Dydrogesterone versus levonorgestrel-releasing intrauterine system for abnormal uterine bleeding: a propensity score-matched analysis of long-term efficacy and recurrence

Objective To compare dydrogesterone versus levonorgestrel-releasing intrauterine system (LNG-IUS) for abnormal uterine bleeding (AUB) management with long-term follow-up and etiology-stratified analysis.

This single-center retrospective cohort study analyzed reproductive-aged women with International Federation of Gynecology and Obstetrics (FIGO)-classified AUB receiving dydrogesterone (n = 131) or LNG-IUS (n = 118). Propensity score matching balanced baseline covariates between groups (n = 104 pairs). Primary outcomes were bleeding control rates (composite: hemoglobin increase ≥ 1.5 g/dL, ≥ 50% Pictorial Blood Loss Assessment Chart [PBAC] reduction, no rescue interventions) at 3 and 12 months. Secondary outcomes included 24-month recurrence and adverse events (AEs).

LNG-IUS demonstrated superior 3-month bleeding control versus dydrogesterone (82.7% vs. 65.4%; relative risk [RR] 1.26, 95% confidence interval [CI] 1.08–1.48; p = 0.002). At 12 months, bleeding control rates were comparable between groups (78.8% vs. 72.1%; RR 1.09, 95% CI 0.96–1.24; p = 0.17). LNG-IUS significantly reduced 24-month recurrence risk (hazard ratio [HR] 0.48, 95% CI 0.32–0.72; p < 0.001), particularly for structural AUB (HR 0.39, 95% CI 0.24–0.64). Subgroup analysis revealed LNG-IUS superiority in adenomyosis at 12 months (88.2% vs. 60.0%; RR 1.47, 95% CI 1.24–1.74; p < 0.001) but comparable efficacy in nonstructural AUB (76.6% vs. 72.9%; RR 0.95, 95% CI 0.78–1.16; p = 0.63). Dydrogesterone exhibited higher systemic AEs (28.2% vs. 13.6%; p = 0.003), while LNG-IUS had device-related events including expulsion (5.1%).

LNG-IUS offers superior short-term bleeding control and sustained long-term recurrence prevention, especially for structural AUB. Dydrogesterone effectively manages nonstructural etiologies but carries higher systemic adverse event rates. Treatment selection should be etiology-guided, balancing immediate symptom control versus recurrence prevention and individualized safety considerations.

Abnormal uterine bleeding (AUB) represents a prevalent gynaecological disorder affecting up to one-third of women during their reproductive years, with significant implications for quality of life, healthcare resource utilisation, and economic productivity [Citation1]. The condition accounts for up to 70% of all gynaecological consultations and imposes a substantial economic burden, with direct annual healthcare costs estimated to be in the tens of billions of dollars in the United States alone [Citation2]. The FIGO PALM-COEIN classification system standardised AUB aetiology diagnosis, highlighting diverse causes including structural anomalies (Polyps, Adenomyosis, Leiomyoma, Malignancy/hyperplasia) and non-structural disorders such as (Coagulopathy, Ovulatory dysfunction, Endometrial disorders, Iatrogenic causes, or Not yet classified) [Citation3]. Among these, uterine leiomyomas (AUB-L) and ovulatory dysfunction (AUB-O) are the most frequently identified causes [Citation4]. The primary goals of AUB management are to control bleeding, restore quality of life, and address the underlying aetiology. While progestogens serve as cornerstone therapies, the treatment landscape is broad. Other first-line medical interventions include non-steroidal anti-inflammatory drugs (NSAIDs), antifibrinolytics like tranexamic acid, and combined oral contraceptives (COCs), particularly for non-structural causes. When medical management is ineffective, contraindicated, or in the presence of significant structural anomalies, procedural interventions such as endometrial ablation, myomectomy, or hysterectomy are considered [Citation5]. First-line management typically involves medical interventions, with progestogens serving as cornerstone therapies. The levonorgestrel-releasing intrauterine system (LNG-IUS) offers localised endometrial suppression with minimal systemic effects, while oral progestins like dydrogesterone provide systemic endocrine modulation [Citation6]. Despite established guidelines, optimal therapeutic selection remains challenging. LNG-IUS demonstrates high efficacy in reducing menstrual blood loss and is recommended for long-term management, yet practical barriers include insertion discomfort, expulsion risks, and contraindications such as uterine anomalies [Citation7]. Conversely, oral dydrogesterone is widely accessible and avoids procedural interventions, but concerns persist regarding long-term adherence, systemic side effects, and comparative effectiveness against intrauterine options [Citation8]. Existing comparative studies are often limited by short follow-up durations, heterogeneous populations, or insufficient control for confounding factors, particularly regarding structural versus non-structural AUB subtypes [Citation9]. Current evidence gaps include a lack of real-world comparisons using rigorous methodologies to address selection bias, and limited data on long-term outcomes beyond 12 months, such as recurrence patterns and therapeutic failure requiring escalation [Citation10,Citation11]. Furthermore, authoritative reviews indicate that the influence of specific FIGO-AUB etiologies on treatment response remains inadequately characterised, highlighting a need for more comparative studies [Citation12]. This study addresses these gaps through a single-centre retrospective cohort analysis employing propensity score matching to minimise confounding. It directly compares dydrogesterone and LNG-IUS over an extended follow-up period exceeding 12 months, incorporating detailed aetiology-based subgroup analyses and comprehensive safety assessment.

