Metformin Exposure in Pregnancy and Childhood Developmental Outcomes

Hannah Gordon; Richard J Hiscock; Sarah Price; Alexis Shub; Alexandra Roddy Mitchell; Jessica Atkinson; Susan Walker; Anna Forsythe; Stephen Tong; Anthea Lindquist; Roxanne Hastie

doi:10.1001/jamanetworkopen.2026.37054

. 2026 Sep 30;9(9):e2637054. doi: 10.1001/jamanetworkopen.2026.37054

Metformin Exposure in Pregnancy and Childhood Developmental Outcomes

Hannah Gordon ^1,^2,^✉, Richard J Hiscock ^1,², Sarah Price ^1,³, Alexis Shub ^1,², Alexandra Roddy Mitchell ^1,², Jessica Atkinson ^1,², Susan Walker ^1,², Anna Forsythe ^1,², Stephen Tong ^1,², Anthea Lindquist ^1,², Roxanne Hastie ^1,²

¹Perinatal Epidemiology Group, Department of Obstetrics, Gynaecology and Newborn Health, University of Melbourne, Melbourne, Victoria, Australia

²Mercy Perinatal, Mercy Hospital for Women, Heidelberg, Victoria, Australia

³The Royal Women’s Hospital, Parkville, Victoria, Australia

Accepted for Publication: August 8, 2026.

Published: September 30, 2026. doi:10.1001/jamanetworkopen.2026.37054

^✉

Corresponding Author: Hannah Gordon, MD, University of Melbourne, Grattan St, Parkville, VIC 3052, Australia (hannah.gordon@unimelb.edu.au).

Author Contributions: Dr Gordon had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Lindquist and Hastie contributed equally as co–senior authors.

Concept and design: Gordon, Shub, Walker, Lindquist, Hastie.

Acquisition, analysis, or interpretation of data: Gordon, Hiscock, Price, Roddy Mitchell, Atkinson, Forsythe, Tong, Lindquist, Hastie.

Drafting of the manuscript: Gordon, Hiscock, Lindquist.

Critical review of the manuscript for important intellectual content: Hiscock, Price, Shub, Roddy Mitchell, Atkinson, Walker, Forsythe, Tong, Lindquist, Hastie.

Statistical analysis: Gordon, Hiscock, Roddy Mitchell, Lindquist.

Obtained funding: Gordon, Lindquist, Hastie.

Administrative, technical, or material support: Gordon, Price, Roddy Mitchell, Atkinson, Forsythe, Lindquist, Hastie.

Supervision: Shub, Walker, Tong, Lindquist, Hastie.

Conflict of Interest Disclosures: Prof Tong reported receiving grants from the National Health and Medical Research Council (NHMRC) of Australia and personal fees from Diamedica Therapeutics, Eli Lilly, Beech Biopharma, and Avilar Therapeutics outside the submitted work and has served as senior author, principal investigator, and coinvestigator of randomized clinical trials evaluating metformin and preeclampsia treatment and prevention. Dr Hastie reported receiving grants from the NHMRC and Medical Research Future Fund during the conduct of the study. No other disclosures were reported.

Funding/Support: This research was funded by an innovation grant from the Norman Beischer Medical Research Foundation and a grant from the Australasian Diabetes in Pregnancy Society. Dr Gordon was additionally supported by a Postgraduate NHMRC Research Scholarship and an Australian Government Research Training Program Scholarship (fee offset). Prof Tong and Drs Lindquist and Hastie were supported by NHMRC fellowship funding.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: This study used data from the Australian Early Development Census (AEDC). The AEDC is funded by the Australian Government Department of Education, Skills and Employment. The findings and views reported are those of the authors and should not be attributed to the department or the Australian government. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity (CCOPMM) provided access to the data used for this project. The conclusions, findings, opinions, and views or recommendations expressed in this article are strictly those of the authors. They do not necessarily reflect those of CCOPMM.

Data Sharing Statement: See Supplement 2.

Additional Contributions: The authors are grateful for the provision of data by the AEDC. They also are grateful to the CCOPMM and for the assistance of the staff at Safer Care Victoria.

Additional Information: These data were sourced from the National Diabetes Services Scheme, an initiative of the Australian government administered with the assistance of Diabetes Australia. Through the National Diabetes Services Scheme, Diabetes Australia provides diabetes self-management products and support services to more than 1 million people with diabetes in Australia.

^✉

Corresponding author.

PMCID: PMC13628313 PMID: 42814460

Key Points

Question

What is the association of metformin use during pregnancy with developmental outcomes among offspring during the first year of full-time school?

Findings

In this cohort study of 177 409 singleton children born in Victoria, Australia, between 2009 and 2020, no association was observed between metformin use during pregnancy and childhood developmental vulnerability (scoring ≤10th percentile on ≥2 of 5 assessed developmental domains using a national assessment tool) after accounting for confounding factors.

Meaning

These findings suggest that metformin in pregnancy is not associated with developmental vulnerability in children aged 4 to 6 years and may provide reassurance for clinicians and women considering metformin use during pregnancy.

This cohort study examines the association of antenatal metformin exposure with child development at school entry.

Abstract

Importance

Metformin is an oral hypoglycemic agent increasingly prescribed during pregnancy. In animal models, metformin influences embryonic cortical development, and it is uncertain whether this translates to clinical implications for humans.

Objective

To examine whether metformin prescription in pregnancy is associated with a risk of developmental vulnerability among offspring at school entry.

Design, Setting, and Participants

This cohort study used population-level pregnancy and birth data from January 1, 2009, to December 31, 2020, linked with the Australian Early Development Census (AEDC), a standardized national assessment of childhood development completed every 3 years during the first year of full-time school (age 4-6 years). The data analysis was performed between March 14, 2025, and January 30, 2026.

Exposure

Metformin prescription dispensation during pregnancy.

Main Outcome and Measures

The association of developmental vulnerability, defined as scoring below the 10th percentile in at least 2 of the 5 assessed domains of the AEDC (physical health and well-being, social competence, emotional maturity, school-based language and cognitive skills, and communication skills and general knowledge), was estimated using inverse probability of treatment weighting combined with regression adjustment.

Results

From 871 627 singleton births in Victoria between 2009 and 2020, 177 409 children with linked developmental outcome data were identified, of whom 1095 (0.6%) were exposed to metformin antenatally (43 [3.9%] aged ≤5 years 0 months, 860 [78.5%] aged 5 years 1 month to 6 years 0 months, and 192 [17.5%] aged ≥6 years 1 month at assessment; 557 male [50.9%]). Developmental vulnerability was observed in 199 children (18.2%) exposed to metformin compared with 24 379 children (13.9%) not exposed. After adjustment for confounders, metformin was not associated with developmental vulnerability among children exposed to metformin (adjusted relative risk, 0.97 [95% CI, 0.74-1.29]) or across any of the individual developmental domains.

Conclusions and Relevance

This cohort study found that metformin use in pregnancy was not associated with an altered risk of developmental vulnerability among children aged 4 to 6 years and may provide reassurance to clinicians and women considering metformin use during pregnancy.

Introduction

Metformin is an oral hypoglycemic medication that is first-line treatment for type 2 diabetes.¹ The use of metformin in pregnancy is increasing.² During pregnancy, metformin crosses the placenta, with the fetus exposed to similar medication concentrations as that in the maternal blood stream,³ raising potential concerns about fetal harm. There are reassuring, short-term perinatal safety data,^4,5 but long-term safety for exposed offspring remains incompletely described.^6,7 Metformin may be associated with increased adiposity in childhood,⁸ but research in this area is conflicting^9,10 and lacks follow-up into adulthood.

Considering neurodevelopmental implications, an animal study suggested that in utero exposure to metformin may cause impaired cortical development,¹¹ but it remains unclear whether this is relevant to human offspring development. A recent systematic review of 7 studies observed no association between in utero metformin exposure and global neurodevelopmental delay in children younger than 2 years (relative risk [RR], 1.09 [95% CI, 0.54-2.17]; 3 studies; 9668 children) and aged 3 to 5 years (RR, 0.90 [95% CI, 0.56-1.45]; 2 studies; 6118 children).⁷ However, in this preschool age range, developmental delays may not yet be apparent. Furthermore, individual studies in this area have been limited by small sample sizes and heterogeneous study populations, with metformin prescribed for diverse indications.^8,12

Larger studies focusing on outcomes for school-aged children are needed to provide evidence for women and their clinicians considering metformin use during pregnancy. Using population-level data from Victoria, Australia, we investigated whether antenatal metformin exposure during pregnancy was associated with an altered risk of developmental vulnerability on a national assessment among offspring during the first year of full-time school.

