Integrating probiotics into a one-day diabetes clinic: outcomes in gestational diabetes Authors: Wei Wang, Nana Cai, Min Ma, Yanmin Cao, Hong Li, Jie Liu, Youxia Zhou, Sijia Liang Published: 2026 ABSTRACT AbstractObjectiveTo investigate the effects of a one-day diabetes clinic management model combined with probiotic supplementation in patients with gestational diabetes mellitus (GDM).MethodsA total of 682 patients diagnosed with GDM who received prenatal care at our hospital between January 2023 and May 2024 were enrolled in this study. All participants underwent oral glucose tolerance test (OGTT) screening at 24–28 weeks of gestation and attended the one-day diabetes clinic management program. Patients were divided into a probiotic group and a control group based on oral probiotic administration. Propensity score matching (PSM) was performed to balance baseline characteristics between the two groups. Generalized estimating equations (GEE) were applied to analyze intergroup differences and temporal trends in blood glucose, lipid metabolism, and inflammatory markers. Pearson correlation analysis was used to examine maternal and neonatal perinatal outcomes.ResultsAfter PSM, no statistically significant differences were observed in baseline characteristics between the two groups (P > 0.05). Generalized estimating equation (GEE) analyses revealed significant group-by-time interaction effects for fasting blood glucose (FPG) and C-reactive protein (CRP). After Benjamini–Hochberg false discovery rate (FDR) correction for multiple testing, lower levels of FPG and CRP were observed in the probiotic group at several follow-up time points, with group differences emerging gradually over the intervention period. Triglycerides showed only a nominally significant group effect without adjustment. Total gestational weight change differed statistically between the two groups. GEE analyses indicated weak crude associations between probiotic supplementation, maternal group B streptococcal colonization (GBS), and neonatal jaundice; however, these associations were no longer statistically significant following FDR correction.ConclusionProbiotic supplementation combined with one-day diabetes clinic management was associated with lower FPG and CRP levels among women with GDM. Weak associations between probiotic use, maternal GBS colonization and neonatal jaundice disappeared after adjustment for multiple testing, and these findings can only be used to generate hypotheses. As an observational study, this analysis cannot confirm clinical benefits of probiotics. All observed findings represent exploratory associations, which require further verification in randomized controlled trials.Trial registrationThis observational study was retrospectively registered with the Chinese Clinical Trial Registry (ChiCTR2500100649) on April 11, 2025. Registration was delayed because the research team was initially unaware that registration is mandatory for observational clinical studies.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12884-026-09866-9.Keywords: Gestational diabetes mellitus, Probiotics, One-day diabetes clinic, Blood glucose, Perinatal outcomes Introduction Gestational diabetes mellitus (GDM) constitutes a significant global public health burden. This condition increases the risk of short-term adverse obstetric complications, including hypertensive disorders of pregnancy, amniotic fluid abnormalities, fetal congenital anomalies, macrosomia, neonatal jaundice, and neonatal hypoglycemia. Furthermore, GDM elevates the lifelong risks of developing type 2 diabetes and obesity for both mothers and their offspring [1, 2]. Current evidence indicates that standardized dietary modifications and appropriate physical activity can effectively regulate glycemic levels and mitigate adverse pregnancy outcomes in women with GDM [3]. The “one-day diabetes clinic” provides personalized dietary interventions and physical activity counseling, alongside multimodal health education that integrates didactic lectures, physical model demonstrations, and experiential learning to equip patients with practical self-management skills. The clinic implements continuous follow-up through standardized diabetic meal demonstrations, guided exercise sessions, and a WeChat-based remote management platform, facilitating timely individualized counseling to optimize glycemic control. Increasing evidence further demonstrates that probiotic supplementation improves multiple clinical indicators in the GDM population, including glycemic profiles, lipid parameters, inflammatory biomarkers, oxidative stress markers, and maternal and neonatal perinatal outcomes [4–6]. However, several randomized controlled trials have reported no statistically significant between-group differences in glycemic regulation and maternal-neonatal outcomes when comparing probiotic and placebo groups among individuals with GDM [7, 8]. Current evidence regarding probiotic interventions for GDM remains inconsistent, and most existing trials have adopted intervention durations of only 4–8 weeks [5]; such short follow-up periods cannot reliably evaluate the long-term sustainability of intervention benefits. By implementing an extended follow-up framework, the present study further explores the impacts of combined one-day diabetes clinic management plus oral probiotic supplementation on glycemic control, lipid metabolism, inflammatory status, and perinatal maternal and neonatal outcomes in women with GDM. Materials and methods Study design This prospective, real-world cohort study enrolled pregnant individuals attending the hospital’s one-day diabetes clinic. Clinical data were collected from pregnant individuals receiving care at the One-Day Diabetes Clinic of the Fourth Hospital of Shijiazhuang, China, with data extraction spanning from January 2023 to May 2024. This manuscript was prepared in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Study participants Participant Selection Pregnant individuals who attended the one-day diabetes clinic at the Fourth Hospital of Shijiazhuang between January 2023 and May 2024 were enrolled as study participants (see Additional file 1). Eligible participants underwent oral glucose tolerance test (OGTT) screening at 24–28 weeks of gestation and received a confirmed diagnosis of GDM. All participants completed standardized one-day diabetes clinic education