The relative effects of staying in school and attending university on smoking, BMI and systolic blood pressure: evidence from multivariable Mendelian randomization

ObjectivesTo investigate which levels of educational attainment affect health. DesignMultivariable Mendelian randomization study (MVMR). SettingUK Biobank. ParticipantsEuropean ancestry participants born in England. Exposure Educationalattainment was defined as leaving school before age 18, leaving school after 18, or getting a university degree. Randomly allocated genetic variants were used as instruments for these traits. Main outcome measuresBody mass index (BMI), smoking initiation, and systolic blood pressure. ResultsThe MVMR estimates provided little evidence that remaining in school to age 18 affected BMI (mean difference=0.04, 95% CI: -0.42, 0.50), but evidence getting a degree reduced BMI by 0.47 standard deviation 95% CI: (0.01 to 0.97). The MVMR estimates provided evidence that remaining in school to age 18 reduced the odds of initiating smoking (odds ratio (OR): 0.48, 95% CI: 0.30 to 0.76), whereas it provided little evidence of effects of getting a degree (OR: 1.14, 95% CI: 0.69 to 1.88). MVMR suggested that both remaining in school to age 18 and getting a degree had similar effects on systolic blood pressure (mean difference=-2.60 95% CI: -3.73 to -1.46 and mean difference=-3.63 95% CI: -4.92 to -2.34, respectively). ConclusionsMultivariable Mendelian randomization can be used to estimate the effects of complex longitudinal exposures such as educational attainment. This approach can help elucidate how and when factors such as educational attainment affect health outcomes. Key messagesO_LIPrevious studies have estimated the average effect an additional year of education has on health, irrespective of educational level. C_LIO_LIHowever, this assumes that each educational level has the same effect, e.g., an additional year of education in primary school has the same effect as an additional year at university; this assumption is implausible. C_LIO_LIMultivariable Mendelian randomization can be used to relax this assumption and estimate the independent effects of educational levels. C_LIO_LIPreviously reported effects of education on smoking initiation appear to be due to remaining in school until age 18. In contrast, effects on BMI are due to attending university, and effects on systolic blood pressure are similar across education levels. C_LI