A clinical diabetes risk prediction model for prediabetic women with prior gestational diabetes

Bernice Man; Alan Schwartz; Oksana Pugach; Yinglin Xia; Ben Gerber

doi:10.1371/journal.pone.0252501

. 2021 Jun 25;16(6):e0252501. doi: 10.1371/journal.pone.0252501

A clinical diabetes risk prediction model for prediabetic women with prior gestational diabetes

Bernice Man ^1,^*, Alan Schwartz ², Oksana Pugach ³, Yinglin Xia ⁴, Ben Gerber ¹

Editor: Andreas Beyerlein⁵

PMCID: PMC8232404 PMID: 34170930

Abstract

Introduction

Without treatment, prediabetic women with a history of gestational diabetes mellitus (GDM) are at greater risk for developing type 2 diabetes compared with women without a history of GDM. Both intensive lifestyle intervention and metformin can reduce risk. To predict risk and treatment response, we developed a risk prediction model specifically for women with prior GDM.

Methods

The Diabetes Prevention Program was a randomized controlled trial to evaluate the effectiveness of intensive lifestyle intervention, metformin (850mg twice daily), and placebo in preventing diabetes. Data from the Diabetes Prevention Program (DPP) was used to conduct a secondary analysis to evaluate 11 baseline clinical variables of 317 women with prediabetes and a self-reported history of GDM to develop a 3-year diabetes risk prediction model using Cox proportional hazards regression. Reduced models were explored and compared with the main model.

Results

Within three years, 82 (25.9%) women developed diabetes. In our parsimonious model using 4 of 11 clinical variables, higher fasting glucose and hemoglobin A1C were each associated with greater risk for diabetes (each hazard ratio approximately 1.4), and there was an interaction between treatment arm and BMI suggesting that metformin was more effective relative to no treatment for BMI ≥ 35kg/m² than BMI < 30kg/m². The model had fair discrimination (bias corrected C index = 0.68) and was not significantly different from our main model using 11 clinical variables. The estimated incidence of diabetes without treatment was 37.4%, compared to 20.0% with intensive lifestyle intervention or metformin treatment for women with a prior GDM.

Conclusions

A clinical prediction model was developed for individualized decision making for prediabetes treatment in women with prior GDM.

Introduction

Gestational diabetes mellitus (GDM) is a common medical complication of pregnancy with a prevalence of 7–9% [1–3]. Although GDM typically resolves after delivery, women remain at high risk of developing type 2 diabetes mellitus, with a cumulative incidence of 30–50% within 5–10 years of the index pregnancy [3, 4]. Women with a history of GDM are more likely to progress to diabetes compared to those without GDM despite the same degree of impaired glucose tolerance at baseline [5]. Thus, the American Diabetes Association (ADA) and American College of Obstetricians and Gynecologists recommend diabetes screening at 4–12 weeks postpartum and every 1–3 years thereafter [6, 7]. For those with prediabetes (with early evidence of abnormal glycemic parameters), screening is recommended annually.

For most adults with prediabetes, treatment involves intensive lifestyle intervention (ILI) and/or metformin. However, the expected treatment response differs based on individual clinical factors (e.g., beta cell function), including history of GDM. Specifically, among those with prior GDM, risk reduction with metformin is comparable to highly-effective ILI (both reducing 3-year risk of progression to diabetes by approximately 50%) [5]. In contrast, metformin is less effective than ILI in the general population, including parous women without a history of GDM [8]. Overall, evidence suggests there is heterogeneity in prediabetes treatment response when GDM history is considered [5, 9].

Individualized risk prediction with estimated treatment response may inform prediabetes treatment decisions [10, 11]. Women with prior GDM can become aware of their future risk of diabetes and potential benefit from metformin and/or ILI. Individualized risk/benefit assessment may improve diabetes risk perception and prediabetes treatment decisions in this high-risk population. Such an assessment may optimize the appropriate use of metformin and ILI, considering the risks and costs, personal preferences, and potential benefits. Individual risk estimation may improve clinical decision making and make diabetes prevention efforts more efficient, cost-effective and patient-centered [10, 11].

There are numerous models available to predict the risk of developing diabetes for the general population [10–15]. However, few use multivariable models to facilitate tailoring preventive interventions to individuals [10, 11, 16, 17]. Of the models specifically developed for women with prior GDM, predictors commonly include measures obtained during or soon after pregnancy (e.g., insulin use during pregnancy or breastfeeding history) [15]. In a multivariable analysis of 174 women with GDM in Sweden, predictors of diabetes within 5 years postpartum include parity, a first-degree family member with diabetes, fasting glucose and hemoglobin A1c (HbA1c) levels during pregnancy [18]. In another GDM-specific model, non-European origin, glucose concentration from the 75 gram, 2-hour oral glucose tolerance test (OGTT) at pregnancy, and body mass index (BMI) at 1–2 years post-partum were predictive of diabetes [19]. Use of models incorporating peripartum measures may be limited because such measures may not be consistently obtained and are not often readily available to different clinicians providing care years later. Additionally, these models do not consider prediabetes treatment response, and may not be generalizable to diverse ethnic minority populations in the U.S. or to women who have already developed prediabetes after pregnancy [18, 19].

Our objective was to develop a clinical diabetes risk prediction model specific for women with prior GDM. The model includes estimation of treatment response to metformin, ILI, or neither (placebo) and can be incorporated into a decision aid for clinicians to use in prediabetes treatment counseling.

Materials and methods

Study design

We conducted a secondary analysis of the Diabetes Prevention Program (DPP) data [8]. The analysis was conducted using the publicly-available data which included no participant identifying information. Our study findings can be replicated in its entirety from the DPP data available only upon request to the National Institute of Diabetes and Digestive and Kidney Diseases Central Repositories (https://repository.niddk.nih.gov/studies/dpp/). The DPP was a randomized, controlled clinical trial comparing the effectiveness of ILI, metformin (850 mg twice daily), and placebo for diabetes prevention over a mean follow-up period of 3.2 years. In addition, standard lifestyle recommendations were provided to all participants randomized to receive metformin or placebo. The ILI consisted of an individualized curriculum focusing on nutrition, exercise and behavioral modification. A weight reduction of at least 7 percent of initial body weight was the goal for participants assigned to ILI. The DPP was conducted from 1996–2001 at 27 sites in the U.S. Racial ethnic minorities and women with prior GDM were prespecified target groups in the DPP recruitment protocol [20]. The study design, rationale, and outcomes have been described previously [21]. The institutional review board of the University of Illinois at Chicago reviewed the study protocol and determined this study exempt from human subjects research oversight.

Study population

Diabetes Prevention Program participants were at least 25 years old, had a BMI ≥ 24 kg/m² (≥ 22 kg/m² if Asian) and had prediabetes. Prediabetes was defined by a fasting glucose 95–125 mg/dL and a 2-hour 75-gram OGTT glucose 140–199 mg/dL. Participants were randomized to treatment with placebo, ILI, or metformin. Of note, women who were pregnant, less than 3 months postpartum, or currently nursing or within 6 weeks of completing nursing were excluded from the DPP trial. Women who answered the question, “Have you ever been told that you had a high sugar level or that you have diabetes?” and selected the answer “Only during pregnancy” were considered to have had GDM. The DPP study initially included a fourth intervention, troglitazone, which was discontinued in 1998 because of the drug’s potential liver toxicity. Women with a history of GDM randomized to troglitazone were excluded from our analysis. Our study cohort included a subset of 317 women with prior GDM and prediabetes (Fig 1).

Candidate predictor variables

Candidate predictor variables for the model included 11 baseline clinical variables known to be associated with diabetes progression: age group, ethnicity, parental (either mother or father) history of diabetes (type not specified), BMI group, waist circumference, waist-to-hip ratio, fasting glucose and triglycerides, HbA1c, self-reported physical activity, and treatment arm (placebo, ILI, or metformin). For variables with multiple readings at baseline (waist circumference and hip girth), an average was used. BMI was grouped as < 30, 30 to < 35, and ≥ 35 kg/m². The age of participants was available in 5-year intervals and collapsed at both extremes: age <40 and age ≥ 65 years. A family history of diabetes was determined if either parent had diabetes. An unknown or negative parental history of diabetes was considered negative. Baseline leisure physical activity was assessed with a comprehensive list of activities in the DPP Modifiable Activity Questionnaire (MAQ), which estimated self-reported past-year activity by duration, frequency and relative intensity or metabolic equivalence task (MET) as expressed by MET-hours per week [22]. Physical activity estimates obtained from the MAQ correlated with measures of obesity and glucose tolerance in the DPP [23]. The 2011 Compendium of Physical Activity was used to determine the MET for each activity reported [24].

Outcome measures

The outcome measure was the development of diabetes as defined by the ADA fasting glucose diagnostic cut off value, which in June 1997 was lowered to 126 mg/dL from 140 mg/dL [25]. Diagnostic criteria for diabetes was defined by a fasting plasma glucose ≥ 140 mg/dL (until June 23, 1997) or ≥ 126 mg/dL (on or after June 24, 1997), or a 2-hour 75-gram oral glucose tolerance test (OGTT) ≥ 200 mg/dL [21]. The diagnosis of diabetes was confirmed if consecutive testing with the same criteria, usually within 6 weeks, was met. The data was explored for potential misclassification of diabetes due to the change in diagnostic criteria and we found the ultimate outcome of diabetes at three years was unaffected. All seventeen women who had two or more consecutive glucose readings between 126–140 mg/dL before June 23, 1997 were ultimately diagnosed with diabetes during the three-year follow-up period. Of the women who did not develop diabetes, none had two consecutive fasting glucoses ≥ 126 mg/dL. Clinical assessments were performed routinely every six months during the average three years of monitoring.

Statistical analyses

Bivariate analyses

Descriptive statistics for continuous variables were expressed as median with interquartile range (IQR) due to our small sample size. Categorical variables were described as frequency and proportion. All statistical tests were two-sided. Fisher’s exact and Kruskal Wallis were used, as appropriate for categorical and continuous variables, respectively, to test differences.

