Comparing the Safety Profiles of Vaginally Administered Misoprostol and Dinoprostone: A Retrospective Pharmacovigilance Study

Summary

This study compares the safety profiles of vaginally administered misoprostol and dinoprostone using FAERS data from 2004 to 2023.

Abstract

Misoprostol and dinoprostone are the most commonly used agents for cervical ripening during labour induction. However, comparative safety data remain limited, particularly regarding rare adverse events and off-label use. Data were obtained from the FDA Adverse Event Reporting System (FAERS) database for the period from the first quarter of 2004 to the fourth quarter of 2023. Adverse event (AE) reports involving vaginal administration of misoprostol and dinoprostone were extracted, followed by deduplication and exclusion of cases without a specified route of administration. A total of 3,925 misoprostol-related and 1,191 dinoprostone-related AEs were included for analysis. Clinical characteristics were then compared between the two groups. Pharmacovigilance signal detection was performed at both the system organ class (SOC) and preferred term (PT) levels using disproportionality and Bayesian approaches, including reporting odds ratios (RORs), proportional reporting ratios (PRRs), information components (ICs), and empirical Bayesian geometric means (EBGMs). Distinct safety profiles were observed between the two agents. Dinoprostone was primarily associated with pregnancy-related complications, including stronger signals for uterine rupture and postpartum haemorrhage. In contrast, misoprostol was more frequently associated with pregnancy termination-related events, including incomplete abortion, as well as gastrointestinal adverse effects such as abdominal pain. Misoprostol also demonstrated a stronger signal for off-label use. These findings suggest that vaginally administered misoprostol and dinoprostone exhibit different safety profiles, which may inform clinical decision-making during labour induction by highlighting potential risks associated with each agent. However, because FAERS is a spontaneous reporting system subject to underreporting, reporting bias, and the inability to establish causality, the results should be interpreted with caution. Further prospective studies are warranted to validate these safety observations.

Introduction

Labour induction plays a crucial role in childbirth management, particularly in cases where labour needs to be initiated. The promotion of cervical ripening is essential to ensuring a smooth delivery process¹. Insufficient cervical ripening can lead to prolonged labour, increasing the risk of maternal and neonatal complications. Therefore, the use of appropriate pharmacological agents to facilitate cervical ripening enhances both the safety and efficiency of labour induction². However, existing induction methods and pharmacological agents exhibit significant variations in efficacy and safety³, necessitating further research to optimise and standardise their clinical application.

Prostaglandins are widely used for labour induction, with misoprostol and dinoprostone being the most commonly employed agents for cervical ripening, particularly via vaginal administration^4,5. However, these drugs differ significantly in their mechanisms of action, pharmacokinetic properties, and safety profiles^6,7. Misoprostol, a prostaglandin E1 (PGE1) analogue, acts on prostaglandin receptors in the cervix and uterine smooth muscle when administered vaginally, promoting cervical softening and dilation while simultaneously inducing uterine contractions⁸. Its rapid onset of action and flexible administration routes may support its use in selected labour induction settings⁹. In contrast, dinoprostone, a prostaglandin E2 (PGE2) analogue, is primarily administered as a sustained-release vaginal formulation, facilitating collagen degradation in the cervical stroma, thereby increasing cervical softness and dilation, while also stimulating uterine contractions^10,11. Due to its prolonged pharmacological effect, dinoprostone may be useful in clinical settings where sustained cervical ripening and controlled administration are preferred.

In clinical practice, drug safety and tolerability are critical factors in decision-making, particularly in obstetric medicine, where maternal and neonatal health is directly influenced by pharmacological interventions. Although previous studies have reported common adverse events (AEs) and potential risks associated with misoprostol and dinoprostone, further analysis and comparison of their safety profiles are warranted¹². Existing studies indicate that the most common adverse reactions associated with misoprostol are gastrointestinal in nature, including nausea, vomiting, diarrhoea, and fever^13,14. Additionally, misoprostol may cause excessive uterine contractions, increasing the risk of uterine rupture, particularly in women with a history of caesarean section¹. On the other hand, reported AEs of dinoprostone include headache, abdominal pain, nausea, and vaginal bleeding¹⁵. Furthermore, studies have suggested that dinoprostone may induce excessive uterine contractions and fetal heart rate abnormalities¹⁶. Compared with dinoprostone, oral misoprostol has been associated with lower rates of caesarean section, uterine hyperstimulation, and foetal heart rate changes, though fewer vaginal deliveries occur within 24 h¹⁷. However, most evidence is derived from clinical trials or single-center studies with limited sample sizes and insufficient ability to detect rare events, which may not fully reflect real-world safety. In addition, direct comparative data on vaginal use remain limited. Further studies are therefore needed to better characterize AE profiles and associated risks. This is particularly relevant for rare but serious obstetric AEs, which are difficult to detect adequately in trials or single-center studies because of limited sample size and restricted follow-up.

