Variable specification (relative deprivation/income) | |||||||||

Continuous/continuous (primary analysis) | Continuous/quartile | Quartile/continuous | Quartile/quartile | ||||||

HR (95% CI) | VIF | HR (95% CI) | VIF | HR (95% CI) | VIF | HR (95% CI) | VIF | ||

Relative deprivation defined by education | |||||||||

Continuous | 1.15 (1.01 to 1.31) | 2.66 | 1.13 (0.96 to 1.33) | 4.14 | |||||

Quartile 1 (least deprived) | Reference | Reference | |||||||

Quartile 2 | 1.12 (0.78 to 1.59) | 2.67 | 0.96 (0.68 to 1.34) | 2.39 | |||||

Quartile 3 | 1.27 (0.83 to 1.93) | 4.27 | 1.10 (0.73 to 1.64) | 4.06 | |||||

Quartile 4 | 1.66 (1.02 to 2.72) | 5.07 | 1.55 (0.94 to 2.54) | 5.66 | |||||

Absolute income | 0.98 (0.84 to 1.13) | 2.69 | |||||||

Continuous | 1.01 (0.84 to 1.21) | 3.95 | |||||||

Quartile 1 | 1.18 (0.70 to 2.00) | 4.96 | 1.06 (0.63 to 1.80) | 4.77 | |||||

Quartile 2 | 1.11 (0.78 to 1.57) | 2.85 | 1.08 (0.70 to 1.65) | 3.83 | |||||

Quartile 3 | 1.24 (0.96 to 1.60) | 1.86 | 1.29 (0.93 to 1.79) | 2.87 | |||||

Quartile 4 (most affluent) | Reference | Reference | |||||||

Relative deprivation defined by area of residence, sex and age | |||||||||

Continuous | 1.13 (0.99 to 1.29) | 2.70 | 1.10 (0.93 to 1.30) | 4.47 | |||||

Quartile 1 (least deprived) | Reference | Reference | |||||||

Quartile 2 | 1.23 (0.88 to 1.71) | 2.48 | 1.13 (0.82 to 1.54) | 2.20 | |||||

Quartile 3 | 1.29 (0.89 to 1.87) | 3.43 | 1.20 (0.84 to 1.70) | 3.15 | |||||

Quartile 4 | 1.55 (0.99 to 2.42) | 4.33 | 1.34 (0.86 to 2.08) | 4.59 | |||||

Absolute income | |||||||||

Continuous | 0.95 (0.82 to 1.10) | 2.59 | 0.97 (0.83 to 1.14) | 3.16 | |||||

Quartile 1 | 1.33 (0.79 to 2.23) | 5.08 | 1.34 (0.84 to 2.13) | 3.78 | |||||

Quartile 2 | 1.17 (0.82 to 1.65) | 2.82 | 1.13 (0.78 to 1.63) | 3.03 | |||||

Quartile 3 | 1.29 (1.00 to 1.67) | 1.82 | 1.25 (0.93 to 1.66) | 2.31 | |||||

Quartile 4 (most affluent) | Reference | Reference |

VIF, variance inflation factor; HR, hazard ratio; CI, confidence intervals. HRs of continuous variables are per one standard deviation increase. All models are adjusted for age, marital status, medical care utilisation and education. Two models with the largest and smallest effect sizes of relative deprivation are selected for sensitivity analysis. VIFs higher than 10 are commonly considered to be problematic.