Skip to main content
. 2017 Sep 26;318(12):1182–1184. doi: 10.1001/jama.2017.12141

Table. Autism Spectrum Disorder Heritability Model Comparisons and Parameter Estimates.

Modelsb Model Comparison Measures Estimated Variance (95% CI)a
No. of Model Parameters −2LL Diff
 − 2LL
P Valuec Additive Genetic (Narrow-Sense Heritability) Nonadditive Genetic Environment Total Genetic (Broad-Sense Heritability)
Shared Nonshared
ACDE 14 146 836 NA NA 0.69 (0.40-0.86) 0.10 (0.00-0.38) 0.04 (0.00-0.14) 0.16 (0.05-0.30) 0.80 (0.59-0.95)
ACE 13 146 836 0.4 .52 0.77 (0.58-0.87) NA 0.03 (0.00-0.13) 0.20 (0.13-0.30) 0.77 (0.58-0.87)
ADE 13 146 836 0.8 .38 0.80 (0.68-0.87) 0.05 (0.00-0.26) NA 0.15 (0.05-0.21) 0.85 (0.79-0.95)
CDE 13 146 856 20.9 <.001 NA 0.64 (0.48-0.75) 0.25 (0.21-0.29) 0.11 (0.03-0.24) 0.64 (0.48-0.75)
AE 12 146 836 0.9 .64 0.83 (0.79-0.87) NA NA 0.17 (0.13-0.21) 0.83 (0.79-0.87)
DE 12 147 100 264 <.001 NA 0.99 (0.97-1.00) NA 0.01 (0.00-0.03) 0.99 (0.97-1.00)
CE 12 146 897 61 <.001 NA NA 0.39 (0.37-0.41) 0.61 (0.59-0.63) NA
E 11 147 996 1160 <.001 NA NA NA 1.00 (1.00-1.00) NA

Abbreviations: 2LL, 2 × logarithm of the likelihood; Diff − 2LL, 2 × difference in log-likelihood between the model and the full model; NA, not applicable.

a

The 95% CIs are 2-sided CIs. Variances are based on the tetrachoric correlations. The unadjusted tetrachoric correlation (SD) was estimated to 0.87 (0.08) and 0.40 (0.10) for monozygotic and dizygotic twins; 0.41 (0.01) for full siblings; 0.22 (0.03) and 0.17 (0.04) for maternal and paternal half siblings.

b

All models adjusted for sex and birth cohort. The genetic terms for each model are shown in each row, which include additive genetic effect (A; inherited additive effects of different alleles), shared environmental factors (C; nongenetic influences contributing to similarity within sibling pairs), nonadditive (dominant) genetic factors (D; interaction effects between alleles at the same locus), and nonshared environmental factors (E; making siblings dissimilar).

c

P value for testing the hypothesis: the parameters not in the model but in the full model are all equal to 0.