Key Points
Question
Is glucagon-like peptide-1 receptor agonist (GLP-1RA) use before pregnancy associated with gestational weight gain?
Findings
In this cohort study, compared with individuals who were not prescribed GLP-1RAs before pregnancy, those who were prescribed GLP-1RAs with discontinuation before or in early pregnancy had a 3.3-kg greater gestational weight gain, a statistically significant difference.
Meaning
Prior GLP-1RA use with discontinuation for pregnancy is associated with greater gestational weight gain.
Abstract
Importance
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are contraindicated in pregnancy. Discontinuation of GLP-1RAs proximal to pregnancy could affect gestational weight gain and pregnancy outcomes.
Objective
To compare gestational weight gain and pregnancy outcomes with and without exposure to GLP-1RAs before or during early pregnancy.
Design, Setting, and Participants
Retrospective cohort study of 149 790 singleton pregnancies delivered between June 1, 2016, and March 31, 2025, within a single academic health system.
Exposure
A GLP-1RA order between 3 years before and 90 days after conception, with propensity score matching of each exposed pregnancy to 3 unexposed pregnancies.
Main Outcomes and Measures
The primary outcome was gestational weight gain. Secondary outcomes were excess gestational weight gain, large and small for gestational age birth weight, birth weight percentile for gestational age and sex, birth length, preterm delivery, cesarean delivery, gestational diabetes, and hypertensive disorders of pregnancy.
Results
Among 149 790 pregnancies during the study period, 1792 (448 exposed and 1344 unexposed) were matched for the primary analysis. Exposed pregnancies had mean maternal age of 34.0 years (SD, 4.7 years) and prepregnancy body mass index of 36.1 (SD, 6.5; calculated as weight in kilograms divided by height in meters squared); 378 of 448 (84%) had obesity and 104 of 448 (23%) had preexisting diabetes; 136 (30%) were Hispanic, 49 (11%) were non-Hispanic Black, and 223 (50%) were non-Hispanic White; and 43 (10%) had public insurance. The GLP-1RA–exposed pregnancies had greater gestational weight gain (mean, 13.7 kg [SD, 9.2]) than propensity score–matched unexposed pregnancies (mean, 10.5 kg [SD, 8.0]), a difference of 3.3 kg (95% CI, 2.3-4.2; P < .001). The GLP-1RA–exposed group had a higher risk of excess gestational weight gain (65% vs 49%; risk ratio [RR], 1.32; 95% CI, 1.19-1.47), greater mean birth weight percentile (58.4% vs 54.8%; difference, 3.6%; 95% CI, 0.2%-6.9%), and higher risk of preterm delivery (17% vs 13%; RR, 1.34; 95% CI, 1.06-1.69), gestational diabetes (20% vs 15%; RR, 1.30; 95% CI, 1.01-1.68), and hypertensive disorders of pregnancy (46% vs 36%; RR, 1.29; 95% CI, 1.12-1.49). There was no difference in birth length, risk of large or small for gestational age birth weight, or cesarean delivery.
Conclusions and Relevance
In a cohort composed primarily of women with obesity, GLP-1RA use with subsequent prepregnancy or early pregnancy discontinuation was associated with more gestational weight gain and a higher risk of preterm delivery, gestational diabetes, and hypertensive disorders of pregnancy.
This cohort study examines gestational weight gain and pregnancy outcomes after glucagon-like peptide-1 (GLP-1) receptor agonist discontinuation.
Introduction
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are therapeutic options for obesity and type 2 diabetes that improve glycemia, weight, and cardiovascular outcomes.1 In 2019, 29% of births in the United States occurred in women with prepregnancy obesity, an increase from 26% in 2016.2 The prevalence of type 2 diabetes in pregnancy increased from 1.8 to 7.3 per 1000 births between 2000 and 2019.3 Diabetes and obesity increase the risk of pregnancy complications, as well as long-term metabolic disease for the offspring.4,5,6,7,8,9 Thus, optimization of hyperglycemia and weight before pregnancy is strongly recommended.4,10 GLP-1RAs are increasingly used to reduce both weight and glucose levels in women of reproductive age. However, GLP-1RA use is not recommended in pregnancy because of animal toxicity studies demonstrating fetal structural abnormalities, intrauterine growth restriction, and embryofetal mortality,5,6,7,8,9 leading to GLP-1RA discontinuation before conception or on learning of pregnancy.
A recent large multinational observational study found no difference in the risk of major congenital malformation in individuals with diabetes and periconceptional exposure to GLP-1RAs compared with insulin.11 Data on associations between GLP-1RA exposure and adverse pregnancy outcomes beyond birth defects remain limited, conflicting, and subject to residual confounding by indication or selection and recall bias.12,13,14,15 Outside of pregnancy, discontinuation of GLP-1RA therapy is associated with weight regain and worsening hyperglycemia.16,17 During pregnancy, excess gestational weight gain is associated with an increased risk of adverse maternal and neonatal outcomes, including large for gestational age birth weight,18 preterm delivery,19 gestational diabetes,19 hypertensive disorders of pregnancy,20 and need for cesarean delivery.21 Excess gestational weight gain is also associated with adverse consequences for the future metabolic health of both the mother and offspring.22 Thus, if GLP-1RA discontinuation before or during early pregnancy leads to excess gestational weight gain, negative outcomes for the parent-child dyad could result. We therefore sought to evaluate gestational weight gain and related secondary outcomes in individuals exposed to GLP-1RAs in the prepregnancy and early pregnancy period compared with unexposed individuals with similar characteristics.
