Clinical pregnancy, live birth, and twin pregnancy outcomes of nine frozen-thawed embryo transfer strategies across female age groups: a real-world cohort study

Frozen-thawed embryo transfer (FET) has become one of the most widely used treatment approaches in assisted reproductive technology (ART) ( 1 , 2 ). A key consideration in selecting an FET strategy is how to improve the probability of pregnancy within a single transfer cycle while minimizing the risk of multiple pregnancy, particularly twin pregnancy, and its associated obstetric complications and adverse perinatal outcomes ( 3 , 4 ). Double-embryo transfer may increase the pregnancy rate per transfer cycle but also substantially increases the risk of twin pregnancy, which is closely associated with preterm birth, low birth weight, and maternal complications. By contrast, single-embryo transfer can reduce the risk of twin pregnancy, although its reproductive outcomes may be influenced by factors such as female age and embryo quality ( 1 , 3 , 4 ). Embryo-related factors affecting FET outcomes can be broadly categorized into three dimensions (1): embryo developmental stage, namely the choice between day-3 (D3) cleavage-stage embryos and day-5 (D5) or day-6 (D6) blastocysts (2); embryo quality, including the distinction between high-quality embryos and non-high-quality usable embryos; and (3) the number of embryos transferred, involving the balance between single- and double-embryo transfer ( 4 – 7 ). Although these individual factors have been widely investigated, large-scale real-world studies systematically comparing the combined transfer strategies derived from these three dimensions across different female age groups remain limited ( 8 ). In clinical practice, two common decision-making scenarios remain insufficiently supported by evidence. First, when both cleavage-stage embryos and blastocysts are available, the optimal transfer strategy remains uncertain. Previous studies have suggested that blastocyst transfer may reduce the risk of ectopic pregnancy (0.8% vs 1.8%) and that single-blastocyst transfer may maintain favorable implantation potential while reducing the risk of multiple pregnancy ( 9 , 10 ). Second, for women of advanced reproductive age with diminished ovarian reserve, a limited number of embryos, and only cryopreserved D3 cleavage-stage embryos without blastocyst culture, the appropriate number of embryos to transfer remains unclear. Previous evidence suggests that double cleavage-stage embryo transfer in women aged ≥38 years may improve the probability of pregnancy to some extent, although the risk of multiple pregnancy must also be considered ( 3 ). Another study indicated that thawing cryopreserved D3 embryos and continuing culture to the blastocyst stage before transfer may result in better pregnancy outcomes than direct D3 embryo transfer ( 11 ). Using real-world FET data from our center, this study compared transfer strategies defined by embryo developmental stage, morphological quality, and number of embryos transferred across different female age groups. We specifically evaluated the associations of these strategies with clinical pregnancy, live birth, and twin pregnancy outcomes in the above-mentioned clinical scenarios, with the aim of providing evidence to support individualized selection of FET strategies.

Female age, male age, body mass index (BMI), duration of infertility, anti-Müllerian hormone (AMH) level, antral follicle count (AFC), number of oocytes retrieved, number of usable embryos, and endometrial thickness on the day of transfer differed significantly across the four female age groups (all *P*<0.05). With increasing age, AMH, AFC, the number of oocytes retrieved, and the number of usable embryos generally decreased, whereas BMI showed an overall increasing trend. The distributions of infertility factor category, polycystic ovary syndrome (PCOS), tubal/pelvic factors, ovulatory dysfunction, diminished ovarian reserve, fertilization method, and endometrial preparation regimen also differed significantly across age groups (all *P*<0.05). No statistically significant between-group difference was observed for endometriosis/uterine factors (*P*=0.075) ( Table 2 ). Baseline and cycle characteristics of patients across female age groups. This table includes 2,143 cycles assigned to the nine prespecified embryo transfer strategies. Continuous variables were non-normally distributed and are presented as median (Q1, Q3); comparisons were performed using the Kruskal-Wallis H test. Categorical variables are presented as n (%) and were compared using the Pearson chi-square test. A