Methods

This cohort study included all singleton births in Victoria, Australia, between January 1, 2009, and December 31, 2020, with linked pregnancy and school-aged developmental outcome data. Ethical approval was received from the Mercy Health Ethics Committee and the Australian Institute of Health and Welfare. A waiver of informed consent was approved for this study. This research adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

The following 5 state-level and national-level databases were linked for this study by the Centre for Victorian Data Linkage and the Australian Institute of Health and Welfare: (1) Victorian Perinatal Data Collection (VPDC), (2) National Diabetes Service Scheme (NDSS), (3) Pharmaceutical Benefits Scheme (PBS), (4) Australian Early Development Census (AEDC), and (5) Victorian Admitted Episodes Dataset. Pregnancy data were sought from the VPDC, which collects statewide data related to pregnancy, birth, and perinatal outcomes, including pregnancy and birth complications, mode of birth, gestational age at birth, birth weight, and the presence of congenital anomalies. It is mandatory for every birth occurring after 20 weeks’ gestation to be recorded in the VPDC, with the database regularly audited to ensure data accuracy and completeness.¹³ Maternal medical conditions, including preeclampsia and polycystic ovary syndrome, were identified using the VPDC (eTable 1 in Supplement 1).

Birth data were linked with diabetes diagnoses from the NDSS (gestational diabetes, type 1 diabetes, type 2 diabetes). The NDSS is an Australia-wide registry including 80% to 90% of people diagnosed with diabetes¹⁴ and is further enriched by diabetes diagnoses identified in the VPDC. Each record in the NDSS is validated by an authorized health professional to ensure accuracy.¹⁵

Exposure

Metformin exposure was identified through dispensation of at least 1 metformin prescription during pregnancy by the PBS. The PBS is an Australian government initiative that completely or partially subsidizes the cost of more than 900 medications for Australian citizens, including metformin. Under the PBS, metformin is dispensed in boxes of between 60 and 120 tablets and prescribed for 1 to 3 times a day, with a typical prescription supplying between 1 and 6 months of medication.¹⁶

Metformin use in pregnancy was defined as dispensing a prescription any time from the first day of the last normal menstrual period until birth. The first day of the last normal menstrual period was calculated by subtracting gestational days at birth from the child’s date of birth. Information related to trimester of exposure and duration of exposure were recorded.

Outcome

The AEDC is a standardized assessment performed across the country every 3 years by a child’s school teacher during the first year of full-time school (at age 4-6 years).¹⁷ It is a validated assessment in this age group, with good concurrent validity and internal reliability.¹⁸ Only children who commenced school during 1 of the years the AEDC was performed (2015, 2018, or 2021) were eligible for linkage. Children were assessed across 5 domains of development: (1) physical health and well-being, (2) social competence, (3) emotional maturity, (4) school-based language and cognitive skills, and (5) communication skills and general knowledge.

Our primary outcome was developmental vulnerability, defined as scoring below the 10th percentile in 2 or more of the 5 AEDC domains. We also examined the association between metformin and childhood scores below the 10th percentile on each of the 5 individual developmental domains. Developmental vulnerability on the AEDC provides an important estimation of emotional and behavioral problems throughout primary school and literacy and numeracy delays in high school.¹⁹

Exclusions

Children born with congenital anomalies identified by International Statistical Classification of Diseases, Tenth Revision, Clinical Modification codes Q00 to Q99 were excluded. We additionally excluded women with uncertain diabetes diagnoses (eg, unknown timing of diabetes onset, 2 concurrent diabetes diagnoses) and those prescribed insulin without a recorded diabetes diagnosis.

Statistical Analysis

The data analysis was performed between March 14, 2025, and January 30, 2026. The association between metformin exposure and childhood developmental vulnerability was estimated using inverse probability of treatment weighting combined with regression adjustment to correct for both nonrandom exposure to metformin and postexposure covariate effects. Data are presented as RR and percentage risk differences with 95% CIs. Statistical significance was inferred if the 95% CI excluded the null (1.00 for relative measures and 0 for absolute measures), corresponding to a 2-sided α = .05. The data analysis was completed using StataMP, version 18 (StataCorp LLC) in accordance with a prespecified statistical analysis plan agreed upon by all coauthors prior to commencing the study (eAppendix in Supplement 1).

Demographic and clinical characteristics are presented as frequencies and percentages for categorical variables and as mean (SD) or median (IQR) for normally distributed and skewed continuous variables, respectively. The association between metformin exposure and childhood developmental vulnerability was estimated using inverse probability–weighted regression adjustment (IPWRA) and presented as RR and percentage risk differences.

Missing Data

We compared missing data across variables and stratified by metformin exposure status. Children who did not receive AEDC scores but were identified by their school as having special educational needs had outcome data deterministically imputed to represent developmental vulnerability across all 5 domains.^20,21 Missing outcome data for children not identified as having special needs and missing covariate data were considered missing at random and were imputed using multiple imputation with chained equations.

To obtain compatibility between the imputation process and analysis, we used the von Hippel-Bartlett approach in which 500 bootstrap resamples were first generated, with 2 imputations performed within each bootstrap sample, resulting in 1000 completed sets.²² Patterns of imputed data were compared with those in the original dataset to assess imputation adequacy, with no material differences identified. There were no missing exposure data.

IPWRA

In brief, IPWRA involves the creation of a propensity score based on covariates identified a priori as influencing metformin prescription during pregnancy, informed by a directed acyclic graph generated by our expert authorship team (eFigures 1 and 2 in Supplement 1). These covariates included maternal age, body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared), maternal education, Socio-Economic Indexes for Areas, diabetes in pregnancy, and use of assisted reproductive technology (ART). The propensity score specifications included main-effects terms for ART and diabetes and an interaction between them modeled using a fully saturated set of exposure indicators representing all ART-by-diabetes combinations. This specification improved covariate balance compared with additive main-effects models and was therefore used in the final analysis. Not all variables specified in the causal framework were available in the dataset. Potential proxies (eg, country of birth for race and ethnicity) were considered but not included as evidence suggested poor validity.²³

Propensity scores were inversely weighted, and postweighting covariate balance was assessed using absolute standardized mean differences (SMDs) within each of the 1000 imputed and bootstrapped datasets. The absolute SMDs and their corresponding standard deviations were then calculated across all datasets (Figure 1; eTable 2 in Supplement 1). An SMD of less than 0.1 was considered acceptable.²⁴ There was complete global overlap of propensity scores across the datasets. To account for clustering of multiple births per woman, standard errors were adjusted using a unique maternal identifier.

Dot plot of absolute S M D by covariate, comparing unweighted and weighted values. Horizontal dot plot with a single panel. The horizontal axis label at the bottom reads Absolute S M D, with tick marks from zero to zero point seven in steps of zero point one. A vertical dotted reference line intersects the plot at approximately zero point one. Along the left side, covariate names appear as grouped row labels, each followed by two indented rows labeled Unweighted and Weighted, separated by thin light gray horizontal rules. The covariate groups, from top to bottom, read Maternal body mass index; Interaction with a superscript a; S E I F A; Maternal age; and Maternal education. Each Unweighted row contains an orange square marker and each Weighted row contains a dark teal square marker, positioned horizontally according to the absolute S M D value. For Maternal body mass index, the orange square lies near zero point six eight and the dark teal square lies near zero point zero eight. For Interaction a, the orange square lies near zero point two six and the dark teal square lies near zero point zero four. For S E I F A, the orange square lies near zero point two zero and the dark teal square lies near zero point zero nine. For Maternal age, the orange square lies near zero point one six and the dark teal square lies near zero point zero seven. For Maternal education, the orange square sits on or very near the zero point one reference line and the dark teal square lies slightly left of it near zero point zero eight. No legend box is visible; color and marker meaning are implied by the repeated Unweighted and Weighted row labels. — Propensity score covariate balance was assessed using standardized mean differences (SMDs) within each of the 1000 imputed datasets used in this study, with the average SMD across the imputed datasets presented. SEIFA indicates Socio-Economic Indexes for Areas.

