Vitamin D is a secosteroid hormone that plays a central role in calcium homeostasis and bone metabolism. Accumulating evidence suggests that vitamin D is also involved in several biological processes relevant to female reproduction, including ovarian steroidogenesis, follicular development, endometrial receptivity, immune modulation, and implantation [1-3].

Experimental studies have shown that vitamin D modulates decidualization, cytokine balance, and steroidogenic pathways in granulosa and endometrial cells, supporting its potential relevance to implantation competence and early embryonic development [3-5]. The expression of vitamin D receptors and vitamin D-metabolizing enzymes in granulosa cells, endometrial tissue, and the placenta provides biological plausibility for a role of vitamin D in fertility and early pregnancy [1,2].

Given its role in implantation and placentation, vitamin D may influence early pregnancy events detectable only by biochemical markers, such as serum β-human chorionic gonadotropin (β-hCG). Biochemical pregnancy is an early indicator of implantation and may provide additional insight into peri-implantation effects in assisted reproduction. Accordingly, numerous observational studies have investigated associations between serum vitamin D status and in vitro fertilization (IVF) outcomes [6-12]. Several cohort studies have reported higher implantation or clinical pregnancy rates among women with sufficient serum 25-hydroxyvitamin D (25(OH)D) levels [7-10]. In particular, Yu et al. [6] reported a nonlinear positive correlation between serum vitamin D concentrations and pregnancy outcomes in a large IVF cohort. However, other studies have found no significant association, particularly after adjustment for relevant demographic and treatment-related confounders or when euploid embryos were transferred [11,12].

Recent systematic reviews and meta-analyses have suggested a possible beneficial association between vitamin D sufficiency and IVF outcomes, although effect estimates vary across studies [13-15]. However, subsequent comprehensive reviews and well-adjusted models have yielded conflicting findings, often reporting that positive associations are attenuated or no longer statistically significant after adjustment for baseline clinical covariates. Substantial heterogeneity in study design, ethnic background, seasonal variation, vitamin D assay methods, and definitions of vitamin D deficiency has limited definitive conclusions [14,15] and contributed to persistent uncertainty about the clinical relevance of vitamin D status in IVF practice. Moreover, no universal consensus exists regarding the optimal serum 25(OH)D threshold for reproductive outcomes, and some authors have proposed potential nonlinear or threshold-dependent associations [1,6]. Despite statistical adjustment in prior studies, residual confounding and heterogeneity across treatment cycles may still influence observed associations.

Methodological differences across studies may further contribute to conflicting findings. Many previous investigations included multiple IVF cycles per woman, potentially introducing bias related to cumulative treatment exposure and cycle-dependent clinical decision-making [8,11]. In addition, inadequate adjustment for IVF-specific procedural factors, such as embryo transfer day and anti-Müllerian hormone (AMH), may confound observed associations between vitamin D status and reproductive outcomes [10,11]. Cohort analyses published in Reproductive BioMedicine Online reported no significant association between vitamin D deficiency and clinical pregnancy outcomes after IVF or intracytoplasmic sperm injection (ICSI) treatment [12], underscoring the importance of rigorous methodology when evaluating the clinical relevance of vitamin D in assisted reproduction. This pattern is further supported by recent large-scale cohorts and systematic reviews, including those by Antunes et al. [7] and Meng et al. [13], which reported null or attenuated associations after extensive multivariable adjustment. However, even large registry-based studies may retain clinical heterogeneity because they include repeated cycles and embryos of varying quality, leaving residual confounding related to embryo competence incompletely addressed. To address these limitations, the present study restricted the analysis to each patient’s first IVF cycle and included only high-quality embryo transfers. This approach was intended to reduce heterogeneity related to cumulative treatment exposure and embryo competence and to better isolate associations between the maternal vitamin D environment and reproductive outcomes. In addition, this relationship was evaluated in a Korean population with a high baseline prevalence of vitamin D deficiency; recent national surveys indicate that more than 80% to 90% of young reproductive-aged women fall below the sufficiency threshold [16]. Therefore, this study aimed to investigate whether pretreatment serum vitamin D status was independently associated with reproductive outcomes, including biochemical and clinical pregnancy, after rigorous adjustment for relevant confounders in this highly controlled cohort.