Methods
We performed a retrospective cohort study of singleton pregnancies with delivery dates between June 1, 2016, and March 31, 2025, at Mass General Brigham, an academic health system with 15 institutions serving the greater Boston area in Massachusetts. The Mass General Brigham institutional review board approved the use of electronic health record (EHR) data and waived the requirement for informed consent, given that minimal risk was involved. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.23
Data acquisition and exposure, covariate, and outcome ascertainment are described below, with more details available in the eMethods in Supplement 1. Maternal medication, anthropometric, and delivery data were obtained from the Mass General Brigham Enterprise Data Warehouse; additional maternal medication, insurance, diagnostic, and demographic data were obtained from the Mass General Brigham Research Patient Data Registry.24
The primary exposure was defined as at least 1 GLP-1RA medication order in the EHR between 3 years before and 90 days after conception.25 The study exposure window was chosen to capture GLP-1RA use that could affect weight patterns in the months to years after discontinuation.16 Individuals with no GLP-1RA medications recorded between 3 years before pregnancy and 90 days after conception were classified as unexposed. The exposed and unexposed classifications were validated through manual medical record review of a subset of exposed participants and a key word search of unstructured note text from a subset of unexposed participants most likely to be matched. The date of the last medication order before delivery was used to estimate discontinuation time in relation to conception. Duration of use was defined as the time between the first medication order and the last medication order before delivery, including orders outside the defined exposure window.
Multiple-gestation pregnancies and pregnancies in individuals with a history of bariatric surgery were excluded because these factors could influence gestational weight gain and other pregnancy outcomes. Additionally, pregnancies with missing gestational age at delivery or with only implausible weight and height information were excluded. A single pregnancy was included for each individual to avoid matching people to themselves in other pregnancies. The included pregnancy was selected at random, although exposed pregnancies were prioritized.
The primary outcome was gestational weight gain, defined as the last weight recorded within 1 week before delivery minus the prepregnancy weight. Prepregnancy weight was self-reported and documented in the EHR; typically, clinicians recorded this information at the initiation of prenatal care. If this information was unavailable, measured prepregnancy weights (between 3 months before and 4 weeks after the conception date) were obtained from the EHR.
Secondary outcomes included excess gestational weight gain, neonatal birth weight outcomes, and pregnancy outcomes. Excess gestational weight gain was determined according to prepregnancy body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), calculated gestational weight gain, and cutoffs as defined by the Institute of Medicine 2009 guidelines.26 Neonatal birth weight outcomes were determined with EHR delivery documentation of measurements, sex, and gestational age at birth and included birth weight percentile,27 large for gestational age birth weight (>90th percentile for gestational age and sex), small for gestational age birth weight (<10th percentile for gestational age and sex), and birth length. Pregnancy outcomes included preterm delivery (28-36 weeks’ gestation), cesarean delivery, gestational diabetes, and hypertensive disorders of pregnancy. Preterm delivery was determined from gestational age recorded in delivery records; spontaneous and medically indicated preterm deliveries were also examined separately. Gestational diabetes was defined with oral glucose tolerance test results; hypertensive disorders of pregnancy were defined according to American College of Obstetricians and Gynecologists criteria using blood pressure data.28,29 In cases of missing blood pressure or laboratory data, International Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes were used to identify clinician diagnoses of gestational diabetes or hypertensive disorders of pregnancy.
Three analytic cohorts were defined. The gestational weight gain cohort was used for analyzing gestational weight gain and excess gestational weight gain and was restricted to term pregnancies with observed weight gain. The birth weight cohort was used for analyzing large for gestational age, small for gestational age, birth weight percentile, and birth length and was restricted to term pregnancies with observed birth weight and neonatal sex. The obstetric cohort was used for analyzing pregnancy outcomes (preterm and cesarean delivery, gestational diabetes, and hypertensive disorders of pregnancy) and included pregnancies with observed mode of delivery at greater than or equal to 28 weeks’ gestation. For the gestational diabetes analysis, pregnancies with preexisting diabetes were excluded. For the hypertensive disorders of pregnancy analysis, pregnancies with chronic hypertension were excluded.
A propensity score model30,31 for the likelihood of GLP-1RA exposure before or in early pregnancy was used to account for confounding when comparing GLP-1RA–exposed vs GLP-1RA–unexposed pregnancies. The model included clinical factors that could have influenced the decision to start treatment, as well as social factors that may affect GLP-1RA access: prepregnancy BMI, maternal age, parity (nulliparous vs multiparous), race and ethnicity (Hispanic, non-Hispanic Asian, non-Hispanic Black, non-Hispanic White, multiracial [anyone who selected >1 race], and none of the above [Native Hawaiian and Pacific Islander, Indian or Alaska Native, and race not listed or unavailable]), insurance type (public vs nonpublic as defined in the eMethods in Supplement 1), preexisting diabetes (except in the gestational diabetes analysis because this group was excluded), preexisting chronic hypertension (except in the hypertensive disorders of pregnancy analysis because this group was excluded), year of delivery, and delivery location within the health system. Race and ethnicity were self-reported and documented in the EHR. They were considered potential confounders because they may affect access to GLP-1RAs32 and are associated with differences in pregnancy outcomes. Hemoglobin A1c values and ICD-10 codes were used to identify preexisting diabetes. Blood pressure, antihypertensive medication orders, and ICD-10 codes were used to identify chronic hypertension. GLP-1RA–exposed pregnancies were matched to unexposed pregnancies by propensity score in a 1:3 ratio using nearest neighbor matching,33 allowing estimation of the average treatment effect among treated individuals (ie, the effect of GLP-1RA exposure among those who were exposed).34
Several secondary analyses were conducted. In the first, exposed pregnancies were restricted to recent exposure to GLP-1RAs (last EHR order after 6 months before conception). In the second, exposure to liraglutide and semaglutide were studied separately; unexposed matches were drawn from the same pool of pregnancies as in the primary analysis. In the third, pregnancies were stratified by preexisting diabetes status. Sensitivity analyses included adjustment for preexisting diabetes, prepregnancy BMI, and chronic hypertension to resolve any residual imbalances despite propensity score matching and exclusion of pregnancies with preexisting diabetes or hypertension.
