Skip to main content
CandipediaResearch

ResearchEvidence map

Oral fluconazole in pregnancy: an evidence map

Compare evidence on birth defects, miscarriage, stillbirth, and neonatal death after pregnancy exposure to oral fluconazole, including the limits of each outcome.

Answer

The evidence needs to be read by outcome. A 2024 review found no significant pooled association with overall major malformations, while pooled observational estimates showed an association with miscarriage. Studies of stillbirth and neonatal death were imprecise and did not show a statistically significant increase. These group estimates are not a personal risk calculation. For a care decision, use the linked treatment guidance and its local authority record; those directions are recommendations from the relevant authority, not conclusions inferred from pooled cohort estimates.

  • Overall major-malformation estimates were near the null in the 2024 review and other large cohorts, although some studies reported signals for particular, uncommon malformation subtypes.
  • Several observational analyses found an association between fluconazole exposure and miscarriage. The studies use different designs and comparators, and an association does not establish that fluconazole caused a loss.
  • Stillbirth and neonatal-death estimates were imprecise, with low absolute event rates; these data do not establish absence of risk.
Confidence
High confidence in the reported direction and limits of the named study estimates; low confidence in causal attribution and in precise estimates for uncommon outcomes.
Why
The evidence is observational, exposure is often measured from prescriptions rather than confirmed ingestion, and studies define populations, comparators, doses, and outcomes differently. Relative estimates for pregnancy loss, live-birth malformations, stillbirth, and neonatal death cannot be combined as if they measured one outcome.
Who this applies to
  • Readers comparing population-level studies of oral fluconazole exposure in pregnancy across malformations, miscarriage, stillbirth, and neonatal death.
  • People tracing what each observational estimate measured and where its limits apply.
Who needs a different route
  • An individual prediction of pregnancy outcome or a calculation of risk for a particular exposure, dose, timing, condition, or medical history.
  • A comparison of oral and topical medicines for treating vaginal yeast infection during pregnancy.
  • A replacement for the relevant locale's treatment guidance or authority recommendation.

Next steps

  1. Read the treatment guidance — Use the same-locale treatment page and its linked authority record for the applicable pregnancy action.
  2. Read the Medicaid cohort summary — See what the first-trimester malformation cohort measured among live births and what it did not measure.
  3. Read the small topical-treatment trial summary — See the separate trial of two vaginal azoles; it did not test oral fluconazole or rare pregnancy safety.

Safety and when to get care

This route is based on the observable trigger and does not identify a diagnosis.

  • Care tier: Routine care

    Observable trigger
    The person is pregnant, might be pregnant, or becomes pregnant while considering or taking treatment for vaginal yeast symptoms.
    What to do
    Contact the pregnancy-care clinician before self-treating; do not use the exact fluconazole products in DailyMed setid 737e0fac-a2d2-4053-a0d7-edbfb6248f69 unless the clinician directs their use, and report pregnancy during use promptly.
    Exact population
    • pregnant or possibly pregnant people in the United States with vaginal yeast symptoms
    • people who become pregnant while taking the exact referenced fluconazole tablet or suspension products
    Not covered here
    • nonpregnant people not planning pregnancy
    • users of other products whose labels were not checked by this record
    Source

In this pathway

Parent problem or section
Vaginal yeast treatment
Current topic
Oral fluconazole in pregnancy: an evidence mapCurrent

Page trust and updates

First published
2026-10-08
Evidence as of
2026-10-08
Last updated
2026-10-08
Authorship
Candipedia Research
What changed
Created an initial synthesis that separates malformations, miscarriage, stillbirth, and neonatal death; no local care action changes.

Evidence map

Maintained claims

Claim: For first-trimester oral fluconazole exposure, adjusted observational evidence does not establish an increase in overall major congenital malformations; dose-specific and malformation-subtype risks remain uncertain.