two-sided P < 0.05 was considered statistically significant. Q1, first quartile; Q3, third quartile; AMH, anti-Müllerian hormone; AFC, antral follicle count; IVF, in vitro fertilization; ICSI, intracytoplasmic sperm injection; PCOS, polycystic ovary syndrome. Among the 2,143 transfer cycles, 1,053 resulted in clinical pregnancy, corresponding to an overall clinical pregnancy rate of 49.1%. Among the 1,802 cycles with known live birth outcomes, 533 resulted in live birth, yielding a live birth rate of 29.6%. Within each of the four age groups (<30, 30–34, 35–39, and ≥40 years), the overall clinical pregnancy rates differed significantly across the nine transfer strategies. Significant between-strategy differences in live birth rates were observed in the 35–39-year and ≥40-year groups (*P*=0.002 and *P*=0.043, respectively), whereas no significant differences were observed in the younger age groups (both *P*>0.05) ( Table 3 ). Clinical pregnancy and live birth outcomes by embryo transfer strategy and female age group. Clinical pregnancy rates were calculated using all transfer cycles within each age group and transfer strategy as the denominator. Live birth was defined as the delivery of at least one live-born infant. Cycles without pregnancy, with biochemical pregnancy, with a documented miscarriage, or with stillbirth only were coded as no live birth. Cycles with a clinical pregnancy but without a documented live birth, miscarriage, stillbirth, or delivery outcome were classified as having an unknown live birth outcome and were excluded from the live birth denominator. The 95% CIs were calculated using the Wilson method. Overall comparisons among the nine strategies within each age group were performed using the Fisher-Freeman-Halton exact test with 200,000 Monte Carlo repetitions. A total of 2,143 cycles were included; no D6 double high-quality blastocyst transfers occurred in women aged ≥40 years. D5 double high-quality blastocyst transfer showed relatively high clinical pregnancy and live birth rates in most age groups. In the D5 single high-quality blastocyst group, the clinical pregnancy rate decreased from 64.2% among women aged <30 years to 41.0% among those aged ≥40 years, while the corresponding live birth rate decreased from 44.4% to 23.5% ( Table 3 ). A total of 174 twin pregnancies and three triplet pregnancies were observed among all included cycles. The twin pregnancy rate among clinical pregnancies was 16.5%, whereas the per-transfer-cycle risk of twin pregnancy was 8.1%. In the <30-, 30–34-, and 35–39-year groups, both the twin pregnancy rate among clinical pregnancies and the per-transfer-cycle twin pregnancy risk differed significantly across transfer strategies (all *P*<0.001). Among women aged ≥40 years, the overall difference in the twin pregnancy rate among clinical pregnancies did not reach statistical significance (*P*=0.063), whereas the between-strategy difference in per-transfer-cycle twin pregnancy risk remained statistically significant (*P*=0.007) ( Table 4 ). Across all four age groups, the per-transfer-cycle twin pregnancy risk was lower with D5 single high-quality blastocyst transfer than with D5 double high-quality blastocyst transfer. Overall, twin pregnancy rates among clinical pregnancies were higher with double-embryo transfer strategies and lower with single-embryo transfer strategies ( Table 4 ). Twin pregnancy outcomes by embryo transfer strategy and female age group. Triplet pregnancies were reported separately and were not included in the numerator for twin pregnancy. The twin pregnancy rate among clinical pregnancies used the number of clinical pregnancy cycles within each age group and transfer strategy as the denominator; the twin pregnancy risk per transfer cycle used all transfer cycles as the denominator. Ultrasound fetal-number data were available for all clinical pregnancy cycles. The 95% CIs were calculated using the Wilson method. Overall comparisons among the nine strategies within each age group were performed using the Fisher-Freeman-Halton exact test with 200,000 Monte Carlo repetitions. No D6 double high-quality blastocyst transfers occurred in women aged ≥40 years. Using D5 single high-quality blastocyst transfer as the reference, and after adjustment for actual female age, BMI, duration of infertility, infertility factor category, PCOS, AMH, number of usable embryos, fertilization method, endometrial preparation regimen, and endometrial thickness on the day of transfer, D5 double high-quality blastocyst transfer was associated with higher odds of clinical pregnancy (*P*=0.026). D5 single usable blastocyst