and received comprehensive prenatal and delivery care at our institution. A total of 952 eligible individuals were initially screened for inclusion. Propensity score matching (PSM) at a 1:1 ratio between the intervention group and the control group was applied to balance baseline covariates, including age, height, family history of diabetes, pre-pregnancy weight, pre-pregnancy body mass index (BMI), weight at enrollment, BMI at enrollment, and gestational age at recruitment. After PSM, the final analytic sample consisted of 682 participants. Individuals who received oral probiotic supplementation during gestation were allocated to the probiotic group (n = 341), while those who did not take any oral probiotic agents formed the control group (n = 341). The participant exclusion and matching flowchart is presented in (Fig. 1). Flowchart of participant screening and exclusion. Abbreviations: OGTT, oral glucose tolerance test; PSM, propensity score matching Inclusion criteria Reproductive-aged individuals aged 20 to 40 years; Reproductive-aged individuals aged 20 to 40 years; Spontaneous singleton gestation; Spontaneous singleton gestation; Gestational age of 24-28 weeks with a confirmed diagnosis of GDM via OGTT; Gestational age of 24-28 weeks with a confirmed diagnosis of GDM via OGTT; Completion of all prenatal care and delivery at our institution; Completion of all prenatal care and delivery at our institution; Provision of written informed consent and willingness to participate in this prospective observational study. Provision of written informed consent and willingness to participate in this prospective observational study. Diagnostic criteria for GDM A 75 g anhydrous glucose OGTT was administered at 24–28 weeks of gestation. GDM was diagnosed if any of the following thresholds were met: (1) fasting plasma glucose (FPG) ≥ 5.1 mmol/L; (2) 1-hour post-load glucose ≥ 10.0 mmol/L; (3) 2-hour post-load glucose ≥ 8.5 mmol/L [9]. Exclusion criteria Pre-gestational diabetes mellitus; Pre-gestational diabetes mellitus; Pre-existing concurrent obstetric complications prior to enrollment; Pre-existing concurrent obstetric complications prior to enrollment; Pre-pregnancy chronic conditions, including hyperlipidemia and hypertension; Pre-pregnancy chronic conditions, including hyperlipidemia and hypertension; Incomplete prenatal medical records; Incomplete prenatal medical records; Concurrent participation in other clinical trials; Concurrent participation in other clinical trials; Poor adherence to scheduled clinic visits and follow-up, or loss to follow-up during the study period; Poor adherence to scheduled clinic visits and follow-up, or loss to follow-up during the study period; Inability or refusal to provide informed consent for study participation. Inability or refusal to provide informed consent for study participation. Diagnostic criteria for pre-gestational diabetes mellitus Diagnoses were established according to the American Diabetes Association Standards of Medical Care in Diabetes-2024. Pre-gestational diabetes was confirmed if any of the following criteria were satisfied at the initial prenatal visit: FPG ≥ 7.0 mmol/L, glycated hemoglobin (HbA1c) ≥ 6.5%, or random glucose ≥ 11.1 mmol/L accompanied by classic hyperglycemic symptoms [10]. These diagnostic cutoffs were uniformly applied to all participants enrolled between January 2023 and May 2024 in the present study. Study variables and definitions Exposure variable Exposure variable Probiotic group: Pregnant individuals who met all inclusion and exclusion criteria and received oral live combined Bifidobacterium triple viable enteric-coated capsules from the time of GDM diagnosis until delivery. The administration regimen consisted of two capsules taken twice daily. Control group: Eligible pregnant individuals who met the inclusion and exclusion criteria and did not receive any probiotic preparations throughout gestation. The probiotic preparation used in this study was Bifidobacterium Triple Viable Bacteria Enteric-coated Capsules (Beifeida; Shanxi Jincheng Haishi Pharmaceutical Co., Ltd., National Medicine Approval No. S19993065). Each capsule contained no less than 2.1×10⁵ CFU each of Bifidobacterium longum, Lactobacillus acidophilus, and Enterococcus faecalis. Participants received 4 capsules daily (2 capsules twice daily), corresponding to a total viable bacteria count of ≥2.52×10⁶ CFU per day, and the product was stored refrigerated at 2-10°C. Participants in both groups received standardized one-day diabetes clinic management and followed identical glycemic control targets: FPG ≤ 5.3 mmol/L and 2-hour postprandial blood glucose (2hPG) ≤ 6.7 mmol/L. If glycemic targets remained unmet following dietary and physical activity interventions, insulin therapy was initiated by endocrinologists. Insulin regimens, dosage titration, and discontinuation criteria were implemented in accordance with the *Clinical Practice Guidelines for the Diagnosis and Management of Hyperglycemia in Pregnancy (2022)* [9]. 2)Outcome measures Outcome measures Primary endpoint: FPG measured at 6–8 weeks postpartum (time point T4). 3)Sample size calculation Sample size calculation The primary outcome was abnormal FPG at 6–8 weeks postpartum (time point T4). Based on historical data from our center and published literature, the expected incidence of this endpoint was estimated to be 9.5% in the probiotic group and 16.7% in the control group [11]. A two-sided significance level of α = 0.05 and statistical power 1 − β = 0.80 were specified, with a 1:1 allocation ratio between the intervention and control arms. Sample size calculation was performed using the formula for comparing proportions between two independent groups, yielding a required sample size of n = 341 participants per group. After accounting for an anticipated 15% loss-to-follow-up rate, the pre-study target enrollment was set at 400 participants in the probiotic group and 400 in the control group, for a total target sample size of N = 800. In actual clinical recruitment, 400 participants were enrolled into the probiotic group and 380 into the control group, with an overall raw cohort of 780 subjects and an observed loss-to-follow-up rate of 2.5%. After PSM, the final analytical sample contained 341 participants in each group, totaling 682 subjects for statistical analyses. Methods Study procedures Research