Linear regression was used for the primary outcome of gestational weight gain and for the secondary outcomes of birth weight percentile and birth length. Risk ratios (RRs) for other secondary outcomes (excess gestational weight gain, large for gestational age, small for gestational age, preterm delivery, cesarean delivery, gestational diabetes, and hypertensive disorders of pregnancy) were estimated with log-binomial regression. Cluster-robust SEs were used throughout to account for matching.33 Statistical tests were 2-sided. Missing data for maternal age, race and ethnicity, and insurance type were addressed using multiple imputation by chained equations,35 with 25 imputations. Missing prepregnancy BMI was multiply imputed only in the birth weight and obstetric analytic cohorts. A simulation-based power calculation was performed to estimate the minimum detectable effect size with the final primary analysis sample size of 448 GLP-1RA–exposed pregnancies. Assuming a matching ratio of 1:3, we had 80% power to detect a difference in gestational weight gain of 1.11 kg at a significance level of .05, assuming an SD for gestational weight gain of 6.80 kg. Propensity score matching and multiple imputation were performed with the MatchThem36 and mice37 packages in R, respectively. Cluster-robust SEs were estimated with the survey38 package. Analyses were conducted in R version 4.4.0 (R Foundation for Statistical Computing).
Results
Among 149 790 pregnancies delivered between June 2016 and March 2025, 655 pregnancies were exposed to GLP-1RAs between 3 years before and 90 days after conception (Figure 1). After exclusions and propensity score matching, 448 exposed pregnancies were matched to 1344 unexposed pregnancies in the gestational weight gain cohort (Figure 1), 442 exposed pregnancies were matched to 1326 unexposed pregnancies in the birth weight cohort (Figure 1), and 566 exposed pregnancies were matched to 1698 unexposed pregnancies in the obstetric cohort (eFigure 1 in Supplement 1).
Figure 1. Derivation of the Gestational Weight Gain and Birth Weight Cohorts Exposed to Glucagon-Like Peptide-1 Receptor Agonists.
GWG indicates gestational weight gain; MGB, Mass General Brigham.
aPregnancies were chosen randomly per individual. In individuals with both exposed and unexposed pregnancies, 1 of the exposed pregnancies was randomly selected.
Table 1 shows the baseline characteristics of the 82 790 pregnancies in the gestational weight gain unmatched cohort and the 1792 matched pregnancies. The characteristics of the birth weight and obstetric cohorts were similar to those in the gestational weight gain cohort (eTables 1 and 2 in Supplement 1), with the exception of a higher public insurance rate in the birth weight and obstetric cohorts. Individuals in the unmatched gestational weight gain cohort had a mean age of 33.2 years (SD, 4.9 years), BMI of 26.2 (SD, 5.9), and gestational weight gain of 13.9 kg (SD, 6.3 kg); 972 (1.2%) had preexisting diabetes, 2389 (3%) had chronic hypertension, 6794 (8%) had a gestational diabetes diagnosis, 15 443 (19%) developed hypertensive disorders of pregnancy, and 26 539 (32%) had a cesarean delivery. A total of 12 739 individuals (15%) were Hispanic/Latino; 1344 (2%) were multiracial; 9197 (11%) were non-Hispanic Asian; 5742 (7%) were non-Hispanic Black; 51 018 (62%) were non-Hispanic White; and 1999 (2%) were none of the above. Pregnancies exposed to GLP-1RAs occurred in individuals who were older (mean maternal age, 34.0 years [SD, 4.7 years]), had a higher prepregnancy BMI (36.1 [SD, 6.5]), and had a greater prevalence of obesity (378 of 448 [84%]) compared with unmatched cohorts. The GLP-1RA–exposed pregnancies also had higher rates of preexisting diabetes (104 of 448 [23%]) and chronic hypertension (94 of 448 [21%]) (Table 1). A higher proportion of individuals identified as Black or Latino and a lower proportion had public insurance among GLP-1RA–exposed pregnancies compared with the unmatched cohorts. Most of the exposed pregnancies delivered in later study years, congruent with an increase in GLP-1RA use across the study period (Figure 2). The most frequently ordered GLP-1RA was semaglutide. Two hundred fifty-nine (58%) of the exposed group in the primary analysis had GLP-1RA treatment duration less than 6 months (Table 1). Two hundred ninety-four (66%) of the exposed group had a last GLP-1RA order within 6 months of conception (Table 1; eTable 3 in Supplement 1). Medical record review of a random sample (N = 50) of GLP-1RA–exposed pregnancies suggested that GLP-1RA therapy was discontinued before or during early pregnancy for all individuals and that the most common reason for discontinuation was pregnancy or planning pregnancy (eMethods in Supplement 1). The exposed and matched unexposed groups appeared well balanced, with standardized mean difference less than 0.1 for all matching variables (Table 1; eFigures 2-4 in Supplement 1).
Table 1. Characteristics of the Gestational Weight Gain Analytic Cohort.