transfer, D3 double high-quality cleavage-stage embryo transfer, D3 single high-quality cleavage-stage embryo transfer, and sequential transfer were associated with lower odds of clinical pregnancy. No statistically significant differences from the reference strategy were observed for D6 single high-quality blastocyst transfer, D5 double blastocyst transfer comprising one high-quality and one usable blastocyst, or D6 double high-quality blastocyst transfer ( Table 5A ; Figure 2 ). Logistic regression analyses of the associations between embryo transfer strategy and clinical pregnancy or live birth. D5 single high-quality blastocyst transfer was the reference group. The crude model included embryo transfer strategy only. Model 1 adjusted for actual female age, body mass index, duration of infertility, infertility factor category, and PCOS. Model 2 additionally adjusted for AMH, number of usable embryos, fertilization method, endometrial preparation regimen, and endometrial thickness on the day of transfer. The clinical pregnancy analysis included all 2,143 cycles. The live birth analysis included 1,802 cycles with known outcomes and excluded 341 cycles with unknown live birth outcomes. OR, odds ratio; aOR, adjusted odds ratio; CI, confidence interval; AMH, anti-Müllerian hormone; PCOS, polycystic ovary syndrome. Forest plot of multivariable-adjusted odds ratios for clinical pregnancy and live birth across 9 embryo transfer strategies. D5 single high-quality blastocyst transfer was used as the reference group. Points represent adjusted odds ratios (aORs), and horizontal lines indicate 95% confidence intervals (CIs). The models were adjusted for actual female age, body mass index (BMI), duration of infertility, infertility factor category, polycystic ovary syndrome (PCOS), anti-Müllerian hormone (AMH), number of usable embryos, fertilization method, endometrial preparation regimen, and endometrial thickness on the day of transfer. The clinical pregnancy model included 2,143 transfer cycles, whereas the live birth model included 1,802 cycles with known live birth outcomes. Among the 1,802 cycles with known live birth outcomes, D5 double high-quality blastocyst transfer was associated with higher odds of live birth. D5 single usable blastocyst transfer, D3 single high-quality cleavage-stage embryo transfer, and sequential transfer were associated with lower odds of live birth. The associations of D3 double high-quality cleavage-stage embryo transfer and D6 single high-quality blastocyst transfer with live birth were no longer statistically significant after full adjustment ( Table 5B ; Figure 2 ). Because no twin pregnancy events occurred in the D5 single usable blastocyst group or the D3 single high-quality cleavage-stage embryo group, Firth bias-reduced logistic regression was applied. After exclusion of the three triplet pregnancies, the primary analysis included 2,140 transfer cycles, of which 174 resulted in twin pregnancy. Using D5 single high-quality blastocyst transfer as the reference and adjusting for actual female age, BMI, infertility factor category, PCOS, AMH, number of usable embryos, and endometrial preparation regimen, D5 double high-quality blastocyst transfer, D5 double blastocyst transfer comprising one high-quality and one usable blastocyst, D3 double high-quality cleavage-stage embryo transfer, D6 double high-quality blastocyst transfer, and sequential transfer were all associated with a significantly increased per-transfer-cycle risk of twin pregnancy (all *P*<0.001). No statistically significant differences from the reference strategy were observed for D5 single usable blastocyst transfer, D3 single high-quality cleavage-stage embryo transfer, or D6 single high-quality blastocyst transfer. The sensitivity analysis restricted to clinical pregnancy cycles yielded results in the same direction, with a significant overall effect of transfer strategy (*P*<0.001) ( Table 6 ). Firth logistic regression analyses of the association between embryo transfer strategy and twin pregnancy risk. The primary analysis included all transfer cycles, whereas the sensitivity analysis was restricted to clinical pregnancy cycles. The n/N (%) values were calculated using all 2,143 transfer cycles as the denominator. After excluding three triplet pregnancies, the primary regression analysis included 2,140 cycles, including 174 twin pregnancies. The sensitivity analysis included 1,050 clinical pregnancy cycles, including 174 twin pregnancies. The adjusted model included embryo transfer strategy, actual female age, body mass index, infertility factor category, PCOS, AMH, number