Benefit verdict
Not enough evidence
Evidence strength
Unclear
Why: The 2024 systematic review found no increase in the adjusted pooled overall estimate, and large cohorts likewise did not establish an overall increase. These overall estimates do not resolve dose-specific or rare-subtype signals: Zhu found a small absolute musculoskeletal-malformation association and Bérard and Mølgaard-Nielsen reported subtype findings based on few exposed cases. Observational exposure and outcome ascertainment limits prevent a safety guarantee.
Safety status
Elevated risk

Claim: Oral fluconazole exposure in early pregnancy is associated with miscarriage in observational studies, but a causal effect and absolute excess risk are not established.

Benefit verdict
Supported
Evidence strength
Moderate
Why: A systematic review and several large registry studies consistently report an observational association, including a comparison with topical azole users and a newer analysis that addressed time-related biases. The newer study remains a systemic-antimicrobial analysis rather than a VVC-specific comparison, and its bias analysis does not establish that confounding explains or causes the association. Association estimates are not a causal or individual absolute-risk estimate.
Safety status
Elevated risk

Claim: Available observational studies do not establish an increase in stillbirth after oral fluconazole exposure during pregnancy; estimates remain imprecise, including by dose.

Benefit verdict
Not enough evidence
Evidence strength
Unclear
Why: Registry studies from Denmark, Quebec, and a matched Nordic cohort did not establish an overall increase in stillbirth. Confidence intervals are wide, and an isolated high-dose association in the Danish study was not replicated in the later Nordic cohort; this does not establish safety at any dose.
Safety status
Elevated risk

Claim: One large Nordic observational analysis did not establish an increase in neonatal death after oral fluconazole exposure during pregnancy; the estimate is imprecise and does not establish absence of risk.

Benefit verdict
Not enough evidence
Evidence strength
Unclear
Why: The matched Nordic analysis found 13 neonatal deaths among 10,640 exposed pregnancies and 177 among 106,387 unexposed pregnancies (RR 0.73, 95% CI 0.42–1.29). This single registry study had few exposed events and cannot establish that the risk is absent.
Safety status
Elevated risk

Where evidence agrees

  • Large cohorts and the 2024 review do not show a consistent increase in overall major congenital malformations. The review's adjusted pooled odds ratio was 1.02 (95% CI 0.98–1.07) across six studies; the 2013 and 2019 studies also did not report an increase in their overall-malformation analyses.
  • Multiple observational analyses report an association between fluconazole exposure and miscarriage, including a 2024 pooled odds ratio of 1.60 (95% CI 1.06–2.42). These associations do not establish causation, and the studies differ in design, dose grouping, and comparator.
  • The named stillbirth and neonatal-death studies do not show statistically significant increases, but their estimates are imprecise and the absolute event rates are low. A non-significant result is not proof that risk is absent.

Where evidence disagrees

  • The 2024 review's pooled estimate for overall major malformations was 1.02 (95% CI 0.98–1.07), while the S002 cohort reported a musculoskeletal-malformation relative risk of 1.30 (95% CI 1.09–1.56) and adjusted risk difference of 12.15 per 10,000 pregnancies (95% CI 3.69–20.60). These estimate different outcomes: an overall-malformation pool versus one uncommon subtype. Other studies also reported signals for selected subtypes, despite no significant pooled subtype associations; definitions, dose groups, and exposed case counts differ.
  • Stillbirth estimates point in different directions across cohorts, but the reported confidence intervals include no association and the studies use different comparisons. The available data do not resolve whether exposure changes stillbirth risk.

Open gaps

  • The S002 Medicaid study identifies malformations in claims through 90 days after delivery among live births. It cannot estimate miscarriage, stillbirth, or neonatal death, so its malformation results must not be treated as estimates for those outcomes.
  • Prescription fills do not confirm ingestion, exact timing, or dose. Observational studies also cannot fully separate a medicine's effect from the infection and other factors related to treatment; a 2026 systemic-antimicrobial analysis found that the fluconazole-miscarriage association moved toward the null only in its sensitivity analysis for infections with high miscarriage risk, which does not prove that confounding explains the association.
  • Signals for specific malformation subtypes involve uncommon outcomes and small exposed case counts, including seven exposed tetralogy-of-Fallot cases in the 2013 study and 13 high-dose cardiac-septal-anomaly cases in the 2019 study. More precise, consistently defined subtype estimates remain limited.
  • The studies report relative associations and population averages, not a validated individual risk estimate. Baseline risk, indication, exposure timing, dose, and clinical history cannot be resolved for an individual from these map-level results.
  • Evidence on pregnancy exposure outcomes does not establish comparative treatment effectiveness for vaginal yeast infection. The topical-azole comparator in observational cohorts concerns pregnancy outcomes, while the small Thai randomized trial compared two topical regimens and did not test oral fluconazole.