| No. (%) | Standardized mean difference after matching | |||
|---|---|---|---|---|
| Gestational weight gain analytic cohort before matching (n = 82 790) | Matched | |||
| GLP-1RA exposed (n = 448) | Unexposed (n = 1344) | |||
| Maternal age at delivery, mean (SD), y | 33.2 (4.9) | 34.0 (4.7) | 33.9 (5.1) | 0.011 |
| Missing | 2392 (3) | |||
| Pregravid BMI, mean (SD) | 26.2 (5.9) | 36.1 (6.5) | 36.3 (9.3) | −0.038 |
| Pregravid BMI categories | ||||
| 18-24.9 | 42 993 (52) | 9 (2) | 142 (11) | |
| 25-29.9 | 22 285 (27) | 61 (14) | 219 (16) | |
| 30-34.9 | 10 152 (12) | 141 (31) | 285 (21) | |
| 35-39.9 | 4520 (5) | 127 (28) | 265 (20) | |
| ≥40 | 2840 (3) | 110 (25) | 432 (32) | |
| Gestational age at delivery, mean (SD), wk | 39.4 (1.2) | 38.7 (1.1) | 38.9 (1.2) | |
| Nulliparous | 41 806 (50) | 192 (43) | 565 (42) | 0.008 |
| Preexisting diabetes | 972 (1) | 104 (23) | 232 (17) | 0.060 |
| Chronic hypertension | 2389 (3) | 94 (21) | 266 (20) | 0.012 |
| Year of delivery | ||||
| Jun 2016-May 2020a | 35 072 (42) | 15 (3) | 32 (2) | |
| Jun 2020-May 2021 | 8914 (11) | 17 (4) | 47 (3) | 0.003 |
| Jun 2021-May 2022 | 9896 (12) | 26 (6) | 68 (5) | 0.008 |
| Jun 2022-May 2023 | 9736 (12) | 67 (15) | 190 (14) | 0.008 |
| Jun 2023-May 2024 | 10 250 (12) | 131 (29) | 399 (30) | −0.005 |
| Jun 2024-Mar 2025b | 8922 (11) | 192 (43) | 608 (45) | −0.024 |
| Race and ethnicity | ||||
| Hispanic/Latino | 12 739 (15) | 136 (30) | 415 (31) | −0.007 |
| Multiracialc | 1344 (2) | 6 (1) | 19 (1) | −0.0009 |
| Non-Hispanic Asian | 9197 (11) | 29 (6) | 78 (6) | 0.007 |
| Non-Hispanic Black | 5742 (7) | 49 (11) | 145 (11) | 0.003 |
| Non-Hispanic White | 51 018 (62) | 223 (50) | 674 (50) | −0.004 |
| Missing | 751 (1) | |||
| None of the aboved | 1999 (2) | 5 (1) | 12 (0.9) | 0.002 |
| Public insurance | 9008 (11) | 43 (10) | 127 (9) | .001 |
| Missing | 3557 (4) | |||
| GLP-1RA medication | ||||
| Liraglutide | 118 (26) | |||
| Dulaglutide | 92 (21) | |||
| Semaglutide | 202 (45) | |||
| Tirzepatide | 31 (7) | |||
| Other GLP-1RAse | 5 (1) | |||
| Duration of GLP-1RA treatment, mo | ||||
| <6 | 259 (58) | |||
| 6 to <12 | 59 (13) | |||
| ≥12 | 130 (29) | |||
| Timing of last GLP-1RA order | ||||
| ≥12 mo before conception | 77 (17) | |||
| 6 to <12 mo before conception | 77 (17) | |||
| <6 mo to 0 d before conception | 220 (49) | |||
| 1-90 d after conception | 70 (16) | |||
| >90 d after conception | 4 (0.9) | |||
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); GLP-1RA, glucagon-like peptide-1 receptor agonist. Blank cells indicate not applicable.
Represents 4 years during the study period.
Represents only 10 months.
Anyone who selected more than 1 race.
Includes Native Hawaiian and Pacific Islander, Indian or Alaska Native, and race not listed or unavailable.
Other GLP-1RAs include exenatide and lixisenatide.
Figure 2. Prevalence of Exposure by Year in the Gestational Weight Gain Cohort.
Exposure was defined as the presence of at least 1 glucagon-like peptide-1 receptor agonist (GLP-1RA) order between 3 years before conception and 90 days after conception. Rates of GLP-1RA exposure are reported per 1000 term deliveries in the gestational weight gain analytic cohort after all exclusions and before propensity score matching.
In the propensity score–matched gestational weight gain cohort, mean gestational weight gain was 13.7 kg (SD, 9.2 kg) in the GLP-1RA–exposed group and 10.5 kg (SD, 8.0 kg) in the unexposed group, a difference of 3.3 kg (95% CI, 2.3-4.2; P < .001) (Figure 3). The risk of excess gestational weight gain was greater in the exposed compared with the unexposed group (65% vs 49%; RR, 1.32; 95% CI, 1.19-1.47) (Figure 3).
Figure 3. Associations Between Glucagon-Like Peptide-1 Receptor Agonist (GLP-1RA) Exposure and Gestational Weight Gain, Birth Weight, and Obstetric Outcomes.
The GLP-1RA–exposed pregnancies were compared with propensity-matched unexposed pregnancies. All: exposure defined as GLP-1RA order between 3 years before and 90 days after conception. Recent: exposure defined as GLP-1RA order between 6 months before and 90 days after conception. Semaglutide and liraglutide analyses considered only these specific GLP-1RAs. Excess gestational weight gain: as per Institute of Medicine 2009 guidelines.26 Large for gestational age birth weight: >90th percentile for gestational age. Small for gestational age birth weight: <10th percentile for gestational age.
Results in the secondary gestational weight gain analyses (recent exposure with prescription within 6 months of conception, separate analysis of liraglutide and semaglutide exposure, and with and without diabetes) were similar to those of the main analysis (Figure 3). In analyses examining semaglutide and liraglutide exposure separately, the point estimate for the gestational weight gain effect was lower in the liraglutide group than in the semaglutide group, albeit with overlapping CIs (Figure 3).
In the propensity score–matched birth weight cohort, the risks of large and small for gestational age birth weight were similar between the exposed and unexposed groups (large for gestational age: 16% vs 15%; RR, 1.09; 95% CI, 0.83-1.42; small for gestational age: 8% vs 8%; RR, 0.97; 95% CI, 0.65-1.44) (Figure 3). There was a higher mean birth weight percentile in the GLP-1RA–exposed group (58.4% vs 54.8%; difference, 3.6%; 95% CI, 0.2%-6.9%) but no difference in birth length (49.8 vs 49.9 cm; difference, −0.1; 95% CI, −0.4 to 0.1). Results were similar for birth weight outcomes in secondary analyses (recent exposure, liraglutide and semaglutide, and with or without diabetes) (Figure 3; eTables 4 and 5 in Supplement 1).
