. Author manuscript; available in PMC: 2012 Oct 22.

Published in final edited form as: JAMA. 2009 Oct 21;302(15):1651–1657. doi: 10.1001/jama.2009.1487

Table 3.

Primary and Secondary Depression and Anxiety Outcomes

Outcome	Statistical Model¹	Parameter²	ITT Estimate (95% CI)³	Test Statistic⁴	P-Value

Beck Depression Inventory II	ANCOVA I	Between-Group Difference in Post-Means	−1.39 (−4.48, 1.71)	t₁₁₆ = −.89	.38
	ANCOVA II	Between-Group Difference in Post-Means	−1.26 (−4.48, 1.97)	t₁₁₆ = −.77	.44

Hamilton Rating Scale for Depression	ANCOVA II	Between-Group Difference in Post-Means	.14 (−2.15, 2.44)	t₁₁₅ = .12	.90
Beck Anxiety Inventory	ANCOVA II	Between-Group Difference in Post-Means	.59 (−2.31, 3.49)	t₁₁₃ = .40	.69

Remission (BDI-II ≤ 8)	LOGISTIC	Intercept	.40 (.22, .71)	t₁₀₆ = −3.16	.002
Remission (BDI-II ≤ 8)		Treatment Group	.96 (.43, 2.15)	t₁₁₃ = −.11	.91
Improvement (>50% reduction in BDI-II from baseline)		Intercept	.91 (.54, 1.55)	t₁₀₁ = −.35	.73
Improvement (>50% reduction in BDI-II from baseline)		Treatment Group	1.06 (.51, 2.19)	t₁₁₂ = .15	.88

An explanation of the statistical models is provided in the statistical analysis section.

Between-group differences (BGD) reflect placebo minus omega-3 group means. Estimates from both ANCOVA models are adjusted for the baseline score. For the remission and improvement outcomes, the placebo arm is the reference group.

Estimated odds ratios are used to describe remission and improvement outcomes.

⁴

Empirical degrees of freedom (EDFs) were computed from the multiple imputation inferential procedure.