Skip to main content
. 2017 Aug 28;27(3):030501. doi: 10.11613/BM.2017.030501

Figure 4.

Figure 4

Predictive model proposed by Vuistiner et al. (34) for 25(OH)D concentrations.
The model was set according to the Bayesian Information Criterion (BIC). Body-mass index (BMI) was transformed using an inverse square-root function t(x) = − 1/√x. Model’s coefficients were as follows: β0 = − 2.754 (P = 0.295), β1 = − 1.077 (P < 0.001), β2 = − 0.756 (P < 0.001), β3 = 0.188 (P < 0.001), β4 = 0.025 (P = 0.368), β5 = − 81.08 (P = 0.002), β6 = − 165.6 (P = 0.014), β7 = − 1174 (P ≤ 0.001), β8= 0.218 (P < 0.001), and β9 = 0.164 (P = 0.003). The model explained 29% of the variance.