Prediction risk model development and evaluation

All models were developed by using multivariable Cox proportional hazards regression and informed by clinical knowledge and previous research [26, 27]. Continuous variables were standardized. Given the extreme right skewness of the fasting triglyceride data, the values were log-transformed to prevent large values from influencing the results. We determined, a priori, that a single model instead of a separate model for each treatment arm would be developed to maximize use of the subsampled DPP data. Residual diagnostic was performed for all models. Proportional hazard assumption for all covariates in all models were checked and satisfied. The 11 candidate variables, including treatment arm (placebo, ILI, or metformin), as well as interactions between treatment arm and BMI group, were entered into an initial model (Model 1). The interaction of treatment arm with BMI group was included because of observed heterogeneity in treatment effects with different BMI groups in the DPP study [8, 28]. We explored reduced models which included known, highly sensitive screening clinical variables (i.e. fasting glucose and HbA1c) and additional commonly accepted predictors (i.e. BMI group, fasting triglycerides) [10, 11, 29–31]. Finally, a parsimonious model (Model 2) was developed that only included significant predictors at 0.05 significance level from the bivariate analysis. Model 2 included screening clinical variables (HbA1c, fasting glucose), BMI group, treatment arm, and the BMI group by treatment arm interaction. To measure the discriminatory performance of the final parsimonious risk model, Harrell’s C-statistic was computed with bias correction [32]. Internal validation was performed with 10-fold cross validation by randomly separating the data into 10 subsets, and estimating the parameters after omitting one of the 10 subsets. The model was applied to the omitted subset and Harrell’s C statistics were calculated for discrimination for each omitted subset. All statistical analyses were performed using R version 3.6.1 (Survival, rms, Survminer, Caret, and survivalROC packages) and can be accessed through https://github.com/bsgerber/dpp-dppos.

Results

Bivariate analyses

Among 3665 adults with prediabetes, 317 (8.6%) women reported a history of GDM with treatment of placebo, ILI, or metformin. Table 1 shows the baseline characteristics of women with prior GDM by treatment arm. Among women with prior GDM: 61.5% were of reproductive age (< 45 years old) and 40.7% self-identified as non-Caucasian. Baseline characteristics were similar among the three treatments with the exception of waist-to-hip ratio (p = 0.02). Three years after randomization, 82 (25.9%) women developed diabetes (Table 2). Diabetes free survival curves for each treatment are provided in S1 Fig. The estimated incidence of diabetes without treatment (placebo group) was 37.4% for women with prior GDM, compared with 20% for either ILI or metformin (p < 0.01).

Table 1. Baseline characteristics by treatment assignment.

	Lifestyle (N = 105)	Metformin (N = 105)	Placebo	p
	Lifestyle (N = 105)	Metformin (N = 105)	(N = 107)	p
Age Group, N (%)				0.13
<40	38 (36.2%)	38 (36.2%)	39 (36.4%)
40–44	29 (27.6%)	28 (26.7%)	23 (21.5%)
45–49	23 (21.9%)	21 (20.0%)	30 (28.0%)
50–54	5 (4.8%)	11 (10.5%)	14 (13.1%)
55–59	5 (4.8%)	2 (1.9%)	1 (0.9%)
60+	5 (4.8%)	5 (4.8%)	0 (0.0%)
Ethnicity, N (%)				0.77
Caucasian	64 (61.0%)	66 (62.9%)	58 (54.2%)
African American	18 (17.1%)	19 (18.1%)	26 (24.3%)
Hispanic, of any race	18 (17.1%)	16 (15.2%)	20 (18.7%)
All other	5 (4.8%)	4 (3.8%)	3 (2.8%)
Smoking Status, N (%)				0.28
Current	8 (7.6%)	4 (3.8%)	5 (4.7%)
Former	27 (25.7%)	22 (21.0%)	34 (31.8%)
≤ 100 cig lifetime	70 (66.7%)	79 (75.2%)	68 (63.6%)
PCOS History, N (%)				0.37
Yes	2 (1.9%)	0 (0.0%)	2 (1.9%)
No	103 (98.1%)	105 (100.0%)	105 (98.1%)
BMI Group kg/m², N (%)				0.55
< 30	29 (27.6%)	30 (28.6%)	25 (23.4%)
30 to < 35	33 (31.4%)	41 (39.0%)	37 (34.6%)
35+	43 (41.0%)	34 (32.4%)	45 (42.1%)
Family History of Diabetes, N (%)	57 (54.3%)	53 (50.5%)	71 (66.4%)	0.05
Live Births, Median (IQR)	2.0 (2.0, 3.0)	2.0 (2.0, 3.0)	2.0 (2.0, 3.0)	0.99
Waist Circumference (cm), Median (IQR)	102.8	98.3	102.5	0.18
Waist Circumference (cm), Median (IQR)	(93.0, 111.5)	(91.8, 107.3)	(91.1, 108.9)	0.18
Waist to Hip Ratio, Median (SD)	0.881 ((0.842, 0.929)	0.867 ((0.823, 0.912)	0.872 (0.836, 0.920)	0.07
Fasting Glucose (mg/dL), Median (IQR)	105	105	106	0.49
Fasting Glucose (mg/dL), Median (IQR)	(100, 110)	(100, 112)	(101, 112)	0.49
Hemoglobin A1c (%), Median (IQR)	5.9 (5.5, 6.1)	5.8 (5.6, 6.1)	5.9 (5.5, 6.2)	0.66
MET-hours/week, Median (IQR)	10.6 (4.2, 18.6)	8.9 (4.0, 20.0)	8.5 (4.1, 16.2)	0.58
Systolic BP (mmHg), Median	118	115	117	0.60
(IQR)	(109, 126)	(108, 124)	(109, 126)	0.60
Diastolic BP (mmHg), Median (IQR)	75 (70, 80)	76 (70, 81)	75 (70, 82)	0.98
Triglyceride (mg/dL), Median (IQR)	136 (97, 204)	119 (94, 179)	130 (98, 193)	0.46

Open in a new tab

Column percentages are presented.

Abbreviations: Cig cigarettes; PCOS Polycystic Ovarian Syndrome; BMI Body Mass Index; MET Metabolic Equivalent Task; IQR Interquartile Range; BP Blood Pressure.

Table 2. Baseline characteristics by diabetes outcome.

	No Diabetes (N = 235)	Diabetes (N = 82)	p
Age Group, N (%)			0.44
<40	88 (37.4%)	27 (32.9%)
40–44	59 (25.1%)	21 (25.6%)
45–49	55 (23.4%)	19 (23.2%)
50–54	18 (7.7%)	12 (14.6%)
55–59	6 (2.6%)	2 (2.4%)
60+	9 (3.8%)	1 (1.2%)
Ethnicity, N (%)			0.14
Caucasian	148 (63.0%)	40 (48.8%)
African American	44 (18.7%)	19 (23.2%)
Hispanic, of any race	35 (14.9%)	19 (23.2%)
All other	8 (3.4%)	4 (4.9%)
Smoking Status, N (%)			0.97
Current	13 (5.5%)	4 (4.9%)
Former	61 (26.0%)	22 (26.8%)
≤100 cig lifetime	161 (68.5%)	56 (68.3%)
PCOS History, N (%)			0.23
Yes	4 (1.7%)	0 (0.0%)
No	231 (98.3%)	82 (100.0%)
BMI Group (kg/m²), N (%)			0.19
<30	59 (25.1%)	25 (30.5%)
30 to <35	89 (37.9%)	22 (26.8%)
35+	87 (37.0%)	35 (42.7%)
Family History of Diabetes, N (%)	133 (56.6%)	48 (58.5%)	0.76
Live Births, Median (IQR)	2.0 (2.0, 3.0)	2.0 (2.0, 3.0)	0.88
Waist Circumference (cm), Median (IQR)	100.0 (92.0, 107.8)	103.9 (91.9, 112.6)	0.47
Waist to Hip Ratio, Median (SD)	0.876 (0.835, 0.914)	0.884 (0.837, 0.935)	0.43
Fasting Glucose (mg/dL), Median (IQR)	104 (100, 110)	111 (106, 118)	< 0.01
Hemoglobin A1c (%), Median (IQR)	5.8 (5.5, 6.1)	6.1 (5.7, 6.3)	< 0.01
MET-hours/week, Median (IQR)	9.0 (4.0, 18.5)	9.4 (4.8, 19.8)	0.58
Systolic BP (mmHg), Median (IQR)	116 (108, 126)	118 (111, 127)	0.20
Diastolic BP (mmHg), Median (IQR)	76 (70, 80)	74 (70, 83)	0.58
Triglyceride (mg/dL), Median (IQR)	129 (98, 186)	112 (91, 209)	0.70
Treatment Arm			<0.01
Placebo	67 (28.5%)	40 (48.8%)
Lifestyle	84 (35.7%)	21 (25.6%)
Metformin	84 (35.7%)	21 (25.6%)

Open in a new tab

Column percentages are presented, row percentages are available in S1 Table.

Abbreviations: Cig cigarettes; PCOS Polycystic Ovarian Syndrome; BMI Body Mass Index; MET Metabolic Equivalent Task; IQR Interquartile Range; BP Blood Pressure.

Risk models

Cox proportional hazard models for Models 1 and 2 are presented in Table 3. Using model 1, fasting glucose and HbA1C were always significant predictors of higher hazard. Women with a BMI ≥ 35 kg/m² had the greatest risk of progressing to diabetes in the placebo arm (S2 Fig). Both models were evaluated and found to have fair discriminative performance corrected with ten-fold cross validation (Table 3). Using likelihood-ratio tests, we assessed the goodness of fit of Models 1 and 2 (rms package) and found no significant differences in discriminative performance. We include the diabetes prediction equation based on Model 2 in Table 4. Additionally, time dependent receiver operating characteristic (ROC) curves were produced for both models (S3 Fig).

Table 3. Two models predicting progression to diabetes at three years for women with prior gestational diabetes.