staff screened all individuals with GDM attending the one-day diabetes clinic, explained the study objectives and follow-up schedule, and obtained written informed consent from all eligible participants. Baseline data were collected for both the probiotic and control groups, including study ID, age, height, pre-pregnancy weight, weight at enrollment, gestational age at recruitment, and family history of diabetes. Biochemical laboratory measurements were recorded at five predefined time points: baseline enrollment (T0), one month after intervention initiation (T1), 34–36 weeks of gestation (T2), prior to delivery (T3), and 6–8 weeks postpartum (T4). Assessed biomarkers included FPG, 2hPG, HbA1c, fasting insulin (FINS), triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and C-reactive protein (CRP). Changes in body weight and maternal and neonatal perinatal outcomes were also documented. Homeostatic model assessment for insulin resistance (HOMA-IR) was calculated for all participants using the following formula: Specimen collection and Laboratory testing All participants underwent venous blood sampling after an overnight fast between 7:00 and 8:00 a.m., with strenuous physical activity restricted prior to collection. Blood specimens were collected according to standardized protocols by the hospital’s phlebotomy center and delivered to the biochemical and immunology laboratory for unified automated testing. Laboratory biomarkers measured included FPG, HbA1c, FINS, lipid profiles, and CRP. For postprandial glucose testing, timing began upon ingestion of the first bite of food, and capillary blood was drawn 2 h afterward to measure 2hPG. All 2hPG measurements at every follow-up time point were performed using capillary blood testing. Detailed information regarding testing methods, analytical instruments, measurement units, and reference ranges for each biomarker is provided in Additional file 2. Statistical analysis This prospective observational cohort study was registered after complete data collection. To mitigate confounding bias inherent to observational data, PSM was applied to balance baseline characteristics between groups. One-to-one nearest-neighbor matching without replacement was performed using the MatchIt (v4.7.2), tableone (v0.13.2), and survey (v4.5) packages in R software (version 4.4.0), with a caliper width set to 0.2. Covariate balance after matching was assessed via standardized mean differences (SMD); an SMD < 0.1 indicated satisfactory balance (see Fig. 3). Standardize Mesn Differences(SMD). Abbreviations: GA, gestational age; FPG, fasting plasma glucose; 2hPG, 2-hour postprandial plasma glucose; HbA1c, glycated hemoglobin; FINS, fasting insulin; CRP, C-reactive protein; TG, triglycerides; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol. General statistical analyses were conducted using SPSSAU. The Shapiro-Wilk test combined with Q-Q plots (Supplementary Fig. 1) was used to evaluate the normality of continuous variables. A Shapiro-Wilk P > 0.05 indicated adherence to the normal distribution, which guided the selection of subsequent analytical approaches. Normally distributed continuous data were presented as mean ± standard deviation and compared between groups via independent samples t-tests; the corresponding t statistic and P-value were reported, with P < 0.05 defined as statistically significant. Non-normally distributed continuous variables were summarized as median (25th percentile, 75th percentile) and compared using the Mann-Whitney U test; the U statistic and P-value were recorded, with statistical significance set at P < 0.05. Effect sizes were presented as Cohen’s d (standardized mean difference based on pooled standard deviation) with approximate 95% CIs using Hedges’ correction. Categorical and ordinal variables were analyzed using the χ² test, with the χ² statistic and P-value reported; P < 0.05 indicated a statistically significant difference. Pearson correlation analysis was performed for binary outcome variables to calculate correlation coefficients and corresponding P-values. A positive coefficient denoted a positive correlation, while a negative coefficient indicated a negative correlation; correlation strength increased as the absolute value of the coefficient approached 1, with P < 0.05 considered statistically significant. Given the large number of statistical tests performed, we implemented procedures to control for multiple comparisons. All tests were divided into four pre-specified hypothesis families: (A) group-by-time interaction effects for 10 biomarkers from primary longitudinal GEE analyses; (B) pairwise comparisons across 50 biomarker–timepoint combinations; (C) associations for binary maternal and fetal outcomes; (D) outcome-group correlation analyses. Within each family, we applied the Benjamini–Hochberg false discovery rate (FDR) correction. As sensitivity analyses, we further performed global FDR correction across all tests, as well as family-specific and global Bonferroni correction. An effect was deemed statistically significant if it remained significant after within-family FDR correction (q < 0.05). Effect sizes with corresponding 95% CIs for statistically meaningful findings were reported in the text. Generalized estimating equations (GEE) and data visualization were performed on the matched dataset using the geepack (v1.3.13), ggplot2 (v4.0.3), and patchwork (v1.3.2) packages in R (version 4.4.0) to estimate the effects of probiotic supplementation on glycemic, inflammatory, and lipid biomarkers. The dependent variable in the GEE model was HbA1c. Fixed effects included group assignment (probiotic intervention vs. control), categorical follow-up time (five time points), and the group-by-time interaction term. Individual participant ID was specified as the subject identifier, and follow-up time points were treated as within-subject repeated measures. An exchangeable correlation structure was adopted, and robust standard errors were used for statistical inference. Model outputs were reported as regression coefficients (β) with 95% confidence intervals (95% CIs). Statistical significance was evaluated via the Wald test, and two-sided P < 0.05 was regarded as statistically meaningful. Individual longitudinal trajectory plots are presented in Supplementary Fig. 2, and a forest plot illustrating the core interaction effects of the