Study design This single-centre retrospective cohort study utilised anonymized electronic medical records (EMR) from Beijing Huairou Hospital to compare the real-world efficacy and safety of dydrogesterone versus the LNG-IUS in managing AUB, with a parallel comparison to a historical cohort receiving standard medical therapy. Data from consecutively treated patients between January 1, 2018, and December 31, 2023, were analysed. The study protocol received approval from the Beijing Huairou Hospital Institutional Review Board (IRB Approval No: 20240416), and a waiver of informed consent was granted due to the anonymized retrospective nature of the data analysis, consistent with local regulations, the Declaration of Helsinki, and institutional privacy protection protocols. Patient selection Inclusion criteria Patients were eligible if they met all the following: (1) Reproductive-aged women (18–50 years); (2) Confirmed diagnosis of AUB according to FIGO PALM-COEIN classification criteria [Citation3]; (3) Initiation of index treatment (dydrogesterone, LNG-IUS, or standard medical therapy) between January 2018 and December 2023; (4) Minimum follow-up duration of 6 months with documented clinical outcomes; (5) Availability of complete baseline data (haemoglobin levels, transvaginal ultrasound, Pictorial Blood Loss Assessment Chart [PBAC] score) and ≥ 1 follow-up assessment at 3 and 12 months. Exclusion criteria Patients were excluded for any of the following: (1) Malignant gynaecological pathology (endometrial, cervical, or ovarian cancer confirmed histopathologically); (2) Severe hepatic impairment (Child-Pugh class B/C) or renal dysfunction (eGFR < 30 mL/min/1.73 m²) [Citation13,Citation14]; (3) Contraindications to LNG-IUS (uterine anomalies, active pelvic infection, unexplained vaginal bleeding, or acute liver disease per manufacturer guidelines [Bayer AG. Mirena®]); (4) Current anticoagulant therapy or diagnosed coagulopathy; (5) Pregnancy, lactation, or planned conception within 12 months; (6) Prior endometrial ablation or hysterectomy. Cohort assembly and sample size The cohort was derived from systematic EMR queries at Beijing Huairou Hospital using International Classification of Diseases, Tenth Revision (ICD-10) codes for AUB (N92.0, N92.1, N93.8, N93.9) [Citation15], procedure codes for LNG-IUS insertion, and pharmacy dispensing records for dydrogesterone and standard therapies (combined oral contraceptives [COCs]: G03AA07, G03AB; progestins: G03DB04). Sample size was calculated a priori using PASS 15.0, targeting 110 patients per treatment arm to detect a 15% difference in bleeding control rates (α = 0.05, power = 90%), accounting for 20% attrition. The historical control group (2015–2017) was assembled using identical selection criteria. The rationale for employing a historical control from an earlier period was twofold: (1) To provide a benchmark representing the standard-of-care medical therapy (COCs or cyclic MPA) prevalent before the widespread adoption and direct comparison of dydrogesterone and LNG-IUS in our clinical practice; (2) To circumvent the substantial selection bias that would arise if a concurrent standard-therapy group were constructed within the 2018–2023 period. Patients eligible for but not receiving either of the two index treatments during this later period would likely represent a distinct population with contraindications, strong preferences, or access issues, making them a non-comparable control for the primary PSM analysis. Grouping Initial screening identified 883 potential candidates (2018–2023). After applying inclusion/exclusion criteria, 305 patients qualified for the primary analysis (dydrogesterone: n = 131; LNG-IUS: n = 118; historical controls: n = 56): (1) Dydrogesterone group: Cyclic oral dydrogesterone (10 mg twice daily from day 11 to day 25 of each menstrual cycle for a total of 14 days); (2) LNG-IUS group: Levonorgestrel-releasing intrauterine system (52 mg, 20 μg/day); (3) Historical control group: Standard medical therapy (combined oral contraceptives containing 30 μg ethinylestradiol/150 μg levonorgestrel or cyclic medroxyprogesterone acetate 10 mg/day). Propensity score matching (PSM) implementation To address confounding bias in non-randomised treatment allocation, PSM was applied between dydrogesterone and LNG-IUS groups (1:1 ratio). Propensity scores were derived via logistic regression incorporating clinically relevant covariates: age, body mass index (BMI), FIGO-AUB subtype (structural vs. non-structural) [Citation3], baseline haemoglobin, endometrial thickness, and PBAC score. Nearest-neighbour matching with a caliper width of 0.2 SD achieved balance (standardised mean differences < 0.1 for all covariates) [Citation16], yielding 104 matched pairs (n = 208) for primary efficacy analysis. Historical controls were analysed separately with covariate adjustment. Given the temporal gap, we employed inverse probability weighting (IPW) to adjust for potential temporal confounders between the historical (2015−2017) and contemporary (2018−2023) cohorts. The IPW model incorporated variables capturing temporal shifts, including treatment epoch (2015−2017 