Baseline Predictors^†	Hazard Ratio (95% CI)
Baseline Predictors^†	Model 1	Model 2
Treatment effect in normal BMI group
Placebo
Lifestyle	0.404 (0.121, 1.346)	0.527 (0.163, 1.709)
Metformin	1.108 (0.447, 2.751)	1.210 (0.510, 2.869)
Fasting glucose, mg/dL	1.537^* (1.233, 1.914)	1.465^* (1.198, 1.793)
Hemoglobin A1c, %	1.449^* (1.096, 1.917)	1.379^* (1.083, 1.755)
BMI group, kg/m², effect in placebo group
< 30
≥ 30 to < 35	0.678 (0.251, 1.829)	1.049 (0.442, 2.486)
≥ 35	0.537 (0.180, 1.605)	1.212 (0.564, 2.607)
Log triglycerides, mg/dL	0.953 (0.740, 1.228)
Race/Ethnicity
Caucasian
African American	0.746 (0.372, 1.498)
Hispanic	1.723 (0.934, 3.179)
Other	1.647 (0.539, 5.031)
Age group, years
< 40
40–44	1.022 (0.521, 2.005)
45–49	1.370 (0.733, 2.560)
50–54	1.483 (0.702, 3.133)
55–59	1.144 (0.220, 5.938)
60+	0.731 (0.093, 5.739)
Family History	0.918 (0.575, 1.466)
Waist circumference, cm	1.426 (0.955, 2.130)
Waist-to-hip ratio	1.066 (0.820, 1.385)
MET-hours/week	1.166 (0.931, 1.460
BMI ≥ 30 to < 35 kg/m²
Lifestyle vs Placebo	1.627 (0.345, 7.676)	1.246 (0.274, 5.667)
Metformin vs Placebo	0.291 (0.066, 1.289)	0.257 (0.061, 1.083)
BMI ≥ 35 kg/m²
Lifestyle vs Placebo	1.175 (0.286, 4.830)	1.017 (0.251, 4.115)
Metformin vs Placebo	0.281 (0.072, 1.095)	0.247^* (0.071, 0.858)
R²	0.191	0.151
Bias corrected C	0.6577	0.6868

Open in a new tab

^†continuous variables are standardized.

*p <0.05

Abbreviations: CI confidence intervals; BMI body mass index.

Table 4. Model equation^† to calculate probability of developing diabetes at 3 years.

Probability of progression to DM = 1-S0(t)^exp(f(x))
S0(3 years) = 0.656
F(X) =
- 0.640 x TREATMENT_L
+ 0.191 x TREATMENT_M
+ 0.047 x BMI₁
+ 0.193 x BMI₂
+ 0.382 x ((FASTING GLUCOSE– 107.1293) / 7.4786)
+ 0.321 x ((HEMOGLOBIN A1c- 5.8427) /0.4834)
+ 0.220 x TREATMENT_L x BMI₁
+ 0.017 x TREATMENT_L x BMI₂
- 1.358 x TREATMENT_M x BMI₁
- 1.400 x TREATMENT_M x BMI₂

Open in a new tab

^†S0 = 3-year survival probability for a woman with the reference covariate pattern where the categorical covariates are set at their reference (placebo, BMI group < 30) and continuous variables are set at the mean (continuous variables are standardized).

TREATMENT_L 1 if treatment is lifestyle, 0 otherwise

TREATMENT_M 1 if treatment is metformin, 0 otherwise

BMI₁ 1 if BMI ≥ 30 to <35 kg/m², 0 otherwise

BMI₂ 1 if BMI ≥ 35 kg/m², 0 otherwise

Discussion

A diabetes risk prediction model using four commonly-available clinical measures was successfully developed for women with prior GDM. Our model performed with similar discrimination as other diabetes prediction models developed from the DPP for the general population [10, 11]. Similar to other models derived from the general population, our parsimonious model includes fasting glucose, HbA1c, BMI, and treatment (ILI, metformin, or placebo) [33]. These commonly available measures offer great value to clinicians in evaluating diabetes risk. Measurement of HbA1c is a guideline-recommended, routinely-assessed screening test for diabetes, and is known to predict diabetes in the DPP cohort and in other large community-based cohorts [25, 34, 35]. In addition, a woman’s BMI is a modifiable, strongly-predictive risk factor for diabetes after a GDM pregnancy [19]. Although measures of abdominal obesity, waist circumference and waist-to-hip ratio correlate with cardiometabolic risk, neither added predictive value in our models [33, 36]. Furthermore, the addition of the triglyceride level, a clinical marker of metabolic syndrome and insulin resistance, did not improve discrimination [29, 30, 37, 38].

Other models have identified similar risk factors for diabetes among women with GDM [18, 19, 39]. In a prospective case-control study of 150 Australian women with GDM and 72 overweight women with normal glucose tolerance, GDM women with a high risk profile (elevated BMI, blood pressure, glucose, insulin, triglyceride, and lower high-density lipoprotein levels) were more likely to develop diabetes compared to women with a low-risk profile in a cluster analysis [40]. Among 1,263 Chinese women with prior GDM, waist circumference, body fat and BMI were all associated with an increased risk of diabetes, with waist circumference and body fat better indicators for diabetes than BMI [41]. Women with previous GDM have a high-risk profile similar to metabolic syndrome where there may be greater benefit from risk reduction therapy [42].

Although a number of risk factors for diabetes are well known, risk estimation is not commonly used in clinical practice. Risk prediction models can facilitate medical decision making [43], and may be more accurate or, at least, less biased than subjective predictions. An evidence-based GDM risk estimate can provide individualized health information to women and clinicians to guide prediabetes treatment discussion. Model equations (Table 4) may help clinicians estimate the risk of progression to diabetes and magnitude of risk reduction with different prediabetes treatment options (i.e., metformin and/or ILI). To realize the benefits of risk prediction, clinical implementation of decision-making activities including aids must be further investigated. Barriers to utilizing risk calculators have been described and may include concern for generalizability and lack of added value to clinical judgment [44, 45].

It is estimated that women with prior GDM have at least a seven-fold increased risk of developing type 2 diabetes compared with those with normoglycemic pregnancies.[46] The model’s modifiable clinical measures appear most relevant to consider in diabetes risk assessment of women with prior GDM, particularly when incorporated into treatment decision making activities. We have developed a decision aid targeting women with prior GDM to assist in decisions regarding diabetes prevention therapy. Models such as those derived from the DPP may be incorporated into decision aids to provide users with calculated diabetes risk scores, if treated with ILI, metformin, or standard lifestyle recommendations.

We recognize our study has a number of limitations. First, this model was developed from DPP participants, and may not be representative of the general population of women with GDM. Adults with certain chronic medical diseases and those taking common chronic medications were excluded from the DPP trial. The diagnosis of GDM was based on self-reporting; however, studies show women recall their GDM diagnosis and treatment accurately [47, 48]. Pregnant women with undiagnosed, preexisting diabetes may be inadvertently diagnosed with GDM, if early first trimester screening is not performed [49]. However, women with prepregnancy diabetes will likely have diabetes after delivery and would have been excluded from the DPP. In addition, women with prior GDM with subsequent prediabetes may not have selected the answer “only during pregnancy” when asked about a history of hyperglycemia; these women were not in our sample.

Women with GDM are at greatest risk of developing diabetes within the postpartum period and first five years of their index pregnancy; the mean time from the index pregnancy was 12 years for GDM women in the DPP [50]. Women who were in the immediate postpartum period and/or breastfeeding were excluded and the mean time from the index pregnancy was 12 years for GDM women in the DPP [50]. Thus, women enrolled in the DPP may not have represented women with the greatest risk for diabetes, and the model may underestimate the predicted risk of diabetes in women with GDM. Second, prediabetes and incident diabetes were defined by fasting blood glucose and 75-gram OGTT criteria in the DPP. In clinical practice, however, screening and diagnosis are more commonly performed with HbA1c. Interestingly, risk reduction by metformin and lifestyle were similar, 44% and 49% respectively, when diabetes incidence was defined by HbA1c ≥ 6.5% (48 mmol/mol) for the overall (men and women) DPP cohort [35]. Third, estimates of risk reduction with an ILI are based on the specific DPP intensive lifestyle program. The risk reduction benefit may not apply to other less-intensive lifestyle interventions. Fourth, our model does not account for treatment adherence (metformin or lifestyle), which would contribute to heterogeneity in treatment effects. Anticipated adherence to therapy is important in decision making, as lifestyle adherence is more effective than metformin in promoting regression to normal glucose regulation [11]. Lastly, our model does not account for peripartum related predictors (e.g. pregnancy OGTT levels or postpartum BMI) which may be more predictive of diabetes risk, though these measures are not always readily available at future medical encounters. Finally, the model was developed utilizing only GDM participants in the DPP. Given the relatively small sample meeting our criteria, external validation is necessary before clinical use.

Conclusion

We have developed and internally validated a clinically applicable prediction model which includes fasting glucose, HbA1c, BMI, treatment arm, and BMI by treatment arm interaction for women with prior GDM. Incorporating individualized diabetes risk prediction into prediabetes treatment decision making may improve understanding the potential benefit of ILI and/or metformin in diabetes prevention.

Supporting information

S1 Fig. Kaplan Meier curves for probability of diabetes free with follow up time of 3 years of women with prediabetes and a history of gestational diabetes in placebo, intensive lifestyle, and metformin arms.

(TIF)

Click here for additional data file.^{(290.7KB, tif)}

S2 Fig. Predicted probability of not progressing to diabetes by BMI group and treatment.

(TIF)

Click here for additional data file.^{(310.8KB, tif)}

S3 Fig. Time dependent receiver operating characteristic (ROC) curves for model 1 (full) and model 2 (parsimonious).

(TIF)

Click here for additional data file.^{(521.6KB, tif)}

S1 Table. Baseline characteristics by diabetes outcomes with row percent.

(DOCX)

Click here for additional data file.^{(19.6KB, docx)}

Acknowledgments

The Diabetes Prevention Program (DPP) was conducted by the DPP Research Group and supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), the General Clinical Research Center Program, the National Institute of Child Health and Human Development (NICHD), the National Institute on Aging (NIA), the Office of Research on Women’s Health, the Office of Research on Minority Health, the Centers for Disease Control and Prevention (CDC), and the American Diabetes Association. The Diabetes Prevention Program dataset supporting the conclusions of this article is available upon request in the National Institute of Diabetes and Digestive and Kidney Diseases Central Repositories, https://repository.niddk.nih.gov/studies/dpp/. This manuscript was not prepared under the auspices of the DPP and does not represent analyses or conclusions of the DPP Research Group, the NIDDK Central Repositories, or the National Institute of Health.

Data Availability

This secondary analysis used publicly accessible data from the Diabetes Prevention Program which is maintained by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Central Repository https://repository.niddk.nih.gov/studies/dpp/. The NIDDK Central Repository collects and maintains data from designated NIDDK studies to facilitate new analyses after a study's original data coordinating center closes. Our analysis used only publicly released data, without any special access privileges from the NIDDK central repository. Formed data (S03, S05 and q03) and non-formed data (basedata, lab and events) were used for our analysis. The Requests can be made by registration through https://repository.niddk.nih.gov/user/login/?next=/requests/data-request/. Data requests must be accompanied by completion of a data agreement form which includes a brief description of the research, research objective and design, analysis plan, and statement of public use.

Funding Statement

B.M. is funded by the UIC Center for Research on Women and Gender, UIC College of Medicine and UIC Department of Medicine and Division of Academic Internal Medicine & Geriatrics.

