Furthermore, the odds of sharing a family meal on four or more days a week were lower among Glaswegian youth. Odds of being in a physical fight were also greater (1.40) among those from Glasgow, whereas Glaswegian girls were less likely to plan to go onto further or higher education. However, the odds of weekly alcohol consumption were lower among those living in Glasgow. In addition, Glaswegian girls reported watching 2 h or more of TV on one more day a week than those living outside Glasgow and played computer games for .2 h on 0.73 days more per week (Table 4). Girls from Glasgow had an average GHQ-12 negative score of 2.45 greater than elsewhere (indicating illness); however, young people from Glasgow also had a GHQ-12 positive score of 2.79 greater than those from elsewhere. Further adjustment for family affluence for the most part reduced the odds ratios, but only marginally (Tables 3 and 4). Odds of girls’ daily consumption of fruit and chips and of
Downloaded from http://jpubhealth.oxfordjournals.org/ at The Chinese University of Hong Kong on November 14, 2015
Data set representativeness Respondents with missing data for age or family affluence were excluded (33 cases, 1 from Glasgow and 32 from the rest of Scotland with missing age; 148, 12 from Glasgow and 136 from elsewhere with missing FAS), leaving a sample of 6593, of which 633 were from Glasgow. Where outcome data were missing, participants were only excluded from relevant analyses. The proportion of missing cases per variable was generally ,5% with the majority of ,2% for both Glasgow and the rest of Scotland, with no discernible pattern difference between Glasgow and the rest of Scotland. The only exception was for the survey item on sexual activity, which had a higher proportion of missing data in Glasgow (18% compared with 6% in the rest of Scotland). Model procedures adjusting only for age described below were re-run including the 148 coded as ‘missing’ FAS data, with results suggesting no obvious bias arising from the omission of these cases. Comparison with school roll statistics22 showed that there was over-representation of children attending independent schools in P7 in the sample. Primary schools in Glasgow in 2010 had on average 39% of pupils receiving free school meals (FSM) (with a range of 2 –76%), whereas secondary schools had 32% with FSM (2– 51%). In the study sample, these figures were 32% (with a range of 3 –65%) and 27% (2 –44%), a slightly lower proportion of low SES children than that of the population of Glasgow. For the rest of Scotland sample, the average proportion of children receiving FSM (20% in primary and 12% in secondary) was in line with population figures although the range also suggested a higher SES bias in the sample.
N O M EA N C I TY
261
Table 1 Prevalence of adolescent health and risk behaviours for boys and girls in Glasgow and the rest of Scotland Boys
Girls
Glasgow
Rest of Scotland
Glasgow
Rest of Scotland
N
300
2931
333
3029
Grade: P7
102 (31.1%)
912 (31.1%)
90 (27.0%)
Grade: S2
82 (27.3%)
946 (32.3%)
80 (24.0%)
933 (30.8%)
Grade: S4
116 (38.7%)
1073 (36.6%)
163 (48.9%)
1148 (37.9%)
FAS: Low FAS
Demographics 948 (31.3%)
920 (31.4%)
167 (50.2%)
936 (30.9%)
87 (29.0%)
960 (32.8%)
98 (29.4%)
1011 (33.4%)
FAS: High FAS
68 (22.7%)
1051 (35.9%)
68 (20.4%)
1082 (35.7%)
% (SE)a
Outcome measures
P-valueb
% (SE)a
P-valueb
Eating habits Daily consumption of fruit
30.6 (4.5)
33.9 (1.2)
0.469
31.2 (3.1)
40.4 (1.2)
0.006
Daily consumption of vegetables
24.1 (3.1)
34.4 (1.1)
0.002
28.8 (2.7)
40.5 (1.1)