vs. 2018−2023), evolution in diagnostic practices (e.g. standardised adoption of the FIGO PALM-COEIN [Citation3] classification), and availability of advanced imaging modalities, aiming to balance the cohorts on these factors. Efficacy and safety assessment Primary efficacy endpoint The primary endpoint was bleeding control effectiveness at 3 and 12 months, defined using a composite outcome adapted from FIGO recommendations [Citation3]: (1) Objective response: Haemoglobin (Hb) elevation ≥ 1.5 g/dL from baseline (confirmed by two consecutive tests within ± 14 days); (2) Symptomatic response: ≥ 50% reduction in PBAC score or absolute PBAC < 75 [Citation17]; (3) Clinical response: Absence of rescue interventions (hormonal therapy escalation, surgical procedures, or blood transfusion). Complete response (CR) required meeting all three criteria; partial response (PR) met two criteria; non-response (NR) met ≤ 1 criterion. Effectiveness rate = (CR + PR) cases/total evaluable cases. PBAC scores were extracted from structured nursing notes documenting pictorial chart completion during clinic visits. Haemoglobin values used the last measurement within ±14 days of each timepoint (Sysmex XN-550 analyser, uniform calibration). Secondary efficacy endpoints Secondary outcomes included: (1) Haemoglobin recovery rate: Proportion achieving Hb > 12 g/dL at 12 months (WHO anaemia threshold) [Citation18]; (2) Endometrial thickness (EMT) dynamics: Maximal EMT reduction on transvaginal ultrasound (TVUS), measured by two blinded radiologists using standardised FIGO protocol [Citation3]; (3) Recurrence rate: Defined as therapeutic re-escalation (e.g. added medications, LNG-IUS replacement < 5 years) or unscheduled surgery after initial success; (4) Patient-reported outcomes: EuroQol 5-Dimension 5-Level (EQ-5D-5L) utility scores from validated Chinese questionnaires [Citation19]. All efficacy analyses adjusted for propensity score covariates. Baseline data collection Standardised baseline covariates were systematically captured within 7 days preceding treatment initiation, encompassing demographic parameters (age, body mass index, parity, smoking status), clinical characteristics including FIGO-AUB subtype validated by transvaginal ultrasound or hysteroscopy, baseline haemoglobin levels measured via ISO 15189:2012-accredited Sysmex XN-550 analysers [Citation20], and endometrial thickness derived from triplicate measurements; bleeding severity was quantified using clinician-verified Pictorial Blood Loss Assessment Chart scores, while comorbidities were classified per ICD-10 codes for hypertension (I10), diabetes (E11), and coagulopathies (D68.9), alongside documentation of prior failed medical therapies and endometrial biopsy results, with 87% of variables sourced from structured electronic medical record fields and 13% from scanned clinical documents undergoing double-entry validation. Safety evaluation Adverse events (AEs) were captured through four complementary sources to minimise underreporting bias: (1) Clinician-documented AEs in progress notes coded per Medical Dictionary for Regulatory Activities (MedDRA) v25.0 [Citation21]; (2) System-triggered alerts for laboratory abnormalities (ALT/AST > 3 × ULN [Citation13], serum creatinine > 1.5 × baseline [Citation14]); (3) Treatment discontinuation records with AE attribution; (4) Emergency visits flagged by ICD-10 codes for treatment-related complications (e.g. T38.5x5A). Severity was graded via CTCAE v5.0 [Citation22], with causality assessed using WHO-UMC criteria [Citation23]. Data quality control To address retrospective data limitations, rigorous validation protocols were implemented: PBAC scores required EMR-embedded photographic documentation; TVUS images underwent centralised re-measurement for 20% random sample (inter-rater κ = 0.91); haemoglobin trends mandated ≥ 2 concordant measurements; AE ascertainment cross-verified against pharmacy discontinuation logs and patient hotline records; all endpoint adjudications were performed by an independent committee blinded to treatment allocation. Statistical analysis The primary comparative analyses utilised the propensity score-matched cohort (dydrogesterone vs. LNG-IUS groups), employing McNemar’s test for categorical efficacy endpoints (bleeding control rate) and paired t-tests for continuous variables (haemoglobin change, endometrial thickness reduction), with Kaplan-Meier methodology applied to compare 24-month recurrence-free survival (events defined as therapeutic re-escalation or unscheduled surgery) using log-rank testing stratified by FIGO-AUB subtype; multivariable logistic regression incorporating all baseline covariates (age, BMI, FIGO classification, baseline PBAC, and treatment group) identified independent predictors of treatment response, while safety analyses employed Fisher’s exact test for adverse event incidence comparisons across unmatched cohorts, with prespecified subgroup analyses for aetiology-specific effects and sensitivity analyses using alternative matching algorithms (kernel matching, caliper = 0.1), all conducted with SPSS 26.0, two-sided p-values < 0.05 were considered statistically significant.