^aAn interaction between diabetes in pregnancy and assisted reproductive therapy was introduced in the propensity score model after initial diagnostics indicated residual imbalance.

Inverse probability weighting was then combined with regression adjustment for preeclampsia, maternal education, language background other than English, Socio-Economic Indexes for Areas, birth weight, age at time of AEDC testing, year of AEDC, gestational age at birth, and child sex assigned at birth, chosen using prespecified direct acyclic graphs (eAppendix in Supplement 1). For each of the 1000 bootstrap-imputed datasets, an IPWRA model was fitted. Within each analysis, the propensity score was reestimated using the prespecified covariates, and inverse probability weights were generated and checked before estimation of the regression-adjusted treatment effect. The resulting treatment estimates were subsequently pooled using the von Hippel-Bartlett approach.²² A detailed description of this analysis technique is described in the eAppendix in Supplement 1.

Sensitivity Analyses

Metformin is most commonly prescribed during pregnancy for the management of diabetes.²⁵ In light of the association between diabetes and other factors that may influence childhood development,²⁶ we performed the analysis restricting the cohort to include only women with type 2 diabetes or gestational diabetes to eliminate the potential for confounding by indication. Early pregnancy is a critical period for fetal neurodevelopment,²⁷ and we additionally examined whether first trimester metformin prescription was associated with an altered risk of overall childhood developmental vulnerability. Our primary analysis was repeated, excluding children with imputed outcome data, including those flagged as having special educational needs and those with incomplete outcome data for an unknown reason. Finally, alternative analytic approaches were explored. Regression adjustment analyses including postexposure variables were performed across 1000 bootstraps and imputations, with results again pooled using the von Hippel-Bartlett approach. A complete-case multivariable logistic regression using the same covariates was conducted as a sensitivity analysis for the primary IPWRA model, estimating the direct effect of exposure. Inverse probability weighting alone was performed across the same 1000 datasets to estimate the total effect.

Results

Demographic and Clinical Characteristics

From 871 627 singleton births in Victoria between 2009 and 2020, we identified 177 409 children with linked developmental outcome data, of whom 1095 (0.6%) were exposed to metformin antenatally (43 [3.9%] aged ≤5 years 0 months, 860 [78.5%] aged 5 years 1 month to 6 years 0 months, and 192 [17.5%] aged ≥6 years 1 month at assessment; 557 male [50.9%] and data suppressed for female to not unmask the number of children with missing sex) (Figure 2; Table 1). The 1095 mothers exposed to metformin during pregnancy were a mean (SD) age at delivery of 31.3 (5.3) years, and 309 (28.2%) were born outside Australia. Metformin exposure was common in the first trimester (703 women [64.2%]), with 141 women (12.9%) commencing metformin in the second trimester and 251 (22.9%) commencing metformin in the third trimester. A total of 467 women (42.6%) were dispensed more than 1 metformin prescription during pregnancy.

Flowchart of Victorian singleton births, exclusions, and linked metformin groups. A vertical flow diagram with light blue rectangular nodes outlined in gray and connected by gray arrows. At the top center, a wide rectangle contains bold number 871627 followed by text Singleton births in Victoria 2009-2020. From this top box, a gray arrow points downward to a middle wide rectangle reading bold 867690 followed by text Included. From the top box, a separate gray arrow points rightward to a large light blue rectangle on the upper right. This right-side rectangle begins with bold 3937 followed by text Excluded, then four indented lines each starting with a bold number and a reason: bold 1514 Congenital anomalies; bold 127 Diabetes with unknown onset timing; bold 2 Concurrent type 1 and type 2 diagnoses recorded; bold 2294 Insulin prescribed without diabetes diagnosis. From the Included box, a gray arrow points downward to a bottom wide rectangle. The bottom rectangle begins with bold 177409 followed by text Linked with A E D C data. Beneath that, two indented lines list subgroup counts: bold 176314 No metformin and bold 1095 Metformin. No axes, scales, legends, or panel labels appear. — AEDC indicates Australian Early Development Census.

Table 1. Demographic and Clinical Characteristics by Metformin Exposure.

Characteristic	Individuals, No. (%)
Characteristic	Not exposed to metformin (n = 176 314)	Exposed to metformin (n = 1095)
Maternal
Age at delivery, mean (SD), y	31.3 (5.4)	32.2 (5.1)
Missing	0	0
BMI, mean (SD)	26.0 (5.8)	32.3 (9.1)
Missing	13 997 (7.9)	55 (5.0)
Parity
Nulliparous	77 402 (43.9)	482 (44.0)
Multiparous	98 878 (56.1)	613 (56.0)
Missing	34 (<0.1)	0
Born outside Australia	49 063 (27.8)	309 (28.2)
Missing	5450 (3.1)	35 (3.2)
SEIFA quintile
1 (Most deprivation)	27 996 (15.9)	231 (21.1)
2	31 951 (18.1)	239 (21.8)
3	39 625 (22.5)	233 (21.3)
4	41 349 (23.4)	237 (21.6)
5 (Least deprivation)	35 187 (19.9)	155 (14.2)
Missing	206 (0.1)	0
Highest education level
Year ≤10	10 441 (5.9)	83 (7.6)
High school	20 586 (11.7)	114 (10.4)
Trade, TAFE, diploma, or certificate	67 653 (38.4)	470 (42.9)
Undergraduate university degree or higher	70 453 (40.0)	384 (35.1)
Missing	7181 (4.1)	44 (4.0)
Comorbidities
Polycystic ovary syndrome	4086 (2.3)	250 (22.8)
Preeclampsia	4341 (2.5)	66 (6.0)
Gestational diabetes	14 620 (8.3)	429 (39.2)
Type 1 diabetes	175 (0.1)	10 (0.9)
Type 2 diabetes	876 (0.5)	196 (17.9)
Assisted reproductive technology (yes)	7559 (4.3)	128 (11.7)
Missing	599 (0.3)	DS (<0.5)
Child
Sex assigned at birth
Female	86 147 (48.9)	NA^a
Male	90 022 (51.1)	557 (50.9)
Missing	145 (0.1)	DS (<0.1)
Gestational age at birth, median (IQR), wk	39.6 (38.6-40.4)	38.6 (37.9-39.6)
Missing	599 (0.3)	0
Preterm birth (yes)	10 419 (5.9)	137 (12.5)
Missing	599 (0.3)	0
Mode of birth
Unassisted vaginal	93 954 (53.3)	440 (40.2)
Assisted vaginal	25 955 (14.7)	119 (10.9)
Emergency caesarean delivery	26 707 (15.2)	277 (25.3)
Elective caesarean delivery	29 623 (16.8)	259 (23.7)
Missing	75 (<0.1)	0
Birth weight, mean (SD), g	3404 (533)	3336 (594)
Missing	853 (0.5)	13 (1.1)
Birth weight >97th percentile^b	3261 (1.9)	69 (6.3)
Missing	1451 (0.8)	13 (1.2)
Birth weight <10th percentile^b	19 328 (11.0)	113 (10.3)
Missing	1451 (0.8)	13 (1.2)
Language background other than English	35 985 (20.4)	242 (22.1)
Missing	0	0
Age at assessment
≤5 y 0 mo	2861 (1.6)	43 (3.9)
5 y 1 mo to 6 y 0 mo	138 944 (78.8)	860 (78.5)
≥6 y 1 mo	34 509 (19.6)	192 (17.5)
Missing	0	0
Special educational needs	8980 (5.1)	82 (7.5)
Incomplete AEDC data	1036 (0.6)	DS (<0.5)

Open in a new tab

Abbreviations: AEDC, Australian Early Development Census; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); DS, data suppressed due to small numbers; NA, not available; SEIFA, Socio-Economic Indexes for Areas; TAFE, Technical and Further Education.

^{^a}

Number of female children not reported per the data custodian as it would unmask the number of children with missing sex.

^{^b}

Hadlock percentiles.

Demographic and clinical characteristics were broadly comparable between children with and without linked developmental data. However, women with linked data compared with those without were less commonly born outside Australia (49 372 [27.8%] vs 239 608 [34.7%]) and had a lower incidence of gestational diabetes (15 049 [8.5%] vs 80 085 [11.6%]) (eTable 3 in Supplement 1).