intervention is shown in (Fig. 4). Forest Plot: Time×Group Interaction Effects. Abbreviations: FPG, fasting plasma glucose; 2hPG, 2-hour postprandial plasma glucose; HbA1c, glycated hemoglobin; FINS, fasting insulin; CRP, C-reactive protein; TG, triglycerides; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol Results Propensity score matching balance Following PSM, SMDs for all baseline covariates, including age, gestational age at enrollment, height, pre-pregnancy weight, weight at baseline, family history of diabetes, parity, baseline FPG, baseline 2hPG, baseline HbA1c, baseline FINS, baseline CRP, baseline TC, baseline TG, baseline HDL-C, and baseline LDL-C, were constrained within the range of -0.1 to 0.1 (Figs. 2 and 3). Dstribution of propensity scores Comparison of baseline characteristics between the two groups After one-to-one PSM, no statistically significant intergroup differences were observed for all baseline covariates (P > 0.05) (Table 1). Table 1Baseline characteristics of the two groupsCharacteristicsGROUP Mean ± SD; median (P25, P75); %t / U (Mann–Whitney U test) / χ²p☐Probiotic group(n = 341)Control group(n = 341)Age (years)31.18 ± 3.9431.09 ± 3.740.3190.750GA at enrollment (weeks)25.51 ± 1.1625.44 ± 1.070.8250.409GA at delivery (weeks)38.53 ± 1.3238.46 ± 1.350.7440.457Birth weight (g)3262.51 ± 480.073241.06 ± 447.250.6040.546Height (cm)161.50 ± 4.48161.51 ± 4.77-0.0170.987Baseline weight (kg)64.15 ± 11.0664.53 ± 11.43-0.4380.661BMI24.59 ± 4.0824.72 ± 4.14-0.4200.674Pre-pregnancy weight (kg)71.63 ± 11.0872.22 ± 11.60-0.6720.502Pre-delivery weight (kg)76.17 ± 11.6877.25 ± 11.23-1.2250.221FPG baseline (mmol/L)5.34 ± 0.515.36 ± 0.59-0.2790.7802hPG baseline (mmol/L)6.65 ± 0.796.67 ± 0.86-0.2920.770HbA1c baseline (%)5.53 ± 0.565.53 ± 0.39-0.0240.981FINS baseline (µU/mL)13.200(7.3,18.4)12.600(8.0,19.3)56533.0000.532HOMA-IR2.918(1.7,4.3)2.888(1.9,4.5)56554.0000.537CRP baseline (mg/L)5.800(3.7,8.3)5.620(2.6,9.7)56554.0000.537TG baseline (mmol/L)3.100(2.4,4.0)3.170(2.5,3.9)57107.0000.688TC baseline (mmol/L)6.250(5.1,7.2)6.160(5.2,7.2)57522.0000.810LDL-C baseline (mmol/L)3.630(2.9,4.5)3.600(2.5,4.8)57097.0000.685HDL-C baseline (mmol/L)2.290(1.8,2.6)2.210(1.9,2.6)56330.0000.482Family history of diabetes No228(49.89)229(50.11)0.0070.935 Yes113(50.22)112(49.78)Parity 1191(49.48)195(50.52)0.2870.962 2131(51.17)125(48.83) 318(47.37)20(52.63) 41(50.00)1(50.00)Mode of delivery Vaginal delivery179(51.59)168(48.41)1.0800.583 Caesarean section150(48.86)157(51.14) Assisted vaginal delivery12(42.86)16(57.14)Abbreviations: GA gestational age, BMI body mass index, FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, HOMA-IR homeostasis model assessment of insulin resistance, CRP C-reactive protein, TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol Baseline characteristics of the two groups Abbreviations: GA gestational age, BMI body mass index, FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, HOMA-IR homeostasis model assessment of insulin resistance, CRP C-reactive protein, TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol Analyses of glycemic, lipid, and inflammatory biomarkers GEE models were constructed to evaluate the main effects of group assignment, follow-up time, and their group-by-time interaction on all outcome biomarkers. The group main effect was statistically significant for TG (β = -0.259, 95% CI: −0.528, 0.508, P = 0.025*) (Table 2), whereas the group-by-time interaction for TG was non-significant (P > 0.05). No significant group main effects were detected for the remaining biomarkers (P > 0.05). Table 2Group effectsOutcomeGroup Effect (β, 95% CI)Group P-valueFPG(mmol/L)-0.17 (-0.587, 0.446)0.4012hPG(mmol/L)-0.392 (-0.529, 0.678)0.377HbA1c(%)-0.217 (-0.449, 0.644)0.401FINS(µU/mL)0.366 (-0.614, 0.482)0.285CRP(mg/L)-0.142 (-0.327, 0.42)0.333TG(mmol/L)-0.259 (-0.528, 0.508)0.025*TC(mmol/L)0.312 (-0.516, 0.696)0.325LDL-C(mmol/L)0.475 (-0.599, 0.479)0.274HDL-C(mmol/L)0.085 (-0.559, 0.439)0.294*P < 0.05 Abbreviations: FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, CRP C-reactive protein, TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol Group effects *P < 0.05 Abbreviations: FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, CRP C-reactive protein, TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol The main effect of time was non-significant for FPG, 2hPG, HbA1c, FINS, CRP, TG, TC, LDL-C, and HDL-C (all P > 0.05) (Table 3). Table 3Time effectsOutcomeTime Effect (β, 95% CI)Time P-valueFPG(mmol/L)-0.445 (-0.658, 0.358)0.1362hPG(mmol/L)-0.428 (-0.565, 0.445)0.493HbA1c(%)0.482 (-0.405, 0.463)0.198FINS(µU/mL)0.285 (-0.421, 0.647)0.170CRP(mg/L)0.13 (-0.534, 0.61)0.108TG(mmol/L)-0.048 (-0.505, 0.534)0.475TC(mmol/L)0.213 (-0.608, 0.515)0.274LDL-C(mmol/L)0.336 (-0.494, 0.338)0.446HDL-C(mmol/L)0.354 (-0.432, 0.59)0.081Abbreviations: FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, CRP, C-reactive protein,TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol Time effects Abbreviations: FPG fasting plasma glucose, 2hPG 2-hour postprandial plasma glucose, HbA1c glycated hemoglobin, FINS fasting insulin, CRP, C-reactive protein,TG triglycerides, TC total cholesterol, LDL-C low-density lipoprotein cholesterol, HDL-C high-density lipoprotein cholesterol Nevertheless, significant group-by-time interaction effects were observed for multiple biomarkers, with all corresponding effect sizes visualized in the forest plot (Fig. 4). The group-by-time interaction term for FPG indicated that participants in the probiotic group experienced a significantly greater longitudinal reduction in FPG across follow-up visits compared to the control group (β = 0.161, 95% CI: 0.055, 0.267, P = 0.003). Within the probiotic group, the longitudinal decline in FPG from baseline to subsequent time points was also statistically significant (β = 0.209, 95% CI: 0.083, 0.335, P = 0.001). Similarly, the probiotic group demonstrated a substantially larger time-dependent decrease in CRP compared to controls (β = 0.799, 95% CI: 0.003, 1.594, P = 0.049). The within-group longitudinal improvement in CRP from baseline to all follow-up assessments was likewise statistically significant for individuals receiving probiotic supplementation (β = 1.204, 95% CI: 0.421, 1.988, P = 0.003). No significant group-by-time interactions were found for 2hPG, HbA1c, FINS, TG, TC, LDL-C, and HDL-C (all