In this study, the initially observed association between vitamin D status and clinical pregnancy was attenuated and was no longer statistically significant after adjustment for confounders. This finding suggests that the apparent positive association in unadjusted analyses may not reflect an independent biological effect. Additional adjustment for seasonal variation did not materially change the findings. By restricting the analysis to first IVF cycles with high-quality embryo transfer, we minimized key sources of heterogeneity related to embryo competence and treatment exposure.

Maternal age remained the strongest independent predictor of clinical pregnancy, consistent with established determinants of IVF success and cumulative live birth outcomes [17]. To further evaluate potential confounding due to the baseline imbalance in maternal age, we performed an additional age-stratified analysis using the clinical threshold of 35 years. Consistent with our primary findings, no statistically significant association between vitamin D status and clinical pregnancy was observed in either the cohort under 35 years (aOR, 2.33; 95% CI, 0.60 to 9.03; p=0.220) or the cohort aged 35 years or older (aOR, 3.33; 95% CI, 0.66 to 16.85; p=0.145). Although both point estimates remained positive, the associations did not reach statistical significance within individual strata, likely because of the limited sample size and reduced statistical power in this sub-analysis. In addition, when serum 25(OH)D was evaluated as a continuous variable as an alternative analytical approach, the results remained consistent with the primary findings, showing no statistically significant association with clinical pregnancy in either the unadjusted or multivariable model. Therefore, although the present study focused on early pregnancy outcomes in a highly controlled cohort, a follow-up analysis of cumulative live birth rates and long-term neonatal outcomes is planned to further evaluate the clinical relevance of baseline vitamin D status.

We additionally evaluated biochemical pregnancy as an early indicator of implantation. No significant association was observed between serum vitamin D status and biochemical pregnancy in either crude or adjusted analyses. This lack of association suggests that vitamin D status may have limited influence on early implantation in this cohort. Notably, the study population consisted of Korean women, a group reported to have among the highest rates of vitamin D deficiency globally. Despite the high prevalence of insufficiency, the lack of a significant association in this well-controlled cohort suggests that baseline vitamin D status may have a smaller-than-expected impact on assisted reproductive technology outcomes in this specific population. Alternatively, this finding may reflect a floor effect. Because baseline 25(OH)D concentrations were low across most of the cohort, limited between-person variation may have masked biological differences between groups and precluded detection of a clear clinical threshold. Together, these findings suggest that the apparent positive signals observed in unadjusted analyses may be partly explained by residual confounding rather than a direct biological effect.

Our findings are consistent with previous cohort studies and meta-analyses reporting no significant association after adjustment [13-15,18,19]. Recent high-powered meta-analyses and cohorts, including those by Chu et al. [15] and Eller et al. [19], similarly showed that the crude clinical association between vitamin D and reproductive outcomes often diminishes or disappears after key procedural covariates are controlled. However, although these large studies achieved substantial statistical power through large sample sizes, they also pooled clinically heterogeneous populations by including repeated cycle attempts and embryos of varying quality.

In contrast, earlier reports suggesting improved pregnancy rates among vitamin D-replete women often included heterogeneous treatment populations or limited adjustment for IVF-specific confounders [20]. Differences in study design, inclusion of repeated IVF cycles, and cumulative treatment exposure may partially explain these discrepancies. Restricting our analysis to first IVF cycles minimized bias related to protocol modification and cycle-dependent clinical decision-making [17].

Although clinical pregnancy is frequently reported, live birth is the most clinically meaningful endpoint in assisted reproduction. Live birth data were not yet available for the most recent treatment cycles included in this cohort. Recent large cohort studies evaluating both fresh and frozen embryo transfer cycles have produced mixed findings regarding associations between vitamin D status and live birth rates [21,22]. Some studies have reported modest associations, whereas others have found attenuation after multivariable adjustment, suggesting that any influence of vitamin D may be subtle and context dependent rather than a robust independent predictor of reproductive success. Differences between fresh and frozen embryo transfer cycles may influence implantation dynamics, although findings remain inconsistent [22].

Embryo transfer day differed between vitamin D groups in unadjusted comparisons but was not independently associated with clinical pregnancy after adjustment. This observation highlights the need to consider IVF-specific procedural factors, such as embryo transfer timing, when evaluating adjunctive biomarkers in reproductive medicine [23].