No clear pattern of missingness was observed. Specifically, BMI missingness was 7.9% and 8.2% for developmental vulnerability and nonvulnerability, respectively (P = .33).

Women using metformin had a higher mean (SD) BMI of 32.3 (9.1) compared with women not using metformin (26.0 [5.8]) (Table 1). A higher proportion of women prescribed metformin conceived via ART (128 [11.7%] vs 7559 [4.3%]) and more commonly had comorbidities, including polycystic ovary syndrome (250 [22.8%] vs 4086 [2.3%]), type 2 diabetes (196 [17.9%] vs 876 [0.5%]), and gestational diabetes (429 [39.2%] vs 14 620 [8.3%]).

Women exposed to metformin compared with those not exposed more commonly gave birth pre term (<37 weeks’ gestation) (137 [12.5%] vs 10 419 [5.9%]) and by caesarean delivery (536 [49.0%] vs 56 330 [32.0%]). Infants exposed to metformin more commonly had birth weights above the 97th percentile compared with those not exposed (69 [6.3%] vs 3261 [1.9%]). The proportion of infants with birth weights below the 10th percentile was similar between groups (113 [10.3%] vs 19 328 [11.0%]).

Developmental Vulnerability

Among children exposed to metformin in utero, 199 (18.2%) were considered developmentally vulnerable, scoring below the 10th percentile across at least 2 of the 5 assessed domains in the AEDC (Table 2) compared with 24 379 children (13.9%) without exposure. This finding includes 9062 children (5.1%) across the cohort who were considered as having special educational needs (Table 1).

Table 2. Developmental Outcomes on the AEDC by Metformin Exposure.

Domain	Individuals, No. (%)		Unadjusted effect estimate		Adjusted effect estimate using IPWRA^a
Domain	Not exposed to metformin (n = 176 314)	Exposed to metformin (n = 1095)	RR (95% CI)	% RD (95% CI)	RR (95% CI)	% RD (95% CI)
Developmental vulnerability (scoring <10th percentile in ≥2 domains)	24 379 (13.9)	199 (18.2)	1.31 (1.16 to 1.49)	4.32 (2.04 to 6.62)	0.97 (0.74 to 1.29)	−0.23 (−4.01 to 3.54)
Missing	443 (1.8)	DS (<1.0)	1.31 (1.16 to 1.49)	4.32 (2.04 to 6.62)	0.97 (0.74 to 1.29)	−0.23 (−4.01 to 3.54)
Domain-specific developmental vulnerability
Physical health and well-being	21 898 (12.4)	196 (17.9)	1.44 (1.27 to 1.63)	5.46 (3.18 to 7.73)	0.98 (0.76 to 1.27)	−0.13 (−3.31 to 3.04)
Missing	297 (1.3)	0	1.44 (1.27 to 1.63)	5.46 (3.18 to 7.73)	0.98 (0.76 to 1.27)	−0.13 (−3.31 to 3.04)
Social competence	22 932 (13.0)	190 (17.4)	1.33 (1.17 to 1.52)	4.32 (2.07 to 6.57)	1.00 (0.74 to 1.36)	0.21 (−3.71 to 4.12)
Missing	303 (1.3)	0	1.33 (1.17 to 1.52)	4.32 (2.07 to 6.57)	1.00 (0.74 to 1.36)	0.21 (−3.71 to 4.12)
Emotional maturity	21 798 (12.4)	167 (15.3)	1.23 (1.07 to 1.42)	2.87 (0.73 to 5.01)	0.84 (0.64 to 1.10)	−1.94 (−4.74 to 0.87)
Missing	895 (3.9)	DS (<5.0)	1.23 (1.07 to 1.42)	2.87 (0.73 to 5.01)	0.84 (0.64 to 1.10)	−1.94 (−4.74 to 0.87)
Language and cognitive skills (school based)	18 828 (10.7)	173 (15.8)	1.48 (1.29 to 1.69)	5.11 (2.94 to 7.28)	1.09 (0.82 to 1.45)	1.08 (−2.28 to 4.44)
Missing	419 (2.2)	DS (<1.0)	1.48 (1.29 to 1.69)	5.11 (2.94 to 7.28)	1.09 (0.82 to 1.45)	1.08 (−2.28 to 4.44)
Communication skills and general knowledge	19 141 (10.9)	164 (15.0)	1.38 (1.20 to 1.59)	4.10 (1.98 to 6.22)	0.90 (0.68 to 1.19)	−1.00 (−3.73 to 1.73)
Missing	308 (1.6)	0	1.38 (1.20 to 1.59)	4.10 (1.98 to 6.22)	0.90 (0.68 to 1.19)	−1.00 (−3.73 to 1.73)

Open in a new tab

Abbreviations: % RD, percentage risk difference; AEDC, Australian Early Development Census; DS, data suppressed due to small numbers; IPWRA, inverse probability–weighted regression adjustment; RR, relative risk; SEIFA, Socio-Economic Indexes of Areas.

^{^a}

Selection model covariates included maternal age, maternal body mass index, maternal education, SEIFA, diabetes in pregnancy, and use of assisted reproductive technology. Outcome model covariates included preeclampsia, maternal education, language background other than English, SEIFA, birth weight, age at time of AEDC testing, year of AEDC, gestational age at birth, and child sex assigned at birth.

For the association between antenatal metformin exposure and overall developmental vulnerability at school entry, the unadjusted RR was 1.31 (95% CI, 1.16-1.49). However, following adjustment via doubly robust inverse probability weighting, this association was no longer observed (adjusted RR [ARR], 0.97 [95% CI, 0.74-1.29]). For each of the 5 domains of the AEDC individually, we found that metformin was also not associated with an altered risk of scoring below the 10th percentile in any developmental domains (physical health and well-being: ARR, 0.98 [95% CI, 0.76-1.27]; social competence: ARR, 1.00 [95% CI, 0.74-1.36]; emotional maturity: ARR, 0.84 [95% CI, 0.64-1.10]; language and cognitive skills [school-based]: ARR, 1.09 [95% CI, 0.82-1.45]; communication skills and general knowledge: ARR, 0.90 [95% CI, 0.68–1.19]) (Table 2).

Sensitivity Analyses

The most common indication for metformin in pregnancy is diabetes.²⁵ To eliminate diabetes as a confounder, we restricted the cohort to include only women with gestational diabetes or type 2 diabetes (n = 16 121). In this group, there were 129 children (20.7%) exposed to metformin considered developmentally vulnerable compared with 2460 children (15.9%) without exposure. Unadjusted, metformin was associated with an increased risk of developmental vulnerability (RR, 1.30 [95% CI, 1.11-1.52]). However, after adjustment for confounders, no association was observed (ARR, 1.08 [95% CI, 0.86-1.35]) (Table 3).

Table 3. Developmental Outcomes on the AEDC by Metformin Exposure Among Women With Gestational Diabetes or Type 2 Diabetes in Pregnancy.