Baseline characteristics and matching efficacy Following systematic screening of 894 EMR, 305 patients met eligibility criteria (dydrogesterone: n = 131; LNG-IUS: n = 118; historical controls: n = 56), with primary exclusions due to insufficient follow-up (n = 312), malignant pathology (n = 58), LNG-IUS contraindications (n = 46), and incomplete baseline data (n = 173); propensity score matching (1:1) generated 104 well-matched pairs (n = 208) with all baseline covariates achieving standardised mean differences < 0.08, confirming optimal balance, while the historical control group demonstrated comparable characteristics after inverse probability weighting adjustment (). Primary efficacy outcomes in the PSM cohort In the PSM cohort (104 pairs, n = 208), the levonorgestrel-releasing intrauterine system (LNG-IUS) demonstrated significantly higher bleeding control rates than dydrogesterone at 3 months (82.7% [86/104] vs. 65.4% [68/104]; RR 1.26, 95% CI 1.08−1.48, p = 0.002), with this advantage attenuating by 12 months (78.8% [82/104] vs. 72.1% [75/104]; RR 1.09, 95% CI 0.96−1.24, p = 0.17). Objective haemoglobin improvement (≥1.5 g/dL) and subjective PBAC reduction (≥50%) consistently favoured LNG-IUS at both timepoints. Endometrial thickness reduction was comparable between groups at 3 months (mean difference –0.2 mm, 95% CI –0.45 to 0.05; p = 0.12) and 12 months (mean difference –0.2 mm, 95% CI –0.49 to 0.09; p = 0.18) (, Supplementary figure 1). Historical control sensitivity analysis with inverse probability weighting After inverse probability weighting adjustment for temporal covariates (treatment era, diagnostic criteria evolution), the historical control group (standard medical therapy, n = 56) demonstrated significantly lower bleeding control rates compared to both contemporary interventions at 3 months (48.2% [27/56] vs. dydrogesterone: RR 1.36, 95% CI 1.08−1.71, p = 0.009; vs. LNG-IUS: RR 1.71, 95% CI 1.32−2.23, p < 0.001), with this disadvantage persisting at 12 months (52.7% vs. 72.1%/78.8%, both p < 0.01), confirming the robustness of contemporary treatment superiority after accounting for temporal confounders (). Multivariable logistic regression analysis for treatment response predictors Multivariable logistic regression incorporating all treatment groups and baseline covariates identified the levonorgestrel-releasing intrauterine system (LNG-IUS) as the strongest independent predictor for 12-month bleeding control (aOR 3.28, 95% CI 2.15−5.01, p < 0.001), followed by non-structural AUB aetiology (aOR 2.17, 95% CI 1.42−3.32, p 10.5 g/dL (aOR 1.89, 95% CI 1.24−2.88, p = 0.003), while dydrogesterone showed intermediate efficacy (aOR 1.76 vs. historical controls, 95% CI 1.18−2.63, p = 0.006), with age ≥ 40 years and adenomyosis diagnosis negatively impacting response (Supplementary tables 1, 2). Subgroup analysis by aetiology and age stratification Stratified analysis of the PSM cohort (n = 208) revealed significant treatment-by-aetiology interactions: LNG-IUS demonstrated superior bleeding control versus dydrogesterone in structural AUB at both 3 months (85.7% vs. 61.4%, RR 1.40, p = 0.003) and 12 months (83.9% vs. 68.4%, RR 1.23, p = 0.03), particularly for adenomyosis (AUB-A: 88.2% vs. 60.0%, p 0.30). Age stratification identified enhanced LNG-IUS benefits in women ≥ 40 years (RR 1.35 at 12 months, p = 0.01), while dydrogesterone performed better in younger patients with ovulatory dysfunction (AUB-O: RR 0.92, p = 0.61) (Supplementary tables 3, Supplementary figure 2). Safety analysis in the full cohort Safety assessment of the full treatment-naïve cohort (n = 305) identified significantly higher treatment-related adverse events (AEs) with dydrogesterone (28.2% [37/131]) versus LNG-IUS (13.6% [16/118], p = 0.003) and historical controls (19.6% [11/56], p = 0.18), primarily driven by progestogenic side effects: breast tenderness (15.3% vs. 5.1%), headache (12.2% vs. 3.4%), and mood disturbances (8.4% vs. 1.7%). LNG-IUS exhibited higher device-related AEs (expulsion: 5.1%, pelvic pain: 6.8%), while severe AEs (CTCAE ≥ 3) were rare across groups (<3%) (Supplementary tables 4). Long-term recurrence analysis with Kaplan-Meier estimates In the PSM cohort (n = 208), LNG-IUS demonstrated significantly superior recurrence-free survival compared to dydrogesterone over the 24-month follow-up period (log-rank p < 0.001). The cumulative recurrence rates were 18.3% (19/104) for LNG-IUS versus 34.6% (36/104) for dydrogesterone (HR 0.48, 95% CI 0.32−0.72). Time-to-event curves showed progressive divergence starting at 6 months, with dydrogesterone exhibiting substantially higher recurrence rates across all timepoints: 8.7% (9/104) vs 3.8% (4/104) at 6 months (HR 0.43, 95% CI 0.22−0.85, p = 0.01), and 19.2% (20/104) vs 7.7% (8/104) at 12 months (HR 0.40, 95% CI 0.24−0.67, p < 0.001). Multivariable Cox regression confirmed LNG-IUS as an independent protective factor after adjusting for age, BMI, FIGO-AUB subtype, and baseline haemoglobin (aHR 0.41, 95% CI 0.27−0.63). Subgroup analyses revealed pronounced benefits in structural AUB (HR 0.39, 95% CI 0.24−0.64), particularly for adenomyosis. Following recurrence, dydrogesterone-treated patients more frequently required surgical interventions (55.6% vs 31.6%, p = 0.03), while LNG-IUS patients predominantly received medical retreatment (63.2% vs 41.7%, p = 0.04) (Supplementary tables 5, Supplementary figure 3).