The FDA Adverse Event Reporting System (FAERS) is a large spontaneous reporting database widely used for post-marketing pharmacovigilance¹⁸. Despite inherent limitations such as underreporting, reporting bias, and lack of causal inference, it remains a key tool for signal detection and complements clinical evidence by providing real-world safety data. Unlike traditional clinical trials, FAERS may help identify rare, unexpected, or underrecognized AE patterns after drug use in broader clinical settings, although the detected signals should be interpreted as reporting associations rather than causal evidence. Comparative evidence on the safety of vaginal misoprostol and dinoprostone in FAERS remains limited. Therefore, differences may exist between the two agents in their AE profiles, as well as in the distribution and strength of safety signals across system organ classes and event types. Using FAERS data from 2004 to 2023, this study systematically compared AEs associated with vaginal misoprostol and dinoprostone to address gaps in existing evidence and provide a more comprehensive assessment of their safety. The findings may help clinicians recognize potential safety signals during labour induction, guide closer monitoring of serious maternal and fetal AEs, and support further validation in prospective studies and other real-world data sources.

Protocol

This research used publicly available data and did not involve animal or human subjects. Therefore, ethical approval was not required.

Data source
This retrospective chart review study utilised publicly available AE reports from the FAERS database to perform a comparative analysis of AEs associated with the vaginal administration of misoprostol and dinoprostone. As a global reporting system for drug-related AEs, the FAERS database serves as a crucial resource for regulatory agencies and researchers in monitoring and evaluating drug safety in real-world settings. For this analysis, comprehensive data on vaginally administered misoprostol and dinoprostone were extracted from FAERS, covering the period from the first quarter of 2004 to the fourth quarter of 2023. This dataset included key information such as demographic details, drug-specific data, AE reports, patient outcomes, and sources of reports. These extensive data facilitated a thorough investigation of AEs related to the vaginal administration of misoprostol and dinoprostone, contributing to a detailed safety assessment of these drugs. All AEs were coded according to the Medical Dictionary for Regulatory Activities (MedDRA, version 26.1). Data extraction, cleaning, and signal detection analyses were performed using R software (version 4.5.1).

AE reports and drug identification
As FAERS does not standardise drug names, both brand and generic names were used to identify records associated with the target drugs. During the data mining process, text string searches were conducted in the FDA public database using both brand and generic names to identify reports related to the two drugs. AE reports explicitly marked as “misoprostol,” “dinoprostone,” or their respective brand names were extracted as the primary suspected cases (see Supplementary Table 1 and Supplementary Table 2 for the specific drug names used in screening). To identify vaginal administration records, reports were screened using route-of-administration information available in FAERS. Only reports indicating vaginal administration of misoprostol or dinoprostone were included in the final analysis. Reports with missing, unknown, or non-vaginal routes of administration were excluded.

Supplementary Table 1: Generic and brand names used to identify misoprostol-related reports in the FAERS database. Please click here to download this file.

Supplementary Table 2: Generic and brand names used to identify dinoprostone-related reports in the FAERS database. Please click here to download this file.