P > 0.05). Simple effect analyses were further performed for FPG and CRP (Table 4). Baseline FPG and CRP levels were well balanced between the two groups. No intergroup differences in FPG or CRP were observed at the early follow-up time point T1 (P > 0.05). Starting from T2, FPG and CRP concentrations were significantly lower in the probiotic group than in the control group, and these intergroup disparities persisted at T3 and the postpartum T4 visit (all P < 0.05). Table 4Simple effect analysis of FPG and CRPGROUP Mean ± SD; median (P25, P75); %t / U (Mann–Whitney U test) ☐p ☐Probiotic group(n = 341)Control group(n = 341)FPG baseline (mmol/L)5.34 ± 0.515.36 ± 0.59-0.2790.780FPGT1 (mmol/L)5.19 ± 0.485.22 ± 0.48-0.8250.409FPGT2 (mmol/L)5.00 ± 0.595.17 ± 0.52-4.0550.000**FPGT3 (mmol/L)4.82 ± 0.464.92 ± 0.50-2.5840.010**FPGT4 (mmol/L)5.25 ± 0.655.47 ± 0.71-4.2290.000**CRP baseline (mg/L)5.800(3.7,8.3)5.620(2.6,9.7)56554.0000.537CRPT1 (mg/L)5.970(3.3,8.4)5.750(3.1,9.2)56779.5000.597CRPT2 (mg/L)5.090(2.9,8.0)6.460(3.2,9.3)49766.0000.001**CRPT3 (mg/L)4.560(2.5,6.9)5.950(3.4,8.6)46632.0000.000**CRPT4 (mg/L)4.450(2.3,7.3)5.770(3.4,8.2)46527.5000.000****P < 0.01 Abbreviations: FPG, fasting plasma glucose; CRP, C-reactive protein Simple effect analysis of FPG and CRP **P < 0.01 Abbreviations: FPG, fasting plasma glucose; CRP, C-reactive protein Longitudinal trajectories of glycemic, lipid, and inflammatory biomarkers across all follow-up time points for both groups are illustrated in (Fig. 5). Changes in glycolipid metabolism and inflammatory indicators: probiotic group vs. control group. Abbreviations: FPG, fasting plasma glucose; 2hPG, 2-hour postprandial plasma glucose; HbA1c, glycated hemoglobin; FINS, fasting insulin; HOMA-IR, homeostasis model assessment of insulin resistance; CRP, C-reactive protein; TG, triglycerides; TC, total cholesterol; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol After FDR correction, six out of 50 biomarker comparisons across the five follow-up time points remained statistically significant (q < 0.05). All six significant findings showed consistent direction, with lower levels observed in the probiotic group. These outcomes included CRP at T2, T3 and T4 (Cohen’s d = − 0.266, 95% CI [− 0.417, − 0.116], q = 0.008; Cohen’s d = − 0.387, 95% CI [− 0.539, − 0.236], q < 0.001; Cohen’s d = − 0.399, 95% CI [− 0.550, − 0.247], q < 0.001), FPG at T2 and T4 (Cohen’s d = − 0.311, 95% CI [− 0.462, − 0.160], q = 0.001; Cohen’s d = − 0.324, 95% CI [− 0.475, − 0.173], q = 0.001), and 2hPG at T2 (Cohen’s d = − 0.216, 95% CI [− 0.367, − 0.066], q = 0.045). No statistically significant between-group differences after correction were detected for the remaining biomarkers, including HbA1c, insulin and lipid-related parameters (Fig. 6). Biomarkers × time points (n = 50) volcano plot: effect size vs. significance. Abbreviations: FPG, fasting plasma glucose; 2hPG, 2-hour postprandial plasma glucose; CRP, C-reactive protein Comparison of gestational weight gain and insulin use between groups Total gestational weight gain differed significantly between the two groups. The median total gestational weight gain was 4.5 (2.0, 7.0) in the probiotic group versus 5.5 (2.5, 8.5) in the control group, with a Mann-Whitney U statistic of 51,109.500 (P = 0.006) (Table 5). Table 5Comparison of weight changes and insulin usage between the two groupsGROUP median (P25, P75); % ☐U (Mann–Whitney U test) / χ²p ☐Probiotic group(n = 341)Control group(n = 341)Weight change from pre-pregnancy to enrollment7.000(5.0,10.0)6.000(4.0,11.0)55292.5000.268Weight change from enrollment to delivery12.000(8.3,15.5)12.400(9.0,16.9)54419.5000.148Total gestational weight gain4.500(2.0,7.0)5.500(2.5,8.5)51109.5000.006**Insulin use, n (%)No insulin330(50.38)325(49.62)1.0440.791Aspart insulin1(33.33)2(66.67)Detemir insulin3(42.86)4(57.14)Aspart plus detemir insulin7(41.18)10(58.82)**P < 0.01 Comparison of weight changes and insulin usage between the two groups **P < 0.01 No significant intergroup difference was observed regarding insulin use (P > 0.05) (Table 5). Perinatal outcomes in the two groups Among the 16 maternal and fetal outcomes examined, significant between-group differences were observed for maternal group B streptococcus (GBS) colonization and neonatal jaundice (P < 0.05). No statistically significant differences (P > 0.05) were detected for the remaining outcomes, including gestational hypertension, preeclampsia, premature rupture of membranes, oligohydramnios/polyhydramnios, placental abruption, postpartum hemorrhage, chorioamnionitis, intrapartum fever, macrosomia, preterm birth, fetal distress, fetal growth restriction, term low birth weight infant, and neonatal hypoglycemia (Table 6). Table 6Comparison of perinatal outcomes between the two groupsPerinatal outcomesOutcome statusGROUP(%)Totalχ2 ☐p ☐Probiotic group(n = 341)Probiotic group(n = 341)Gestational hypertensionNo326(50.46)320(49.54)6461.0560.304Yes15(41.67)21(58.33)36PreeclampsiaNo319(50.00)319(50.00)6380.0001.000Yes22(50.00)22(50.00)44Premature rupture of membranesNo282(50.18)280(49.82)5620.0400.841Yes59(49.17)61(50.83)120Polyhydramnios / OligohydramniosNo336(50.60)328(49.40)6643.6520.056Yes5(27.78)13(72.22)18Placental abruptionNo340(50.00)340(50.00)6800.0001.000Yes1(50.00)1(50.00)2Postpartum hemorrhageNo337(50.07)336(49.93)6730.1130.737Yes4(44.44)5(55.56)9ChorioamnionitisNo304(50.00)304(50.00)6080.0001.000Yes37(50.00)37(50.00)74Intrapartum feverNo340(50.30)336(49.70)6762.6900.101Yes1(16.67)5(83.33)6Maternal GBS colonizationNo332(50.92)320(49.08)6525.0210.025*Yes9(30.00)21(70.00)30MacrosomiaNo334(50.45)328(49.55)6621.8540.173Yes7(35.00)13(65.00)20Preterm birthNo325(50.31)321(49.69)6460.4690.493Yes16(44.44)20(55.56)36Fetal distressNo323(49.69)327(50.31)6500.5250.469Yes18(56.25)14(43.75)32Fetal growth restrictionNo331(49.70)335(50.30)6661.0240.312Yes10(62.50)6(37.50)16Term low birth weightNo338(49.93)339(50.07)6770.2010.654Yes3(60.00)2(40.00)5Neonatal hypoglycemiaNo327(50.15)325(49.85)6520.1390.709Yes14(46.67)16(53.33)30Neonatal jaundiceNo307(51.77)286(48.23)5935.6990.017*Yes34(38.20)55(61.80)89*P < 0.05Abbreviations: GBS group B streptococcus Comparison of perinatal outcomes between the two groups *P < 0.05 Abbreviations: GBS group B streptococcus After FDR correction, no statistically significant between-group differences were observed across the 16 maternal