Skip to main content
. 2016 Mar 8;352:i582. doi: 10.1136/bmj.i582

Table 3.

Associations between higher BMI and five measures of socioeconomic, using linear or logistic regression and instrumental variable analysis

Socioeconomic status measures and subcategories No Observational* Genetic† Genetic: Egger‡
Change in socioeconomic status (95%CI) per SD higher BMI P value Change in socioeconomic status (95%CI) per SD higher BMI P value Change in socioeconomic status (95%CI) per SD higher BMI P value
Age completed full time education:
 All 82 543 −0.08 (−0.08 to −0.07) <1×10−15 −0.01 (−0.07 to 0.04) 0.63
 Men only 38 342 −0.07 (−0.08 to −0.06) <1×10−15 0.00 (−0.09 to 0.09) 0.98
 Women only 44 201 −0.08 (−0.09 to −0.07) <1×10−15 −0.02 (−0.09 to 0.05) 0.56
Degree level education:
 All 118 565 OR: 0.83 (0.82 to 0.84) <1×10−15 0.94 (0.85 to 1.03) 0.18
 Men only 56 111 OR: 0.82 (0.81 to 0.84) <1×10−15 0.94 (0.81 to 1.09) 0.43
 Women only 62 454 OR: 0.83 (0.82 to 0.84) <1×10−15 0.93 (0.82 to 1.06) 0.28
Job class (skilled/unskilled):
 All 76 404 OR: 0.91 (0.89 to 0.92) <1×10−15 0.90 (0.79 to 1.02) 0.10
 Men only 37 608 OR: 0.93 (0.91 to 0.95) 8×10−9 0.88 (0.73 to 1.08) 0.22
 Women only 38 796 OR: 0.89 (0.87 to 0.91) <1×10−15 0.91 (0.76 to 1.08) 0.29
Annual household income:
 All 103 327 −0.06 (−0.06 to −0.05) <1×10−15 −0.05 (−0.10 to −0.00) 0.041 −0.03 (−0.11 to 0.05) 0.58
 Men only 50 862 −0.01 (−0.02 to −0.00) <1×10−15 0.06 (−0.02 to 0.14) 0.15
 Women only 52 465 −0.09 (−0.10 to −0.08) <1×10−15 −0.14 (−0.20 to −0.08) 1×10−5 −0.17 (−0.25 to −0.05) 0.004
Townsend deprivation index:
 All 119 519 0.08 (0.07 to 0.08) <1×10−15 0.05 (0.01 to 0.10) 0.024 −0.00 (−0.08 to 0.08) 0.96
 Men only 56 582 0.05 (0.04 to 0.05) <1×10−15 −0.01 (−0.08 to 0.06) 0.78
 Women only 62 937 0.10 (0.09 to 0.11) <1×10−15 0.10 (0.04 to 0.16) 0.001 0.10 (−0.01 to 0.21) 0.08

BMI=body mass index; OR=odds ratio.

For age completed full time education, annual household income, and Townsend deprivation index, changes reported are standard deviation. For degree and job class, odds ratios are shown, representing odds of higher socioeconomic status per SD higher BMI.

*Age, assessment centre, and sex adjusted associations.

†Uses instrumental variable analysis, via ivreg2 command in Stata for continuous variables and two step approach for binary outcomes, using BMI genetic risk score. F statistic for all participants is ≥1257 for each socioeconomic status measure; in men only, F statistic is ≥591 for each socioeconomic status measure; in women only, F statistic is ≥666 for each socioeconomic status measure.

‡Alternative genetic approach detailed in Bowden et al 2015,24 used as sensitivity analysis when instrumental variable was P<0.05.