Data extraction
For data analysis from the FAERS database, data processing was performed according to the FDA-recommended deduplication procedure. Initially, 19,842,493 demographic records (DEMO) were extracted. After removal of 3,174,322 duplicate records, 16,668,171 unique records were retained. Additionally, 80,716,249 drug-related records and 48,387,071 reaction records were retrieved. Among these, AE reports explicitly identifying dinoprostone as the primary suspect drug (PS) totalled 1,316, corresponding to 4,068 preferred term (PT)-level AE records associated with dinoprostone-induced AEs. Similarly, AE reports explicitly identifying misoprostol as the PS totalled 9,884, corresponding to 37,297 PT-level AE records linked to misoprostol-induced AEs. These numbers represent the initial drug-specific identification before further restriction by administration route. Reports with missing case identifiers, duplicate records, or incomplete drug information were excluded during data cleaning. When multiple versions of the same case were identified, only the most recent case version was retained according to FDA recommendations. After further excluding cases without specified vaginal administration, a total of 3,925 AEs related to vaginally administered misoprostol and 1,191 AEs related to vaginally administered dinoprostone were included in the final statistical analysis. The detailed case identification, deduplication, inclusion, and exclusion procedures are illustrated in the workflow diagrams (Figure 1 and Figure 2).

Figure 1: The flow diagram of selecting dinoprostone-related AE reports​ from the FAERS database. Flow diagram showing the identification, deduplication, and signal-detection process for dinoprostone-related AE reports from FAERS. Please click here to view a larger version of this figure.

Figure 2: The flow diagram of selecting misoprostol-related AE reports​ from the FAERS database. Flow diagram showing the identification, deduplication, and signal-detection process for misoprostol-related AE reports from FAERS. Please click here to view a larger version of this figure.

To facilitate reproducibility and allow readers to verify the intermediate outputs of each step, key checkpoints were also provided in Supplementary Table 3. These included the numbers of initially extracted records, records retained after deduplication, primary suspect drug reports, PT-level AE records, and final vaginal administration reports included in the analysis (Supplementary Table 3).

Supplementary Table 3: Computational workflow and intermediate verification checkpoints for FAERS data processing and signal detection. Please click here to download this file.

Statistical analysis
This study employed multiple signal detection methods based on the FAERS database, including the reporting odds ratio (ROR) and proportional reporting ratio (PRR), to assess the strength of association between AEs in the exposed and non-exposed populations. ROR and PRR were selected because they are among the most widely used disproportionality analysis methods in pharmacovigilance and have demonstrated good performance for signal detection in spontaneous reporting databases. Higher ROR and PRR values indicate stronger disproportionality signals and a higher reporting frequency of a specific AE for the target drug. A positive signal was defined as: (1) ROR lower 95% confidence interval (ROR025) > 1 with at least three reports; (2) PRR ≥ 2, χ² ≥ 4, and number of reports ≥ 3; (3) IC025 > 0; and (4) EBGM05 > 2. The formulas for calculating ROR, PRR, and EBGM are as follows:

‘a’ represents the number of reports for the target drug and target AE.
‘b’ represents the number of reports for the target drug and non-target AEs.
‘c’ represents the number of reports for non-target drugs and the target AE.
‘d’ represents the number of reports for non-target drugs and non-target AEs.

‘O’ represents observed frequency of AE reports for a given drug.
‘E’ expected frequency based on background data.

Results

Descriptive analysis
Based on FAERS data from the first quarter of 2004 to the fourth quarter of 2023, a total of 1,567 AE reports related to misoprostol and 388 AE reports related to dinoprostone were included after deduplication and exclusion of cases without specified vaginal administration.

The temporal trend illustrated in Figure 3 shows a significant phase-specific fluctuation in AE reports for misoprostol. Between 2004 and 2007, the number of reports sharply declined from 299 per year to 49 per year (an 83.6% decrease). In the subsequent 16 years, the number of reports remained relatively low, ranging between 22 and 82 cases annually. In contrast, the annual report volume for dinoprostone remained relatively stable, consistently ranging between 8 and 36 cases per year.

Figure 3: Trends in AE​ reports for misoprostol and dinoprostone from Q1 2004 to Q4 2023. Please click here to view a larger version of this figure.

Table 1 presents the clinical characteristics of patients who experienced AEs following vaginal administration of misoprostol or dinoprostone. The majority of AEs for both drugs were concentrated in women of reproductive age, with a median age of 28 years (interquartile range: 23-34) for misoprostol-related cases and 32 years (interquartile range: 27–35) for dinoprostone-related cases. Regarding reporter type, misoprostol-related AEs were predominantly reported by other healthcare professionals (1,974 cases, 50.29%), whereas dinoprostone-related AEs were primarily reported by pharmacists (335 cases, 28.13%).