of usable embryos, and endometrial preparation regimen. Because no twin pregnancy events occurred in the D5 single usable blastocyst and D3 single high-quality cleavage-stage embryo groups, Firth bias-reduced logistic regression was used. The 95% CIs were based on the penalized information matrix, and P values were obtained using penalized likelihood-ratio tests. OR, odds ratio; aOR, adjusted odds ratio; CI, confidence interval; AMH, anti-Müllerian hormone; PCOS, polycystic ovary syndrome. Age-by-transfer-strategy interaction analyses showed no statistically significant interaction for either clinical pregnancy or live birth ( Table 7A ). In sensitivity analyses, replacing AMH with AFC or replacing the overall infertility factor category with four specific infertility factors produced associations that were generally consistent in direction and magnitude with those observed in the primary models. After exclusion of women with PCOS, the clinical pregnancy findings remained consistent, and the overall effect of transfer strategy remained statistically significant (*P*<0.001). The overall effect of transfer strategy on live birth also remained significant (*P*=0.007), although the confidence intervals for some individual comparisons crossed 1 because of the reduced sample size. When all 341 cycles with unknown live birth outcomes were conservatively classified as no live birth, D5 double high-quality blastocyst transfer remained associated with higher odds of live birth (aOR=1.60, 95% CI: 1.10–2.33). The associations for several other strategies were attenuated, but the overall effect of transfer strategy remained statistically significant (*P*<0.001) ( Table 7B ). Interaction and sensitivity analyses of female age and embryo transfer strategy. All models used D5 single high-quality blastocyst transfer as the reference group. The primary model adjusted for actual female age, body mass index, duration of infertility, infertility factor category, PCOS, AMH, number of usable embryos, fertilization method, endometrial preparation regimen, and endometrial thickness on the day of transfer. Interactions were assessed using likelihood-ratio tests comparing nested logistic regression models, with age entered either as a continuous variable or as four age groups. Sensitivity analyses included replacing AMH with AFC, replacing the overall infertility factor category with four specific infertility factors, excluding patients with PCOS, and coding all 341 cycles with unknown live birth outcomes as no live birth. The primary live birth analysis included 1,802 cycles with known outcomes, whereas the conservative sensitivity analysis included all 2,143 cycles. No D6 double high-quality blastocyst transfers occurred in women aged ≥40 years. Values in part B are aORs (95% CIs). aOR, adjusted odds ratio; CI, confidence interval; AMH, anti-Müllerian hormone; AFC, antral follicle count; PCOS, polycystic ovary syndrome. In the fully adjusted main-effects models, the predicted probabilities of clinical pregnancy and live birth generally decreased with increasing age across all transfer strategies. Further age-by-transfer-strategy interaction tests were not statistically significant. Because no D6 double high-quality blastocyst transfer cycles occurred among women aged ≥40 years, the predicted value for this age–strategy combination was based on model extrapolatio and should therefore be interpreted with caution ( Figures 3A, B ; Supplementary Figures 1, 2 ). (A) Age-stratified adjusted predicted probabilities of clinical pregnancy. Age was categorized as <30, 30–34, 35–39, and ≥40 years. The vertical axis represents the multivariable-adjusted predicted probability of clinical pregnancy derived from the multivariable logistic regression model. Points indicate predicted probabilities, and error bars represent 95% CIs. The model was adjusted for the same covariates as Model 2 in Table 5 . HQ, high-quality.In the “1 HQ + 1 usable” group, “usable” refers to a non-high-quality usable blastocyst. (B) Age-stratified adjusted predicted probabilities of live birth. Age was categorized as <30, 30–34, 35–39, and ≥40 years. The vertical axis represents the multivariable-adjusted predicted probability of live birth derived from the multivariable logistic regression model. Points indicate predicted probabilities, and error bars represent 95% CIs. The live birth analysis included 1,802 transfer cycles with known live birth outcomes, and the model was adjusted for the same covariates as Model 2 in Table 5 .HQ, high-quality.In the “1 HQ + 1 usable” group, “usable” refers to a non-high-quality usable blastocyst.

