Table 3.

Associations between candidate correlating factors and percent change in CA125 levels among premenopausal women (N = 1157) by univariable and multivariable linear regressions

Variables	Univariable linear regressionsa			Multivariable linear regressionsb
	Effect estimate	CA125 change	P value	Effect estimate	Standard error	CA125 change	P value
Smoking
Currentd	−0.124	−11.66%	0.047	−0.287	0.093	−24.95%	0.008
Former	0.065	6.72%	0.330	0.014	0.194	1.41%	0.943
None	ref
EtOH
> 2 drinks/day	−0.049	−4.78%	0.653
1–2 drinks/day	0.002	0.20%	0.972
None	ref
Caffeine	−0.0001	−0.01%	0.404
Age of menarche
> 12	0.039	3.98%	0.479
≤ 12	ref
Breastfeedc	0.093	9.75%	0.078	0.110	0.121	11.63%	0.378
CHC use	−0.155	−14.36%	0.218	−0.108	0.108	−10.24%	0.337
OCP duration
10+ years	−0.111	−10.51%	0.310
5–10 years	0.057	5.87%	0.502
2–5 years	0.113	11.96%	0.278
< 2 years	ref
CRP	0.061	6.29%	0.138	0.041	0.078	4.19%	0.607
Ferritin	−0.0003	− 0.03%	0.453
Osteoporosis	−0.373	−31.13%	0.049
Osteoarthritis	0.312	36.62%	0.067
Salpingectomy	−0.121	−11.40%	0.092	0.132	0.194	14.11%	0.506
Endometriosis	0.234	26.36%	0.115	0.057	0.271	5.87%	0.837
Fibroids	0.153	16.53%	0.184	−0.130	0.171	−12.19%	0.460
CAD	−0.689	−49.79%	0.001
HLD	−0.037	−3.63%	0.700
Non-ovarian cancerd	0.216	24.11%	0.106	0.341	0.156	40.64%	0.045

EtOH Alcohol, CHC Current hormonal contraception, OCP Oral contraceptive pill, CRP C reactive protein, CAD Coronary artery disease, HLD Hyperlipidemia

^a All univariable regressions were adjusted for baseline variables including age, race/ethnicity, BMI, and parity

^b Multivariable regressions included the baseline variables (age, race/ethnicity, BMI, parity) and eligible variables from univariable analyses with cut-off p value < 0.25

^c For the “Breastfeed” variable, the reference for parity is one instead of none

^d These variables demonstrated significant associations with CA125 levels in the multivariable linear regression model