Delayed pregnancy detection and return to fertility with extended levonorgestrel-containing oral contraceptives: A real world setting cohort study.

PURPOSE: As part of the European risk management plan of a 91-day extended levonorgestrel-containing combined oral contraceptive (COCLNG ), a study was performed to assess its safety. This analysis was conducted to examine delayed pregnancy detection and return to fertility with extended combined oral contraceptives (COC).

METHODS

We conducted a retrospective cohort study in new users of 91-day COCLNG or 28-day COCLNG within a US-based healthcare claims database from 2006 to 2017. Delayed pregnancy detection during current COCLNG exposure was defined as the time between estimated pregnancy start and first prenatal care encounter. Additionally, the time between estimated pregnancy start and COCLNG discontinuation was measured. To measure return to fertility, pregnancy rates were estimated among females who discontinued treatment. 91-day COCLNG users were propensity score-matched to 28-day COCLNG users. Hazard ratio for pregnancy was calculated using Cox proportional hazards models.

RESULTS

The 91-day and 28-day COCLNG users had 25 593 and 76 586 treatment episodes, respectively. The median time to pregnancy detection was 64.5 and 61.0 days (p = 0.24) in users of 91-day COCLNG and 28-day COCLNG . The median exposure time to treatment after estimated pregnancy start was 54.0 and 38.0 days (p < 0.01). In the fertility analysis, pregnancy rates were 54.82 (95% CI, 50.05-59.93) and 69.30 (95% CI, 64.98-73.82) per 1000 person-years in extended COCLNG discontinuers and 28-day COCLNG discontinuers. The adjusted hazard ratio of pregnancy was 0.77 (95% CI, 0.69-0.85).

CONCLUSIONS

Small differences were observed for pregnancy rates and delayed pregnancy detection between 91-day extended COCLNG and 28-day COCLNG , which may be related to the longer days' supply of extended COCLNG . Differences in the fertility analysis may be related to unmeasured residual confounding.