This retrospective cohort study utilising propensity score matching demonstrates distinct therapeutic profiles for dydrogesterone and the LNG-IUS in managing AUB. Our primary findings indicate significantly superior short-term (3-month) bleeding control with LNG-IUS compared to dydrogesterone (82.7% vs. 65.4%), consistent with the potent local endometrial suppression exerted by continuous intrauterine levonorgestrel delivery. This rapid efficacy aligns with randomised trials showing LNG-IUS achieves faster amenorrhoea rates than oral progestins, attributable to direct endometrial effects bypassing systemic metabolism [Citation24]. The attenuation of this advantage by 12 months (78.8% vs. 72.1%, p = 0.17) suggests progressive therapeutic benefit with dydrogesterone over time, possibly reflecting cumulative endometrial stabilisation or patient adaptation to therapy. This finding contrasts with some shorter-term studies reporting sustained LNG-IUS superiority but parallels real-world analyses indicating oral progestins can achieve comparable control beyond 6 months, particularly in non-structural AUB [Citation6]. Crucially, our analysis revealed significant treatment-by-aetiology interactions. LNG-IUS demonstrated pronounced superiority in structural AUB, particularly adenomyosis (AUB-A), where 12-month control rates were 88.2% versus 60.0% for dydrogesterone. This likely stems from the LNG-IUS delivering a high local concentration of levonorgestrel, which induces profound endometrial decidualization and atrophy. This high local dose is crucial for overcoming the relative progesterone resistance often observed in adenomyotic tissue. Mechanistically, LNG suppresses endometrial cell proliferation, promotes apoptosis, and significantly down-regulates key angiogenic factors like VEGF and inflammatory mediators such as COX-2 [Citation25], mechanisms less robustly addressed by cyclic oral progestins [Citation26]. Conversely, efficacy was comparable between treatments for non-structural AUB (76.6% vs. 72.9%), especially ovulatory dysfunction (AUB-O). This suggests dydrogesterone, acting via central and peripheral progesterone receptor modulation to restore endocrine balance, may suffice for functional disorders without anatomical pathology [Citation27]. Specifically, by providing predictable negative feedback on the hypothalamic-pituitary-ovarian axis, oral progestins regulate uncontrolled oestrogenic effects and transform the proliferative endometrium into a secretory one, leading to organised withdrawal bleeding and cycle regulation [Citation28]. Furthermore, dydrogesterone showed relatively better performance in women under 40 years, potentially reflecting differing aetiological distributions or hormonal milieu influencing treatment response. This observation likely reflects two interconnected factors. First, the aetiological distribution of AUB varies with age; younger women more commonly present with ovulatory dysfunction (AUB-O), for which systemic hormonal regulation is highly effective. In contrast, the prevalence of structural pathologies like adenomyosis, where LNG-IUS is superior, increases with age. Second, the hormonal milieu itself may play a role; younger women may exhibit greater endometrial sensitivity to progestogens, whereas progressive progesterone resistance can develop in older women, particularly in the context of underlying structural disease [Citation29,Citation30]. This age-specific effect warrants consideration in therapeutic selection. The long-term recurrence analysis provided critical insights: LNG-IUS significantly reduced recurrence risk over 24 months (HR 0.48), particularly for structural AUB (HR 0.39). This sustained protection aligns with the established 5-year efficacy profile of LNG-IUS and its ability to induce profound endometrial quiescence [Citation31]. The progressive divergence in recurrence-free survival curves, starting at 6 months, underscores LNG-IUS as a preferred strategy for preventing disease progression and reducing subsequent interventions. Notably, recurrence in the dydrogesterone group more frequently necessitated surgical management (55.6% vs. 31.6%), highlighting a potential clinical consequence of oral therapy failure. This finding reinforces LNG-IUS as a first-line option for patients seeking long-term solutions or at high risk of disease progression, such as those with adenomyosis [Citation32]. Safety profiles differed substantially. Dydrogesterone was associated with higher overall treatment-related adverse events (28.2% vs. 13.6%), primarily systemic progestogenic effects like breast tenderness and headache, consistent with its oral administration and pharmacokinetics [Citation33]. While LNG-IUS had lower systemic side effects, it carried device-specific risks like expulsion (5.1%) and pelvic pain (6.8%), corroborating real-world safety surveillance data [Citation34]. These divergent safety considerations necessitate personalised risk-benefit discussions, particularly for patients with contraindications or preferences regarding device placement. These findings have direct implications for clinical practice, enabling a more nuanced and evidence-based approach to individualising AUB therapy. For patients with structural AUB, particularly adenomyosis, and for those whose primary therapeutic goals are rapid bleeding control and long-term prevention of recurrence and surgical intervention, the LNG-IUS should be strongly recommended as a first-line option. Conversely, for patients with non-structural AUB (e.g. ovulatory dysfunction), especially