Scottish Health Survey - topic report: The Glasgow Effect

Appendix 2: Anxiety model development

Many models are discussed here; all odds ratios given are for the sample with complete data on all variables contained in any of the 'final' models, allowing them to be directly compared. This sample size was 1,134, as some final models contained variables only collected from participants who had a blood sample taken. The weights for the blood sample were therefore used in all the analyses, even if they did not contain any blood variables, as the sample was restricted to those who provided a blood sample. McFadden's pseudo R ²s are compared for the models.

In the initial logistic regression model (Model 1) containing age, sex and residence, residents of Greater Glasgow and Clyde had an odds ratio of 2.92 for suffering from anxiety, meaning their odds were almost three times those of the rest of Scotland. McFadden's pseudo R ² for this model was 0.057. When SIMD was added to the model (Model 2) the odds ratio for residence in Greater Glasgow and Clyde dropped slightly to 2.82, showing that only a small amount of the higher levels of anxiety found in Greater Glasgow and Clyde was explained by the SIMD. McFadden's pseudo R ² for this model was 0.079, showing this model to be a better fit than the previous model.

The socio-economic variables were then added to the model and backward selection performed (Model 3). The variables which remained in the model were age, sex, residence, economic activity and marital status. The odds ratio for residence in Greater Glasgow and Clyde was reduced slightly further to 2.55, showing that socio-economic variables did not fully explain the increased levels of anxiety found in Greater Glasgow and Clyde. McFadden's pseudo R ² for this model was 0.107, showing this model to be a better fit than either of the previous models.

The next model (Model 4) started with all the socio-economic and behavioural variables, and after backward selection contained age, sex, residence, economic activity, potential problem drinking, abstaining from alcohol and physical activity level. Residents of Greater Glasgow and Clyde had over twice the odds of anxiety compared to the rest of Scotland, with an odds ratio of 2.21. This odds ratio was lower than in any of the previous models, but the socio-economic and behavioural variables did not fully explain the increased odds of anxiety for residents of Greater Glasgow and Clyde. McFadden's pseudo R ² for this model was 0.115, showing this model to be a slightly better fit than the previous model.

Model 5 started with all the socio-economic, behavioural and biological variables, and after backward selection only residence, fibrinogen quintile and lung function remained in the model. Residents of Greater Glasgow and Clyde again had over twice the odds of anxiety compared to the rest of Scotland, with an odds ratio of 2.34. McFadden's pseudo R ² for this model was 0.066, meaning this model did not fit the data as well as the previous model. Model 4 would therefore be chosen in preference to this model.

The last model, Model 6, added relationships and social mobility to the variables entered into the previous model, and after backward selection it contained equivalised income, fibrinogen quintiles and lung function. McFadden's pseudo R ² for this model was 0.063, meaning this model did not fit the data as well as Model 4.