References

1.DeSisto CL, Kim SY, Sharma AJ. Prevalence estimates of gestational diabetes mellitus in the United States, Pregnancy Risk Assessment Monitoring System (PRAMS), 2007–2010. Prev Chronic Dis. 2014;11:E104. doi: 10.5888/pcd11.130415 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Man B, Turyk ME, Kominiarek MA, Xia Y, Gerber BS. Diabetes Screening in US Women With a History of Gestational Diabetes, National Health and Nutrition Examination Survey, 2007–2012. Prev Chronic Dis. 2016;13:E124. doi: 10.5888/pcd13.160106 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Zhu Y, Zhang C. Prevalence of Gestational Diabetes and Risk of Progression to Type 2 Diabetes: a Global Perspective. Curr Diab Rep. 2016;16(1):7. doi: 10.1007/s11892-015-0699-x [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Kim C, Berger DK, Chamany S. Recurrence of gestational diabetes mellitus: a systematic review. Diabetes Care. 2007;30(5):1314–9. doi: 10.2337/dc06-2517 [DOI] [PubMed] [Google Scholar]
5.Ratner RE, Christophi CA, Metzger BE, Dabelea D, Bennett PH, Pi-Sunyer X, et al. Prevention of diabetes in women with a history of gestational diabetes: effects of metformin and lifestyle interventions. J Clin Endocrinol Metab. 2008;93(12):4774–9. doi: 10.1210/jc.2008-0772 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.ACOG Practice Bulletin No. 190: Gestational Diabetes Mellitus. Obstet Gynecol. 2018;131(2):e49–e64. doi: 10.1097/AOG.0000000000002501 [DOI] [PubMed] [Google Scholar]
7.Management of Diabetes in Pregnancy: Standards of Medical Care in Diabetes-2018. Diabetes Care. 2018;41(Suppl 1):S137–s43. doi: 10.2337/dc18-S013 [DOI] [PubMed] [Google Scholar]
8.Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346(6):393–403. doi: 10.1056/NEJMoa012512 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Aroda VR, Christophi CA, Edelstein SL, Zhang P, Herman WH, Barrett-Connor E, et al. The effect of lifestyle intervention and metformin on preventing or delaying diabetes among women with and without gestational diabetes: the diabetes prevention program outcomes study 10-year follow-up. J Clin Endocrinol Metab. 2015;100(4):1646–53. doi: 10.1210/jc.2014-3761 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Sussman JB, Kent DM, Nelson JP, Hayward RA. Improving diabetes prevention with benefit based tailored treatment: risk based reanalysis of Diabetes Prevention Program. BMJ. 2015;350:h454. doi: 10.1136/bmj.h454 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Herman WH, Pan Q, Edelstein SL, Mather KJ, Perreault L, Barrett-Connor E, et al. Impact of Lifestyle and Metformin Interventions on the Risk of Progression to Diabetes and Regression to Normal Glucose Regulation in Overweight or Obese People With Impaired Glucose Regulation. Diabetes Care. 2017. doi: 10.2337/dc17-1116 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Noble D, Mathur R, Dent T, Meads C, Greenhalgh T. Risk models and scores for type 2 diabetes: systematic review. BMJ. 2011;343:d7163. doi: 10.1136/bmj.d7163 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Bang H, Edwards AM, Bomback AS, Ballantyne CM, Brillon D, Callahan MA, et al. Development and validation of a patient self-assessment score for diabetes risk. Ann Intern Med. 2009;151(11):775–83. doi: 10.7326/0003-4819-151-11-200912010-00005 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Lindström J, Tuomilehto J. The diabetes risk score: a practical tool to predict type 2 diabetes risk. Diabetes Care. 2003;26(3):725–31. doi: 10.2337/diacare.26.3.725 [DOI] [PubMed] [Google Scholar]
15.Köhler M, Ziegler AG, Beyerlein A. Development of a simple tool to predict the risk of postpartum diabetes in women with gestational diabetes mellitus. Acta Diabetol. 2016;53(3):433–7. doi: 10.1007/s00592-015-0814-0 [DOI] [PubMed] [Google Scholar]
16.Lindstrom J, Peltonen M, Eriksson JG, Aunola S, Hamalainen H, Ilanne-Parikka P, et al. Determinants for the effectiveness of lifestyle intervention in the Finnish Diabetes Prevention Study. Diabetes Care. 2008;31(5):857–62. doi: 10.2337/dc07-2162 [DOI] [PubMed] [Google Scholar]
17.Costa B, Barrio F, Cabre JJ, Pinol JL, Cos X, Sole C, et al. Delaying progression to type 2 diabetes among high-risk Spanish individuals is feasible in real-life primary healthcare settings using intensive lifestyle intervention. Diabetologia. 2012;55(5):1319–28. doi: 10.1007/s00125-012-2492-6 [DOI] [PubMed] [Google Scholar]
18.Ekelund M, Shaat N, Almgren P, Groop L, Berntorp K. Prediction of postpartum diabetes in women with gestational diabetes mellitus. Diabetologia. 2010;53(3):452–7. doi: 10.1007/s00125-009-1621-3 [DOI] [PubMed] [Google Scholar]
19.Ignell C, Ekelund M, Anderberg E, Berntorp K. Model for individual prediction of diabetes up to 5 years after gestational diabetes mellitus. Springerplus. 2016;5:318. doi: 10.1186/s40064-016-1953-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Rubin RR, Fujimoto WY, Marrero DG, Brenneman T, Charleston JB, Edelstein SL, et al. The Diabetes Prevention Program: recruitment methods and results. Control Clin Trials. 2002;23(2):157–71. doi: 10.1016/s0197-2456(01)00184-2 [DOI] [PubMed] [Google Scholar]
21.The Diabetes Prevention Program. Design and methods for a clinical trial in the prevention of type 2 diabetes. Diabetes Care. 1999;22(4):623–34. doi: 10.2337/diacare.22.4.623 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Kriska AM, Knowler WC, LaPorte RE, Drash AL, Wing RR, Blair SN, et al. Development of questionnaire to examine relationship of physical activity and diabetes in Pima Indians. Diabetes Care. 1990;13(4):401–11. doi: 10.2337/diacare.13.4.401 [DOI] [PubMed] [Google Scholar]
23.Kriska AM, Edelstein SL, Hamman RF, Otto A, Bray GA, Mayer-Davis EJ, et al. Physical activity in individuals at risk for diabetes: Diabetes Prevention Program. Med Sci Sports Exerc. 2006;38(5):826–32. doi: 10.1249/01.mss.0000218138.91812.f9 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Compendium of Physical Activities 2018. Available from: https://sites.google.com/site/compendiumofphysicalactivities/home. [Google Scholar]
25.Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2018. Diabetes Care. 2018;41(Suppl 1):S13–s27. doi: 10.2337/dc18-S002 [DOI] [PubMed] [Google Scholar]
26.Sauerbrei W, Perperoglou A, Schmid M, Abrahamowicz M, Becher H, Binder H, et al. State of the art in selection of variables and functional forms in multivariable analysis—outstanding issues. Diagnostic and Prognostic Research. 2020;4. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Heinze G, Wallisch C, Dunkler D. Variable selection—A review and recommendations for the practicing statistician. Biom J. 2018;60(3):431–49. doi: 10.1002/bimj.201700067 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Aroda VR, Knowler WC, Crandall JP, Perreault L, Edelstein SL, Jeffries SL, et al. Metformin for diabetes prevention: insights gained from the Diabetes Prevention Program/Diabetes Prevention Program Outcomes Study. Diabetologia. 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Navarro-Gonzalez D, Sanchez-Inigo L, Pastrana-Delgado J, Fernandez-Montero A, Martinez JA. Triglyceride-glucose index (TyG index) in comparison with fasting plasma glucose improved diabetes prediction in patients with normal fasting glucose: The Vascular-Metabolic CUN cohort. Prev Med. 2016;86:99–105. doi: 10.1016/j.ypmed.2016.01.022 [DOI] [PubMed] [Google Scholar]
30.Hjellvik V, Sakshaug S, Strom H. Body mass index, triglycerides, glucose, and blood pressure as predictors of type 2 diabetes in a middle-aged Norwegian cohort of men and women. Clin Epidemiol. 2012;4:213–24. doi: 10.2147/CLEP.S31830 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Stern MP, Williams K, Haffner SM. Identification of persons at high risk for type 2 diabetes mellitus: do we need the oral glucose tolerance test? Ann Intern Med. 2002;136(8):575–81. doi: 10.7326/0003-4819-136-8-200204160-00006 [DOI] [PubMed] [Google Scholar]
32.Harrell FE Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361–87. doi: [DOI] [PubMed] [Google Scholar]
33.Edelstein SL, Knowler WC, Bain RP, Andres R, Barrett-Connor EL, Dowse GK, et al. Predictors of progression from impaired glucose tolerance to NIDDM: an analysis of six prospective studies. Diabetes. 1997;46(4):701–10. doi: 10.2337/diab.46.4.701 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Leong A, Daya N, Porneala B, Devlin JJ, Shiffman D, McPhaul MJ, et al. Prediction of Type 2 Diabetes by Hemoglobin A1c in Two Community-Based Cohorts. Diabetes Care. 2018;41(1):60–8. doi: 10.2337/dc17-0607 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Diabetes Prevention Program Research Group. HbA1c as a predictor of diabetes and as an outcome in the diabetes prevention program: a randomized clinical trial. Diabetes Care. 2015;38(1):51–8. doi: 10.2337/dc14-0886 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Vazquez G, Duval S, Jacobs DR Jr., Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev. 2007;29:115–28. doi: 10.1093/epirev/mxm008 [DOI] [PubMed] [Google Scholar]
37.Grundy SM. Hypertriglyceridemia, insulin resistance, and the metabolic syndrome. Am J Cardiol. 1999;83(9b):25f–9f. doi: 10.1016/s0002-9149(99)00211-8 [DOI] [PubMed] [Google Scholar]
38.Boursier G, Sultan A, Molinari N, Maimoun L, Boegner C, Picandet M, et al. Triglycerides and glycated hemoglobin for screening insulin resistance in obese patients. Clin Biochem. 2018;53:8–12. doi: 10.1016/j.clinbiochem.2017.12.002 [DOI] [PubMed] [Google Scholar]
39.Göbl CS, Bozkurt L, Prikoszovich T, Winzer C, Pacini G, Kautzky-Willer A. Early possible risk factors for overt diabetes after gestational diabetes mellitus. Obstet Gynecol. 2011;118(1):71–8. doi: 10.1097/AOG.0b013e318220e18f [DOI] [PubMed] [Google Scholar]
40.Barden A, Singh R, Walters B, Phillips M, Beilin LJ. A simple scoring method using cardiometabolic risk measurements in pregnancy to determine 10-year risk of type 2 diabetes in women with gestational diabetes. Nutr Diabetes. 2013;3:e72. doi: 10.1038/nutd.2013.15 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Wang L, Liu H, Zhang S, Leng J, Liu G, Zhang C, et al. Obesity index and the risk of diabetes among Chinese women with prior gestational diabetes. Diabet Med. 2014;31(11):1368–77. doi: 10.1111/dme.12532 [DOI] [PubMed] [Google Scholar]
42.Akinci B, Celtik A, Yener S, Yesil S. Prediction of developing metabolic syndrome after gestational diabetes mellitus. Fertil Steril. 2010;93(4):1248–54. doi: 10.1016/j.fertnstert.2008.12.007 [DOI] [PubMed] [Google Scholar]
43.Edwards AG, Naik G, Ahmed H, Elwyn GJ, Pickles T, Hood K, et al. Personalised risk communication for informed decision making about taking screening tests. Cochrane Database Syst Rev. 2013;2:CD001865. doi: 10.1002/14651858.CD001865.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Muller-Riemenschneider F, Holmberg C, Rieckmann N, Kliems H, Rufer V, Muller-Nordhorn J, et al. Barriers to routine risk-score use for healthy primary care patients: survey and qualitative study. Arch Intern Med. 2010;170(8):719–24. doi: 10.1001/archinternmed.2010.66 [DOI] [PubMed] [Google Scholar]
45.Collins GS, Mallett S, Omar O, Yu LM. Developing risk prediction models for type 2 diabetes: a systematic review of methodology and reporting. BMC Med. 2011;9:103. doi: 10.1186/1741-7015-9-103 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Bellamy L, Casas JP, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet. 2009;373(9677):1773–9. doi: 10.1016/S0140-6736(09)60731-5 [DOI] [PubMed] [Google Scholar]
47.Hinkle SN, Rawal S, Zhu Y, Grewal J, Albert PS, Zhang C. Validation of Self-reported Diagnosis of Gestational Diabetes at 6-weeks Postpartum. Epidemiology. 2017;28(5):747–52. doi: 10.1097/EDE.0000000000000695 [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Hosler AS, Nayak SG, Radigan AM. Agreement between self-report and birth certificate for gestational diabetes mellitus: New York State PRAMS. Matern Child Health J. 2010;14(5):786–9. doi: 10.1007/s10995-009-0529-3 [DOI] [PubMed] [Google Scholar]
49.Battarbee AN, Cavallini M, Keller C, Boggess KA. Missed Opportunities for Early Diabetes Screening in Pregnancy. Am J Perinatol. 2021;38(1):23–7. doi: 10.1055/s-0039-1694727 [DOI] [PubMed] [Google Scholar]
50.Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: a systematic review. Diabetes Care. 2002;25(10):1862–8. doi: 10.2337/diacare.25.10.1862 [DOI] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0252501.r001

