Table 2 Mixed-effects meta-regression of mental disorder PRS effects on social-behavioural symptoms.

		Non-adjusted for cross-disorder effects				Adjusted for cross-disorder effects
	Parameter		θ (SE)	Z-value	P-value		θ (SE)	Z-value	P-value
ADHD-PRS		R² = 0.88				R² = 0.83
	Intercept (Age 4, parent-reported, low prosociality)		−0.015 (0.010)	−1.49	0.14		−0.017 (0.011)	−1.50	0.13
	Age (Centred at 4 years)		0.0025 (0.00089)	2.86	0.0042		0.002 (0.001)	1.96	0.05
	Reporter (Teacher-reported)		0.044 (0.0085)	5.16	2.5 × 10⁻⁷		0.046 (0.0083)	5.56	2.6 × 10⁻⁸
	Trait (Peer problems)		0.03 (0.0089)	3.38	7.3 × 10⁻⁴		0.03 (0.0058)	5.20	2.0×10⁻⁷
ASD-PRS		R² = 0.58				R² = 0.67
	Intercept (Low prosociality)		0.021 (0.0063)	3.36	7.7 × 10⁻⁴		0.017 (0.0049)	3.49	4.8 × 10⁻⁴
	Trait (Peer problems)		0.037 (0.0083)	4.47	7.9 × 10⁻⁶		0.025 (0.0054)	4.51	6.3 × 10⁻⁶
BP-PRS		R² = 0.00				R² = 0.00
	Intercept		0.0054 (0.0056)	0.98	0.33		0.00056 (0.0054)	0.10	0.92
MD-PRS		R² = 0.84				R² = 0.81
	Intercept (Age 4, low prosociality)		−0.018 (0.011)	−1.66	0.096		−0.021 (0.012)	−1.68	0.09
	Age (Centred at 4 years)		0.0035 (0.00095)	3.74	1.9 × 10⁻⁴		0.0027 (0.0011)	2.38	0.02
	Trait (Peer problems)		0.048 (0.0093)	5.14	2.8 × 10⁻⁷		0.051 (0.0068)	7.46	8.8 × 10⁻¹⁴
Schizophrenia-PRS		R² = 0.45				R² = 0.70
	Intercept (Age 4, low prosociality)		0.017 (0.011)	1.55	0.12		0.026 (0.014)	1.91	0.06
	Age (Centred at 4 years)		0.0018 (0.00096)	1.86	0.063		0.0013 (0.0013)	0.99	0.32
	Trait (Peer problems)		−0.027 (0.0094)	−2.94	0.0033		−0.043 (0.0076)	−5.72	1.1 × 10⁻⁸

PRS(C + T) association effects for ADHD, ASD, BP, MD and schizophrenia risk on social behaviour (negative binominal model) were combined across 29 social symptoms (14 ALSPAC-based + 15 TEDS-based; at P_T ≤ 0.1) for each disorder using mixed-effects meta-regressions, accounting for phenotypic correlations between social scores. Here, the most parsimonious models are shown with fixed effect predictors (θ) of PRS effect heterogeneity including age- (centred at 4 years), reporter- (parent versus teacher reports), and trait-specific differences in social behaviour (low prosociality versus peer problems). The most parsimonious model for each disorder was identified using meta-regression combining univariate PRS effects that were non-adjusted for cross-disorder PRS effects. For comparison, corresponding estimates are given here for meta-regressions combining univariate PRS effects that were adjusted for cross-disorder PRS effects. R² is defined as the ratio of explained variance to total variance in univariate PRS effects β.
ADHD Attention-Deficit/Hyperactivity Disorder, ALSPAC Avon Longitudinal study of Parents and Children, ASD Autism spectrum disorder, BP Bipolar disorder, C + T clumping and thresholding, MD Major depression, PRS Polygenic risk scores, P_T PRS threshold, TEDS Twins Early Development Study.
Multiple-testing corrected P-value: P ≤ 0.01.

Search