Quality of life outcomes in women with endometriosis are highly influenced by recruitment strategies

STUDY QUESTION: To what extent are outcome measures in endometriosis-related quality of life studies influenced by the setting in which patient recruitment is performed? SUMMARY ANSWER: Quality of life outcomes in women with endometriosis are highly influenced by recruitment strategies. WHAT IS KNOWN ALREADY: Most studies on quality of life in women with endometriosis are conducted in tertiary care centres or patient associations. It is conceivable that the setting in which patient recruitment is performed influences the quality of life results. This has not been investigated before. STUDY DESIGN, SIZE, DURATION: Retrospective questionnaire based cohort study (part of the World Endometriosis Research Foundation (WERF) EndoCost study). The investigated women were recruited in three settings: a tertiary care centre for endometriosis (n = 135); five secondary care centres (n = 63); an endometriosis patient association (n = 291). PARTICIPANTS/MATERIALS, SETTING, METHODS: The secondary and tertiary care population included women with a laparoscopic and/or histological diagnosis of endometriosis. The patient association population consisted of women with a self-reported diagnosis of surgically confirmed endometriosis. MAIN RESULTS AND THE ROLE OF CHANCE: The populations did not differ in terms of age, co-morbidities and education level. Delay of diagnosis was the longest in the patient association (median 7 years) (tertiary care 2 years; secondary care 1.5 years) (P < 0.001). The tertiary care population reported more laparotomies (64%) than the other populations (secondary care 43%; patient association 47%) (P = 0.002). Affected job was least prevalent in the secondary care setting (35%) (patient association 64%; tertiary care 56%) (P < 0.001). Affected relationships were most prevalent in the patient association setting (52%) (tertiary care 38%; secondary care 22%) (P < 0.001). Chronic pain was least prevalent in patients in secondary care (44%) (tertiary care 65%; patient association 61%) (P = 0.009). Substantial differences in quality of life were detected between secondary care (median physical component 50.4, mental component 49.6); tertiary care (physical component 46.2, mental component 46.2) and the patient association (physical component 45.0, mental component 44.6) (P < 0.001, P = 0.018). LIMITATIONS, REASONS FOR CAUTION: The response ratewas relatively low (35%). Analysis of the hospital populations revealed that non-responders and responders did not differ with respect to age or revised American Fertility Society classification, indicating that the non-responder bias is limited. However, other factors, such as social and marital status or symptomatology, might be different for non-responders. Missing values were analysed as if the symptomwas not present. Missing values never exceeded 10%, except for one value. Therefore, it can be expected that the effect of missing data on the outcome is negligible. Twenty-five patients belonged to more than one category. A sensitivity analysis showed that the influence of assigning patients to another category was limited. WIDER IMPLICATIONS OF THE FINDINGS: Outcomes regarding quality of life are highly influenced by recruitment strategy. None of the groups appeared to be a representative selection of the total population of women with endometriosis. An alternative strategy for creating a representative population for cost and quality of life studies is probably to recruit women who live in a specific geographic area rather than women that visit a specific hospital or are a member of a patient association.