Table 2.
ENID cohort. Association between LTL (T/S mean z-score) and seasonality using Fourier regression models.
| N | LRT χ2 (2 df) | LRT p-value | |
|---|---|---|---|
| Model 1 – crude seasonality modelled with one pair of Fourier termsa Baseline: intercept only | 487 | 4.24 | 0.12 |
| Model 2 – adjusted for age and sex 1 pair Fourier terms + age + sex Baseline: age and sex only | 487 | 4.24 | 0.11 |
| Model 3 – adjusted for age, sex and estimated white cell composition 1 pair Fourier terms + age + sex + cell composition Baseline: age + sex + cell composition only | 204 | 0.87 | 0.65 |
| Model 4 – adjusted for age, sex and maternal BMI 1 pair Fourier terms + age + sex + maternal BMI Baseline: age + sex + maternal BMI only | 345 | 5.48 | 0.06 |
| Model 5 – adjusted for age + sex + birthweight 1 pair Fourier terms + age + sex + birthweight Baseline: age + sex + birthweight | 395 | 5.86 | 0.05 |
aIn each case, model fit is assessed by comparing the full and baseline models using a likelihood ratio test (LRT). LRT confirmed that two/three pairs of Fourier terms did not improve model fit. BMI, Body Mass Index.