younger women, and for those who prioritise avoiding a procedural intervention or have contraindications to an intrauterine device, dydrogesterone represents an effective oral alternative, albeit with a higher likelihood of systemic side effects and a greater need for vigilance regarding potential recurrence. The choice between these therapies should thus be guided by a shared decision-making process that carefully weighs the specific AUB aetiology, the patient’s age, the urgency of symptom control versus the importance of long-term disease modification, and individual tolerance for different risk profiles (systemic side effects vs. device-related events). The comparison with historical controls receiving standard medical therapy (combined oral contraceptives or cyclic medroxyprogesterone acetate) confirmed the superiority of both contemporary interventions. After inverse probability weighting adjustment, bleeding control rates were significantly lower in historical controls at both 3 and 12 months. This reinforces the clinical value of both dydrogesterone and LNG-IUS over older progestin regimens, potentially reflecting better tolerability and adherence profiles or optimised dosing strategies [Citation35,Citation36]. The age range of 18–50 years was specified for this study to define a patient population within the primary reproductive lifespan. This approach aimed to create a more homogeneous cohort for comparison by excluding adolescents, in whom the hypothalamic-pituitary-ovarian axis may be immature [Citation37], and perimenopausal women, who exhibit a different hormonal environment and etiologies for AUB [Citation38]. The exclusion of these groups was intended to minimise potential confounding variables and reduce outcome heterogeneity. Our findings contribute novel evidence through several methodological strengths. This is the first study employing propensity score matching to directly compare dydrogesterone and LNG-IUS over an extended follow-up (>12 months), mitigating confounding inherent in retrospective designs. The inclusion of detailed aetiology-based subgroup analyses provides granularity often missing in trials, offering clinically actionable insights for personalised therapy. Furthermore, the integration of long-term recurrence data and comprehensive safety assessment from multiple sources enhances the real-world applicability of our findings. The inclusion of a historical control group receiving standard medical therapy (COCs/cyclic MPA) from an earlier period (2015−2017) served a contextual rather than a primary comparative purpose. Its main relevance lies in demonstrating that both dydrogesterone and the LNG-IUS, when compared after rigorous adjustment for temporal trends via IPW, were associated with significantly superior bleeding control rates compared to the earlier standard regimen. This finding reinforces the clinical value of the two focal interventions of this study within the therapeutic evolution of AUB management. However, this comparison must be interpreted with caution due to inherent temporal biases. Although we statistically adjusted for measurable confounders like diagnostic criteria and treatment era, unmeasured evolutions in supportive care, patient awareness, or subtle shifts in clinical management philosophy could persist. Therefore, the historical comparison primarily offers a benchmark for progress, while the core conclusions of this study rely on the contemporaneous, PSM-based comparison between dydrogesterone and LNG-IUS, which is not subject to these temporal biases. However, several limitations warrant acknowledgement. Firstly, the retrospective single-centre design introduces potential selection bias and limits generalisability. While PSM balanced measured confounders, unmeasured factors (e.g. socioeconomic status, adherence motivation, exact bleeding patterns) could influence outcomes. Secondly, the sample sizes within specific aetiological subgroups (e.g. AUB due to coagulopathy [AUB-C]) were modest, which limits the precision of estimates and the statistical power for those particular comparisons; findings in these smaller subgroups should be interpreted with caution and require validation in larger, dedicated cohorts. Thirdly, bleeding assessment relied partly on PBAC scores, which, despite rigorous validation protocols including photographic documentation, remain subjective and can be influenced by patient reporting accuracy. Fourthly, the historical control comparison, while adjusted for temporal trends, cannot fully account for potential evolutions in diagnostic practices or supportive care over time. Finally, detailed data on bleeding patterns and quality-of-life metrics beyond EQ-5D-5L were not comprehensively captured, restricting a more nuanced understanding of patient experience. Future research should prioritise prospective, multi-centre randomised trials directly comparing dydrogesterone and LNG-IUS with pre-specified subgroup analyses based on FIGO-AUB classification and patient age. For less common etiologies (e.g. AUB-C), collaborative studies with larger sample sizes are needed to draw definitive conclusions. Investigations into biomarkers predictive of treatment response could further personalise therapy selection. Studies with longer follow-up (>5 years) are needed to definitively establish the durability of LNG-IUS benefits and evaluate endometrial safety with prolonged dydrogesterone use. Furthermore, incorporating patient preference and cost-effectiveness analyses would greatly inform healthcare decision-making.

