Can Preconception Metformin Improve Hyperemesis Gravidarum?

Highlights

– An observational cohort from the MotherToBaby Pregnancy Studies compared 80 pregnancies with preconception metformin exposure to 4,411 unexposed pregnancies and found a numerically lower rate of hyperemesis gravidarum (HG) in the exposed group (1.25% vs 2.20%).

– Adjusted risk ratio for HG was 0.50 (95% CI 0.07–3.39), a non-significant result with wide uncertainty due to small exposed sample size.

– Rates and mean duration of nausea and vomiting in pregnancy (NVP) were similar between groups (adjusted RR for NVP 1.06; 95% CI 0.94–1.19).

Background and clinical context

Nausea and vomiting of pregnancy (NVP) affect a large majority of pregnant women and range from mild, self-limited symptoms to the severe, life-disrupting condition hyperemesis gravidarum (HG). HG is typically defined by persistent vomiting resulting in >5% pre-pregnancy weight loss, dehydration, electrolyte disturbance, and need for intravenous fluids or hospitalization. Although HG is uncommon—the commonly cited prevalence is 0.3–2%—it has disproportionate morbidity, including maternal weight loss, need for acute care, work loss, and marked reductions in quality of life. Effective prevention strategies are lacking, and current management focuses on symptom control after onset using dietary measures and pharmacologic antiemetics; no established prophylactic pharmacotherapy exists.

Study design: MotherToBaby observational cohort

The study by Sillis et al. (BJOG 2025) used data from the MotherToBaby Pregnancy Studies (United States and Canada), a prospective observational registry that conducts structured maternal interviews and collects medical records during pregnancy. The analytic cohort included live-born singleton pregnancies enrolled 2012–2023 with HG diagnostic data available. The exposure of interest was preconception metformin use (any metformin exposure before conception), compared with no metformin exposure before or during pregnancy. Outcomes were HG diagnosis, occurrence of any NVP, and mean duration of NVP symptoms. Risk ratios (RR) and 95% confidence intervals (CI) were calculated, with adjustment for available covariates.

Key findings

In total, 80 pregnancies with preconception metformin exposure and 4,411 unexposed pregnancies were included. The principal results were:

• Hyperemesis gravidarum: 1 of 80 (1.25%) in the preconception-metformin group versus 97 of 4,411 (2.20%) in the unexposed group. The adjusted RR was 0.50 (95% CI 0.07–3.39). The point estimate suggests fewer HG diagnoses among metformin-exposed pregnancies, but confidence intervals are wide and include both potential benefit and no effect.
• Nausea and vomiting of pregnancy (any NVP): 32/80 (88.9%) in the exposed group versus 1,664/4,011 (82.6%) in the unexposed group; adjusted RR 1.06 (95% CI 0.94–1.19), indicating no meaningful difference in overall NVP frequency between groups.
• Duration of NVP symptoms: no clinically important differences in mean duration were observed between the exposed and unexposed groups.

Safety and secondary maternal–fetal outcomes were not the primary focus of this report; the authors report that available outcome data did not indicate unexpected safety signals but emphasize that the sample of exposed pregnancies was small.

Interpretation: what this study does and does not show

Strengths of the study include the prospective data collection, combination of maternal interview and medical record data, and a geographically diverse North American registry. The principal appeal of the work is that it provides the first reasonably large observational estimate suggesting a potential association between preconception metformin use and a reduced rate of clinically diagnosed HG.

However, the study does not provide definitive evidence that metformin prevents HG. Important limitations to bear in mind:

• Small exposed sample and sparse events: only 80 exposed pregnancies and a single HG event in that group yield imprecise estimates (wide CIs including the null).
• Observational design and confounding: preconception metformin is most commonly used for polycystic ovary syndrome (PCOS), diabetes, or prediabetes. Women using metformin differ from unexposed women in ways that could independently influence HG risk (e.g., BMI, parity, comorbidities, health-care-seeking behaviour). Although adjusted analyses were performed, residual confounding is likely.
• Exposure definition: the exposure window was preconception metformin use; the paper does not provide granular information on dose, duration, whether metformin was continued into pregnancy, or timing relative to conception—each of which could affect any biological plausibility for prophylaxis.
• Selection and information bias: MotherToBaby is a voluntary registry; participants may differ from the general pregnant population. Self-reported NVP is subjective and may be influenced by recall and reporting bias, though corroboration with medical records strengthens case ascertainment for HG.