Using real-world cohort data, this study systematically compared nine embryo transfer strategies defined by embryo developmental stage, embryo quality, and the number of embryos transferred, and further evaluated the potential modifying effect of female age on the associations of these strategies with clinical pregnancy and live birth. The principal findings showed that D5 single high-quality blastocyst transfer achieved favorable clinical pregnancy rates across age groups while maintaining a relatively low per-transfer-cycle risk of twin pregnancy. D5 double high-quality blastocyst transfer was associated with higher odds of clinical pregnancy and live birth, but also with a markedly increased per-transfer-cycle risk of twin pregnancy. The outcomes of several D3 cleavage-stage embryo, D6 blastocyst, and sequential transfer strategies were inferior to, or not significantly better than, those of D5 single high-quality blastocyst transfer. No statistically significant age-by-transfer-strategy interaction was observed, suggesting that the available data were insufficient to demonstrate that the relative effects of the transfer strategies varied substantially with female age. Therefore, selection of an embryo transfer strategy should incorporate female age, embryo developmental stage and quality, the number of usable embryos, and the risk of multiple pregnancy, rather than being based solely on the unadjusted pregnancy rate within a particular age group. D5 single high-quality blastocyst transfer was the most frequently used strategy in this study, and its clinical pregnancy rate generally declined with increasing female age. The twin pregnancy rates among clinical pregnancies across the four age groups were 0.9%, 1.4%, 4.7%, and 0, respectively, while the corresponding per-transfer-cycle twin pregnancy risks were 0.6%, 0.8%, 2.5%, and 0. These findings indicate that this strategy was associated with a generally low risk of multiple pregnancy, although twin pregnancy was not completely eliminated. This observation is consistent with current international guidelines advocating elective single-blastocyst transfer (eSET) to reduce multiple pregnancy and its associated maternal and neonatal complications ( 10 ). Although some studies have suggested that embryo developmental stage may have a limited effect on reproductive outcomes, most available evidence indicates that blastocyst transfer may provide greater implantation potential and a higher probability of live birth when a usable high-quality blastocyst is available ( 12 , 13 ). Previous studies have also shown that, among young women with a favorable prognosis, single high-quality blastocyst transfer can maintain satisfactory reproductive outcomes while reducing the risks of multiple pregnancy and preterm birth ( 4 , 10 ). In the present study, women aged <35 years who underwent D5 single high-quality blastocyst transfer achieved a clinical pregnancy rate exceeding 60%, while the per-transfer-cycle twin pregnancy risk remained below 1%, suggesting that this strategy may provide a favorable balance between effectiveness and safety. Although the clinical pregnancy and live birth rates decreased to 53.5% and 27.4%, respectively, among women aged 35–39 years, the per-transfer-cycle twin pregnancy risk remained relatively low at 2.5%. Previous research has similarly shown that single high-quality blastocyst transfer can maintain a low twin pregnancy rate in this age group, substantially below that observed after double-embryo transfer ( 8 ). Among women aged ≥40 years, although the unadjusted clinical pregnancy and live birth rates were relatively high in the D5 double high-quality blastocyst group, the per-transfer-cycle twin pregnancy risk was also markedly increased, and the number of cycles was limited. Therefore, decisions regarding whether to increase the number of embryos transferred in women of advanced reproductive age should take into account embryo quality, previous treatment history, patient preferences, and the risk of multiple pregnancy. Compared with D5 single high-quality blastocyst transfer, D5 double high-quality blastocyst transfer remained associated with higher odds of clinical pregnancy and live birth after full adjustment, suggesting that the addition of a second high-quality D5 blastocyst may increase the probability of achieving pregnancy and live birth within a single transfer cycle. However, this potential benefit was accompanied by a marked increase in the per-transfer-cycle risk of twin pregnancy. Firth logistic regression showed that the adjusted odds of twin pregnancy in the D5 double high-quality blastocyst group were approximately 33-fold higher than those in the reference group. The direction of the association remained consistent in the sensitivity analysis restricted to clinical pregnancy cycles. Previous studies have likewise shown that double-embryo transfer may increase the probability of pregnancy within a single cycle but substantially increases the risks of multiple pregnancy and associated maternal and neonatal complications ( 3 , 4 , 14 ). The present findings further emphasize that any potential reproductive benefit obtained by increasing the number of embryos transferred should not be evaluated independently of the accompanying risk of twin pregnancy. Notably, no significant interaction between female age and transfer