Table 3.

Effect of Selected Confounders on Association with CRP and PHQ-9/DAST.

	PHQ-9	Association	beta	DAST	Association	beta
All	None		0.124	None		0.014
	Percent Body Fat	Decrease	0.045	Nicotine Use	Increase	−0.017
	Body Mass Index	Decrease	0.057	Percent Body Fat	Increase	0.064
	Waist-to-Hip Ratio	Decrease	0.099	Body Mass Index	Decrease	−0.010
	QIDS-SR	Decrease	0.065	Waist-to-Hip Ratio	Increase	−0.042
Female	None		0.091	None		−0.048
	Percent Body Fat	Decrease	−0.002	Nicotine Use	Increase	−0.062
	Body Mass Index	Decrease	0.016	Percent Body Fat	Increase	−0.061
	Waist-to-Hip Ratio	Decrease	0.050	Body Mass Index	Increase	−0.089
	QIDS-SR	Decrease	0.047	Waist-to-Hip Ratio	Increase	−0.117
				Contraceptive Use	Decrease	−0.030
Male	None		0.154	None		0.120
	Age	Decrease	0.122	Nicotine Use	Decrease	0.060
	Percent Body Fat	Decrease	0.069	Percent Body Fat	Increase	0.150
	Body Mass Index	Decrease	0.096	Waist-to-Hip Ratio	Decrease	0.095
	Waist-to-Hip Ratio	Decrease	0.109
	QIDS-SR	Decrease	0.074

To identify the variables that modulate the association between CRP and psychiatric symptoms, we used beta coefficients from the unadjusted linear regression model (only PHQ-9 and DAST) as the baseline. After progressively adding one new variable into the model, we compared these new beta coefficients for PHQ-9 and DAST with the baseline. Table 3 shows both linear and logistic regression models, which indicated that an unadjusted model including only PHQ-9 or DAST was significantly associated with CRP. For PHQ-9, All (PBF, BMI, WHR, and QIDS-SR), Female (PBF, BMI, WHR, and QIDS-SR), and Male (age, PBF, BMI, WHR, and QIDS-SR)] variables affected CRP concentrations. For DAST, All (nicotine use, PBF, BMI, and WHR), Female (nicotine use, PBF, BMI, WHR, and OC use), and Male (nicotine use, PBF, WHR)] variables affected associations between PHQ-9 or DAST and CRP.