Domain	Women, No. (%)		Unadjusted effect estimate		Adjusted effect estimate using IPWRA^a
Domain	Not exposed to metformin (n = 15 496)	Exposed to metformin (n = 625)	RR (95% CI)	% RD (95% CI)	RR (95% CI)	% RD (95% CI)
Developmental vulnerability (scoring <10th percentile in ≥2 domains)	2460 (15.9)	129 (20.7)	1.30 (1.11 to 1.52)	4.75 (1.52 to 7.98)	1.08 (0.86 to 1.35)	1.35 (−2.53 to 5.23)
Missing	43 (0.3)	DS (<1.0)	1.30 (1.11 to 1.52)	4.75 (1.52 to 7.98)	1.08 (0.86 to 1.35)	1.35 (−2.53 to 5.23)
Domain-specific developmental vulnerability
Physical health and well-being	2183 (14.1)	125 (20.0)	1.41 (1.20 to 1.66)	5.84 (2.68 to 9.00)	1.11 (0.88 to 1.39)	1.65 (−2.01 to 5.31)
Missing	33 (0.2)	0	1.41 (1.20 to 1.66)	5.84 (2.68 to 9.00)	1.11 (0.88 to 1.39)	1.65 (−2.01 to 5.31)
Social competence	2260 (14.6)	123 (19.7)	1.37 (1.16 to 1.60)	5.37 (2.21 to 8.53)	1.16 (0.90 to 1.50)	2.54 (−1.63 to 6.71)
Missing	30 (0.2)	0	1.37 (1.16 to 1.60)	5.37 (2.21 to 8.53)	1.16 (0.90 to 1.50)	2.54 (−1.63 to 6.71)
Emotional maturity	2089 (13.6)	103 (16.6)	1.23 (1.03 to 1.47)	3.18 (0.22 to 6.14)	0.94 (0.74 to 1.20)	−0.70 (−3.82 to 2.42)
Missing	90 (0.6)	DS (<1.0)	1.23 (1.03 to 1.47)	3.18 (0.22 to 6.14)	0.94 (0.74 to 1.20)	−0.70 (−3.82 to 2.42)
Language and cognitive skills (school based)	1906 (12.3)	103 (16.5)	1.37 (1.15 to 1.64)	4.55 (1.59 to 7.52)	1.03 (0.80 to 1.33)	0.48 (−2.73 to 3.71)
Missing	41 (0.3)	DS (<1.0)	1.37 (1.15 to 1.64)	4.55 (1.59 to 7.52)	1.03 (0.80 to 1.33)	0.48 (−2.73 to 3.71)
Communication skills and general knowledge	2079 (13.4)	108 (17.3)	1.30 (1.09 to 1.55)	4.03 (1.03 to 7.04)	1.07 (0.84 to 1.36)	1.04 (−2.47 to 4.55)
Missing	32 (0.2)	0	1.30 (1.09 to 1.55)	4.03 (1.03 to 7.04)	1.07 (0.84 to 1.36)	1.04 (−2.47 to 4.55)

Open in a new tab

^{^a}

Similarly, investigating each of the 5 domains of the AEDC individually in this cohort with gestational diabetes or type 2 diabetes, metformin was not associated with an altered risk of scoring below the 10th percentile in any developmental domain (physical health and well-being: ARR, 1.11 [95% CI, 0.88-1.39]; social competence: ARR, 1.16 [95% CI, 0.90-1.50]; emotional maturity: ARR, 0.94 [95% CI, 0.74-1.20]; language and cognitive skills [school-based]: ARR, 1.03 [95% CI, 0.80-1.33]; communication skills and general knowledge: ARR, 1.07 [95% CI, 0.84-1.36]) (Table 3).

We performed the same analysis while considering only children with complete outcome data and excluding children with special educational needs (9062 [5.1%]). We observed that 117 children (11.6%) exposed to metformin were considered developmentally vulnerable compared with 15 399 children (9.2%) not exposed. After adjustment for confounders, we observed no association between metformin exposure and developmental risk in this group (ARR, 0.80 [95% CI, 0.53-1.23]) (eTable 4 in Supplement 1).

Considering whether first trimester exposure was associated with developmental outcomes, 121 children (17.2%) with first trimester metformin exposure were considered developmentally vulnerable compared with 24 379 children (13.9%) not exposed. After adjustment for covariates, we observed no association between metformin exposure and developmental vulnerability in this cohort (ARR, 1.07 [95% CI, 0.77-1.48]) (eTable 4 in Supplement 1). Finally, results were robust to alternative specifications of the model (eTable 5 in Supplement 1).

Discussion

This cohort study of 177 409 children in Victoria, Australia, found no association between metformin prescription in pregnancy and developmental vulnerability in the first year of full-time school. No association was observed across all assessed developmental domains from the AEDC and with restriction of the cohort to include only women with gestational diabetes or type 2 diabetes in pregnancy.

To date, there have only been 2 other observational studies that examined the association of metformin use in pregnancy with childhood development.^28,29 The first included a cohort of 1681 children born to women with gestational diabetes, which observed no association between metformin exposure and teacher-assessed offspring behavioral difficulties prior to school entry.²⁸ Our analysis supports this finding in an older group of children both after restricting the cohort to include only women with gestational diabetes or type 2 diabetes and in our primary analysis, which included children whose mothers were prescribed metformin for any indication. This finding is particularly relevant given the off-label prescription of metformin for nondiabetes indications, including treatment for polycystic ovary syndrome and obesity.^30,31

The second study observed no association between metformin exposure and challenges in offspring motor-social development at various ages (2114 children exposed to metformin at age 3-5 years, 899 at age 6-8 years, and 290 at age 9-11 years) based on International Statistical Classification of Diseases, Tenth Revision codes.²⁹ Our findings are also consistent with this study, though our study considers development across a range of domains rather than by diagnostic criteria. Children from lower socioeconomic backgrounds are less able to access and receive developmental diagnoses,^32,33 and using a nationally performed assessment tool, as we have done, ensures that these children are not misclassified as developmentally on track.

The 5 randomized clinical trials in this area were small and typically used short follow-up (eg, into infancy) but, reassuringly, also demonstrated no association between metformin use and adverse offspring neurodevelopment.^{34,35,36,37,38} Overall, our study is in keeping with current research observing no association between in utero metformin exposure and adverse childhood neurodevelopment.

Strengths and Limitations

The strength of this study lies in the large number of children exposed antenatally to metformin followed up to the first year of full-time school. We used the NDSS enriched with the VPDC to identify a comprehensive cohort of women with diabetes in pregnancy. Furthermore, we considered childhood development across a range of physical and intellectual domains using a national and validated assessment tool. We used a doubly robust estimator combined with multiple imputation and bootstrapping to address some of the bias typically associated with observational data. Indication for metformin was not directly available in our study, though maternal comorbidities provided insight into the potential reason for metformin prescription and were included as covariates.

This study also had some limitations. There remains the potential for residual confounding, including by race and ethnicity. Self-reported race and ethnicity data were not available, and while maternal country of birth was considered as a potential proxy, it was not included in the final models due to high heterogeneity and limited validity as a measure of race and ethnicity in an Australian population.²³ Furthermore, while residual confounding could not be entirely excluded, its influence would have been expected to bias estimates toward a seemingly harmful association. The absence of an association after adjustment is hence particularly notable and reassuring.

Another limitation is that we were unable to assess medication adherence. However, we considered women as exposed to metformin only if they had been dispensed the medication from a pharmacy, which provides a more important estimation of adherence than medication prescription alone. There were 467 women (42.6%) who were dispensed more than 1 metformin prescription during pregnancy, and despite the large cohort, planned sensitivity analyses investigating the association of multiple prescriptions and second and third trimester metformin commencement were underpowered to support doubly robust or fully adjusted models. A small number of women may have been dispensed a metformin prescription immediately prior to conception and continued treatment into early pregnancy without requiring an additional dispensing during pregnancy. As our exposure definition relied on dispensing records during pregnancy, these women would have been misclassified as not exposed to metformin. In this instance though, the duration of unrecorded exposure may have been short and therefore may not have materially influenced longer-term offspring outcomes. Similarly, we could not ascertain metformin dose from our dataset. With the rising prescription of metformin during pregnancy,² future studies may be better placed to investigate this.

Conclusions

This cohort study found no evidence between in utero metformin exposure and developmental vulnerability in early primary school–aged children. Alongside previously published research in this area, the finding may provide reassurance for clinicians and women considering metformin use during pregnancy. Further research is still needed to examine the association of antenatal metformin use with offspring growth and cardiometabolic health, with current research in this area conflicting.^9,10,39

Supplement 1.

eTable 1. Methods Used to Ascertain Clinical Characteristics

eTable 2. Propensity Score Covariate Balance Assessment

eTable 3. Demographic and Clinical Characteristics for Participants With and Without Linked Developmental Outcome Data

eTable 4. Sensitivity Analyses: Childhood Developmental Vulnerability on the Australian Early Development Census by Subgroup

eTable 5. Sensitivity Analyses: Childhood Developmental Vulnerability on the Australian Early Development Census by Analysis Type

eAppendix. Does In Utero Exposure to Metformin Alter the Risk of Childhood Neurodevelopment at School Entry? A Statistical Analysis Plan

eFigure 1. Directed Acyclic Graph: Metformin in Pregnancy and Offspring Development

eFigure 2. Directed Acyclic Graph: Sensitivity Analysis—Metformin in Pregnancy Among Women With Diabetes (Type 2, Gestational Diabetes) and Offspring Development

eReferences.

jamanetwopen-e2637054-s001.pdf^{(972.6KB, pdf)}

Supplement 2.