Among cases with clearly documented clinical outcomes, serious outcomes-including hospitalisation, disability, life-threatening events, or death-were observed in 54.47% of dinoprostone cases and 69.64% of misoprostol cases. Hospitalisation was the most frequently reported serious outcome, accounting for 34.91% of misoprostol-related cases and 23.96% of dinoprostone-related cases. Outcome analyses were based on available clinical outcome information recorded in the FAERS OUTC table. Reports without documented outcomes were treated as missing and excluded from outcome-specific calculations. Outcome proportions were calculated using available outcome data only. Because a single FAERS report could include multiple outcome codes, the reported proportions do not represent true incidence rates.

Geographical distribution analysis revealed that dinoprostone-related AEs from non-US regions were primarily concentrated in Germany (9.06%), while misoprostol-related reports were predominantly from Canada (24.62%). These findings highlight the need for further investigation into the geographic heterogeneity of pharmacovigilance signals and the potential impact of reporting biases on safety assessments. These geographic patterns should be interpreted with caution, as they may reflect reporting biases, differences in regional pharmacovigilance practices, and accessibility to reporting systems.

Table 1: Clinical characteristics of patients with AEs following vaginal administration of misoprostol and dinoprostone. Please click here to download this Table.

Disproportionality analyses
Table 2 shows the SOCs associated with Dinoprostone and Misoprostol. The SOCs with information components (IC > 0) reveal five main systems commonly associated with both drugs: "Pregnancy, Puerperium and Perinatal Conditions" (Dinoprostone: 359 cases; Misoprostol: 981 cases, "Injury, Poisoning and Procedural Complications" (206 cases vs 543 cases), "Reproductive System and Breast Disorders" (72 cases vs 163 cases), "Vascular Disorders" (47 cases vs 401 cases), and "Blood and Lymphatic System Disorders" (38 cases vs 138 cases). In addition, dinoprostone is specifically associated with "Cardiac Disorders" (43 cases), while misoprostol shows a unique signal for "Congenital, Familial and Genetic Disorders" (18 cases).

Signal strength analysis should be interpreted as disproportionality signals rather than evidence of causality. In pregnancy-related SOCs, Dinoprostone has significantly higher signal strength compared to Misoprostol (ROR = 68.93 vs 53.74; PRR = 48.46 vs 40.56; EBGM05 = 48.35 vs 40.33), with a Chi-square value of 16,752.45, indicating a strong association. In reproductive system disorders, dinoprostone also maintains a higher risk signal (ROR = 6.24 vs 4.20, EBGM05 = 5.92 vs 4.07). Notably, misoprostol shows a more significant signal in vascular diseases (ROR = 5.19 vs 1.87), while it also has a slightly stronger signal in blood and lymphatic system disorders (ROR = 2.43 vs 2.19). No significant signals were detected in other SOC categories, suggesting differences in AE reporting patterns between drugs rather than confirmed organ-specific toxicity. Special attention should be given to the cardiovascular toxicity of dinoprostone and the potential risks of vascular-related complications with misoprostol.

To explore the PT differences between the two drugs, reflecting their distinct safety profiles in clinical use, Table 3 indicates that we ranked AEs by their frequency for both drugs and selected the top 15 AEs across 10 SOC categories for further analysis.

For dinoprostone, AE signals were predominantly associated with pregnancy and labour complications, particularly excessive uterine activity, placental abruption, and haemorrhage. For instance, placental abruption exhibited a strong signal (ROR = 621.85, 95% CI = 430.68–897.88, EBGM05 = 589.80), indicating a potential risk of premature placental separation with dinoprostone use that warrants further investigation. Furthermore, uterine rupture (ROR = 476.03, 95% CI = 321.43–704.99, EBGM05 = 455.92), a severe complication likely linked to excessive uterine tachysystole induced by dinoprostone (ROR = 2774.51, EBGM05 = 2436.81). Notably, heightened uterine hypertonus (ROR = 4821.76, EBGM05 = 3843.10) and uterine hyperstimulation (ROR = 13760.22, EBGM05 = 8313.82) further support the possibility of excessive contractions. Additionally, postpartum haemorrhage displayed an extremely high signal strength (ROR = 826.54, 95% CI = 578.10–1181.75, EBGM05 = 775.70), suggesting a heightened risk of severe postpartum bleeding. Dinoprostone was also associated with low haemoglobin levels (ROR = 13.21, EBGM05 = 12.95), indicating potential risks of excessive blood loss or haematological effects. However, owing to the inherent limitations of the FAERS database, these findings reflect disproportionality-based reporting associations rather than confirmed causal relationships between dinoprostone exposure and the observed AEs.

