Table 2.

Polygenic prediction of schizophrenia measures in the GRAS patient sample

		Years of educationa	Age at prodrome	Age at disease onset	Premorbid IQa	GAFb	CGI-Sb	PANSS positiveb	PANSS negativeb
Baseline model
SZ_all	Stand. beta	0.001	−0.041	−0.056	−0.063	−0.024	0.041	0.033	0.043
	P value	0.976	0.297	0.129	0.090	0.510	0.249	0.364	0.253
EA_all	Stand. beta	0.182**	0.005	−0.002	0.149**	0.068*	−0.057	0.001	−0.051
	P value	4.4 × 10−09	0.884	0.961	7.2 × 10−6	0.029	0.065	0.981	0.107
	Adj. R²	0.0612	0.0023	0.0047	0.0417	0.0655	0.0816	0.0711	0.0243
	ΔAdj. R²c	0.0312	−0.0010	−0.0009	0.0209	0.0035	0.0023	−0.0010	0.0015
Split model
Concordant	Stand. beta	−0.013	−0.019	−0.031	−0.043	−0.096*	0.050	0.079	0.125**
	P value	0.751	0.665	0.456	0.326	0.022	0.232	0.059	0.0036
Discordant	Stand. beta	0.014	−0.030	−0.035	−0.034	0.066	<0.001	−0.039	−0.072
	P value	0.730	0.515	0.409	0.437	0.112	0.996	0.351	0.090
EA_all	Stand. beta	0.191**	0.002	−0.002	0.153**	0.122**	−0.074	−0.039	−0.118**
	P value	1.0 × 10−06	0.965	0.953	2.7×10−4	0.002	0.058	0.319	0.003
	Adj. R²	0.0604	0.0012	0.0037	0.0406	0.0694	0.0811	0.0728	0.0306
	ΔAdj. R²c	0.0304	−0.0021	−0.0019	0.0198	0.0074	0.0018	0.0007	0.0078
n		1039	915	1043	903	1010	1014	1009	1002
ΔR² (Split−baseline model)		−0.0008	−0.0011	−0.0010	−0.0011	0.0039	−0.0005	0.0017	0.0063
P value from F testd		0.698	0.907	0.968	0.891	0.023*	0.479	0.098	0.007**

Notes: Linear regression using the first ten genetic principal components as control variables ^aAge of onset was included as covariate ^bMedication was included as covariate ^cChange in Adj. R² of the models compared to a model that only contains the SZ_all score and the control variables ^dP value from F test refers to improvement in split model compared to baseline model *Significance at P < 0.05 **Significance at P < 0.01 Stand standardized.