In the propensity score–matched obstetric cohort, risks of preterm delivery (17% vs 13%; RR, 1.34; 95% CI, 1.06-1.69), gestational diabetes (20% vs 15%; RR, 1.30; 95% CI, 1.01-1.68), and hypertensive disorders of pregnancy (46% vs 36%; RR, 1.29; 95% CI, 1.12-1.49) were higher in the GLP-1RA–exposed group (Figure 3). There was no association between GLP-1RA use and the risk of cesarean delivery (Figure 3). When examined separately, the point estimates for the associations between GLP-1RA use and spontaneous and indicated preterm delivery were similar to that of the full preterm analysis, but the 95% CIs overlapped and included the null value. Semaglutide and liraglutide analyses for the obstetric cohort are shown in eTable 5 in Supplement 1.
Sensitivity analyses showed findings similar to those of the main analyses (Table 2).
Table 2. Sensitivity Analyses Adjusting for Body Mass Index, Chronic Hypertension, Preexisting Diabetes, and Gestational Age at Delivery.
| Outcomes | GLP-1RA | Adjusted (95% CI) | ||||
|---|---|---|---|---|---|---|
| Exposed | Unexposed | Relative risk | Mean difference | |||
| No. of pregnancies | No. (%) | No. of pregnancies | No. (%) | |||
| Matched gestational weight gain cohort, mean (SD), kg | ||||||
| Mean gestational weight gain adjusted for BMI and preexisting diabetesa | 448 | 13.74 (9.2) | 1344 | 10.49 (8.0) | 3.28 (2.32 to 4.24) | |
| Mean gestational weight gain adjusted for gestational age at delivery | 448 | 13.74 (9.2) | 1344 | 10.49 (8.0) | 3.36 (2.38 to 4.33) | |
| Matched birth weight cohort | ||||||
| Large for gestational age adjusted for BMI category, preexisting diabetes, and chronic hypertensionb,c | 442 | 71 (16) | 1326 | 196 (15) | 1.03 (0.79 to 1.35) | |
| Small for gestational age adjusted for BMI, preexisting diabetes, and chronic hypertensiona,d | 442 | 34 (8) | 1326 | 105 (8) | 0.98 (0.66 to 1.45) | |
| Birth weight percentile adjusted for BMI, preexisting diabetes, and chronic hypertension, mean (SD), %a | 442 | 58.4 (29.5) | 1326 | 54.8 (29.3) | 3.27 (−0.07 to 6.61) | |
| Birth length adjusted for BMI, preexisting diabetes, and chronic hypertension, mean (SD), cma | 442 | 49.8 (2.3) | 1326 | 49.9 (2.4) | −0.12 (−0.38 to 0.14) | |
| Matched obstetric cohort | ||||||
| Preterm delivery adjusted for obesity, preexisting diabetes, and chronic hypertensione | 566 | 99 (17) | 1698 | 223 (13) | 1.21 (0.96 to 1.53) | |
| Cesarean delivery adjusted for obesity, preexisting diabetes, and chronic hypertensione | 566 | 263 (46) | 1698 | 790 (47) | 0.94 (0.85 to 1.05) | |
| Gestational diabetes adjusted for obesity and chronic hypertensione | 413 | 83 (20) | 1239 | 191 (15) | 1.23 (0.95 to 1.58) | |
| Hypertensive disorders of pregnancy adjusted for obesity and preexisting diabetese | 432 | 200 (46) | 1296 | 465 (36) | 1.15 (1.00 to 1.32) | |
| Matched obstetric cohort excluding preexisting hypertension and diabetes | ||||||
| Preterm delivery adjusted for BMI categoryc | 324 | 36 (11) | 972 | 78 (8) | 1.37 (0.87 to 2.16) | |
| Gestational diabetes adjusted for BMI categoryc | 324 | 63 (19) | 972 | 129 (13) | 1.38 (0.98 to 1.94) | |
| Hypertensive disorders of pregnancy adjusted for BMI categoryc | 324 | 136 (42) | 972 | 317 (33) | 1.19 (0.99 to 1.45) | |
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); GLP-1RA, glucagon-like peptide-1 receptor agonist. Blank cells indicate not applicable.
Body mass index was adjusted for as a continuous variable.
Large for gestational age: birth weight greater than 90th percentile for gestational age and sex.
When the model adjusting for continuous BMI failed to converge, adjustment for BMI category was performed (defined as BMI <25, ≥25 to <30, ≥30 to <35, ≥35 to <40, or ≥40).
Small for gestational age: birth weight less than 10th percentile for gestational age and sex.
When the models adjusting for continuous BMI and categorical BMI failed to converge, adjustment for obesity was performed (BMI <30 or ≥30).
Discussion
In this retrospective EHR-based cohort study, GLP-1RA exposure with prepregnancy or early pregnancy discontinuation was associated with greater gestational weight gain and a higher risk of certain adverse pregnancy outcomes (preterm delivery, gestational diabetes, and hypertensive disorders of pregnancy) compared with propensity-matched unexposed pregnancies. Despite a small detectable increase in birth weight of newborns from GLP-1RA–exposed pregnancies, there was no increased risk of large or small for gestational age birth weight or cesarean delivery associated with prepregnancy or early pregnancy GLP-1RA use. These findings are consistent with those of prior studies that observed weight regain after discontinuation of GLP-1RA therapy outside pregnancy16,17 and may guide obstetric risk assessment for women who discontinue GLP-1RAs for pregnancy.
