Risk factors for and outcomes of ring expulsions with a 1-year contraceptive vaginal system

Background: The US Food and Drug Administration–approved segesterone acetate and ethinyl estradiol ring-shaped contraceptive vaginal system, known as Annovera (Sever Pharma Solutions/QPharma, Malmö, Sweden), was inserted and removed under a woman's control for a 21 day in and 7 day out regimen for up to 13 cycles of use. Objective: We aimed to describe the patterns of ring expulsion over time, to identify potential predictors of expulsion, and to evaluate the impact of expulsions on method discontinuation and pregnancy risk. Study Design: Using data from 2064 participants who were enrolled in 2 multinational phase 3 clinical trials on the use of this contraceptive vaginal system, we examined data from participants’ daily diaries for documentation of complete ring expulsion. We modeled the odds of reported expulsions over time with adjustment for background and demographic characteristics using mixed-effects logistic regression models with random intercepts. We compared the probability of continuation between those who did and those who did not report expulsions in the first cycle of use using survival analysis and hazards modeling. To determine if expulsions during the first cycle of use affected the risk for pregnancy, we calculated Pearl Indices. Results: Most participants (75%) never experienced any expulsions during any cycle of use, and 91% to 97% did not experience an expulsion during any 1 cycle. The incidence of expulsion was highest in cycle 1 (9%). The odds of experiencing expulsions decreased by half in cycles 2 to 8 when compared with cycle 1 (0.48; 95% confidence interval, 0.40–0.58), and in cycles 9 to 13, expulsions were about a third of that in cycle 1 (0.32; 95% confidence interval, 0.26–0.41). Of those who did experience expulsions, most (62%–84%) experienced ≤2 expulsions per cycle. Participants from study sites in Latin America vs those in the United States had higher odds of not experiencing an expulsion (odds ratio, 1.95; 95% confidence interval, 1.45–2.63). Women with a higher education level had higher odds of experiencing an expulsion. Notably, parity, age, and body mass index were not associated with expulsion. Participants who experienced any expulsions in cycle 1 were more likely to discontinue use early (hazard ratio, 1.28; 95% confidence interval, 1.14–1.43) than participants who did not have an expulsion. The Pearl Index for participants who had expulsions during cycle 1 was 3.99 (95% confidence interval, 1.29–9.31), which was higher than that among participants who reported no expulsions (Pearl Index, 2.39; 95% confidence interval, 1.61–3.41), but the overlapping confidence intervals indicate that there is not sufficient evidence to demonstrate an association between expulsions and pregnancy risk. Conclusion: Expulsions were infrequent overall, decreased with subsequent cycles of use, and were not associated with body mass index or parity. Early discontinuation of product use was higher among participants who experienced an expulsion during cycle 1. Although it is unclear whether pregnancy risk was associated with expulsions, early recognition of expulsions among users may identify those at higher risk for discontinuation and may highlight when enhanced anticipatory counselling and guidance may be advantageous.