Data Sharing Statement

jamanetwopen-e2637054-s002.pdf^{(10.1KB, pdf)}

References

1.The Royal Australian College of General Practitioners . Management of Type 2 Diabetes: A Handbook for General Practice. Royal Australian College of General Practitioners; 2020. [Google Scholar]
2.Charlton RA, Klungsøyr K, Neville AJ, et al. Prescribing of antidiabetic medicines before, during and after pregnancy: a study in seven European regions. PLoS ONE. 2016;11(5):e0155737. doi: 10.1371/journal.pone.0155737 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Charles B, Norris R, Xiao X, Hague W. Population pharmacokinetics of metformin in late pregnancy. Ther Drug Monit. 2006;28(1):67-72. doi: 10.1097/01.ftd.0000184161.52573.0e [DOI] [PubMed] [Google Scholar]
4.Goh JEL, Sadler L, Rowan J. Metformin for gestational diabetes in routine clinical practice. Diabet Med. 2011;28(9):1082-1087. doi: 10.1111/j.1464-5491.2011.03361.x [DOI] [PubMed] [Google Scholar]
5.Tarry-Adkins JL, Ozanne SE, Aiken CE. Impact of metformin treatment during pregnancy on maternal outcomes: a systematic review/meta-analysis. Sci Rep. 2021;11(1):9240. doi: 10.1038/s41598-021-88650-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Prescribing medicines in pregnancy database. Australian Government, Department of Health, Disability and Ageing, Therapeutic Goods Administration. May 19, 2026. Accessed May 19, 2026. https://www.tga.gov.au/prescribing-medicines-pregnancy-database
7.Gordon HG, Atkinson JA, Tong S, et al. Metformin in pregnancy and childhood neurodevelopmental outcomes: a systematic review and meta-analysis. Am J Obstet Gynecol. 2024;231(3):308-314.e6. doi: 10.1016/j.ajog.2024.02.316 [DOI] [PubMed] [Google Scholar]
8.Mayadunne T, Saadati S, Asmelash D, et al. Long-term effects of metformin on offspring health: a review of current evidence and future directions. Diabetes Obes Metab. 2025;27(suppl 3):48-63. doi: 10.1111/dom.16418 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Tarry-Adkins JL, Aiken CE, Ozanne SE. Neonatal, infant, and childhood growth following metformin versus insulin treatment for gestational diabetes: a systematic review and meta-analysis. PLoS Med. 2019;16(8):e1002848. doi: 10.1371/journal.pmed.1002848 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Fu J, Tabbara N, Tomlinson G, Murphy KE, Hamilton J, Feig DS. Intrauterine metformin exposure and adiposity outcomes in children: a systematic review and meta-analysis. BMJ Open. 2025;15(1):e088653. doi: 10.1136/bmjopen-2024-088653 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Oner M, Chen MC, Cheng PT, et al. Impact of metformin on neocortical development during pregnancy: Involvement of ERK and p35/CDK5 pathways. Chemosphere. 2024;358:142124. doi: 10.1016/j.chemosphere.2024.142124 [DOI] [PubMed] [Google Scholar]
12.Paavilainen E, Nyman A, Niinikoski H, et al. Metformin versus insulin for gestational diabetes: cognitive and neuropsychological profiles of children aged 9 years. J Dev Behav Pediatr. 2023;44(9):e642-e650. doi: 10.1097/DBP.0000000000001233 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Flood MM, McDonald SJ, Pollock WE, Davey MA. Data accuracy in the Victorian Perinatal Data Collection: results of a validation study of 2011 data. Health Inf Manag. 2017;46(3):113-126. doi: 10.1177/1833358316689688 [DOI] [PubMed] [Google Scholar]
14.Australian Institute of Health and Welfare (AIHW) . Diabetes: Australian Facts. Australian Government, Australian Institute of Health and Welfare. Updated May 27, 2026. Accessed May 30, 2026. https://www.aihw.gov.au/reports/diabetes/diabetes/contents/summary
15.Registration of people with diabetes with the NDSS. National Diabetes Services Scheme. August 16, 2022. Accessed February 6, 2026. https://www.ndss.com.au/health-professionals/registration/
16.Metformin. In: Australian Medicines Handbook. Australian Medicines Handbook Pty Ltd; 2024. Accessed April 15, 2025. https://amhonline.amh.net.au
17.About the AEDC. Australian Early Development Census . 2022. Accessed January 28, 2025. https://www.aedc.gov.au/about-the-aedc
18.Gregory T, Harman-Smith Y, Brinkman S. About the Reliability and Validity of the Australian Version of the Early Development Instrument (AEDC Fact Sheet). Australian Government; 2021. Accessed June 28, 2026. https://www.aedc.gov.au/docs/default-source/default-document-library/aedc-fact-sheet---validity-and-reliability-of-the-avedi.pdf?sfvrsn=2197a0ad_1
19.Centre for Adolescent Health, Murdoch Children’s Research Institute . The Effects on Schooling Outcomes of Early Developmental Vulnerabilities in Children. Australian Government Department of Education, Skills and Employment; 2018. [Google Scholar]
20.Gordon HG, Hiscock RJ, Shub A, et al. Diabetes in pregnancy and school-age developmental outcomes for offspring: a statewide retrospective cohort study. Diabetes Care. 2025;48(9):1612-1621. doi: 10.2337/dc25-0678 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Kennedy AL, Vollenhoven BJ, Hiscock RJ, et al. School-age outcomes among IVF-conceived children: a population-wide cohort study. PLoS Med. 2023;20(1):e1004148. doi: 10.1371/journal.pmed.1004148 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.von Hippel PT, Bartlett JW. Maximum likelihood multiple imputation: faster imputations and consistent standard errors without posterior draws. Stat Sci. 2021;36(3):400-420. doi: 10.1214/20-STS793 [DOI] [Google Scholar]
23.Porter M, Todd AL, Zhang LY. Ethnicity or cultural group identity of pregnant women in Sydney, Australia: Is country of birth a reliable proxy measure? Women Birth. 2016;29(2):168-171. doi: 10.1016/j.wombi.2015.10.001 [DOI] [PubMed] [Google Scholar]
24.Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat. 1985;39(1):33-38. doi: 10.1080/00031305.1985.10479383 [DOI] [Google Scholar]
25.Lindsay RS, Loeken MR. Metformin use in pregnancy: promises and uncertainties. Diabetologia. 2017;60(9):1612-1619. doi: 10.1007/s00125-017-4351-y [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Zhang C, Rawal S, Chong YS. Risk factors for gestational diabetes: is prevention possible? Diabetologia. 2016;59(7):1385-1390. doi: 10.1007/s00125-016-3979-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Cole TJ. Growth and organ development. In: Goldberg G, Prentice A, Prentice A, Filteau S, Simondon K, eds. Breast-Feeding: Early Influences on Later Health. Springer Netherlands; 2009:1-13. doi: 10.1007/978-1-4020-8749-3_1 [DOI] [Google Scholar]
28.Landi SN, Radke S, Engel SM, et al. Association of long-term child growth and developmental outcomes with metformin vs insulin treatment for gestational diabetes. JAMA Pediatr. 2019;173(2):160-168. doi: 10.1001/jamapediatrics.2018.4214 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Brand KMG, Saarelainen L, Sonajalg J, et al. ; CLUE Study Group . Metformin in pregnancy and risk of adverse long-term outcomes: a register-based cohort study. BMJ Open Diabetes Res Care. 2022;10(1):e002363. doi: 10.1136/bmjdrc-2021-002363 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Dodd JM, Louise J, Deussen AR, et al. Effect of metformin in addition to dietary and lifestyle advice for pregnant women who are overweight or obese: the GRoW randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol. 2019;7(1):15-24. doi: 10.1016/S2213-8587(18)30310-3 [DOI] [PubMed] [Google Scholar]
31.Løvvik TS, Carlsen SM, Salvesen Ø, et al. Use of metformin to treat pregnant women with polycystic ovary syndrome (PregMet2): a randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol. 2019;7(4):256-266. doi: 10.1016/S2213-8587(19)30002-6 [DOI] [PubMed] [Google Scholar]
32.Thomas P, Zahorodny W, Peng B, et al. The association of autism diagnosis with socioeconomic status. Autism. 2012;16(2):201-213. doi: 10.1177/1362361311413397 [DOI] [PubMed] [Google Scholar]
33.Kelly B, Williams S, Collins S, et al. The association between socioeconomic status and autism diagnosis in the United Kingdom for children aged 5-8 years of age: findings from the Born in Bradford cohort. Autism. 2019;23(1):131-140. doi: 10.1177/1362361317733182 [DOI] [PubMed] [Google Scholar]
34.Tertti K, Eskola E, Rönnemaa T, Haataja L. Neurodevelopment of two-year-old children exposed to metformin and insulin in gestational diabetes mellitus. J Dev Behav Pediatr. 2015;36(9):752-757. doi: 10.1097/DBP.0000000000000230 [DOI] [PubMed] [Google Scholar]
35.Greger HK, Hanem LGE, Østgård HF, Vanky E. Cognitive function in metformin exposed children, born to mothers with PCOS - follow-up of an RCT. BMC Pediatr. 2020;20(1):60. doi: 10.1186/s12887-020-1960-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Ijäs H, Vääräsmäki M, Saarela T, Keravuo R, Raudaskoski T. A follow-up of a randomised study of metformin and insulin in gestational diabetes mellitus: growth and development of the children at the age of 18 months. BJOG. 2015;122(7):994-1000. doi: 10.1111/1471-0528.12964 [DOI] [PubMed] [Google Scholar]
37.Wouldes TA, Battin M, Coat S, Rush EC, Hague WM, Rowan JA. Neurodevelopmental outcome at 2 years in offspring of women randomised to metformin or insulin treatment for gestational diabetes. Arch Dis Child Fetal Neonatal Ed. 2016;101(6):F488-F493. doi: 10.1136/archdischild-2015-309602 [DOI] [PubMed] [Google Scholar]
38.Deussen AR, Louise J, Dodd JM. Childhood follow-up of the GRoW randomized trial: Metformin in addition to dietary and lifestyle advice for pregnant women with overweight or obesity. Pediatr Obes. 2023;18(1):e12974. doi: 10.1111/ijpo.12974 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Hanem LGE, Salvesen Ø, Juliusson PB, et al. Intrauterine metformin exposure and offspring cardiometabolic risk factors (PedMet study): a 5-10 year follow-up of the PregMet randomised controlled trial. Lancet Child Adolesc Health. 2019;3(3):166-174. doi: 10.1016/S2352-4642(18)30385-7 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eTable 1. Methods Used to Ascertain Clinical Characteristics

