Final analytic model of fatigue using two-part regression modeling

Variables within hierarchical modelsFirst part—logistic regression
(n = 600)
Any fatigue (binary)
Second part—OLS regression
(n = 576)
Magnitude of fatigue
Overall model stats
ORSE95% CIPPseudo R2bSE95% CIPR2AICLR test
Base covariates< 0.040.074< 0.00010.0864,192Base
Age1.000.020.96–1.030.83–0.070.03–0.12–0.010.02
Gender (ref. female)
    Male0.720.390.25–2.070.54–0.940.92–2.73–0.860.31
Education (ref. ≤ HS)
    Bachelor’s degree0.140.140.02–1.070.06–2.480.82–4.09–0.880.002
    Master’s degree or above0.160.170.02–1.250.08–4.510.81–6.10––2.92< 0.001
DMT (ref. no therapy)
    First line0.600.410.16–2.270.45–1.340.97–3.23–0.560.17
    Second line1.841.450.39–8.580.440.250.97–1.65–2.150.80
MS subtype (ref. RRMS)
    SPMS1.501.220.31–7.350.622.550.990.62–4.480.01
Childhood stressors0.010.105< 0.00010.1384,160< 0.0001
Child stressor count0.510.190.24–1.070.07–1.230.59–2.38––0.08< 0.04
Child stressor severity1.240.121.02–1.510.030.470.130.21–0.74< 0.001

Categories within variables dropped from the model based on collinearity: 1) transgender, non-binary, gender non-conforming, or other, 2) associate degree or some college, 3) PPMS, 4) PRMS, 5) unsure. ≤ HS: high school equivalency or below; PPMS: primary progressive MS; PRMS: progressive-relapsing MS; ref.: reference; SPMS: secondary progressive MS. Blank cells indicate not applicable to individual variables