Observational, MR, and longitudinal association: obesity and the
mediators
Figure 2 depicts the associations between BMI z-scores and mediators in
both observational and MR 2SLS analyses. According to the MR analysis, a
one-unit increase in the BMI z-score was associated with a decrease in
FEV1/FVC ratio (β, −2.17; 95% CI, −3.92 to −0.42), physical fitness
index (β, −0.71; 95% CI, −1.30 to −0.13), and an increased risk of
early puberty (OR, 1.09; 95% CI, 1.02 to 1.17). Both observational and
MR 2SLS analyses consistently supported the causal association of
obesity with FEV1/FVC, physical fitness, and early puberty, with a
similar causality direction. Although the observational association
between obesity and SDB was significant (β, 0.21; 95% CI, 0.06–0.35),
the MR 2SLS analysis did not support causal relationships. Furthermore,
most of the results of the MR sensitivity analysis (IVW and weighted
median method) confirmed the direction of causal relationships (Figure
E2). No evidence supported the association of individual BMI-SNPs with
the mediators (Tables E5 and E6). The GRSs of the 28 BMI-SNPs showed
strong evidence of an association of obesity with low FEV1/FVC, low
physical fitness, high SDB scores, and high early puberty risk (Table
E1). Prospective cohort analyses supported the finding of Figure 2 that
obesity was associated with a decline in FEV1/FVC, physical fitness, and
increased risk of early puberty (Table 2).