Decision Letter 0

Andreas Beyerlein

30 Dec 2020

PONE-D-20-34391

A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes

PLOS ONE

Dear Dr. Man,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Feb 13 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

We look forward to receiving your revised manuscript.

Kind regards,

Andreas Beyerlein

Academic Editor

PLOS ONE

Additional Editor Comments:

- I appreciate it very much that the authors aim to make their data fully available. However, the link of the data repository seems to require a personal registration beforehand. Personally, I think this is OK, but it should not be labelled as "data are fully available without restriction".

- Furthermore, the authors should also make their data analysis code available.

- Abstract: It does not get clear that this was a secondary analysis of a clinical trial. For example, the fact that 82 women developed diabetes within three years and the estimated incidences in the groups with and without treatment are not a result of this particular study and should therefore be mentioned in the Methods. It should also be mentioned there how treatment and control groups were defined. Further, the results should be more specific with respect to the associations between the individual predictors and the diabetes risk.

- Was there no information about GDM treatment and breastfeeding available? Both were relevant predictors of postpartum diabetes in an analysis of Köhler et al., Acta Diabetol 2016.

- Further I wonder why the mean time between the first (or last) pregnancy with GDM and study enrolment and the number of diabetic pregnancies were not investigated as potential predictors? At least, both variables should be mentioned in table 1.

- How was family history of diabetes defined? Could it be differed between GDM, type 1 and type 2 diabetes?

- l. 168-170: I suggest to use median and IQR throughout and not to mix it with mean / SD.

- l. 171-173: Fisher's exact test should be used instead of Chi-Square tests when expected cell numbers are too low.

- l. 177-178: Please give another rationale for log-transformation of fasting triglycerides as normal distribution is not a requirement for predictor variables.

- l. 181-191: This approach makes no sense to me. A more common approach would be to start with either all potentially relevant variables (i.e. model 1) or only those variables which were significant in the bivariate analyses (i.e. start with model 3) and then (in either case) reduce the number of variables through variable selection.

- l. 195: Where exactly was cross validation applied and for which purpose?

- Figure 1: Please explain the meaning of the arrow "Troglitazone"

- Tables 1 and 2: It should be mentioned that the % values are column %. However, I think row % might be more appropriate in table 2.

- Table 3: I doubt that the interaction terms of BMI and intervention groups are of use here because the 95% confidence intervals of the respective hazard ratios are so wide. Furthermore, the hazard ratios of the basic variables treatment arm and BMI cannot be intepreted for themselves when the models include the respective interaction terms.

- l. 217-218: How was significance determined in this sentence?

- Table 4 needs more explanation, e.g. what is the meaning of S0, or where do the factors in the fasting glucose and HbA1c lines come from?

- Suggest to show a Kaplan-Meier plot in Suppl. Figure 1.

- Please dicuss how the risk score compares to those from other studies in terms of predicitive performance.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. Please amend your list of authors on the manuscript to ensure that each author is linked to an affiliation. Authors’ affiliations should reflect the institution where the work was done (if authors moved subsequently, you can also list the new affiliation stating “current affiliation:….” as necessary).

4. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: A secondary analysis of the DPP trial which included 317 women with gestational diabetes.

It is an intriuging study but with some caveats: First the risk model is practical and simple and may be used on a primary conselling setting. the simplicity is maybe also the piont of where exact values of risk are not to be trusted: It was a sample of study not geared for pregnancy and the dioagnosis of GDM is based on self-reported statements on what was told during pregnancy. Knowing that glucosuria is a copmmon recurrence in normal pregnancy , too, some misclassification could be present. For second the relative few women to make a model on 5 variables seem to stretch the creditability. Unless, of course, the correlations are that strong that even in each substrata varibales exerts its effect on the outcome. A hint that the sample of 317 may be too skewed is the non-significance of ethnicity. I think too many caucasians with certain traits have been included in the DPP and, consequently, fewer of the ethnic mixed with other diabetogenic backgrounds. Nevertheless, it may be true and work under these circumstances!

Smaller issues:

They introduce the subject by claiming equipotency of metformin and ILI in that both have 50 % reduction diabetes risk; well may be, if applied on similar time shortly before diabetes would manifest itself: while LIL may reduce the risk anytime, metformin is most potent right before the beta-cell goes dry and diabetes would manifest itself

titel page: the affilations denoted as superscripts on the authors names (1-5) do not correspond to the institutions (a-d)

Reviewer #2: This paper presents important information that adds to the GDM knowledge pool, especially in the risk prediction model of subsequent prediabetes or diabetes. The paper is well written, with a clear text, easy to read. However, I have few comments sated as follows:

1. Method section of the abstract: Though, developed and validated a 3-year diabetes prediction model using Cox proportional hazards regression acceptable but the risk prediction model using 11 baseline clinical variables independently and/ or in combined for risk of prediabetes would be computed by the area under the ROC curve (discriminative power). In addition, a risk score calculation based on a combination of the most pertinent clinical variables model prediction for prediabetes is highly suggested.

2. As insulin also given for GDM, why the prediction model is not included about insulin management or why authors focused only lifestyle intervention vs metformin treatment?

3. There is controversies in some statement in the use of prediabetes vs diabetes eg. the title A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes. Meaning that the main outcome focused on “prediabetes”, However the result section of the abstract “ Within three years, 82 (25.9%) women developed diabetes” and Conclusion section of the abstract also “clinical prediction model was developed for individualized decision making for prediabetes treatment in women with prior GDM” . As Prediabtes and Diabetes has different definition and cutoff point of blood glucose levels, the use of the two terms should be managed thorough out the entire document.

4. The introduction section page 6 line 91-105: What is the relevance of the stated sentences about model comparison here? I suggested to remove this and described in detail in the method section.

5. Page 7 line 111 “ ------ neither (placebo)” and result section page 13 Table 1 “Placebo (N=107)” How could the GDM patient in placebo. They should get at least the minimum standard treatment. It has also the ethical concerns?? Clarification is needed

6. Method section Page 7 line 120 “ --- placebo on the development of diabetes over an average of 3.2 years” Confusing ???

7. Page 8 line 134-136:“Women who answered the question, “Have you ever been told that you had a high sugar level or that you have diabetes” and selected the answer “Only during pregnancy” were considered to have had GDM”. What if there is undiagnosed preexisting diabetes mellitus?

8. As authors stated “ The outcome measure was the development of diabetes as defined by ADA criteria.( Clinical assessments were performed routinely every six months during the average three years of monitoring. Diagnostic criteria for diabetes was defined by a fasting plasma glucose ≥ 140 mg/dL (until June 23, 1997) or ≥ 126 mg/dL (on or after June 24, 1997), or a 2-hour 75-gram oral glucose tolerance test (OGTT) ≥ 200 mg/dL.(20) The diagnosis of diabetes was confirmed if consecutive testing with the same criteria, usually within 6 weeks, was met.” The Outcome Measures ascertainment is confusing eg. FPG ≥ 140 mg/dL or ≥ 126 mg/dL ???? This leads missclassifications bias. Clarification is needed

9. Method section: ROC analysis for “Prediction risk model development and evaluation” and its main findings eg. AUC in the result section and supported by figures also suggested.