strategy was observed for either clinical pregnancy or live birth. Therefore, the unadjusted outcome rates in a particular age group do not provide sufficient evidence to conclude that D5 double high-quality blastocyst transfer is appropriate only for women of advanced reproductive age or that its per-transfer-cycle twin pregnancy risk consistently decreases with increasing age. For patients considering double high-quality blastocyst transfer, individualized decision-making should be based on comprehensive counseling regarding the potential benefit within a single transfer cycle, the risk of twin pregnancy, and the subsequent obstetric risks. Transfer of two D5 blastocysts comprising one high-quality blastocyst and one non-high-quality usable blastocyst is a relatively common clinical strategy. Compared with D5 single high-quality blastocyst transfer, this strategy showed no statistically significant advantage in either the adjusted clinical pregnancy or live birth model but was associated with a markedly increased per-transfer-cycle risk of twin pregnancy. Previous studies comparing the transfer of a single high-quality embryo with the transfer of a high-quality embryo plus one low-quality embryo have shown that the additional contribution of the second low-quality embryo to live birth is inconsistent, whereas the risks of multiple pregnancy and adverse neonatal outcomes may be substantially increased ( 15 ). Similarly, the present findings suggest that, when one high-quality D5 blastocyst is already available, the addition of a non-high-quality usable blastocyst should be considered cautiously and should not be routinely undertaken solely to increase the probability of success within a single transfer cycle. Compared with D5 single high-quality blastocyst transfer, D3 double high-quality cleavage-stage embryo transfer was associated with lower odds of clinical pregnancy after full adjustment, whereas the difference in live birth was not statistically significant. These findings suggest that D3 double high-quality cleavage-stage embryo transfer may be less favorable than D5 single high-quality blastocyst transfer in terms of early pregnancy outcomes. However, the unadjusted difference in live birth was attenuated after controlling for age, ovarian reserve, embryo availability, and endometrial factors, indicating that part of the observed difference may be attributable to patient selection and baseline prognosis. Previous studies have suggested that extended culture of thawed D3 embryos to the blastocyst stage may improve pregnancy outcomes in some patients with recurrent implantation failure, while embryo developmental stage and the endometrial preparation regimen may both influence transfer outcomes in women of advanced reproductive age ( 11 , 16 ). It should also be noted that the per-transfer-cycle risk of twin pregnancy was markedly higher after D3 double high-quality cleavage-stage embryo transfer than after D5 single high-quality blastocyst transfer. Therefore, when both D5 high-quality blastocysts and D3 high-quality cleavage-stage embryos are available, the present findings do not indicate a more favorable effectiveness–safety balance for D3 double-embryo transfer. Nevertheless, D3 high-quality cleavage-stage embryos remain a clinically usable resource for patients who do not obtain blastocysts or are not suitable candidates for extended blastocyst culture, and the present results should not be interpreted as indicating that D3 embryos have no clinical value. D3 single high-quality cleavage-stage embryo transfer was associated with lower odds of both clinical pregnancy and live birth. However, this group had a relatively small sample size, and no twin pregnancy events were observed. The less favorable outcomes may have been related to embryo developmental stage, but they may also reflect poorer embryo availability and baseline prognosis among patients selected for single D3 embryo transfer. Accordingly, these results should primarily be interpreted as observational associations rather than as evidence of a direct adverse effect of transferring a single D3 embryo. Compared with D5 single high-quality blastocyst transfer, D6 single high-quality blastocyst transfer yielded adjusted odds ratios below 1 for both clinical pregnancy and live birth; however, the 95% CIs crossed 1, and neither association reached statistical significance. No marked increase in the per-transfer-cycle risk of twin pregnancy was observed. These findings suggest that D6 single high-quality blastocyst transfer may represent an option associated with a relatively low risk of multiple pregnancy, although the available data are insufficient to determine whether its reproductive outcomes are comparable to those of D5 single high-quality blastocyst transfer. A previous age-stratified study comparing D6 high-quality blastocysts with D5 non-high-quality blastocysts likewise found no clear clinical advantage for D6 high-quality blastocysts, indicating that the day of blastocyst development and morphological quality should be evaluated jointly ( 17 ).D6 double high-quality blastocyst transfer showed no clear advantage in clinical pregnancy or live birth but was associated with a markedly increased per-transfer-cycle risk of twin pregnancy. Because this group included only 24 cycles and no such transfers occurred among women aged ≥40 