For misoprostol, AE signals were closely related to pregnancy termination events and gastrointestinal adverse reactions. Incomplete abortion was one of the most prominent AEs associated with misoprostol (ROR = 3274.53, 95% CI = 2943.17–3643.20, EBGM05 = 2020.57), indicating a risk of incomplete expulsion of the gestational sac during pregnancy termination. Additionally, the signal for ongoing pregnancy (ROR = 93.92, 95% CI = 81.59–108.12, EBGM05 = 87.92) demonstrates disproportional reporting patterns consistent with both intended and off-label use. Notably, misoprostol demonstrated strong signals for "product use in unapproved indication" (ROR = 6.43, EBGM05 = 6.32) and off-label use (ROR = 3.92, EBGM05 = 3.79), highlighting potential non-standard clinical applications rather than confirmed safety outcomes.

Regarding gastrointestinal AEs, misoprostol-induced abdominal pain showed a significant signal strength (ROR = 5.12, 95% CI = 4.14–6.34, EBGM05 = 5.03), possibly linked to its uterotonic effects. Furthermore, the drug was associated with pyrexia (ROR = 4.28, EBGM05 = 4.20) and chills (ROR = 5.22, EBGM05 = 5.17), indicating a potential systemic inflammatory response. Additionally, haemorrhage (ROR = 64.08, EBGM05 = 58.33), vaginal haemorrhage (ROR = 8.41, EBGM05 = 8.32), and anaemia (ROR = 9.26, EBGM05 = 9.02) suggest that misoprostol may pose a certain degree of blood loss risk.

Overall, the analysis of 3,925 misoprostol and 1,191 dinoprostone vaginal administration AEs demonstrates distinct disproportionality patterns. Dinoprostone is associated with stronger signals for pregnancy and labour complications, while misoprostol shows higher signals for pregnancy termination-related and gastrointestinal events, as well as off-label use. These findings support the hypothesis that the two drugs have different safety profiles in real-world reporting data. However, the results should be interpreted in light of the limitations inherent to spontaneous reporting systems, including underreporting, reporting bias, and regional variations in pharmacovigilance practices. The present study provides useful insights for clinicians regarding potential AE patterns associated with vaginally administered misoprostol and dinoprostone. However, the identified safety signals should be interpreted as hypothesis-generating findings rather than evidence of causality or absolute risk. Therefore, future studies should integrate FAERS data with prospective registries, electronic health records, large-scale real-world datasets, and multicenter clinical studies to further validate these signals and clarify their clinical relevance.

Table 2: AE signals at the SOC level for dinoprostone and misoprostol. Please click here to download this Table.

Table 3: AE signals at the PT level for dinoprostone and misoprostol. Please click here to download this Table.

DATA AVAILABILITY:
The data for this study were derived from the FDA Adverse Event Reporting System (FAERS). All FAERS datasets are publicly accessible through the FDA Open Data website (https://open.fda.gov/data/faers).

Discussion

This study, based on data analysis from the FAERS database, compared the AE profiles of misoprostol and dinoprostone when administered via the vaginal route. The results revealed significant differences in the association of AEs between the two drugs. These findings support the hypothesis that misoprostol and dinoprostone exhibit distinct AE profiles and help address the limited availability of large-scale real-world comparative safety data for these commonly used cervical ripening agents. However, the observed signals represent reporting associations rather than confirmed causal relationships. The number of AE reports for misoprostol exhibited periodic fluctuations, whereas the annual reports for dinoprostone remained relatively stable. Both misoprostol and dinoprostone were associated with reproductive system and perinatal complications, particularly within the SOC of "Pregnancy, Puerperium, and Perinatal Conditions." Dinoprostone showed a stronger association with severe AEs such as uterine hyperresponsiveness, placental abruption, and haemorrhage, whereas misoprostol was more closely linked to pregnancy termination-related events and gastrointestinal adverse reactions. Additionally, signals for "off-label use and product use in unapproved indication" were notably stronger for misoprostol, suggesting a certain degree of non-standard clinical application. From a methodological perspective, these findings also highlight the importance of rigorous drug-name standardization, deduplication, route-of-administration screening, and primary suspect drug identification when using FAERS data to compare drug safety profiles.