10. Adding the implication of main findings has recommend to make stronger discussion section

11. Discussion section. Page 21 line 252-264. The paragraphs is not related with your findings eg. Plasma lipidomic analysis, common genetic loci…

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Finn Lauszus

Reviewer #2: Yes: Achenef Asmamaw Muche

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Jun 25;16(6):e0252501. doi: 10.1371/journal.pone.0252501.r002

Author response to Decision Letter 0

13 Apr 2021

We have provided this as a separate attachment but included it in its entirety here.

Dear Plos One Reviewers,

Thank you for giving us the opportunity to submit a revised draft of the manuscript “A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes” for publication in Plos One. We appreciate the time and effort that the reviewers dedicated to providing feedback on our manuscript and are grateful for the insightful comments on and valuable improvements to our paper. We have incorporated most of the suggestions made by the reviewers. Those changes are highlighted within the manuscript. Please see below, in blue, for a point-by-point response to the reviewers’ comments and concerns. All page numbers refer to the revised manuscript file with tracked changes.

1)- I appreciate it very much that the authors aim to make their data fully available. However, the link of the data repository seems to require a personal registration beforehand. Personally, I think this is OK, but it should not be labelled as "data are fully available without restriction".

We have modified the data availability statement to include that access to DPP data is available upon request to the National Institute of Diabetes and Digestive and Kidney Diseases Central Repositories (https://repository.niddk.nih.gov/studies/dpp/) in the manuscript (line 101), acknowledgements (line 479) and cover letter.

2)- Furthermore, the authors should also make their data analysis code available.

Currently, our data analysis code is maintained in a private repository on Github. We will change this to a public repository upon acceptance of the manuscript.

3)- Abstract: It does not get clear that this was a secondary analysis of a clinical trial. For example, the fact that 82 women developed diabetes within three years and the estimated incidences in the groups with and without treatment are not a result of this particular study and should therefore be mentioned in the Methods. It should also be mentioned there how treatment and control groups were defined. Further, the results should be more specific with respect to the associations between the individual predictors and the diabetes risk.

For the methods in the abstract (lines 7-13), we have added that this is a secondary analysis of the DPP and defined the treatment and control groups. For the abstract results, we included the variables (both main and conditional effects) of our parsimonious model but have chosen to not include the hazard ratios and 95% confidence intervals for simplicity (line 15-19).

4)- Was there no information about treatment and breastfeeding available? Both were relevant predictors of postpartum diabetes in an analysis of Köhler et al., Acta Diabetol 2016. - Further I wonder why the mean time between the first (or last) pregnancy with GDM and study enrolment and the number of diabetic pregnancies were not investigated as potential predictors? At least, both variables should be mentioned in table 1.

Thank you for the additional reference. We have added the above article to our citations and note the predictors (lines 74, 77-78) in our introduction. There were no DPP survey questions about breast feeding history or GDM treatment for us to explore as potential predictors. While the survey asked about the total number of live births, which is included in Table 1, there was no question about the number of pregnancies with diabetes. In addition, women who were pregnant, less than 3 months postpartum, or currently nursing or within 6 weeks of completing nursing were excluded from the DPP trial. These exclusions were noted in our discussion of study limitations (lines 430-432) previously and we now also note the exclusions in our methods (lines 125-127) .

5)- How was family history of diabetes defined? Could it be differed between GDM, type 1 and type 2 diabetes?

Family history was defined as a binary indicator variable: (1) either mother or father with diabetes or (2) no known family history. We clarified this in the revision (lines 140-141, 147-148) . The type of diabetes a family member had was not specified in the survey. The survey question asked “Did your mother or father have diabetes?”

6)- l. 168-170: I suggest to use median and IQR throughout and not to mix it with mean / SD.

In Table 1, we now use median and IQR to describe waist-to-hip ratio to be consistent with the other continuous variables.

7)- l. 171-173: Fisher's exact test should be used instead of Chi-Square tests when expected cell numbers are too low.

A few variables did have expected cell counts <5. As suggested, we have revised the comparisons for all the categorial variables in Table 1 to use the Fisher’s exact test instead of Chi Square for conservative comparisons.

8)- l. 177-178: Please give another rationale for log-transformation of fasting triglycerides as normal distribution is not a requirement for predictor variables.

We have added further clarification for the log transformation of triglyceride values which were extremely right skewed (lines 186-188).

9)- l. 181-191: This approach makes no sense to me. A more common approach would be to start with either all potentially relevant variables (i.e. model 1) or only those variables which were significant in the bivariate analyses (i.e. start with model 3) and then (in either case) reduce the number of variables through variable selection.

Model 1 was our initial and main model in which we started with available relevant variables informed by clinical knowledge and previous research (lines 184-186). The common approaches to variable selection (i.e., backward elimination or forward selection) are reliable but may suffer from systematic biases.(see references below) Our initial manuscript Models 2 and 3 were reduced models in which we further explored the role of well-established, known risk factors (e.g., race/ethnicity), and commonly available clinical factors (e.g, fasting triglycerides). We have removed model 2 from the manuscript given its development was not based on the traditional approaches for variable selection. However, we include our parsimonious model (revised Model 2), based on bivariate analysis, and its equation for diabetes risk calculation because of its potential clinically utility (lines 201- 211).

References:

Sauerbrei et al. State of the art in selection of variables and functional forms in multivariable analysis—outstanding issues. Diagnostic and Prognostic Research (2020) 4:3 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7114804/

Heinze et al. Variable selection - A review and recommendations for the practicing statistician. Biom J 2018 May;60(3):431-449. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5969114/

10)- l. 195: Where exactly was cross validation applied and for which purpose?

Estimating discriminatory performance (Harrell’s C) without overfitting requires separating the data into model building and model testing subsets, which we did through 10-fold cross validation. Cross validation was applied for internal validation of the models. We randomly separated the data into 10 subsets, and estimated the parameters after omitting 1 of the 10 subsets. Then we applied the model to the omitted subset. Harrell’s C-statistics were calculated for discrimination for each omitted subset. This is clarified in the methods section (lines 213- 217).

11)- Figure 1: Please explain the meaning of the arrow "Troglitazone"

The DPP study initially included a fourth intervention, troglitazone, which was discontinued in 1998 because of the drug's potential liver toxicity. Women with a history of GDM randomized to troglitazone were excluded from our analysis. We have added this exclusion criteria to our description of Figure 1 with respect to subject selection in the methodology (lines 130- 133).

12)- Tables 1 and 2: It should be mentioned that the % values are column %. However, I think row % might be more appropriate in table 2.

We clarify that column percentages are included in Tables 1 and 2. For Table 2, we believe this is preferable as we are comparing the distribution of factors between two populations (based on outcome of whether developed diabetes or not). We have provided a Supplementary Table 1 to show the row percentages for Table 2 (noted in footnotes of Table 2).

Reference: https://www.annalsthoracicsurgery.org/article/S0003-4975(15)01520-9/fulltext

13)- Table 3: I doubt that the interaction terms of BMI and intervention groups are of use here because the 95% confidence intervals of the respective hazard ratios are so wide. Furthermore, the hazard ratios of the basic variables treatment arm and BMI cannot be interpreted for themselves when the models include the respective interaction terms.

The confidence intervals for the interaction terms are appropriate. The interactions for our exploratory models are significant which is why we felt it was better to keep them in the prediction model (Model 2). Although the regression coefficients of the terms that are part of the interaction cannot be interpreted as main effects, instead they should be interpreted as conditional effects. The effect of BMI will vary depending on the treatment group.

14)- l. 217-218: How was significance determined in this sentence?

We clarified the use of likelihood-ratio tests to compare models in discriminative performance (lines 280-282).

15)- Table 4 needs more explanation, e.g. what is the meaning of S0, or where do the factors in the fasting glucose and HbA1c lines come from?

S0 is a baseline survivor function, in our case, the probability of diabetes by 3 years, where baseline is defined as a set of categorical covariates at their reference categories and at zero for continuous covariates. Since all continuous variables were standardized, zero represents the mean of a variable. The factors in the fasting glucose and HbA1c are the sample means and standard deviations of the respective variables. We have added this to the footnote for Table 4.

16)- Suggest show a Kaplan-Meier plot in Suppl. Figure 1.

We have added Supplementary Figure 1, a Kaplan Meier curve, to show the probability of diabetes free with a follow-up of three years.

17)- Please discuss how the risk score compares to those from other studies in terms of predictive performance.

We added to the discussion clarification of how our prediction model compares to those from other studies in terms of predictive performance (lines 337-339). Our model performed with similar discrimination as other diabetes prediction models developed from the DPP for the general population.

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Check file naming

The DPP data is publicly available upon request to the National Institute of Diabetes and Digestive and Kidney Diseases Central Repositories, https://repository.niddk.nih.gov/studies/dpp/.

In your revised cover letter, please address the following prompts:

We will update your Data Availability statement on your behalf to reflect the information you provide.

We have amended acknowledgements to state: The DPP data is publicly available upon request to the National Institute of Diabetes and Digestive and Kidney Diseases Central Repositories, https://repository.niddk.nih.gov/studies/dpp/.

The author affiliations have been amended, updated and linked to the appropriate authors’ names.

Captions for supplementary figures 1, 2, and 3 and table 1 were added to the end of the manuscript.

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: Yes

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

3. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

Reviewer #2: Yes

4. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

Reviewer #2: Yes

5. Review Comments to the Author

Reviewer #1: A secondary analysis of the DPP trial which included 317 women with gestational diabetes.

This was a secondary analysis of 317 women with a history of GDM and current prediabetes in the DPP study. These women were not pregnant at the time. We have addressed potential for misclassifications in our discussion (lines 414-426) and in item 7 below. In our discussion, we note that the diagnosis of GDM was based on self-reporting; however, studies show women recall their GDM diagnosis and treatment accurately. We agree with the reviewer that the lack of significance of ethnicity, a known risk factor, may have been due to our sample size and race/ethnicity distribution.

Smaller issues:

We agree that timing of intervention (i.e., extent of remaining beta cell function) is important in determining relative treatment efficacy. We clarify that treatment response differs based on individual clinical factors (such as remaining beta cell function) in line 52.

titel page: the affilations denoted as superscripts on the authors names (1-5) do not correspond to the institutions (a-d)

We have corrected the affiliations to match with the superscripts.

Your suggestion is appreciated. We have computed the survival ROC curves for our full and parsimonious models (supplementary Figure 3) and provided the AUC for each model as recommended.

2. As insulin also given for GDM, why the prediction model is not included about insulin management or why authors focused only lifestyle intervention vs metformin treatment?