Biological plausibility and mechanistic considerations

Mechanistic rationale for metformin as an HG prophylactic is currently speculative. Potential (but unproven) mechanisms include modulation of metabolic and hormonal pathways that influence placental function and hCG dynamics, insulin-sensitizing effects that alter the maternal hormonal milieu in early pregnancy, or indirect effects mediated through altered body composition or ovulatory physiology in women with PCOS. Notably, metformin commonly causes gastrointestinal side effects—including nausea and diarrhoea—making a direct antiemetic mechanism unlikely. Given the paradox that metformin can cause nausea in some users, any apparent protective association against HG seen in observational data must be carefully interrogated for confounding and bias.

Clinical implications and next steps

At present, this observational signal is hypothesis-generating rather than practice-changing. Clinicians should not prescribe metformin solely to prevent HG based on the available evidence. For women already taking metformin for established indications (e.g., type 2 diabetes, PCOS), this study provides limited reassurance that preconception exposure was not associated with a higher rate of HG in this cohort and may be associated with a lower rate; however, the evidence is not robust.

Key priorities for future research include:

• Randomized controlled trials (RCTs) in at-risk populations: A well-powered RCT of metformin versus placebo in women at high risk for HG (history of HG, certain hormonal or metabolic profiles, or selected PCOS subgroups) would provide the most reliable evidence. Trials should pre-specify clinically meaningful endpoints such as incidence of HG (well-defined diagnostic criteria), hospitalization for HG, PUQE (Pregnancy-Unique Quantification of Emesis) scores, duration of symptoms, weight loss, maternal functional outcomes, and fetal outcomes (birth weight, preterm birth).
• Mechanistic studies: Investigate whether metformin changes early pregnancy biomarkers (hCG kinetics, placental hormones, inflammatory mediators) that might plausibly affect emesis pathways.
• Safety profiling: Trials must assess maternal gastrointestinal tolerance and long-term offspring outcomes, because metformin crosses the placenta and long-term metabolic implications remain an area of ongoing investigation.
• Subgroup and dose–response exploration: Determine whether any effect differs by indication (PCOS, diabetes), BMI, or continuation of metformin into early pregnancy, and identify the minimum effective dose and optimal timing (preconception versus early gestation).

Expert commentary and limitations

From a methodological perspective, the MotherToBaby analysis is hypothesis-generating and useful for informing sample size estimates for future RCTs. The non-significant adjusted RR of 0.50 with a 95% CI spanning 0.07–3.39 highlights that an RCT will need substantially larger sample sizes to detect clinically meaningful reductions in an outcome with low base-rate events such as HG.

Clinicians should weigh current evidence on metformin for established indications in pregnancy (e.g., glucose control in gestational diabetes, metabolic management in PCOS) against this preliminary signal; clinical decisions about metformin use should continue to rest on guideline-based indications and individualized risk–benefit discussion.

Conclusions

Sillis et al. report an intriguing, but imprecise, association between preconception metformin exposure and lower frequency of clinically diagnosed hyperemesis gravidarum in a MotherToBaby observational cohort. The finding is hypothesis-generating rather than definitive. Rates and duration of general NVP were similar between exposed and unexposed pregnancies. Randomized trials and mechanistic studies are needed to determine whether metformin has a true prophylactic effect against HG, to define optimal dosing and timing, and to ensure maternal and fetal safety.

Funding and clinicaltrials.gov

The BJOG article reports data from the MotherToBaby Pregnancy Studies; readers should consult the original paper for specific funding sources. No registered randomized trials of metformin for HG prevention were identified in this report; investigators designing future trials should preregister protocols on clinicaltrials.gov before initiation.

Reference

Sillis L, Heinonen EW, Ceulemans M, Johnson D, Luo Y, Chambers CD. Metformin for the Prevention of Hyperemesis Gravidarum: An Observational Cohort Study. BJOG. 2025 Nov;132(12):1772-1778. doi: 10.1111/1471-0528.18238. Epub 2025 May 29. PMID: 40443168.