Related research

What would trigger a reassessment

Reassess this map when a new systematic review, well-powered cohort with outcome-specific denominators, correction, or change to a relevant authority recommendation alters the estimates or their interpretation for malformations, miscarriage, stillbirth, or neonatal death, or by the next-review date.

Evidence as of 2026-10-08.

Sources behind these conclusions

Claims by outcome and what the estimates mean

Overall major congenital malformations

The 2024 review searched through 2 December 2024 and included nine observational studies covering 3,764,897 pregnancies, of which 116,425 were fluconazole-exposed. Across six studies, Latour and colleagues reported an adjusted pooled odds ratio of 1.02 (95% CI 0.98–1.07) for overall major congenital malformations. The 2013 Danish cohort reported birth defects in 2.86% (210/7,352) of fluconazole-exposed pregnancies and 2.60% (25,159/968,236) of unexposed pregnancies, with an adjusted prevalence odds ratio of 1.06 (95% CI 0.92–1.21). The 2019 Quebec study also did not report an increase in overall major congenital malformations. These overall estimates do not mean that every specific malformation subtype has the same estimate.

Selected malformation subtypes

In the S002 Medicaid cohort, musculoskeletal malformations occurred at 52.1 versus 37.3 per 10,000 pregnancies in the live-birth cohort after first-trimester fluconazole versus topical-azole fills; the adjusted relative risk was 1.30 (95% CI 1.09–1.56), and adjusted risk difference was 12.15 per 10,000 (95% CI 3.69–20.60). Its conotruncal and oral-cleft estimates were not statistically significant. A higher-dose group above 450 mg had a musculoskeletal relative risk of 1.98 (95% CI 1.23–3.17), based on a smaller sample. The 2013 cohort reported an association with tetralogy of Fallot (odds ratio 3.16, 95% CI 1.49–6.71; seven exposed cases); the 2019 study reported an association between high-dose exposure and cardiac septal anomalies (odds ratio 1.81, 95% CI 1.04–3.14; 13 exposed cases). Latour's review found no significant pooled subtype associations, although its sensitivity analysis for exposures above 150 mg retained a malformation signal when studies used valid major-malformation definitions. These subtype findings are uncommon outcomes, differ by definition and study, and do not establish a single overall pattern.

Miscarriage

Latour's review reported an adjusted pooled odds ratio of 1.60 (95% CI 1.06–2.42). Mølgaard-Nielsen's 2016 cohort reported a hazard ratio of 1.48 (95% CI 1.23–1.77) versus no exposure and 1.62 (95% CI 1.26–2.07) versus topical azoles. Bérard reported odds ratios of 2.23 for low-dose and 3.20 for high-dose exposure. Boissiere-O'Neill and colleagues reported a fluconazole-miscarriage hazard ratio of 1.61 (95% CI 1.45–1.78) in a systemic-antimicrobial study that was not specific to vaginal yeast infection. These observational associations are not proof that fluconazole caused a miscarriage. The relative measures arise from different studies and should not be read as one interchangeable estimate.

Stillbirth and neonatal death

Mølgaard-Nielsen's 2016 study reported a stillbirth hazard ratio of 1.32 (95% CI 0.82–2.14) versus no exposure and 1.18 (95% CI 0.64–2.16) versus topical azoles. Bérard reported no stillbirth association. In a matched Nordic cohort, Pasternak and colleagues reported a stillbirth hazard ratio of 0.76 (95% CI 0.52–1.10) and a neonatal-death relative risk of 0.73 (95% CI 0.42–1.29). The corresponding absolute rates were 2.7 versus 3.6 per 1,000 for stillbirth and 1.2 versus 1.7 per 1,000 for neonatal death. These imprecise estimates do not establish no risk.