It is possible that benefits of prepregnancy GLP-1RA use could be negated or attenuated by the adverse effects of GLP-1RA discontinuation. The findings of this study are consistent with those of a recent smaller, single-center study, which found a lower risk of inadequate gestational weight gain associated with prepregnancy or early-pregnancy GLP-1RA use; there was no statistically significant difference in the risk of excess gestational weight gain associated with GLP-1RA exposure, but the point estimate for the risk of excess gestational weight gain was similar to that found in the present study.15 Outside the context of GLP-1RA use, greater gestational weight gain, particularly in the first trimester, is associated with increased risk of hypertensive disorders of pregnancy and gestational diabetes.19,20,21 Given the established relationship between gestational weight gain and adverse pregnancy outcomes, it is possible that greater gestational weight gain contributed to the increased risk of gestational diabetes and hypertensive disorders of pregnancy observed in this study after GLP-1RA discontinuation for pregnancy. Furthermore, gestational diabetes and hypertensive disorders of pregnancy are both associated with preterm birth. Although prior studies have implicated prepregnancy obesity and excess gestational weight gain in the risk of fetal overgrowth,21 this study did not find a higher risk of large for gestational age in GLP-1RA–exposed pregnancies despite slightly higher birth weight percentile for gestational age. Prepregnancy BMI may be a stronger determinant of birth weight than gestational weight gain.39 However, exposure to excess gestational weight gain in utero is independently associated with increased adiposity in midchildhood, highlighting the need for long-term follow-up studies of offspring after GLP-1RA exposure and discontinuation for pregnancy.40
The findings of this study contrast with those from 2 recent large retrospective studies reported in the US Collaborative Network in TriNetX that used medications and diagnostic codes to examine pregnancy complications associated with GLP-1RA exposure. Imbroane et al13 investigated 4267 adults with GLP-1RA exposure within 2 years of pregnancy compared with controls matched by propensity score (based on age, race and ethnicity, a broad categorical overweight/obesity designation, prediabetes, type 2 diabetes, and chronic hypertension), finding a lower risk of gestational diabetes, hypertensive disorders of pregnancy, preterm delivery, and cesarean delivery in those exposed to GLP-1RA. Hanif et al14 used the same dataset and compared 1826 adult women with type 2 diabetes and exposure within 1 year of pregnancy with controls matched by propensity score on age, race, continuous BMI, comorbid conditions obtained from diagnostic codes, laboratory values, and medications and found no difference in hypertensive disorder of pregnancy risk in those who were exposed vs unexposed. These studies used a large, representative dataset but primarily relied on diagnostic codes from administrative data to ascertain covariates and outcomes, potentially resulting in misclassification and residual confounding.41 Neither prior study included insurance status, a socioeconomic indicator and determinant of access to GLP-1RA therapy, in its propensity score model. In contrast, the present study used propensity scoring that incorporated continuous BMI, insurance status, delivery hospital, and comorbidities identified with quantitative clinical data in combination with diagnostic codes. Outcomes in the present study were also largely ascertained via quantitative clinical data (eg, gestational age at delivery, laboratory values, directly measured weights and blood pressure). Differences in outcome and covariate ascertainment, as well as differences in propensity score modeling variables, may account for the differences in findings observed across studies.
Additional studies are needed to inform preconception counseling of individuals currently using GLP-1RA or considering GLP-1RA therapy. In particular, more data are needed on the implications of GLP-1RA discontinuation–associated gestational weight gain for long-term weight retention in mothers and programming effects in offspring that may be independent of birth weight. Moreover, studies that quantify and evaluate the potential benefits of GLP-1RA–induced weight loss before pregnancy could provide further evidence on the balance of benefits and risks associated with initiating GLP-1RA therapy that will be interrupted for pregnancy. Finally, future studies should define clinical approaches to mitigate excess gestational weight gain in individuals with obesity and diabetes who were using GLP-1RAs before pregnancy.
Limitations
This study had several limitations. First, it matched on prepregnancy BMI rather than pre–GLP-1RA treatment BMI and therefore did not address potential benefits of initiating GLP-1RA therapy with the goal of entering pregnancy at a lower weight. Second, as is the case in all observational studies, residual confounding may be present. This study demonstrated minimal residual confounding across the variables used in the propensity score model for GLP-1RA exposure, with less than 0.1 standardized mean difference for all variables, but had some attenuation in relative risks for obstetric outcomes when additional covariate adjustment in the propensity score–matched analyses was used. Third, because all identified exposed pregnancies were matched to a subset of identified unexposed pregnancies, the study was only able to estimate the average treatment effect among treated individuals rather than the overall average treatment effect (ie, the effect of GLP-1RA on all pregnancies) or the average treatment effect among untreated individuals (ie, the effect of GLP-1RA among pregnancies that were actually unexposed).42 Therefore, this study does not exactly emulate a hypothetical clinical trial that randomizes individuals prepregnancy to GLP-1RA vs placebo, which could provide an estimate of the overall average treatment effect. Fourth, inherent to a large retrospective EHR-based study, data are limited to structured fields from clinical encounters. Because the analysis relies on clinician orders, whether patients actually used their prescribed medications could not be confirmed and estimates of discontinuation timing were imperfect. Misclassification of the exposure in the case of discontinuity of care was also possible because the medication order data source was limited to a single health system. However, successful validation of the exposed vs unexposed classifications increases confidence in the robustness of the primary exposure ascertainment. Fifth, given the limited numbers of individuals with GLP-1RA orders in the EHR after conception, this group was not examined separately from individuals with exposure in the last 6 months before conception. Sixth, findings for gestational weight gain and birth weight may be generalizable only to term pregnancies in individuals with obesity. Seventh, the Mass General Brigham health system’s population is older, less likely to have a BMI greater than or equal to 25, less representative of Black and Latino individuals, and less likely to have public insurance than the US birthing population.43
Conclusions
There is limited clinical guidance available on the risks of discontinuation of GLP-1RAs for pregnancy. In a cohort composed primarily of women with obesity, GLP-1RA use with subsequent prepregnancy or early pregnancy discontinuation was associated with more gestational weight gain and a higher risk of preterm delivery, gestational diabetes, and hypertensive disorders of pregnancy.
eMethods. Supplementary Methods
eTable 1. Baseline Characteristics of the Overall and Matched Birthweight Analytic Cohort
eTable 2. Baseline Characteristics of the Overall and Matched Obstetric Analytic Cohort
eTable 3. Duration of Treatment and Time From Discontinuation to Conception in the Gestational Weight Gain Cohort
eTable 4. Secondary Outcomes for the Birthweight and Obstetric Cohorts
eTable 5. Associations Between Liraglutide and Semaglutide and Birthweight and Obstetric Outcomes
eFigure 1. Derivation of the Obstetric Cohort
eFigure 2. Gestational Weight Gain Cohort
eFigure 3. Birthweight Cohort
eFigure 4. Obstetric Cohort
eReferences.