Existing research has demonstrated that misoprostol and dinoprostone are widely utilised in obstetrics and gynaecology, primarily for pregnancy termination, labour induction, and the management of postpartum haemorrhage¹⁹. However, safety data for these agents are largely derived from randomised controlled trials and small-scale clinical observational studies, with systematic analyses of large-scale real-world data remaining relatively limited. International studies indicate that misoprostol is extensively used for medical abortion and pregnancy termination, whereas dinoprostone is primarily employed for labour induction^20,21 . In terms of AEs, studies suggest that misoprostol is predominantly associated with risks related to pregnancy termination, such as incomplete abortion, while dinoprostone is more strongly linked to pregnancy complications, including abnormal uterine contractions, placental abruption, and uterine rupture⁴. The findings of the present study provide a degree of support for these observations.

Regarding specific AEs, dinoprostone demonstrated a stronger association with severe perinatal complications, such as uterine hyperstimulation (ROR = 13,760.22), uterine rupture (ROR = 476.03), and placental abruption (ROR = 621.85). These findings align with a study that indicated dinoprostone may significantly increase the risk of uterine hyperstimulation during labour induction, thereby leading to severe obstetric complications²², which is consistent with its pharmacological role as a potent uterotonic agent²³. From a pharmacological perspective, the stronger effect of dinoprostone may be attributed to its prolonged duration of action and sustained impact on the uterus²⁴. Its sustained-release properties (maintaining PGE2 release for 24 h) continuously activate EP3 receptors in uterine smooth muscle, potentially resulting in excessive uterine tone and an increased risk of uterine rupture, placental abruption, and fetal distress²⁵. This study also identified a significant association between dinoprostone and fetal distress (ROR = 296.5) as well as fetal bradycardia (ROR = 544.05). An analysis demonstrated that, compared with oxytocin, dinoprostone significantly shortened labour duration but concurrently increased the incidence of uterine hyperstimulation and fetal distress²⁶. Similarly, a systematic review and meta-analysis reported that dinoprostone-induced uterine tachysystole and hyperstimulation may contribute to abnormal fetal heart rate patterns, including fetal bradycardia²⁷. Therefore, in clinical practice, careful monitoring of uterine contractions is essential when using dinoprostone, and dosage regimens should be adjusted according to individual patient conditions to minimise the risk of severe complications. Nevertheless, these findings should be interpreted cautiously, as disproportionality signals from FAERS do not establish causality or quantify absolute risk. Additionally, this study found a significant association between dinoprostone and postpartum haemorrhage (ROR = 826.54). While this finding is consistent with most previous studies, some randomised controlled trials have not observed this risk. This discrepancy may stem from the observation period in clinical trials (typically limited to 24 h postpartum), which may not capture delayed postpartum haemorrhage (PPH) events, whereas FAERS data encompass longer follow-up periods. Notably, dinoprostone exhibited slightly stronger signals than misoprostol in relation to haematological disorders, which may be linked to haemorrhage-related complications such as low haemoglobin levels and postpartum haemorrhage.