and fetal outcomes (all q ≥ 0.05). Nominally significant findings with raw P < 0.05 included maternal GBS colonization (OR = 0.413, 95% CI [0.186, 0.916], P = 0.025) and neonatal jaundice (OR = 0.576, 95% CI [0.365, 0.909], P = 0.017); however, these associations lost significance following correction (Fig. 7). Maternal/neonatal outcomes (probiotic vs. control) volcano plot. Abbreviations: GBS, group B streptococcus Pearson correlation analyses were performed to assess associations between group assignment and the 16 maternal and neonatal outcomes. Only two outcomes demonstrated statistically significant correlations with group status: maternal GBS colonization (r = -0.086, P = 0.025) and neonatal jaundice (r = -0.091, P = 0.017). Both correlation coefficients indicated weak negative correlations (Table 7). Table 7Pearson correlation analysis of maternal and fetal perinatal outcomesVariableGestational hypertensionPreeclampsiaPremature rupture of membranesPolyhydramnios / OligohydramniosPlacental abruptionPostpartum hemorrhageChorioamnionitisIntrapartum feverMaternal GBS colonizationMacrosomiaPreterm birthFetal distressFetal growth restrictionTerm low birth weightNeonatal hypoglycemiaNeonatal jaundiceOral probiotic administration(1 = No, 2 = Yes)Correlation coefficient-0.039-0.000-0.008-0.0730.000-0.013-0.000-0.063-0.086-0.052-0.0260.0280.0390.017-0.014-0.091P-value0.3051.0000.8410.0561.0000.7381.0000.1010.0250.1740.4940.4700.3120.6540.7090.017n682682682682682682682682682682682682682682682682Abbreviations: GBS group B streptococc Pearson correlation analysis of maternal and fetal perinatal outcomes Oral probiotic administration (1 = No, 2 = Yes) Abbreviations: GBS group B streptococc After FDR correction, correlations for all 16 maternal and neonatal perinatal outcomes were non-significant (all q ≥ 0.05). Only maternal group B streptococcal colonization (r = 0.086, P = 0.025) and neonatal jaundice (r = 0.091, P = 0.017) yielded raw P < 0.05, yet the corresponding correlation coefficients were very small in magnitude (|r| < 0.1) (Fig. 8). Outcome ~ group correlations (Pearson r) volcano plot. Abbreviations: GBS, group B streptococcus Discussion GDM is defined as abnormal glucose metabolism first identified during pregnancy, and its prevalence varies substantially by diagnostic criteria, geographic region, and study population. Standardized analyses using the International Association of Diabetes in Pregnancy Study Groups (IADPSG) criteria report a global GDM prevalence of 14.0% [12]. A 2019 meta-analysis focusing on mainland Chinese populations reported a pooled prevalence of 14.8%, which exceeds the global average and demonstrates an increasing temporal trend [11]. Accordingly, effective glycemic control and complication prevention among individuals with GDM are critical for safeguarding maternal and neonatal well-being. Medical nutrition therapy and individualized physical activity interventions constitute the cornerstone of GDM management. The localized one-day diabetes clinic program integrates hands-on practical training, food model demonstrations, WeChat-based remote follow-up reminders, and one-on-one personalized counseling. This integrated model addresses the key limitation of poor long-term adherence associated with conventional health education approaches. Prior research has validated that this clinic-based strategy facilitates gestational weight and glycemic control in people with GDM, improves postpartum pancreatic beta-cell function recovery, and reduces the incidence of adverse perinatal outcomes including premature rupture of membranes, macrosomia, and neonatal jaundice [13]. Nevertheless, behavioral interventions alone exert limited effects on alleviating chronic low-grade inflammation and dyslipidemia. This unmet clinical gap provides the theoretical rationale and novel research motivation for the present study to incorporate adjuvant probiotic supplementation. Dysregulation of the gut-metabolic-inflammatory axis, driven by gut microbiota disturbance, has emerged as a major research focus in exploring the pathological mechanisms underlying GDM. Current evidence confirms that individuals with GDM exhibit perturbed gut microbial composition, characterized by reduced abundances of beneficial bacteria such as Bifidobacterium and Lactobacillus, alongside elevated levels of opportunistic pathogens [14, 15]. Such microbial imbalance triggers disorders in glucose and lipid metabolism and the dysregulated production of inflammatory mediators, exacerbating insulin resistance and dyslipidemia, thereby elevating the risk of adverse pregnancy outcomes [16]. Previous clinical and experimental studies have demonstrated that supplementation with probiotic strains, including Lactobacillus casei, Bifidobacterium bifidum, and Lactobacillus acidophilus, can modulate intestinal microecology, suppress proinflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6), alleviate chronic low-grade inflammation, enhance insulin sensitivity, and optimize glycemic regulation [17–19]. Although gut microbiota profiling was not performed in the present study, we reasonably hypothesize that probiotic intervention may exert therapeutic effects through the aforementioned biological pathways based on existing literature. Further research has identified complex interactive relationships among nutritional intake, gut microbiota, metabolic processes, impaired glucose tolerance, and low-grade inflammation in individuals with GDM [20]. One meta-analysis reported that probiotic supplementation effectively improves glycemic profiles, HOMA-IR, lipid concentrations, and high-sensitivity C-reactive protein levels, while reducing the rate of cesarean delivery. No significant effects were observed for neonatal hypoglycemia, preterm birth risk, or birth weight [21]. Nevertheless, a consistent consensus regarding the clinical efficacy of probiotic therapy for GDM has not been established, due to limitations including substantial interstudy heterogeneity, reliance on surrogate endpoints, and insufficient sample sizes. The long-term benefits and generalizability of probiotic interventions remain to be validated. Unlike prior studies that implemented only short-term isolated probiotic supplementation, the present study established a combined intervention model integrating standardized one-day diabetes clinic behavioral counseling with