Data Sharing Statement
References
- 1.American Diabetes Association . 8. Pharmacologic approaches to glycemic treatment. Diabetes Care. 2017;40(suppl 1):S64-S74. doi: 10.2337/dc17-S011 [DOI] [PubMed] [Google Scholar]
- 2.Driscoll AK, Gregory ECW. Increases in prepregnancy obesity: United States, 2016-2019. NCHS Data Brief. 2020;(392):1-8. [PubMed] [Google Scholar]
- 3.Gorsch LP, Wen T, Lonier JY, et al. Trends in delivery hospitalizations with pregestational and gestational diabetes mellitus and associated outcomes: 2000-2019. Am J Obstet Gynecol. 2023;229(1):63.e1-63.e14. doi: 10.1016/j.ajog.2022.12.006 [DOI] [PubMed] [Google Scholar]
- 4.American College of Obstetricians and Gynecologists’ Committee on Practice Bulletins–Obstetrics . ACOG practice bulletin no. 201: pregestational diabetes mellitus. Obstet Gynecol. 2018;132(6):e228-e248. doi: 10.1097/AOG.0000000000002960 [DOI] [PubMed] [Google Scholar]
- 5.US Food and Drug Administration . Ozempic (semaglutide) highlights of prescribing information. Accessed August 6, 2025. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/209637s025lbl.pdf
- 6.US Food and Drug Administration . Mounjaro (tirzepatide) highlights of prescribing information. Accessed August 6, 2025. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/215866s031lbl.pdf
- 7.US Food and Drug Administration . Byetta (exenatide) highlights of prescribing information. Accessed August 6, 2025. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/021773s050lbl.pdf
- 8.US Food and Drug Administration . Victoza (liraglutide) highlights of prescribing information. Accessed August 6, 2025. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/022341s044lbl.pdf
- 9.US Food and Drug Administration . Trulicity (dulaglutide) highlights of prescribing information. Accessed August 6, 2025. https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/125469s063lbl.pdf
- 10.Obesity in pregnancy: ACOG practice bulletin, number 230. Obstet Gynecol. 2021;137(6):e128-e144. doi: 10.1097/AOG.0000000000004395 [DOI] [PubMed] [Google Scholar]
- 11.Cesta CE, Rotem R, Bateman BT, et al. Safety of GLP-1 receptor agonists and other second-line antidiabetics in early pregnancy. JAMA Intern Med. 2024;184(2):144-152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dao K, Shechtman S, Weber-Schoendorfer C, et al. Use of GLP1 receptor agonists in early pregnancy and reproductive safety: a multicentre, observational, prospective cohort study based on the databases of six Teratology Information Services. BMJ Open. 2024;14(4):e083550. doi: 10.1136/bmjopen-2023-083550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Imbroane MR, LeMoine F, Nau CT. Preconception glucagon-like peptide-1 receptor agonist use associated with decreased risk of adverse obstetrical outcomes. Am J Obstet Gynecol. 2025;233(2):116.e1-116.e7. doi: 10.1016/j.ajog.2025.01.019 [DOI] [PubMed] [Google Scholar]
- 14.Hanif M, Hays AG, Nagarajan JS, Sah SP, Weinstock RS, Taub CC. Efficacy and safety of glucagon-like peptide-1 receptor agonist use during the first trimester in pregnant women with type 2 diabetes. Am J Cardiol. 2025;246:10-13. doi: 10.1016/j.amjcard.2025.03.013 [DOI] [PubMed] [Google Scholar]
- 15.Pondugula N, Culhane JF, Lundsberg LS, Partridge C, Merriam AA. Gestational weight gain and hypertensive disorders of pregnancy with prepregnancy and early pregnancy glucagon-like peptide-1 receptor agonist exposure. Obstet Gynecol. Published online July 3, 2025. doi: 10.1097/AOG.0000000000005995 [DOI] [Google Scholar]
- 16.Wilding JPH, Batterham RL, Davies M, et al. ; STEP 1 Study Group . Weight regain and cardiometabolic effects after withdrawal of semaglutide: the STEP 1 trial extension. Diabetes Obes Metab. 2022;24(8):1553-1564. doi: 10.1111/dom.14725 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Rubino D, Abrahamsson N, Davies M, et al. ; STEP 4 Investigators . Effect of continued weekly subcutaneous semaglutide vs placebo on weight loss maintenance in adults with overweight or obesity: the STEP 4 randomized clinical trial. JAMA. 2021;325(14):1414-1425. doi: 10.1001/jama.2021.3224 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Vesco KK, Sharma AJ, Dietz PM, et al. Newborn size among obese women with weight gain outside the 2009 Institute of Medicine recommendation. Obstet Gynecol. 2011;117(4):812-818. doi: 10.1097/AOG.0b013e3182113ae4 [DOI] [PubMed] [Google Scholar]
- 19.Champion ML, Harper LM. Gestational weight gain: update on outcomes and interventions. Curr Diab Rep. 2020;20(3):11. doi: 10.1007/s11892-020-1296-1 [DOI] [PubMed] [Google Scholar]
- 20.Dude AM, Kominiarek MA, Haas DM, et al. Weight gain in early, mid, and late pregnancy and hypertensive disorders of pregnancy. Pregnancy Hypertens. 2020;20:50-55. doi: 10.1016/j.preghy.2020.03.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Langford A, Joshu C, Chang JJ, Myles T, Leet T. Does gestational weight gain affect the risk of adverse maternal and infant outcomes in overweight women? Matern Child Health J. 2011;15(7):860-865. doi: 10.1007/s10995-008-0318-4 [DOI] [PubMed] [Google Scholar]