Intervention studies have not demonstrated consistent improvements in clinical pregnancy or live birth rates [24]. These findings suggest that although vitamin D may influence biological processes relevant to implantation, such effects may not translate into measurable differences in overall IVF success rates. The potential effect of postscreening supplementation should also be considered. In contemporary clinical practice, patients identified as vitamin D-deficient are often prescribed supplementation promptly. Although postscreening supplementation data were not available in this retrospective study, our results reflect real-world clinical outcomes in which baseline deficiency is identified and may be managed. The absence of inferior outcomes in the deficient group despite this common practice suggests that either the impact of baseline deficiency is clinically modest or routine clinical management may sufficiently reduce any potential risk.

Associations between vitamin D status and ovarian response parameters have also been explored. Some studies have evaluated fertilization rates, follicular-fluid vitamin D concentrations, and embryo development markers, with inconsistent results [25]. Overall, correlations between serum vitamin D and ovarian response appear modest at best, underscoring the multifactorial determinants of IVF outcomes and the limited predictive value of isolated biochemical markers.

Although vitamin D has biological roles in reproductive tissues, translation into clinical benefit remains uncertain [3-5]. Ethnic variation in vitamin D metabolism, particularly in East Asian populations with widespread deficiency, and seasonal fluctuations in serum 25(OH)D concentrations represent additional sources of variability [14,15]. Furthermore, the impact of vitamin D may differ across specific clinical subgroups. For example, studies in women with polycystic ovary syndrome (PCOS) have suggested subgroup-specific associations between vitamin D status and reproductive outcomes [26]. A recent systematic review in Clinical and Experimental Reproductive Medicine also highlighted the potential role of vitamin D in improving fertility parameters specifically among patients with PCOS [27]. In the present cohort, patients with a documented diagnosis of PCOS accounted for only 2.3% (n=3) of the total study population. Although this limited subgroup size precluded formal stratified analysis, acknowledging this baseline characteristic is important for contextualizing the findings. These factors may partially explain interstudy inconsistency and highlight the need for standardization in future research.

Several limitations should be acknowledged. First, the retrospective design limits causal inference. Second, the relatively small sample size (n=107) and number of events may have reduced statistical power for subgroup estimates and precluded formal propensity-score matching. Multivariable analyses were performed using complete-case data because AMH levels were missing for some patients, which may have introduced bias. Third, and most notably, unmeasured postscreening vitamin D supplementation between baseline screening and subsequent embryo transfer may have confounded the findings. Because patients identified as deficient may have independently initiated over-the-counter supplementation or multivitamins during the treatment cycle, postbaseline changes in vitamin D status cannot be entirely excluded and could have obscured the true impact of baseline deficiency. Nevertheless, because vitamin D is a fat-soluble hormone precursor that may require prolonged accumulation to exert cellular and genomic effects, short-term supplementation during the active treatment cycle may not have produced immediate changes in oocyte maturation or endometrial receptivity. Fourth, the absence of maternal body mass index (BMI) data is an additional limitation because BMI may influence both baseline vitamin D status and clinical pregnancy outcomes. Finally, as a single-center study, generalizability may be limited.

Despite these limitations, the study has notable strengths, including restriction to first IVF cycles with high-quality embryo transfer, adjustment for key demographic and IVF-related confounders, and evaluation of both biochemical and clinical pregnancy outcomes. The concordant null findings across early and later reproductive endpoints support internal consistency. Given the widespread screening and supplementation of vitamin D in reproductive practice, these findings suggest that routine correction of vitamin D deficiency solely to improve IVF outcomes should be interpreted with caution pending stronger prospective evidence.

In conclusion, the apparent association between vitamin D status and pregnancy outcomes was attenuated after adjustment for confounders, suggesting that the current data do not demonstrate vitamin D status as an independent predictor of IVF success in this cohort. However, given the limited sample size of the sufficient group and the wide CIs, a clinically relevant association cannot be definitively excluded. Previously reported positive associations in less controlled studies may partly reflect residual confounding rather than a direct biological effect on implantation. Maternal age remained the principal determinant of reproductive outcomes. Larger prospective studies with adequate statistical power, standardized vitamin D assessment, and comprehensive IVF procedural data are warranted to further clarify the role of vitamin D in assisted reproduction.