Table 8.
Observational multivariable and one-sample multivariable MR analyses of the effect of BMI and WHR on yearly hospital admission rate in UK Biobank participants of White British ancestry (N = 310,471). Rates and 95% confidence intervals (95% CI) are given. Estimates are provided per exposure SD (SDBMI=4.74 and SDWHR=0.090).
| Observational (adjusteda) | IV (adjustedb) | |
|---|---|---|
| Ratec 95%CI | Ratec 95%CI | |
| BMI (per SD) | 1.029 (1.015–1.043) | 1.035 (0.930–1.153) |
| WHR (per SD) | 1.141 (1.120–1.163) | 1.314 (1.037–1.665) |
BMI = body mass index, CI = confidence interval, IV = instrumental variable, SD = standard deviation, WHR = waist hip ratio
a) The observational analysis regresses the outcome directly on the exposures BMI and WHR simultaneously. The quasi-Poisson regression is adjusted for sex (categorical), age at study entry, alcohol frequency (categorical, from on a daily basis to never), employment (categorical), qualifications (categorical), Townsend deprivation score (categorical in quintiles, where 1 is not deprived and 5 is very deprived), and days of exercise per week (categorical, from 1 to 7);
b) Adjusted for sex, age at study entry, and 40 PCAs
c) Estimates (with corresponding 95% CIs) represent the fold increase in yearly hospital admission rate per BMI SD (4.74 kg/m2) and per WHR SD (0.090). Estimates per BMI and WHR unit are given in Table S7.