Background: Previous studies showed that obesity is negatively associated with health related quality of life (HRQoL), however, this association might be biased by reversed causation and omitted variables. To overcome these limitations, this study aims to estimate the causal effect of body mass index (BMI) on HRQoL using genetic variants as instrumental variables - a method that is often called Mendelian Randomization (MR).
Methods: We use a) individual-level data on BMI, 77 identified single nucleotide polymorphism (SNPs), HRQoL measured by the physical (PCS) and mental component score (MCS) of the SF-12 of 3080 participants from a population-based German study, and b) meta-data on the SNP-BMI association of 77 identified SNPs from a meta-analysis with >300,000 individuals. We apply a) an ordinal least square (OLS) regression model, b) a one sample MR approach in which we fit a two stage least square (2SLS) regression model with a genetic SNP-score as instrument, and c) a two sample MR approach in which we estimate Wald estimates using the techniques of inverse variance weighting, simple and weighted medians, and MR Egger regression.
Results: The OLS model shows a 0.32 point decrease in PCS (p<0.001) and a 0.04 point decrease in MCS (p=0.54) per BMI unit. The 2SLS regression of the one sample MR approach indicates endogeneity (Durbin-Wu-Hausman-Test; p<0.05) and shows a nonsignificant 0.92 point decrease in PCS and a nonsignificant 0.32 point decrease in MCS per BMI unit. The two sample MR models also show consistently larger, but non-significant PCS and MCS decreases per BMI unit than the linear model.
Conclusion: The negative effect of BMI on HRQoL might be higher than previous studies suggested. This highlights the potential value resulting from obesity prevention. MR studies with larger sample sizes are needed to obtain more precise causal effect estimates.
M. Laxy: None.
