(P-378) A Counterfactual Approach to Minimize Channeling Bias in Post-market Safety Surveillance

Background: Elagolix is an approved treatment for moderate to severe endometriosis (EM) pain, but its hypoestrogenic effects may increase risk of mood disorders. EM pain is associated with greater risk of mood disorders. Elagolix initiators may have more sever EM pain than the general population. This channeling bias limits the ability to quantify rates of adverse events in the indicated population to contextualize on-treatment adverse event rates in the post-market setting.

Objectives: To generate an unexposed comparator group for elagolix users (counterfactual group) and quantify the rate of mood disorders in the counterfactual group relative to the general EM population (general EM group).

Methods: The general EM cohort included women, 15-55 years old, with a diagnosis code for EM or pelvic pain in the IBM MarketScan claims database between 01Jul18 and 01Jul2019. These women were followed until the first date of: disenrollment, death, exposure termination, outcome event or end of data (30Sep20). Mood disorders were defined by ICD diagnosis code (self-harm) or ICD diagnosis codes and/or medication use (depression and anxiety). A propensity score to predict elagolix initiation was developed based on comorbidities and medication use in the 365 days prior to initiation (baseline). Women with EM unexposed to elagolix were matched 1:1 with a 1% caliper to elagolix initiators, to generate a counterfactual group (N=1,206). Baseline characteristics were compared between the general EM group (N=134,820) and counterfactual groups. The relative risk (RR) (95%CI) of mood disorders during follow-up comparing the counterfactual v general EM group was estimated and chi-squared tests determined statistical significance.

Results: No significant differences between the counterfactual and elagolix groups were seen after PS matching. At baseline, the prevalence in the counterfactual v general EM groups was 52% v 37% for depression, 44% v 36% for anxiety and 1% v 2% for self-harm. During follow-up, the risk per 1,000 women was 462 for depression, 406 for anxiety and 12 for self-harm in the counterfactual and 327 for depression, 301 for anxiety and 9 for self-harm in the general EM group. The RR (95%CI) comparing the counterfactual v general EM population was 1.42 (1.33, 1.50) for depression, 1.35 (1.26, 1.44) for anxiety, and 1.34 (0.79, 2.27) for self-harm.

Conclusions: Women with EM with clinical profiles similar to elagolix initiators may have a higher baseline risk of mood disorders relative to the general EM population. Channeling bias in post-market surveillance studies may be minimized using a counterfactual comparator group.