years, the estimates were imprecise and accompanied by wide confidence intervals. Previous evidence has suggested that live birth outcomes after D6 blastocyst transfer in frozen-thawed cycles may be inferior to those after D5 blastocyst transfer, although the results may also be influenced by cycle type, embryo quality, and patient characteristics ( 18 ). Therefore, considerable uncertainty remains regarding the effectiveness of D6 double high-quality blastocyst transfer. The absence of a clear reproductive benefit, together with the observed increase in twin pregnancy risk, indicates that increasing the number of D6 blastocysts transferred should be approached cautiously. Sequential transfer involves transferring a D3 cleavage-stage embryo followed by a D5 or D6 blastocyst within the same treatment cycle and may theoretically increase the likelihood of synchrony between embryo development and the implantation window. In the present study, however, sequential transfer was associated with lower odds of clinical pregnancy and live birth after adjustment, while the per-transfer-cycle risk of twin pregnancy was markedly higher than that observed with D5 single high-quality blastocyst transfer. These findings do not support the routine use of sequential transfer to improve FET outcomes. Previous studies of sequential transfer have mainly focused on patients with recurrent implantation failure. Although some studies have reported improvements in clinical pregnancy outcomes, the available findings remain inconsistent, and the populations investigated differ from that of the present study ( 19 ). The less favorable outcomes observed in our cohort may reflect the more complex infertility profiles, previous treatment histories, or embryo-resource characteristics of patients selected for sequential transfer. Moreover, transferring additional embryos may increase the probability of multiple implantations without producing a corresponding improvement in the likelihood of live birth. Although the multivariable models adjusted for several potential confounders, confounding by indication cannot be completely excluded. The clinical value of sequential transfer should therefore be further evaluated in prospective studies involving patients with clearly defined indications, rather than being considered a routine alternative strategy on the basis of the present findings. This study has several limitations. First, the single-center retrospective design could not fully eliminate selection bias or confounding by indication. The transfer strategies were not randomly assigned, and residual confounding related to previous transfer history, clinician preference, and unmeasured embryo characteristics may have persisted despite multivariable adjustment. Second, some transfer strategies and age–strategy combinations included relatively few cycles. In particular, only 24 cycles involved D6 double high-quality blastocyst transfer, and no such transfers occurred among women aged ≥40 years, limiting the precision of the age-interaction analyses and age-specific estimates. Third, live birth outcomes were unknown for 341 cycles. Although a conservative sensitivity analysis was performed, loss-to-follow-up bias remains possible. Fourth, cumulative live birth, time to live birth, preterm birth, low birth weight, and other maternal and neonatal outcomes were not evaluated. Consequently, the long-term overall benefits and risks of the different strategies could not be fully assessed. Previous studies have suggested that embryo developmental stage and multiple pregnancy may be associated with perinatal outcomes ( 20 , 21 ). Further multicenter prospective studies are therefore warranted.

Clinical pregnancy, live birth, and per-transfer-cycle twin pregnancy risk differed across frozen-thawed embryo transfer strategies. D5 single high-quality blastocyst transfer generally yielded favorable reproductive outcomes and a relatively low per-transfer-cycle risk of twin pregnancy and may therefore serve as a clinically relevant reference strategy. D5 double high-quality blastocyst transfer may increase the probability of clinical pregnancy and live birth within a single transfer cycle but also substantially increases the per-transfer-cycle risk of twin pregnancy. Adding one non-high-quality usable blastocyst when a D5 high-quality blastocyst is already available did not confer a clear benefit in clinical pregnancy or live birth and was accompanied by an increased twin pregnancy risk. Compared with D5 single high-quality blastocyst transfer, D3 double high-quality cleavage-stage embryo transfer was associated with lower odds of clinical pregnancy and a higher risk of twin pregnancy. D6 double high-quality blastocyst transfer showed no clear advantage in clinical pregnancy or live birth but was associated with an increased twin pregnancy risk. Sequential transfer was associated with lower odds of clinical pregnancy and live birth and a higher risk of twin pregnancy. No statistically significant age-by-transfer-strategy interaction was identified; therefore, the present findings do not support defining fixed transfer-strategy thresholds based solely on female age. Clinical decisions should be individualized by considering baseline patient prognosis, embryo developmental stage and quality, the number of usable embryos, and the risk of multiple pregnancy.