This study delineates complementary roles for dydrogesterone and LNG-IUS in AUB management. LNG-IUS offers superior short-term bleeding control, particularly for structural etiologies like adenomyosis, and provides significant long-term protection against recurrence, reducing the need for subsequent surgical interventions. Its efficacy profile supports its position as a first-line long-term option for suitable patients. Dydrogesterone demonstrates effective control, especially in non-structural AUB and younger women, albeit with a higher burden of systemic progestogenic side effects and increased recurrence risk over time. It represents a valuable alternative for patients preferring oral therapy, contraindicated for LNG-IUS, or requiring shorter-term intervention. Clinical decision-making should integrate aetiology, patient age, desired treatment duration, individual risk profiles, and patient preferences to optimise outcomes in this heterogeneous condition. Consent for publication Not applicable. Ethics approval and consent to participate The study protocol received approval from the Beijing Huairou Hospital Institutional Review Board (IRB Approval No: 20240416), and a waiver of informed consent was granted due to the anonymized retrospective nature of the data analysis, consistent with local regulations, the Declaration of Helsinki, and institutional privacy protection protocols. Supplemental material Supplementary figures.docx Download MS Word (342.9 KB)Supplementary figures.docxSupplementary tables.docx Download MS Word (31.4 KB)Supplementary tables.docxSTROBE_checklist_cohort.docx Download MS Word (31.6 KB)STROBE_checklist_cohort.docxAcknowledgements Not applicable. Supplemental material Supplemental data for this article can be accessed at https://doi.org/10.1080/09513590.2025.2612154. Disclosure statement The authors report there are no competing interests to declare. Funding The work was not funded by any funding. Data availability statement All data generated or analysed during this study are included in this article. Further inquiries can be directed to the corresponding author, Xiaoli Wei.

- Davis E, Sparzak PB. Abnormal Uterine Bleeding. StatPearls. Treasure Island (FL): StatPearls Publishing; 2024 Jan. PMID: 32809331. - Lou Z, Huang Y, Li S, et al. Global, regional, and national time trends in incidence, prevalence, years lived with disability for uterine fibroids, 1990-2019: an age-period-cohort analysis for the global burden of disease 2019 study. BMC Public Health. 2023;23(1):916. doi: 10.1186/s12889-023-15765-x - Munro MG, Critchley HOD, Fraser IS. The two FIGO systems for normal and abnormal uterine bleeding symptoms and classification of causes of abnormal uterine bleeding in the reproductive years: 2018 revisions [published correction appears in Int J Gynaecol Obstet. 2019 Feb;144(2):237]. Int J Gynaecol Obstet. 2018;143(3):393–408. doi: 10.1002/ijgo.12666 - Sabre A, Serventi L, Nuritdinova D, et al. Abnormal uterine bleeding types according to the PALM-COEIN FIGO classification in a medically underserved American community. J Turk Ger Gynecol Assoc. 2021;22(2):91–96. doi: 10.4274/jtgga.galenos.2021.2020.0228 - Tsakiridis I, Giouleka S, Koutsouki G, et al. Investigation and management of abnormal uterine bleeding in reproductive-aged women: a descriptive review of national and international recommendations. Eur J Contracept Reprod Health Care. 2022;27(6):504–517. doi: 10.1080/13625187.2022.2112169 - Matteson KA, Rahn DD, Wheeler 2nd TL, et al. Nonsurgical management of heavy menstrual bleeding: a systematic review. Obstet Gynecol. 2020;135(4):888–901. doi: 10.1097/AOG.0000000000003760 - National Institute for Health and Care Excellence (NICE). Heavy menstrual bleeding: assessment and management. NICE guideline NG88 2023. - Chen S, Liu J, Peng S, et al. LNG-IUS vs. medical treatments for women with heavy menstrual bleeding: a systematic review and meta-analysis. Front Med (Lausanne). 2022;9:948709. doi: 10.3389/fmed.2022.948709 - Gemzell-Danielsson K, Schellschmidt I, Apter D. A randomized, phase II study describing the efficacy, bleeding profile, and safety of two low-dose levonorgestrel-releasing intrauterine contraceptive systems and Mirena. Fertil Steril. 2021;115(6):1524–1532. doi: 10.1016/j.fertnstert.2021.01.040 - Vannuccini S, Petraglia F, Carmona F, et al. The modern management of uterine fibroids-related abnormal uterine bleeding. Fertil Steril. 2024;122(1):20–30. doi: 10.1016/j.fertnstert.2024.04.041 - Favilli A, Mazzon I, Etrusco A, et al. The challenge of FIGO type 3 leiomyomas and infertility: exploring therapeutic alternatives amidst limited scientific certainties. Int J Gynecol Obstet. 2024;165(3):975–987. doi: 10.1002/ijgo.15260 - MacGregor B, Munro MG, Lumsden MA. Therapeutic options for the management of abnormal uterine bleeding. Int J Gynaecol Obstet. 