The outcome estimands must remain separate: the S002 malformation analysis concerns live births and claims through 90 days after delivery; miscarriage, stillbirth, and neonatal death are distinct outcomes with their own risk sets, timing, and comparisons. A topical-azole comparator used to study pregnancy outcomes is not a trial of comparative treatment effectiveness.

Where studies agree and differ

Overall-malformation estimates in the 2013 and 2019 cohorts and the 2024 pooled analysis do not show a consistent increase. The review's pooled overall-malformation odds ratio was 1.02 (95% CI 0.98–1.07). By contrast, S002 reported a musculoskeletal-malformation relative risk of 1.30 (95% CI 1.09–1.56) and an adjusted risk difference of 12.15 per 10,000 pregnancies (95% CI 3.69–20.60). These estimates concern different outcomes—overall malformations versus one uncommon subtype—not conflicting estimates of the same endpoint. Other individual studies also reported selected subtype signals, while the review found no significant pooled subtype associations; definitions, exposure groups, and case counts differ.

Several studies and the 2024 review report a miscarriage association, but all are observational and use different effect measures and comparators. The 2026 systemic-antimicrobial analysis is not specific to vaginal yeast infection. Its sensitivity analysis moved the estimate toward the null only when the infection was assumed to have high miscarriage risk; this observation does not establish that confounding explains the association.

Stillbirth estimates point in different directions but have confidence intervals that include no association. The neonatal-death estimate is also imprecise. These results leave uncertainty rather than showing that the outcomes are unaffected.

Latest meaningful changes

This draft brings together the listed cohorts, review, and authority recommendations while keeping the four outcome groups distinct. No locale-specific care action is inferred from the pooled estimates or changed by this evidence map. The relevant action remains in the linked treatment guidance and its authority record.

Limitations and gaps

Most of the evidence is observational. Prescription fills do not confirm ingestion, and the studies cannot fully separate medication exposure from the condition and circumstances that led to treatment. The S002 Medicaid study's malformation estimates apply to claims-identified outcomes among live births; they do not cover miscarriage, stillbirth, or neonatal death. Rare subtype outcomes have small exposed case counts, and dose-stratified estimates use smaller groups.

The studies do not provide a validated method for predicting an outcome for one person. Relative measures, group-average absolute rates, and study-specific comparisons are not personal risk estimates. The available observational pregnancy-outcome comparisons also do not establish whether one route of treatment is more effective than another for vaginal yeast infection.

Selected research and practical contribution

The S002 Medicaid cohort adds a large comparison for selected malformations among live births, with an adjusted association and small absolute difference for musculoskeletal outcomes. The 2024 review pools broader observational evidence and reports a near-null overall-malformation estimate alongside a miscarriage association; it did not find significant pooled subtype associations. The 2013 and 2019 cohorts contribute specific, uncommon subtype signals and miscarriage estimates, while Mølgaard-Nielsen 2016 and Pasternak 2018 provide separate stillbirth and neonatal-death estimates. Boissiere-O'Neill 2026 adds a systemic-antimicrobial analysis of miscarriage, not a VVC-specific result.

The linked Thai trial summary concerns two topical vaginal azoles in 96 pregnant adults. It is a separate treatment comparison and does not test oral fluconazole, estimate the outcomes on this map, or establish rare pregnancy safety. CDC's US guidance recommends only seven-day topical azole regimens during pregnancy and advises against fluconazole because of possible miscarriage and congenital anomalies. WHO's 2024 global recommendation separately names seven-day clotrimazole or nystatin for pregnancy. These authority recommendations have their own scope; the linked local treatment page and authority record determine the applicable locale-specific action. None of these statements is a comparative pregnancy cure claim derived from the cohort data.

Sources

View all 10 sources