eTable 2. Propensity Score Covariate Balance Assessment

eTable 3. Demographic and Clinical Characteristics for Participants With and Without Linked Developmental Outcome Data

eTable 4. Sensitivity Analyses: Childhood Developmental Vulnerability on the Australian Early Development Census by Subgroup

eTable 5. Sensitivity Analyses: Childhood Developmental Vulnerability on the Australian Early Development Census by Analysis Type

eAppendix. Does In Utero Exposure to Metformin Alter the Risk of Childhood Neurodevelopment at School Entry? A Statistical Analysis Plan

eFigure 1. Directed Acyclic Graph: Metformin in Pregnancy and Offspring Development

eFigure 2. Directed Acyclic Graph: Sensitivity Analysis—Metformin in Pregnancy Among Women With Diabetes (Type 2, Gestational Diabetes) and Offspring Development

eReferences.

jamanetwopen-e2637054-s001.pdf^{(972.6KB, pdf)}

Supplement 2.

Data Sharing Statement

jamanetwopen-e2637054-s002.pdf^{(10.1KB, pdf)}

[zoi260993r1] 1.The Royal Australian College of General Practitioners . Management of Type 2 Diabetes: A Handbook for General Practice. Royal Australian College of General Practitioners; 2020. [Google Scholar]

[zoi260993r2] 2.Charlton RA, Klungsøyr K, Neville AJ, et al. Prescribing of antidiabetic medicines before, during and after pregnancy: a study in seven European regions. PLoS ONE. 2016;11(5):e0155737. doi: 10.1371/journal.pone.0155737 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r3] 3.Charles B, Norris R, Xiao X, Hague W. Population pharmacokinetics of metformin in late pregnancy. Ther Drug Monit. 2006;28(1):67-72. doi: 10.1097/01.ftd.0000184161.52573.0e [DOI] [PubMed] [Google Scholar]

[zoi260993r4] 4.Goh JEL, Sadler L, Rowan J. Metformin for gestational diabetes in routine clinical practice. Diabet Med. 2011;28(9):1082-1087. doi: 10.1111/j.1464-5491.2011.03361.x [DOI] [PubMed] [Google Scholar]

[zoi260993r5] 5.Tarry-Adkins JL, Ozanne SE, Aiken CE. Impact of metformin treatment during pregnancy on maternal outcomes: a systematic review/meta-analysis. Sci Rep. 2021;11(1):9240. doi: 10.1038/s41598-021-88650-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r6] 6.Prescribing medicines in pregnancy database. Australian Government, Department of Health, Disability and Ageing, Therapeutic Goods Administration. May 19, 2026. Accessed May 19, 2026. https://www.tga.gov.au/prescribing-medicines-pregnancy-database

[zoi260993r7] 7.Gordon HG, Atkinson JA, Tong S, et al. Metformin in pregnancy and childhood neurodevelopmental outcomes: a systematic review and meta-analysis. Am J Obstet Gynecol. 2024;231(3):308-314.e6. doi: 10.1016/j.ajog.2024.02.316 [DOI] [PubMed] [Google Scholar]

[zoi260993r8] 8.Mayadunne T, Saadati S, Asmelash D, et al. Long-term effects of metformin on offspring health: a review of current evidence and future directions. Diabetes Obes Metab. 2025;27(suppl 3):48-63. doi: 10.1111/dom.16418 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r9] 9.Tarry-Adkins JL, Aiken CE, Ozanne SE. Neonatal, infant, and childhood growth following metformin versus insulin treatment for gestational diabetes: a systematic review and meta-analysis. PLoS Med. 2019;16(8):e1002848. doi: 10.1371/journal.pmed.1002848 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r10] 10.Fu J, Tabbara N, Tomlinson G, Murphy KE, Hamilton J, Feig DS. Intrauterine metformin exposure and adiposity outcomes in children: a systematic review and meta-analysis. BMJ Open. 2025;15(1):e088653. doi: 10.1136/bmjopen-2024-088653 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r11] 11.Oner M, Chen MC, Cheng PT, et al. Impact of metformin on neocortical development during pregnancy: Involvement of ERK and p35/CDK5 pathways. Chemosphere. 2024;358:142124. doi: 10.1016/j.chemosphere.2024.142124 [DOI] [PubMed] [Google Scholar]

[zoi260993r12] 12.Paavilainen E, Nyman A, Niinikoski H, et al. Metformin versus insulin for gestational diabetes: cognitive and neuropsychological profiles of children aged 9 years. J Dev Behav Pediatr. 2023;44(9):e642-e650. doi: 10.1097/DBP.0000000000001233 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r13] 13.Flood MM, McDonald SJ, Pollock WE, Davey MA. Data accuracy in the Victorian Perinatal Data Collection: results of a validation study of 2011 data. Health Inf Manag. 2017;46(3):113-126. doi: 10.1177/1833358316689688 [DOI] [PubMed] [Google Scholar]

[zoi260993r14] 14.Australian Institute of Health and Welfare (AIHW) . Diabetes: Australian Facts. Australian Government, Australian Institute of Health and Welfare. Updated May 27, 2026. Accessed May 30, 2026. https://www.aihw.gov.au/reports/diabetes/diabetes/contents/summary

[zoi260993r15] 15.Registration of people with diabetes with the NDSS. National Diabetes Services Scheme. August 16, 2022. Accessed February 6, 2026. https://www.ndss.com.au/health-professionals/registration/

[zoi260993r16] 16.Metformin. In: Australian Medicines Handbook. Australian Medicines Handbook Pty Ltd; 2024. Accessed April 15, 2025. https://amhonline.amh.net.au

[zoi260993r17] 17.About the AEDC. Australian Early Development Census . 2022. Accessed January 28, 2025. https://www.aedc.gov.au/about-the-aedc