In contrast, the AEs associated with misoprostol are more closely related to pregnancy termination and gastrointestinal side effects²⁸. This study found that misoprostol exhibited strong signals for major AEs, including incomplete abortion (ROR = 3,274.53), abdominal pain (ROR = 5.12), and pyrexia (ROR = 4.28), suggesting a potential risk of incomplete embryo expulsion and systemic inflammatory reactions during pregnancy termination, further supporting this conclusion. From a pharmacological perspective, these effects may be linked to its mechanism as a PGE1 analogue. Misoprostol exerts its effects by binding to prostaglandin E (EP) receptors, facilitating uterine contractions and cervical ripening. Additionally, EP receptor activation in the gastrointestinal tract can enhance smooth muscle activity, contributing to the common gastrointestinal adverse reactions associated with misoprostol²⁹. Furthermore, this study found that, compared to dinoprostone, misoprostol exhibited a stronger signal for vascular disorders (ROR = 5.19 vs. 1.87), consistent with previous reports suggesting that misoprostol may increase the risk of thrombosis and hypertension³⁰. Moreover, misoprostol showed strong signals for "off-label use and product use in unapproved indication", indicating a potentially high proportion of off-label applications in clinical practice. Variations in indications, dosage regimens, and patient characteristics associated with off-label use may have influenced the observed reporting patterns. Therefore, strengthened pharmacovigilance is necessary to ensure the rational and regulated use of misoprostol.

This study, based on a large-scale analysis of the FAERS database, has revealed the AE spectrum and safety profile of misoprostol and dinoprostone when administered via the vaginal route. A practical advantage of this workflow is that it provides an efficient and scalable approach for screening large-scale real-world pharmacovigilance data, while remaining applicable to other drugs or administration routes with appropriate modifications. However, as a spontaneous reporting system, the FAERS database has certain limitations. First, the underreporting of mild AEs may lead to an overestimation of severity differences, and FAERS is also subject to reporting bias, duplicate reports, incomplete clinical information, and missing data. Second, because denominator information is unavailable, the true incidence of AEs cannot be calculated. Third, the lack of detailed dosage and exposure information hinders dose-response analysis, and indication bias may affect the interpretation. Furthermore, residual confounding and reporting bias remain unavoidable, and disproportionality analyses cannot fully account for differences in patient characteristics or clinical indications and can only detect potential safety signals rather than establish causal relationships. To mitigate these limitations, we applied the FDA-recommended deduplication procedure, restricted the analysis to reports in which the target drug was recorded as the primary suspect, standardized AEs using MedDRA PTs, and employed multiple signal-detection algorithms to improve the robustness of the findings. Nevertheless, FAERS should be viewed as a complementary signal-generation tool rather than a substitute for randomised controlled trials, electronic health records, claims databases, or prospective registries, which are better suited for estimating incidence, controlling confounding, and validating causal associations. Therefore, future research should integrate electronic health records (EHR), real-world data (RWD), prospective registries, and prospective clinical studies, employing methods such as propensity score matching to control for clinical confounders, identify high-risk populations, and optimise personalised medication strategies to enhance the safety of pregnancy-related drugs. Additionally, the combination of multicentre observational studies and randomised controlled trials would provide a more comprehensive assessment of the safety profiles of misoprostol and dinoprostone, offering stronger evidence-based support for clinical decision-making.

In conclusion, although misoprostol and dinoprostone are both widely used for cervical ripening and labour induction, they showed distinct safety signal profiles in the FAERS database. Dinoprostone-related safety concerns were mainly associated with perinatal management, particularly serious pregnancy-related complications potentially linked to excessive uterine activity, whereas misoprostol showed relatively stronger signals for pregnancy termination-related events and gastrointestinal adverse reactions. These findings suggest that clinicians should carefully monitor uterine activity, fetal heart rate abnormalities, postpartum hemorrhage, and other serious pregnancy-related AEs when using vaginal prostaglandins for cervical ripening and labour induction. Drug selection should be individualized according to maternal and fetal risk profiles, clinical indication, and local practice guidelines. However, given the inherent limitations of spontaneous reporting databases, the identified pharmacovigilance signals should be interpreted as hypothesis-generating rather than causal evidence. Future studies should integrate FAERS data with prospective registries and large-scale real-world analyses to further validate these signals, identify high-risk populations, and optimize obstetric medication safety protocols.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
FAERS dataset (2004-2023)	U.S. FDA	www.fda.gov/drugs/drug-approvals-and-databases/fda-adverse-event-reporting-system-faers-database	Public database used to extract AE reports for misoprostol and dinoprostone.
faersR package	N/A	www.bioconductor.org/packages/faers	faersR package used for FAERS-based disproportionality and Bayesian signal detection analyses, including ROR, PRR, IC, and EBGM metrics
R software (version 4.5.1)	R Foundation for Statistical Computing	RRID:SCR_001905	Statistical computing environment used for all analyses.