2023;162(Suppl 2):43–57. doi: 10.1002/ijgo.14947 - European Association for the Study of the Liver, EASL clinical practice guidelines: drug-induced liver injury. J Hepatol. 2019;70(6):1222–1261. doi: 10.1016/j.jhep.2019.02.014 - Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group, KDIGO 2012 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int Suppl. 2013;3(1):1–150. - World Health Organization International Statistical Classification of Diseases and Related Health Problems, Tenth Revision. 6th Vol. 2 World Health Organization; 2019 Instruction Manual. - Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat. 2011;10(2):150–157. doi: 10.1002/pst.433 - Ko JKY, Lao TT, Cheung VYT. Pictorial blood loss assessment chart for evaluating heavy menstrual bleeding in Asian women. Hong Kong Med J. 2021;27(6):399–404. doi: 10.12809/hkmj208743 - Warner MA, Hanson AC, Frank RD, et al. Prevalence of and recovery from anemia following hospitalization for critical illness among adults. JAMA Netw Open. 2020;3(9):e2017843. doi: 10.1001/jamanetworkopen.2020.17843 - Li DL, Wang ZT, Nie XY, et al. EQ-5D-5L population norms for China derived from a national health survey. Value Health. 2024;27(8):1108–1120. doi: 10.1016/j.jval.2024.04.014 - International Organization for Standardization. ISO 15189:2012. Medical laboratories — Requirements for quality and competence. International Organization for Standardization. 2012. - International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. Medical Dictionary for Regulatory Activities (MedDRA). Version 28.1. ICH. 2025. Accessed December 23, 2025. https://www.meddra.org - National Cancer Institute. Common Terminology Criteria for Adverse Events (CTCAE) v5.0. 2025. Published November 27, 2017. Accessed July 7. https://ctep.cancer.gov/protocoldevelopment/electronic_applications/ctc.htm - World Health Organization. WHO-UMC Causality Assessment. 2025. Published 2018. Accessed July 7. https://www.who.int/docs/default-source/medicines/pharmacovigilance/whocausality-assessment.pdf - Kaunitz AM, Bissonnette F, Monteiro I, et al. Levonorgestrel-releasing intrauterine system or medroxyprogesterone for heavy menstrual bleeding: a randomized controlled trial. Obstet Gynecol. 2010;116(3):625–632. doi: 10.1097/AOG.0b013e3181eb737e - Schrager S, Yogendran L, Marquez CM, et al. Adenomyosis: diagnosis and management. Am Fam Physician. 2022;105(1):33–38. - Pontis A, D'Alterio MN, Pirarba S, et al. Adenomyosis: a systematic review of medical treatment. Gynecol Endocrinol. 2016;32(9):696–700. doi: 10.1080/09513590.2016.1197200 - Patki A, Pawar VC. Modulating fertility outcome in ovulatory disorders by dydrogesterone. Gynecol Endocrinol. 2019;35(sup1):7–11. doi: 10.1080/09513590.2019.1653568 - Grimes DA, Lopez LM, O'Brien PA, et al. Progestin-only pills for contraception. Cochrane Database Syst Rev. 2013;2013(11):CD007541. doi: 10.1002/14651858.CD007541.pub3 - Lebduska E, Beshear D, Spataro BM. Abnormal uterine bleeding. Med Clin North Am. 2023;107(2):235–246. doi: 10.1016/j.mcna.2022.10.014 - Patel BG, Rudnicki M, Yu J, et al. Progesterone resistance in endometriosis: origins, consequences and interventions. Acta Obstet Gynecol Scand. 2017;96(6):623–632. doi: 10.1111/aogs.13156 - Bednarek PH, Jensen JT. Safety, efficacy and patient acceptability of the contraceptive and non-contraceptive uses of the LNG-IUS. Int J Womens Health. 2010;2:45–58. doi: 10.2147/ijwh.s6164 - El Behery MM, El-Khayat W, El-Khayat Z, et al. Levonorgestrel-releasing intrauterine device versus dydrogesterone for management of endometrial hyperplasia without atypia. Reprod Sci. 2015;22(3):329–334. doi: 10.1177/1933719114542014 - Wahabi HA, Fayed AA, Esmaeil SA, et al. Progestogen for treating irregular bleeding in women using progestogen-only contraceptives. Cochrane Database Syst Rev. 2021;8(8):CD013895. doi: 10.1002/14651858.CD013895.pub2 - Gemzell-Danielsson K, Schellschmidt I, Apter D. A randomized, phase II study describing the efficacy, bleeding profile, and safety of two low-dose levonorgestrel-releasing intrauterine contraceptive systems and Mirena. Fertil Steril. 2021;115(6):1524–1532. doi: 10.1016/j.fertnstert.2021.01.040 - Fraser IS, Critchley HOD, Munro MG, et al. Can we achieve international agreement on terminologies and definitions used to describe abnormalities of menstrual bleeding? Hum Reprod. 2007;22(3):635–643. doi: 10.1093/humrep/del478 - Marions L, Lövkvist L, Taube A, et al. Use of the levonorgestrel releasing-intrauterine system in nulliparous women - a non-interventional study in Sweden. Eur J Contracept Reprod Health Care. 2020;25(3):200–205. doi: 10.1080/13625187.2020.1743825 - Kabra R, Fisher M. Abnormal uterine bleeding in adolescents. Curr Probl Pediatr Adolesc Health Care. 2022;52(5):101185. doi: 10.1016/j.cppeds.2022.101185 - Delamater L, Santoro N. Management of the perimenopause. Clin Obstet Gynecol. 2018;61(3):419–432. doi: 10.1097/GRF.0000000000000389