We retrospectively screened frozen-thawed embryo transfer (FET) cycles performed at the Center for Reproductive Medicine, Qinzhou Maternal and Child Health Hospital, between January 2024 and December 2025. For patients with more than one eligible FET cycle during the study period, only the first eligible cycle was included. The inclusion criteria were as follows (1): transfer of a cryopreserved-thawed day-3 (D3) cleavage-stage embryo, day-5 (D5) blastocyst, or day-6 (D6) blastocyst (2); complete data on female age, embryo developmental stage, morphological grade, number of embryos transferred, and clinical pregnancy outcome; and (3) classification into one of the nine predefined embryo transfer strategies. The exclusion criteria were as follows (1): cycles involving preimplantation genetic testing (2); the presence of a confirmed uterine malformation (3); a history of recurrent implantation failure, defined as three or more previous failed embryo transfer attempts; and (4) missing key data on female age, embryo transfer strategy, or clinical pregnancy outcome. A total of 2,143 FET cycles were ultimately included ( Figure 1 ). Study flowchart. This retrospective cohort study was approved by the Ethics Committee of Qinzhou Maternal and Child Health Hospital (Ethics No. QSFYL〔2026〕05002). The requirement for written informed consent was waived because only de-identified data obtained during routine clinical care were retrospectively analyzed. Day-3 cleavage-stage embryos were evaluated according to the consensus criteria for morphological assessment of human embryos ( 5 ). High-quality D3 embryos were defined as embryos with ≥7 blastomeres, <20% fragmentation, and symmetrical blastomeres. Usable D3 embryos were defined as embryos with ≥4 blastomeres and <50% fragmentation. Blastocysts were evaluated using the Gardner grading system ( 6 , 7 ). High-quality blastocysts were defined as those with an expansion grade of ≥3 and with both the inner cell mass (ICM) and trophectoderm (TE) graded A or B. Usable blastocysts were defined as those with an expansion grade of ≥2 and at least one C grade for the ICM or TE. Clinical pregnancy was defined as the presence of an intrauterine gestational sac with detectable fetal cardiac activity on transvaginal ultrasonography 6–7 weeks after embryo transfer. Live birth was defined as the delivery of at least one live-born infant resulting from the index pregnancy. Cycles without pregnancy, with biochemical pregnancy, with a documented spontaneous miscarriage, or with stillbirth only were coded as no live birth. Cycles in which a clinical pregnancy was achieved but no live birth, miscarriage, stillbirth, or delivery outcome was documented were classified as having an unknown live birth outcome. A total of 341 such cycles were excluded from the primary live birth analysis. Twin pregnancy was defined as the ultrasonographic detection of two gestational sacs, both with fetal cardiac activity, during a clinical pregnancy cycle. Triplet pregnancy was defined as the ultrasonographic detection of three gestational sacs, all with fetal cardiac activity. The twin pregnancy rate among clinical pregnancies was calculated using clinical pregnancy cycles as the denominator, whereas the per-transfer-cycle risk of twin pregnancy was calculated using all transfer cycles as the denominator. Nine embryo transfer strategies were defined according to embryo developmental stage or transfer type—D3 cleavage-stage embryo, D5 blastocyst, D6 blastocyst, or sequential transfer—embryo quality—high-quality or non-high-quality usable embryo—and the number of embryos transferred—single or double transfer—as detailed in Table 1 . Female age was categorized into four groups: <30, 30–34, 35–39, and ≥40 years. Definitions and composition of the nine frozen-thawed embryo transfer strategies. D3, D5, and D6 denote days 3, 5, and 6 of embryo development, respectively. Blastocysts were graded using the Gardner scoring system; the numeral indicates the degree of expansion, and the first and second letters indicate the quality of the inner cell mass (ICM) and trophectoderm (TE), respectively. In the “1 high-quality + 1 usable” group, “usable” specifically refers to a non-high-quality usable blastocyst. In the sequential-transfer group, “two usable embryos” refers broadly to two embryos meeting the relevant transferability criteria. The total sample comprised 2,143 transfer cycles, and all morphologic examples were derived from the study database. All continuous variables were non-normally distributed and are presented as medians with first and third quartiles, median (Q1, Q3). Comparisons among groups were performed using the Kruskal–Wallis H test. Categorical variables are presented as numbers and percentages, n (%). The Pearson chi-square test was used when the sample size and expected cell counts were adequate; otherwise, Fisher’s exact test was applied. Logistic regression models were used to evaluate the associations of embryo transfer strategy with clinical pregnancy and live birth, with results reported as odds ratios (ORs), adjusted odds ratios (aORs), and 95% confidence intervals (CIs). Because twin pregnancy events were relatively uncommon and no events occurred in some transfer-strategy groups, Firth penalized logistic regression was used to analyze the risk of twin pregnancy. Potential effect modification by female age was evaluated by including an age-by-transfer-strategy interaction term in the regression models. Sensitivity analyses were performed to assess the robustness of the findings. All tests were two-sided, and a P value <0.05 was considered statistically significant. Statistical analyses were conducted using SPSS version 26.0 and R version 4.4.2.