[zoi260993r18] 18.Gregory T, Harman-Smith Y, Brinkman S. About the Reliability and Validity of the Australian Version of the Early Development Instrument (AEDC Fact Sheet). Australian Government; 2021. Accessed June 28, 2026. https://www.aedc.gov.au/docs/default-source/default-document-library/aedc-fact-sheet---validity-and-reliability-of-the-avedi.pdf?sfvrsn=2197a0ad_1

[zoi260993r19] 19.Centre for Adolescent Health, Murdoch Children’s Research Institute . The Effects on Schooling Outcomes of Early Developmental Vulnerabilities in Children. Australian Government Department of Education, Skills and Employment; 2018. [Google Scholar]

[zoi260993r20] 20.Gordon HG, Hiscock RJ, Shub A, et al. Diabetes in pregnancy and school-age developmental outcomes for offspring: a statewide retrospective cohort study. Diabetes Care. 2025;48(9):1612-1621. doi: 10.2337/dc25-0678 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r21] 21.Kennedy AL, Vollenhoven BJ, Hiscock RJ, et al. School-age outcomes among IVF-conceived children: a population-wide cohort study. PLoS Med. 2023;20(1):e1004148. doi: 10.1371/journal.pmed.1004148 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r22] 22.von Hippel PT, Bartlett JW. Maximum likelihood multiple imputation: faster imputations and consistent standard errors without posterior draws. Stat Sci. 2021;36(3):400-420. doi: 10.1214/20-STS793 [DOI] [Google Scholar]

[zoi260993r23] 23.Porter M, Todd AL, Zhang LY. Ethnicity or cultural group identity of pregnant women in Sydney, Australia: Is country of birth a reliable proxy measure? Women Birth. 2016;29(2):168-171. doi: 10.1016/j.wombi.2015.10.001 [DOI] [PubMed] [Google Scholar]

[zoi260993r24] 24.Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat. 1985;39(1):33-38. doi: 10.1080/00031305.1985.10479383 [DOI] [Google Scholar]

[zoi260993r25] 25.Lindsay RS, Loeken MR. Metformin use in pregnancy: promises and uncertainties. Diabetologia. 2017;60(9):1612-1619. doi: 10.1007/s00125-017-4351-y [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r26] 26.Zhang C, Rawal S, Chong YS. Risk factors for gestational diabetes: is prevention possible? Diabetologia. 2016;59(7):1385-1390. doi: 10.1007/s00125-016-3979-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r27] 27.Cole TJ. Growth and organ development. In: Goldberg G, Prentice A, Prentice A, Filteau S, Simondon K, eds. Breast-Feeding: Early Influences on Later Health. Springer Netherlands; 2009:1-13. doi: 10.1007/978-1-4020-8749-3_1 [DOI] [Google Scholar]

[zoi260993r28] 28.Landi SN, Radke S, Engel SM, et al. Association of long-term child growth and developmental outcomes with metformin vs insulin treatment for gestational diabetes. JAMA Pediatr. 2019;173(2):160-168. doi: 10.1001/jamapediatrics.2018.4214 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r29] 29.Brand KMG, Saarelainen L, Sonajalg J, et al. ; CLUE Study Group . Metformin in pregnancy and risk of adverse long-term outcomes: a register-based cohort study. BMJ Open Diabetes Res Care. 2022;10(1):e002363. doi: 10.1136/bmjdrc-2021-002363 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r30] 30.Dodd JM, Louise J, Deussen AR, et al. Effect of metformin in addition to dietary and lifestyle advice for pregnant women who are overweight or obese: the GRoW randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol. 2019;7(1):15-24. doi: 10.1016/S2213-8587(18)30310-3 [DOI] [PubMed] [Google Scholar]

[zoi260993r31] 31.Løvvik TS, Carlsen SM, Salvesen Ø, et al. Use of metformin to treat pregnant women with polycystic ovary syndrome (PregMet2): a randomised, double-blind, placebo-controlled trial. Lancet Diabetes Endocrinol. 2019;7(4):256-266. doi: 10.1016/S2213-8587(19)30002-6 [DOI] [PubMed] [Google Scholar]

[zoi260993r32] 32.Thomas P, Zahorodny W, Peng B, et al. The association of autism diagnosis with socioeconomic status. Autism. 2012;16(2):201-213. doi: 10.1177/1362361311413397 [DOI] [PubMed] [Google Scholar]

[zoi260993r33] 33.Kelly B, Williams S, Collins S, et al. The association between socioeconomic status and autism diagnosis in the United Kingdom for children aged 5-8 years of age: findings from the Born in Bradford cohort. Autism. 2019;23(1):131-140. doi: 10.1177/1362361317733182 [DOI] [PubMed] [Google Scholar]

[zoi260993r34] 34.Tertti K, Eskola E, Rönnemaa T, Haataja L. Neurodevelopment of two-year-old children exposed to metformin and insulin in gestational diabetes mellitus. J Dev Behav Pediatr. 2015;36(9):752-757. doi: 10.1097/DBP.0000000000000230 [DOI] [PubMed] [Google Scholar]

[zoi260993r35] 35.Greger HK, Hanem LGE, Østgård HF, Vanky E. Cognitive function in metformin exposed children, born to mothers with PCOS - follow-up of an RCT. BMC Pediatr. 2020;20(1):60. doi: 10.1186/s12887-020-1960-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r36] 36.Ijäs H, Vääräsmäki M, Saarela T, Keravuo R, Raudaskoski T. A follow-up of a randomised study of metformin and insulin in gestational diabetes mellitus: growth and development of the children at the age of 18 months. BJOG. 2015;122(7):994-1000. doi: 10.1111/1471-0528.12964 [DOI] [PubMed] [Google Scholar]

[zoi260993r37] 37.Wouldes TA, Battin M, Coat S, Rush EC, Hague WM, Rowan JA. Neurodevelopmental outcome at 2 years in offspring of women randomised to metformin or insulin treatment for gestational diabetes. Arch Dis Child Fetal Neonatal Ed. 2016;101(6):F488-F493. doi: 10.1136/archdischild-2015-309602 [DOI] [PubMed] [Google Scholar]

[zoi260993r38] 38.Deussen AR, Louise J, Dodd JM. Childhood follow-up of the GRoW randomized trial: Metformin in addition to dietary and lifestyle advice for pregnant women with overweight or obesity. Pediatr Obes. 2023;18(1):e12974. doi: 10.1111/ijpo.12974 [DOI] [PMC free article] [PubMed] [Google Scholar]

[zoi260993r39] 39.Hanem LGE, Salvesen Ø, Juliusson PB, et al. Intrauterine metformin exposure and offspring cardiometabolic risk factors (PedMet study): a 5-10 year follow-up of the PregMet randomised controlled trial. Lancet Child Adolesc Health. 2019;3(3):166-174. doi: 10.1016/S2352-4642(18)30385-7 [DOI] [PubMed] [Google Scholar]

PERMALINK

Metformin Exposure in Pregnancy and Childhood Developmental Outcomes

Hannah Gordon, MD

Richard J Hiscock, MBBS

Sarah Price, PhD

Alexis Shub, PhD

Alexandra Roddy Mitchell, MPH

Jessica Atkinson, BBiomedSc (Hons)

Susan Walker, MD

Anna Forsythe, MEnv

Stephen Tong, PhD

Anthea Lindquist, DPhil

Roxanne Hastie, PhD

Key Points

Question

Findings

Meaning

Abstract

Importance

Objective

Design, Setting, and Participants

Exposure

Main Outcome and Measures

Results

Conclusions and Relevance

Introduction

Methods

Exposure

Outcome

Exclusions

Statistical Analysis

Missing Data

IPWRA

Figure 1. Dot Plot of the Propensity Score Covariate Balance for the Primary Cohort.

Sensitivity Analyses

Results

Demographic and Clinical Characteristics

Figure 2. Participant Flow Diagram.

Table 1. Demographic and Clinical Characteristics by Metformin Exposure.

Developmental Vulnerability

Table 2. Developmental Outcomes on the AEDC by Metformin Exposure.

Sensitivity Analyses

Table 3. Developmental Outcomes on the AEDC by Metformin Exposure Among Women With Gestational Diabetes or Type 2 Diabetes in Pregnancy.

Discussion

Strengths and Limitations

Conclusions

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases