Control group matters: Differences in population vs. surgical controls for endometriosis research

Many of the controls included in endometriosis case-control studies are surgically-disconfirmed for endometriosis, who present with various indications for surgery (fibroids/infertility/pelvic pain). Factors/biomarkers associated with endometriosis and pain symptoms may also be associated with these indications. We compared 754 population-based and 282 surgically-disconfirmed controls aged >18 years from two studies. Participants completed surveys on demographics, pain, menstrual, and reproductive factors. We used logistic regression to compare these factors between control groups and between endometriosis cases and controls by control type, adjusting for age, study site, and hormone use. Surgical controls were older (22% vs 6% >40 years), had less current hormone use (24% vs 55%) and more prescription pain medication use (29% vs 6%), compared to population-based controls. Surgical controls were more likely to present with severe dysmenorrhea and general pelvic pain compared to population-based controls (OR:3.89; 95%CI:2.19-6.91 and OR:4.86; 95%CI: 2.24-9.87, respectively). In case-control analyses, the association between endometriosis and severe dysmenorrhea was stronger when compared to population-based (OR:19.4; 95%CI:13.4-28.1), as opposed to surgical (OR:2.20; 95%CI:1.39-3.50), controls. Similar differences by control type were observed for general pelvic pain and dyspareunia. There was also a decreased odds of endometriosis with increasing age at menarche when cases were compared to population-based controls (OR=0.91; 95%CI:0.85-0.98), but not surgical (OR=1.05; 95%CI:0.94-1.17) controls. Control populations differ on several characteristics and pain symptoms. Carefully considering the sampling frame endometriosis cases arose from, and using multiple control group types that are well characterized should aid interpretation of results.