- 22.Voerman E, Santos S, Patro Golab B, et al. Maternal body mass index, gestational weight gain, and the risk of overweight and obesity across childhood: an individual participant data meta-analysis. PLoS Med. 2019;16(2):e1002744. doi: 10.1371/journal.pmed.1002744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577. doi: 10.7326/0003-4819-147-8-200710160-00010 [DOI] [PubMed] [Google Scholar]
- 24.Nalichowski R, Keogh D, Chueh HC, Murphy SN. Calculating the benefits of a research patient data repository. AMIA Annu Symp Proc. 2006;2006:1044. [PMC free article] [PubMed] [Google Scholar]
- 25.Committee opinion no. 700: methods for estimating the due date. Obstet Gynecol. 2017;129(5):e150-e154. doi: 10.1097/AOG.0000000000002046 [DOI] [PubMed] [Google Scholar]
- 26.Rasmussen KM, Yaktine AL, eds; Institute of Medicine (US) and National Research Council (US) Committee to Reexamine IOM Pregnancy Weight Guidelines . Weight Gain During Pregnancy: Reexamining the Guidelines. National Academies Press; 2009. Accessed November 29, 2023. https://www.ncbi.nlm.nih.gov/books/NBK32813/
- 27.Aris IM, Kleinman KP, Belfort MB, Kaimal A, Oken E. A 2017 US reference for singleton birth weight percentiles using obstetric estimates of gestation. Pediatrics. 2019;144(1):e20190076. doi: 10.1542/peds.2019-0076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.ACOG practice bulletin no. 190: gestational diabetes mellitus. Obstet Gynecol. 2018;131(2):e49-e64. doi: 10.1097/AOG.0000000000002501 [DOI] [PubMed] [Google Scholar]
- 29.Gestational hypertension and preeclampsia: ACOG practice bulletin, number 222. Obstet Gynecol. 2020;135(6):e237-e260. doi: 10.1097/AOG.0000000000003891 [DOI] [PubMed] [Google Scholar]
- 30.Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70(1):41-55. doi: 10.1093/biomet/70.1.41 [DOI] [Google Scholar]
- 31.Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal. 2007;15(3):199-236. doi: 10.1093/pan/mpl013 [DOI] [Google Scholar]
- 32.Eberly LA, Yang L, Essien UR, et al. Racial, ethnic, and socioeconomic inequities in glucagon-like peptide-1 receptor agonist use among patients with diabetes in the US. JAMA Health Forum. 2021;2(12):e214182. doi: 10.1001/jamahealthforum.2021.4182 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):13-22. doi: 10.1093/biomet/73.1.13 [DOI] [Google Scholar]
- 34.Williamson E, Morley R, Lucas A, Carpenter J. Propensity scores: from naive enthusiasm to intuitive understanding. Stat Methods Med Res. 2012;21(3):273-293. doi: 10.1177/0962280210394483 [DOI] [PubMed] [Google Scholar]
- 35.White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377-399. doi: 10.1002/sim.4067 [DOI] [PubMed] [Google Scholar]
- 36.Pishgar F, Greifer N, Leyrat C, Stuart E. MatchThem: matching and weighting after multiple imputation. R J. 2021;13(2):228. doi: 10.32614/RJ-2021-073 [DOI] [Google Scholar]
- 37.van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3):1-67. doi: 10.18637/jss.v045.i03 [DOI] [Google Scholar]
- 38.Lumley T. Analysis of complex survey samples. J Stat Softw. 2004;9(8):1-19. doi: 10.18637/jss.v009.i08 [DOI] [Google Scholar]
- 39.Waters TP, Huston-Presley L, Catalano PM. Neonatal body composition according to the revised Institute of Medicine recommendations for maternal weight gain. J Clin Endocrinol Metab. 2012;97(10):3648-3654. doi: 10.1210/jc.2012-1781 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hivert MF, Rifas-Shiman SL, Gillman MW, Oken E. Greater early and mid-pregnancy gestational weight gains are associated with excess adiposity in mid-childhood. Obesity (Silver Spring). 2016;24(7):1546-1553. doi: 10.1002/oby.21511 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Wells BJ, Chagin KM, Nowacki AS, Kattan MW. Strategies for handling missing data in electronic health record derived data. EGEMS (Wash DC). 2013;1(3):1035. doi: 10.13063/2327-9214.1035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Greifer N, Stuart EA. Choosing the causal estimand for propensity score analysis of observational studies. arXiv. Preprint posted online June 19, 2021. doi: 10.48550/arXiv.2106.10577 [DOI]
- 43.Osterman MJK, Hamilton BE, Martin JA, Driscoll AK, Valenzuela CP. Births: final data for 2022. Natl Vital Stat Rep. 2024;73(2):1-56. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods. Supplementary Methods
eTable 1. Baseline Characteristics of the Overall and Matched Birthweight Analytic Cohort
eTable 2. Baseline Characteristics of the Overall and Matched Obstetric Analytic Cohort
eTable 3. Duration of Treatment and Time From Discontinuation to Conception in the Gestational Weight Gain Cohort
eTable 4. Secondary Outcomes for the Birthweight and Obstetric Cohorts
eTable 5. Associations Between Liraglutide and Semaglutide and Birthweight and Obstetric Outcomes
eFigure 1. Derivation of the Obstetric Cohort
eFigure 2. Gestational Weight Gain Cohort
eFigure 3. Birthweight Cohort
eFigure 4. Obstetric Cohort
eReferences.
Data Sharing Statement