The women in the DPP were assigned to ILI or metformin – with participation after a pregnancy, to prevent future diabetes. Both are considered clinical treatment options for diabetes prevention or prediabetes treatment (not insulin). If there is interest in insulin given during GDM to influence future development of diabetes, that information (insulin given during GDM pregnancy) was not collected in the DPP study.

We have reviewed our use of “prediabetes” and “diabetes” throughout the manuscript. For our title, prediabetic is used to describe the population for which the model was developed (in addition to the history of GDM). The model was developed for women who had gestational diabetes during a pregnancy and now have prediabetes. Thus, these women have a high risk for developing diabetes. For women who had GDM and now have prediabetes, ADA guidelines recommend both metformin and ILI as “prediabetes treatment” for diabetes prevention.

Reference: https://care.diabetesjournals.org/content/43/Supplement_1/S32

4. The introduction section page 6 line 91-105: What is the relevance of the stated sentences about model comparison here? I suggested to remove this and described in detail in the method section.

We appreciate your suggestion. In our introduction, we wanted to acknowledge the other models that have been published for predicting diabetes in women with a history of GDM and their limitations for clinical use.

For clarification, these women had prior GDM, and upon enrollment had prediabetes. At the time, the study was conducted 20 years ago, there was uncertainty of ILI and metformin vs. placebo in terms of effective diabetes prevention for these women. Thus, at the time, it was considered clinical equipoise. Standard lifestyle recommendations were provided to all participants randomized to metformin or placebo. We have added this detail to the study design description (lines 104-105).

6. Method section Page 7 line 120 “ --- placebo on the development of diabetes over an average of 3.2 years” Confusing ???

We have changed the statement to “The DPP was a randomized, controlled clinical trial comparing the effectiveness of ILI, metformin (850 mg twice daily), and placebo for diabetes prevention over a mean follow-up period of 3.2 years.” (lines 103-106)

As you point out, it is possible that some women who are diagnosed with GDM may, in fact, have undiagnosed pre-existing diabetes. Women with pregestational diabetes will likely continue to have diabetes on postpartum screening and thereafter; therefore they would not have qualified for the DPP study. At the time of the DPP, there was limited use of first trimester screening compared to now. More likely, an answer with “Only during pregnancy” underestimates the number of women with a prior history of GDM. Women with a history of GDM and prediabetes (prediabetes was a DPP inclusion criteria and diabetes was an exclusion criteria) may not have checked “only during pregnancy” because they may have been diagnosed with glucose intolerance after their GDM pregnancy and therefore we may be underestimating women with prior GDM. We note these potential limitations in our discussion (lines 420-426).

The potential for misclassification was our concern too. In the methods, we clarify that the change in diagnostic criteria during the study is based on a change to the American Diabetes Association (ADA) diagnostic criteria in 1997. However, this change in diagnostic criteria did not ultimately affect the outcome and have added details to the methodology (lines 158-160, 168-175).

9. Method section: ROC analysis for “Prediction risk model development and evaluation” and its main findings eg. AUC in the result section and supported by figures also suggested.

Our response to item #1 addresses this.

10. Adding the implication of main findings has recommend to make stronger discussion section

We currently describe some of the implications in the discussion:

Risk prediction models can facilitate medical decision making, and may be more accurate or, at least, less biased than subjective predictions. An evidence-based GDM risk estimate can provide individualized health information to women and clinicians to guide prediabetes treatment discussion. The model equation (now in Table 4) may help clinicians estimate the risk of progression to diabetes and magnitude of risk reduction with different prediabetes treatment options (i.e., metformin and/or ILI). To realize the benefits of risk prediction, clinical implementation of decision-making activities including aids must be further investigated.

Since the manuscript was submitted, we have piloted a decision aid that targets with women with prior GDM. We added the following sentence to the discussion (lines 405-409) : “We have developed a decision aid targeting women with prior GDM to assist in decisions regarding diabetes prevention therapy. Models such as those derived from the DPP may be incorporated into decision aids to provide users with calculated diabetes risk scores, including with ILI, metformin, or no therapy.”

11. Discussion section. Page 21 line 252-264. The paragraphs is not related with your findings eg. Plasma lipidomic analysis, common genetic loci…

This paragraph has been removed.

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

We have answered “Yes”

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Finn Lauszus

Reviewer #2: Yes: Achenef Asmamaw Muche

Attachment

Submitted filename: Response to reviewers03142021.docx

Click here for additional data file.^{(38.1KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0252501.r003

Decision Letter 1

Andreas Beyerlein

9 May 2021

PONE-D-20-34391R1

A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes

PLOS ONE

Dear Dr. Man,

Please submit your revised manuscript by Jun 23 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Andreas Beyerlein

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments (if provided):

The authors improved their manuscript considerably and responded well to my comments. I am willing to accept it for publication when they will put their complete analysis code together with a data dictionary into a publicly available online repository and mention the respective URL in the Methods section.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #2: Yes

**********

6. Review Comments to the Author

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

PLoS One. 2021 Jun 25;16(6):e0252501. doi: 10.1371/journal.pone.0252501.r004

Author response to Decision Letter 1

14 May 2021

We have provided the URL to the analysis code in the methods section.

Attachment

Submitted filename: Response to reviewers05112021.docx

Click here for additional data file.^{(21.1KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0252501.r005

Decision Letter 2

Andreas Beyerlein

18 May 2021

A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes

PONE-D-20-34391R2

Dear Dr. Man,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Andreas Beyerlein

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0252501.r006

Acceptance letter

Andreas Beyerlein

17 Jun 2021

PONE-D-20-34391R2

A Clinical Diabetes Risk Prediction Model For Prediabetic Women With Prior Gestational Diabetes

Dear Dr. Man:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Andreas Beyerlein

Academic Editor

PLOS ONE

Associated Data

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

Supplementary Materials

(TIF)

Click here for additional data file.^{(290.7KB, tif)}

S2 Fig. Predicted probability of not progressing to diabetes by BMI group and treatment.

(TIF)

Click here for additional data file.^{(310.8KB, tif)}

S3 Fig. Time dependent receiver operating characteristic (ROC) curves for model 1 (full) and model 2 (parsimonious).

(TIF)

Click here for additional data file.^{(521.6KB, tif)}

S1 Table. Baseline characteristics by diabetes outcomes with row percent.

(DOCX)

Click here for additional data file.^{(19.6KB, docx)}

Attachment

Submitted filename: Response to reviewers03142021.docx

Click here for additional data file.^{(38.1KB, docx)}

Attachment

Submitted filename: Response to reviewers05112021.docx

Click here for additional data file.^{(21.1KB, docx)}

Data Availability Statement

[pone.0252501.ref001] 1.DeSisto CL, Kim SY, Sharma AJ. Prevalence estimates of gestational diabetes mellitus in the United States, Pregnancy Risk Assessment Monitoring System (PRAMS), 2007–2010. Prev Chronic Dis. 2014;11:E104. doi: 10.5888/pcd11.130415 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref002] 2.Man B, Turyk ME, Kominiarek MA, Xia Y, Gerber BS. Diabetes Screening in US Women With a History of Gestational Diabetes, National Health and Nutrition Examination Survey, 2007–2012. Prev Chronic Dis. 2016;13:E124. doi: 10.5888/pcd13.160106 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref003] 3.Zhu Y, Zhang C. Prevalence of Gestational Diabetes and Risk of Progression to Type 2 Diabetes: a Global Perspective. Curr Diab Rep. 2016;16(1):7. doi: 10.1007/s11892-015-0699-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref004] 4.Kim C, Berger DK, Chamany S. Recurrence of gestational diabetes mellitus: a systematic review. Diabetes Care. 2007;30(5):1314–9. doi: 10.2337/dc06-2517 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref005] 5.Ratner RE, Christophi CA, Metzger BE, Dabelea D, Bennett PH, Pi-Sunyer X, et al. Prevention of diabetes in women with a history of gestational diabetes: effects of metformin and lifestyle interventions. J Clin Endocrinol Metab. 2008;93(12):4774–9. doi: 10.1210/jc.2008-0772 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref006] 6.ACOG Practice Bulletin No. 190: Gestational Diabetes Mellitus. Obstet Gynecol. 2018;131(2):e49–e64. doi: 10.1097/AOG.0000000000002501 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref007] 7.Management of Diabetes in Pregnancy: Standards of Medical Care in Diabetes-2018. Diabetes Care. 2018;41(Suppl 1):S137–s43. doi: 10.2337/dc18-S013 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref008] 8.Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346(6):393–403. doi: 10.1056/NEJMoa012512 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref009] 9.Aroda VR, Christophi CA, Edelstein SL, Zhang P, Herman WH, Barrett-Connor E, et al. The effect of lifestyle intervention and metformin on preventing or delaying diabetes among women with and without gestational diabetes: the diabetes prevention program outcomes study 10-year follow-up. J Clin Endocrinol Metab. 2015;100(4):1646–53. doi: 10.1210/jc.2014-3761 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref010] 10.Sussman JB, Kent DM, Nelson JP, Hayward RA. Improving diabetes prevention with benefit based tailored treatment: risk based reanalysis of Diabetes Prevention Program. BMJ. 2015;350:h454. doi: 10.1136/bmj.h454 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref011] 11.Herman WH, Pan Q, Edelstein SL, Mather KJ, Perreault L, Barrett-Connor E, et al. Impact of Lifestyle and Metformin Interventions on the Risk of Progression to Diabetes and Regression to Normal Glucose Regulation in Overweight or Obese People With Impaired Glucose Regulation. Diabetes Care. 2017. doi: 10.2337/dc17-1116 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref012] 12.Noble D, Mathur R, Dent T, Meads C, Greenhalgh T. Risk models and scores for type 2 diabetes: systematic review. BMJ. 2011;343:d7163. doi: 10.1136/bmj.d7163 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref013] 13.Bang H, Edwards AM, Bomback AS, Ballantyne CM, Brillon D, Callahan MA, et al. Development and validation of a patient self-assessment score for diabetes risk. Ann Intern Med. 2009;151(11):775–83. doi: 10.7326/0003-4819-151-11-200912010-00005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref014] 14.Lindström J, Tuomilehto J. The diabetes risk score: a practical tool to predict type 2 diabetes risk. Diabetes Care. 2003;26(3):725–31. doi: 10.2337/diacare.26.3.725 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref015] 15.Köhler M, Ziegler AG, Beyerlein A. Development of a simple tool to predict the risk of postpartum diabetes in women with gestational diabetes mellitus. Acta Diabetol. 2016;53(3):433–7. doi: 10.1007/s00592-015-0814-0 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref016] 16.Lindstrom J, Peltonen M, Eriksson JG, Aunola S, Hamalainen H, Ilanne-Parikka P, et al. Determinants for the effectiveness of lifestyle intervention in the Finnish Diabetes Prevention Study. Diabetes Care. 2008;31(5):857–62. doi: 10.2337/dc07-2162 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref017] 17.Costa B, Barrio F, Cabre JJ, Pinol JL, Cos X, Sole C, et al. Delaying progression to type 2 diabetes among high-risk Spanish individuals is feasible in real-life primary healthcare settings using intensive lifestyle intervention. Diabetologia. 2012;55(5):1319–28. doi: 10.1007/s00125-012-2492-6 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref018] 18.Ekelund M, Shaat N, Almgren P, Groop L, Berntorp K. Prediction of postpartum diabetes in women with gestational diabetes mellitus. Diabetologia. 2010;53(3):452–7. doi: 10.1007/s00125-009-1621-3 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref019] 19.Ignell C, Ekelund M, Anderberg E, Berntorp K. Model for individual prediction of diabetes up to 5 years after gestational diabetes mellitus. Springerplus. 2016;5:318. doi: 10.1186/s40064-016-1953-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref020] 20.Rubin RR, Fujimoto WY, Marrero DG, Brenneman T, Charleston JB, Edelstein SL, et al. The Diabetes Prevention Program: recruitment methods and results. Control Clin Trials. 2002;23(2):157–71. doi: 10.1016/s0197-2456(01)00184-2 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref021] 21.The Diabetes Prevention Program. Design and methods for a clinical trial in the prevention of type 2 diabetes. Diabetes Care. 1999;22(4):623–34. doi: 10.2337/diacare.22.4.623 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref022] 22.Kriska AM, Knowler WC, LaPorte RE, Drash AL, Wing RR, Blair SN, et al. Development of questionnaire to examine relationship of physical activity and diabetes in Pima Indians. Diabetes Care. 1990;13(4):401–11. doi: 10.2337/diacare.13.4.401 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref023] 23.Kriska AM, Edelstein SL, Hamman RF, Otto A, Bray GA, Mayer-Davis EJ, et al. Physical activity in individuals at risk for diabetes: Diabetes Prevention Program. Med Sci Sports Exerc. 2006;38(5):826–32. doi: 10.1249/01.mss.0000218138.91812.f9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref024] 24.Compendium of Physical Activities 2018. Available from: https://sites.google.com/site/compendiumofphysicalactivities/home. [Google Scholar]