prolonged probiotic administration. All participants received continuous probiotic supplementation from 24 to 28 weeks of gestation until delivery (a minimum intervention duration of 12 weeks) to evaluate clinical outcomes. This integrated regimen addresses the existing gap in localized combined clinical protocols and long-term longitudinal follow-up data, representing the core novel contribution of this investigation. The probiotic preparation used in this study was a multi-strain capsule containing Bifidobacterium longum, Lactobacillus acidophilus, and Enterococcus faecalis, which is widely used in clinical practice in China. These three strains belong to the genera Bifidobacterium, Lactobacillus, and Enterococcus, and may exert synergistic effects across distinct intestinal niches [22–24]. Previous investigations have demonstrated that this formulation modulates intestinal microbiota composition, reduces inflammatory cytokines, and serves as an adjuvant therapy for GDM [25]. Regarding the biosafety of Enterococcus faecalis in this preparation, existing evidence indicates that pathogenicity is strain-specific. This formulation has been widely administered to vulnerable populations including pregnant women [25], critically ill patients [26], and children [27], with no reports of severe adverse reactions. The present study found that integrating probiotic supplementation into the one-day diabetes clinic model was associated with lower FBG and CRP levels among patients with GDM. These findings partially align with those of previous investigations [5, 6]. Reductions in FBG and CRP exhibited distinct temporal patterns: significant intergroup differences emerged starting at 34–36 weeks of gestation (T2) and persisted until 6–8 weeks postpartum (T4), indicating that a specific cumulative duration of probiotic administration is required to elicit measurable therapeutic effects. Prior studies have validated that 4–8 weeks of probiotic intervention improves glycemic control, lipid profiles, and inflammatory status in patients with GDM [5, 6]. In this study, probiotic supplementation lasted 12 weeks or longer. Intergroup differences tended to stabilize in late pregnancy and persisted postpartum, suggesting that this combined management model was associated with reduced gestational glycemia and low-grade inflammation, alongside trends of postpartum glycemic recovery. This sustained association may stem from potential modulatory effects of probiotics on intestinal microecology. Nevertheless, without intestinal microbiota and mechanistic biomarker data and given the observational study design, definitive causal mechanisms remain unconfirmed [28]. To mitigate false-positive bias arising from multiple comparisons, Benjamini-Hochberg FDR correction was applied in this study. After correction, lower FPG and CRP levels remained evident in the probiotic group; however, the corresponding standardized effect sizes (Cohen’s d) ranged from − 0.216 to -0.399, representing small effect magnitudes overall, indicating detectable statistical differences with limited clinical magnitude. Meanwhile, GEE models revealed significant group-by-time interaction effects for FPG and CRP, demonstrating divergent declining trajectories of these biomarkers between the two groups and further supporting that the associations emerged gradually with prolonged intervention. In addition, primary GEE models revealed a significant main group effect for TG, with no statistically significant group-by-time interaction. Notably, after FDR correction for multiple testing, between-group differences in TG at all follow-up time points lost statistical significance, indicating that this raw positive finding was likely attributable to multiplicity bias and represented only a nominally significant result. From the perspective of intervention characteristics, one-day diabetes clinic care alone exerts favorable regulatory effects on metabolic markers. On the basis of standardized dietary and exercise interventions, probiotic supplementation may struggle to further amplify changes in some metabolic indicators over the short term. Furthermore, although several prior studies have reported that probiotics reduce 2hPG and HbA1c [29], most follow-up time points in the present study showed no statistically significant intergroup differences for these markers after multiple-testing correction. Such heterogeneity may be explained by differences in intervention duration, probiotic dosage, and baseline metabolic profiles of participants. Importantly, this is a prospective cohort study without a placebo control. Although PSM balanced baseline confounders, unmeasured confounding and residual bias cannot be excluded. Combined with the results of multiple-testing correction and small effect magnitudes, all positive associations identified herein constitute exploratory signals lacking causal inference value; we cannot confirm that probiotics effectively improve clinical metabolic outcomes. Accordingly, these findings remain exploratory and require further validation in randomized controlled trials. To date, global studies have yielded inconsistent conclusions regarding the effects of probiotic supplementation on gestational weight gain (GWG) and insulin requirements among individuals with GDM. One meta-analysis demonstrated that multi-strain probiotic administration initiated in early pregnancy reduced mean GWG by 1.25 kg while significantly improving metabolic markers, including FPG, HOMA-IR, and TC [30]. Partially consistent with these findings, the present study observed markedly lower total GWG in the probiotic group relative to the control group. However, no significant intergroup difference was detected in the proportion of participants requiring insulin therapy. We hypothesize that the standardized dietary and physical activity guidance provided via the one-day diabetes clinic program achieved adequate glycemic control, potentially attenuating the auxiliary metabolic benefits conferred by probiotic supplementation. Future multicenter clinical trials with large sample sizes are warranted to further clarify the clinical utility of probiotic interventions for weight management and insulin-sparing effects in individuals with GDM. GBS is a common colonizing organism in the maternal genital tract and a leading pathogen associated with puerperal and neonatal infectious