TABLE 2.

Polygenic risk score and psychosis risk in individuals at clinical high risk and unaffected comparison subjects^a

Group and Ancestry	Clinical High-Risk Subjects (N)	Unaffected Comparison Subjects (N)	Odds Ratio	95% CI	Wald Z	p	AUC	R2 for 2-Year Population Psychosis Risk	R2 for 2-Year Population Psychosis Risk	R2 for 2-Year Population Psychosis Risk
Clinical high risk with psychosis conversion versus clinical high risk with no psychosis conversion
								Risk, 10%	Risk, 20%	Risk, 30%
European	32	92	8.21	1.57, 43.1	2.44	0.015	0.65	0.092	0.112	0.123
Non-European	48	156	1.77	0.63, 4.94	2.00	0.046	0.59	0.035	0.043	0.048
Clinical high risk with psychosis conversion versus unaffected comparison subjects
								Risk, 1%
European	32	70	22.17	3.88, 126	3.16	0.002	0.70	0.117
Non-European	48	146	2.38	0.83, 6.84	2.51	0.012	0.62	0.032
Clinical high risk with no psychosis conversion versus unaffected comparison subjects
								Risk, 5%
European	92	70	1.83	0.67, 5.05	1.33	0.184	0.53	0.007
Non-European	156	146	1.44	0.70, 2.95	1.02	0.309	0.51	<0.001

^aThe main evaluation of polygenic risk score (PRS) risk prediction was prediction of 2-year psychosis in persons at clinical high risk. A secondary analysis compared PRS scores in persons who developed psychosis with those of unaffected comparison subjects. Individuals were considered European if the first principal component was ≥–0.01 and the second was ≤–0.026, and non-European otherwise (see Figure S1 in the online supplement). The PRS odds ratio compared the risk of psychosis conversion in persons in the highest PRS quintiles with the risk in those in the lowest PRS quintiles; p value is for the model containing the PRS alone. Area under the curve (AUC) and R²_liability are corrected for optimism with 1,000 bootstrapped resamples. R²_liability is adjusted for estimated population risk of psychosis.