Table 7. Regression of Testosterone-Change on Aggression-Change in Men on Statins^*.

*	Without Adjustment for LDL						Adjusted for LDL and LDL Change†
	Change in Testosterone			Baseline Testosterone			Change in Testosterone			Baseline Testosterone
	Beta ± SE	95% CI	P	Beta ± SE	95% CI	P	Beta ± SE	95% CI	P	Beta ± SE	95% CI	P
On Placebo N = 162	0.059± 0.20	-0.33, 0.45	0.77	-0.005± 0.21	-0.42, 0.41	0.98	0.063 ± 0.20	-0.33, 0.45	0.75	-0.015 ± 0.21	-0.44, 0.41	0.95
On Statins N = 322	0.65 ± 0.30	0.064, 1.2	0.030	0.83 ± 0.31	0.23, 1.4	0.007	0.66 ± 0.31	0.055, 1.3	0.033	0.85 ± 0.31	0.24, 1.5	0.007
On Simvastatin N = 165	1.3 ± 0.49	0.33, 2.2	0.009	1.5 ± 0.51	0.55, 2.5	0.003	1.3 ± 0.51	0.32, 2.4	0.010	1.6 ± 0.52	0.57, 2.6	0.002

Beta = regression coefficient; CI = confidence interval; LDL = low density lipoprotein cholesterol; OASMa = Overt-Aggression-Scale-Modified–Aggression-Subscale; SE = standard error. Change in aggression: Final on-treatment OASMa minus baseline OASMa.

* All regressions adjust for baseline aggression and baseline testosterone (as well as testosterone-change).

^† LDL Change is unrelated to aggression decline in these models, in which testosterone values are also adjusted: A change in testosterone (independent variable) predicts a change in aggression (dependent variable) on statins and simvastatin (of the same sign, producing a positive coefficient). The testosterone analysis was not robust to exclusion of influential outliers.