[pone.0252501.ref025] 25.Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2018. Diabetes Care. 2018;41(Suppl 1):S13–s27. doi: 10.2337/dc18-S002 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref026] 26.Sauerbrei W, Perperoglou A, Schmid M, Abrahamowicz M, Becher H, Binder H, et al. State of the art in selection of variables and functional forms in multivariable analysis—outstanding issues. Diagnostic and Prognostic Research. 2020;4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref027] 27.Heinze G, Wallisch C, Dunkler D. Variable selection—A review and recommendations for the practicing statistician. Biom J. 2018;60(3):431–49. doi: 10.1002/bimj.201700067 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref028] 28.Aroda VR, Knowler WC, Crandall JP, Perreault L, Edelstein SL, Jeffries SL, et al. Metformin for diabetes prevention: insights gained from the Diabetes Prevention Program/Diabetes Prevention Program Outcomes Study. Diabetologia. 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref029] 29.Navarro-Gonzalez D, Sanchez-Inigo L, Pastrana-Delgado J, Fernandez-Montero A, Martinez JA. Triglyceride-glucose index (TyG index) in comparison with fasting plasma glucose improved diabetes prediction in patients with normal fasting glucose: The Vascular-Metabolic CUN cohort. Prev Med. 2016;86:99–105. doi: 10.1016/j.ypmed.2016.01.022 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref030] 30.Hjellvik V, Sakshaug S, Strom H. Body mass index, triglycerides, glucose, and blood pressure as predictors of type 2 diabetes in a middle-aged Norwegian cohort of men and women. Clin Epidemiol. 2012;4:213–24. doi: 10.2147/CLEP.S31830 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref031] 31.Stern MP, Williams K, Haffner SM. Identification of persons at high risk for type 2 diabetes mellitus: do we need the oral glucose tolerance test? Ann Intern Med. 2002;136(8):575–81. doi: 10.7326/0003-4819-136-8-200204160-00006 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref032] 32.Harrell FE Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361–87. doi: [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref033] 33.Edelstein SL, Knowler WC, Bain RP, Andres R, Barrett-Connor EL, Dowse GK, et al. Predictors of progression from impaired glucose tolerance to NIDDM: an analysis of six prospective studies. Diabetes. 1997;46(4):701–10. doi: 10.2337/diab.46.4.701 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref034] 34.Leong A, Daya N, Porneala B, Devlin JJ, Shiffman D, McPhaul MJ, et al. Prediction of Type 2 Diabetes by Hemoglobin A1c in Two Community-Based Cohorts. Diabetes Care. 2018;41(1):60–8. doi: 10.2337/dc17-0607 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref035] 35.Diabetes Prevention Program Research Group. HbA1c as a predictor of diabetes and as an outcome in the diabetes prevention program: a randomized clinical trial. Diabetes Care. 2015;38(1):51–8. doi: 10.2337/dc14-0886 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref036] 36.Vazquez G, Duval S, Jacobs DR Jr., Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev. 2007;29:115–28. doi: 10.1093/epirev/mxm008 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref037] 37.Grundy SM. Hypertriglyceridemia, insulin resistance, and the metabolic syndrome. Am J Cardiol. 1999;83(9b):25f–9f. doi: 10.1016/s0002-9149(99)00211-8 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref038] 38.Boursier G, Sultan A, Molinari N, Maimoun L, Boegner C, Picandet M, et al. Triglycerides and glycated hemoglobin for screening insulin resistance in obese patients. Clin Biochem. 2018;53:8–12. doi: 10.1016/j.clinbiochem.2017.12.002 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref039] 39.Göbl CS, Bozkurt L, Prikoszovich T, Winzer C, Pacini G, Kautzky-Willer A. Early possible risk factors for overt diabetes after gestational diabetes mellitus. Obstet Gynecol. 2011;118(1):71–8. doi: 10.1097/AOG.0b013e318220e18f [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref040] 40.Barden A, Singh R, Walters B, Phillips M, Beilin LJ. A simple scoring method using cardiometabolic risk measurements in pregnancy to determine 10-year risk of type 2 diabetes in women with gestational diabetes. Nutr Diabetes. 2013;3:e72. doi: 10.1038/nutd.2013.15 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref041] 41.Wang L, Liu H, Zhang S, Leng J, Liu G, Zhang C, et al. Obesity index and the risk of diabetes among Chinese women with prior gestational diabetes. Diabet Med. 2014;31(11):1368–77. doi: 10.1111/dme.12532 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref042] 42.Akinci B, Celtik A, Yener S, Yesil S. Prediction of developing metabolic syndrome after gestational diabetes mellitus. Fertil Steril. 2010;93(4):1248–54. doi: 10.1016/j.fertnstert.2008.12.007 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref043] 43.Edwards AG, Naik G, Ahmed H, Elwyn GJ, Pickles T, Hood K, et al. Personalised risk communication for informed decision making about taking screening tests. Cochrane Database Syst Rev. 2013;2:CD001865. doi: 10.1002/14651858.CD001865.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref044] 44.Muller-Riemenschneider F, Holmberg C, Rieckmann N, Kliems H, Rufer V, Muller-Nordhorn J, et al. Barriers to routine risk-score use for healthy primary care patients: survey and qualitative study. Arch Intern Med. 2010;170(8):719–24. doi: 10.1001/archinternmed.2010.66 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref045] 45.Collins GS, Mallett S, Omar O, Yu LM. Developing risk prediction models for type 2 diabetes: a systematic review of methodology and reporting. BMC Med. 2011;9:103. doi: 10.1186/1741-7015-9-103 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref046] 46.Bellamy L, Casas JP, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet. 2009;373(9677):1773–9. doi: 10.1016/S0140-6736(09)60731-5 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref047] 47.Hinkle SN, Rawal S, Zhu Y, Grewal J, Albert PS, Zhang C. Validation of Self-reported Diagnosis of Gestational Diabetes at 6-weeks Postpartum. Epidemiology. 2017;28(5):747–52. doi: 10.1097/EDE.0000000000000695 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0252501.ref048] 48.Hosler AS, Nayak SG, Radigan AM. Agreement between self-report and birth certificate for gestational diabetes mellitus: New York State PRAMS. Matern Child Health J. 2010;14(5):786–9. doi: 10.1007/s10995-009-0529-3 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref049] 49.Battarbee AN, Cavallini M, Keller C, Boggess KA. Missed Opportunities for Early Diabetes Screening in Pregnancy. Am J Perinatol. 2021;38(1):23–7. doi: 10.1055/s-0039-1694727 [DOI] [PubMed] [Google Scholar]

[pone.0252501.ref050] 50.Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: a systematic review. Diabetes Care. 2002;25(10):1862–8. doi: 10.2337/diacare.25.10.1862 [DOI] [PubMed] [Google Scholar]

PERMALINK

A clinical diabetes risk prediction model for prediabetic women with prior gestational diabetes

Bernice Man

Alan Schwartz

Oksana Pugach

Yinglin Xia

Ben Gerber

Roles

Abstract

Introduction

Methods

Results

Conclusions

Introduction

Materials and methods

Study design

Study population

Fig 1. Flow diagram for selection of women with GDM in the DPP cohort.

Candidate predictor variables

Outcome measures

Statistical analyses

Bivariate analyses

Prediction risk model development and evaluation

Results

Bivariate analyses

Table 1. Baseline characteristics by treatment assignment.

Table 2. Baseline characteristics by diabetes outcome.

Risk models

Table 3. Two models predicting progression to diabetes at three years for women with prior gestational diabetes.

Table 4. Model equation† to calculate probability of developing diabetes at 3 years.

Discussion

Conclusion

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Andreas Beyerlein

Roles

Author response to Decision Letter 0

Decision Letter 1

Andreas Beyerlein

Roles

Author response to Decision Letter 1

Decision Letter 2

Andreas Beyerlein

Roles

Acceptance letter

Andreas Beyerlein

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Table 4. Model equation^† to calculate probability of developing diabetes at 3 years.