Table 3.

Multivariate logistic regression analysis Inuit women with depleted iron stores as the dependent variable^1-2

	Multivariate odds ratio3	95% Confidence interval for OR	P-value
Postmenopausal (1 = yes, 0 = no)	0.120	0.069 – 0.207	<0.001
% Body fat > cut-off(1 = yes, 0 = no)	0.414	0.297 – 0.578	<0.001
hs-CRP2 (1 = 3-10 mg/L, 0 = <3 mg/L)	0.571	0.369 – 0.884	0.012
RBC LC-PUFA,% of total fatty acids	0.896	0.793 – 1.012	0.068
Oral contraceptive use (1 = yes, 0 = no)	0.426	0.262 – 0.693	0.078
Food insecure (1 = yes, 0 = no)	1.37	0.98 – 1.92	0.001
Food insecure* RBC LC-PUFA	0.842	0.721 – 0.984	0.030

¹ Women with hs-CRP > 10 mg/L and women with elevated iron stores were excluded from the model (n = 1062 in final model).

² For highly prevalent outcomes, the odds ratios will tend to exaggerate the true relative risk.

³All variables presented were evaluated together in one model. The within-dwelling variance component was 0.1.