Table 3. Subgroup analyses for evaluating clinicopathologic characteristics of endometriosis-associated ovarian cancer.

				Heterogeneity
Category	No. of studies with references	RR	95% CI	P	I2	Model used
FIGO stage I–II disease
Study design
Case–control	2	1.983	1.683–2.336	0.973	0	Fixed effect
Cohort	4	1.920	1.020–3.616	<0.001	91.975	Radom effects
Adjustment for potential confounding factors
Age	5	1.973	1.297–3.003	<0.001	89.656	Random effects
Grade 1 disease
Study design
Case–control	14	1.354	1.169–1.568	0.743	0	Fixed effect
Quality of study (NOS)
⩾7	11	1.328	1.155–1.528	0.456	0	Fixed effect
<7	4	1.087	0.518–2.280	0.963	0	Fixed effect
Assessment of endometriosis
Histology	3	1.801	0.898–3.610	0.063	63.872	Random effects
Self-report	12	1.303	1.121–1.515	0.945	0	Fixed effect
Adjustment for potential confounding factors
Two factorsa	13	1.330	1.147–1.543	0.858	0	Fixed effect
Eight factorsb	12	1.303	1.121–1.515	0.945	0	Fixed effect
Nulliparity
Assessment of endometriosis
Histology	3	1.648	1.212–2.241	0.150	47.262	Fixed effect
Adjustment for potential confounding factors
Age	3	1.319	1.237–1.407	0.308	15.038	Fixed effect
Optimal debulking surgery
Study design
Case–control	2	1.739	0.630–4.799	<0.001	97.838	Random effects
Cohort	3	1.147	0.973–1.352	0.239	30.192	Fixed effect
Adjustment for potential confounding factors
Age	4	1.376	0.827–2.290	<0.001	94.729	Random effects

Abbreviations: CI=confidence interval; FIGO= International Federation of Gynecology and Obstetrics; NOS=Newcastle–Ottawa Scale; RR=risk ratio.

^aAdjusted for age and race.

^bAge, body mass index, breastfeeding, family history of ovarian cancer, history of tubal ligation, parity, race, and use of oral contraceptive.