Table 2 Comparison of model fit in predicting PTSD symptoms using linear, squared, cubic, and decile terms for testosterone (always including lower order variables), with and without covariates.

Model	Variables in Models	Males				Females
Model	Variables in Models	n	R²	logλ	p-value	n	R²	logλ	p-value
1-1	TT	62,565	0.0004	−153606	1.00E-08	67,906	0.0002	−175902	8.19E-04
1-2	TT+TT²	62,565	0.001	−153589	ref.	67,906	0.0003	−175897	ref.
1–3	TT+TT²+TT³	62,565	0.001	−153589	0.40	67,906	0.0004	−175896	0.23
1–4	Decile TT	62,565	0.0008	−153593	0.32	67,906	0.0004	−175895	0.73
2-1	TT + Covariates	61,631	0.0287	−150211	3.70E-05	66,923	0.029	−172122	1.09E-04
2-2	TT+TT² + Covariates	61,631	0.029	−150202	ref.	66,923	0.0293	−172115	ref.
2-3	TT+TT²+TT³ + Covariates	61,631	0.029	−150202	0.42	66,923	0.0293	−172114	0.43
2–4	Decile TT + Covariates	61,631	0.0289	−150204	0.87	66,923	0.0293	−172111	0.49

Covariates were age, time of collection, BMI, ethnicity, and Townsend deprivation index.
TT total testosterone, Log\(\lambda\) log-likelihood.

Quick links

Search