complications [31]. Previous studies have indicated that probiotic supplementation holds potential benefits for reducing maternal GBS colonization [32] and enhancing neonatal bilirubin metabolism [33]. This study detected weak negative correlations between probiotic supplementation and maternal GBS colonization as well as neonatal jaundice, with correlation coefficients approximately − 0.09. Notably, these associations lost statistical significance after correction for multiple testing. The weak correlations achieved statistical significance merely due to the large sample size and are of negligible clinical relevance. In view of the study design, these findings can only serve as hypothesis-generating signals. We cannot conclude that probiotic intervention reduces maternal GBS colonization or the risk of neonatal jaundice, and these results do not constitute valid evidence for clinical benefits of probiotics. No statistically significant associations between probiotic use and other maternal and neonatal perinatal outcomes were detected, including gestational hypertension, premature rupture of membranes, macrosomia, and neonatal hypoglycemia. Discrepancies between these findings and results from earlier studies may be explained by variations in participant baseline characteristics, probiotic strain formulations, intervention duration, and administered dosages. Statistically significant correlations were restricted exclusively to GBS colonization and neonatal jaundice, implying that the metabolic and genital tract-related effects of probiotics exhibit outcome-specificity, rather than exerting consistent associations across all adverse perinatal endpoints. Limitations This study has several notable limitations. First, it was a single-center, prospective observational investigation, which restricts the generalizability and representativeness of the findings. Second, intestinal microbiota profiles of participants were not measured. We can only identify correlations between probiotic supplementation and various outcomes among women with GDM, and the mechanisms underlying probiotic effects cannot be elucidated. Third, key confounding variables including adherence to probiotic intake, dietary compliance, physical activity levels, antibiotic exposure, socioeconomic factors, and individualized clinical management decisions were not collected. Potential confounding influences of these factors on glucose and lipid metabolism as well as maternal and neonatal outcomes cannot be ruled out, and unavoidable residual confounding bias remains present. Fourth, the probiotic formulation utilized in this trial was constrained by commercial pharmaceutical specifications. The daily viable bacterial count per strain was lower than that adopted in several comparable GDM studies, reflecting only the standard safe dosage routinely prescribed for pregnant women in China. Dose-gradient trials are warranted in future work to determine the optimal intervention dose. Finally, all follow-up was administered via a WeChat-based remote management platform. This digital monitoring approach may independently alter participant adherence and clinical endpoints, making it impossible to isolate the standalone therapeutic effect of probiotics. In summary, all conclusions drawn from this analysis are correlational. Large-scale, multicenter randomized controlled trials are required to validate these results. Conclusion This study established a combined intervention comprising a standardized one-day diabetes clinic and a minimum of 12 weeks of probiotic supplementation, addressing gaps in longitudinal data for localized integrated GDM management. The findings suggest that the combined intervention regimen was associated with lower levels of FPG and CRP, with biomarker reductions emerging gradually over the course of intervention. As a non-randomized, observational cohort study without a placebo control, this analysis cannot confirm therapeutic benefits of probiotics and only demonstrates correlational trends. Weak associations between probiotic supplementation, maternal GBS colonization and neonatal jaundice lost significance after adjustment for multiple testing bias; these nominally significant raw findings exhibited minimal effect sizes and are merely hypothesis-generating. In addition, intestinal and genital tract microbiota as well as mechanistic biomarkers were not measured, so relevant biological mechanisms remain speculative. In summary, all results from this study represent exploratory correlational observations and do not support causal inferences. Further validation in multicenter randomized controlled trials incorporating microbiota testing is warranted. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements Not applicable. Abbreviations Gestational diabetes mellitus Oral glucose tolerance test Propensity score matching Homeostasis Model Assessment of Insulin Resistance Body mass index Fasting Plasma Glucose Glycated hemoglobin Fasting insulin C-reactive protein 2-hour postprandial blood glucose Triglycerides Total cholesterol Lipoprotein cholesterol High-density lipoprotein cholesterol Group B streptococcus Standardized mean differences Generalized estimating equations Gestational weight gain False discovery rate Authors’ contributions WW and YMC designed the study.YMC and MM supervised the work, reviewed and edited the article. YXZ and SJL collected samples and clinical data. HL and JL were responsible for data management and performed data validation.WW and NNC researched the data, performed the statisticalanalysis and wrote the manuscript. All authors have read and approved the final manuscript. Funding This work was supported by Medical Science Research Project of Hebei (Project Number: 20231681). Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study was approved by Medical Ethics Committee of Shijiazhuang Fourth Hospital (Approval No.: 20220114) and performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to data collection. Consent for publication Not applicable(there are no images or other personal or clinical details of participants that compromise anonymity). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Wei Wang and Nana Cai contributed equally to this work. Contributor Information Min Ma, Email: 15930181760@163.com. Yanmin Cao, Email: 13832122410@163.com. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.