Pregnancy loss rate in recurrent pregnancy loss patients according to gestational week: life tables for clinical guidance of patients during supportive care before week 24

Abstract Study question What is the hazard for a new pregnancy loss (PL) each gestational week (GW) reached in RPL patients and does it depend on baseline characteristics? Summary answer The PL rate declines for every GW passed during pregnancy and the hazard rate depends on age, BMI, conception method, and history of late PL. What is known already A few studies in healthy women have provided information about the risk of PL for each GW reached but were highly biased due to variable entry time. No similar study has been performed in recurrent pregnancy loss (RPL) patients who often ask their clinician about their pregnancy prognosis for each GW passed. No tool based on baseline characteristics have been provided to guide patients on the risk of a new PL at a specific GW Study design, size, duration In a prospective cohort study from January 2016 to January 2023, 517 RPL patients with ≥2 previous, consecutive PLs, no uterine or chromosomal abnormalities, and who during follow-up had ≥1 pregnancies after admission to the specialized RPL Clinic of Western Denmark were included. All patients were guided to contact the clinic immediately after a positive urine-hCG test to have an early ultrasound scan from GW 5. All registered pregnancies after admission were included (N = 639). Participants/materials, setting, methods Life table analyses on the risk of a PL according to GW from GW 4-24 were performed on the first and on all subsequent pregnancies after admission, respectively, and hazard ratios (HR) for age (</≥35 years), BMI (</≥30), smoking (no/yes), previous livebirth (no/yes), conception method (natural/IUI versus IVF/ICSI/FET), number of previous PLs (≥/<5), and previous late PL (no/yes) were calculated by univariate log-rank test for equality and a multivariate Cox proportional hazard model. Main results and the role of chance The median GW of PL after admission (n = 169) was 6 (IQR 4-7). In total, 32.1% of patients had ≥1 PL after admission with a cumulative hazard of 0.37 (95%CI 0.32-0.43). The cumulated hazard at GW 7 was 0.27 (0.23-0.32) and 0.34 (0.30-0.41) at GW 12 at which times 76.5% and 94.0% of PLs, respectively, had occurred. The life tables for first pregnancy outcome stratified by baseline characteristics showed that the risk of PL per GW was independent of smoking (HR = 1.14 [0.76-1.73], p = 0.514), prior livebirth (0.77 [0.59-1.01] p = 0.056) and number of previous PL (1.21 [0.84-1.75] p = 0.302), while significantly differing curves and hazard rates were found for maternal age >35 (1.36 [1.00-1.85], p = 0.043), BMI >30 (1.39 [1.00-1.97], p = 0.050), previous late PLs (1.80 [1.20-2.70], p = 0.005), and conception by ART (1.75 [1.39-2.56], p < 0.001), respectively, in comparison to their counterparts (BMI<35, no late PL, and natural conception). The results can be interpreted such as for each week the pregnancy has progressed intact, the risk of PL decreased significantly faster in recurrent PL patients with lower age and BMI, no prior late PL, and pregnancy after natural conceptioncompared with patients in the opposite groups[OBC1] . Comparable results were found when all pregnancies after admission were included. Limitations, reasons for caution For the stratified analyses, the size of each subgroup is small and therefore these life table analyses should be interpreted with caution. The risk of missing registration of early PLs is expectedly small, as patients are encouraged to take contact and have all PL registered to receive best supportive care. Wider implications of the findings The results illustrate that advice to RPL patients about expected pregnancy success at different GW of progressing pregnancy should be modified according to certain baseline characteristics. The time analysis graphs can help clinician inform their patients on the pregnancy prognosis and potentially alleviate the patient’s stress. Trial registration number no applicable as no human specimens were used. Database registered locally by: 2018-5