One-Hour Oral Glucose Tolerance Test for the Postpartum Reclassification of Women With Hyperglycemia in Pregnancy Free

The International Diabetes Federation (IDF) recently published a position statement supporting a 1-h oral glucose tolerance test (OGTT) (1). In this position statement, the IDF working group reviewed an extensive literature demonstrating the value of 1-h plasma glucose over fasting and 2-h glucose measurements for predicting diabetes and associated complications (including mortality) in study populations spanning a broad range of age and ethnicities in both women and men (1–7). On the basis of these data, the IDF working group supported glycemic thresholds by which the 1-h glucose value may identify dysglycemic states of 1) intermediate hyperglycemia (IH, which is analogous to prediabetes) and 2) diabetes (1). Specifically, 1-h glucose ≥8.6 mmol/L defines IH/prediabetes, while a value of ≥11.6 mmol/L is indicative of diabetes (1,2,8,9). Moreover, the working group noted specific advantages of a 1-h OGTT over the traditional 2-h OGTT, including the convenience of the shorter duration of the test and the earlier detection of high-risk individuals (since the rise in 1-h glucose precedes that of 2-h glucose) (1–3,6,10).

A clinical setting in which both of these advantages may be particularly relevant is the postpartum assessment of women with hyperglycemia in pregnancy (11). This patient population has an elevated risk of glycemic deterioration over time, which may lead to prediabetes and ultimately type 2 diabetes (T2D) (11–13). For this reason, women with hyperglycemia in pregnancy are advised to undergo a 2-h 75-g OGTT within 6 months after delivery (11,14,15). While initially performed for postpartum reclassification of their glucose tolerance, the test also serves to identify those at highest risk of subsequent progression to T2D. However, in clinical practice, this recommended test following gestational diabetes mellitus (GDM) is often missed, with completion rates of ∼50% at best (16–18), thereby representing a missed opportunity for the early detection of high-risk women (that could otherwise enable preventive intervention) (19,20). Despite its widespread recognition as such, this missed opportunity has defied simple solution, owing to its multifactorial etiology including uncertainty as to which provider should be responsible for ordering the 2-h OGTT (e.g., obstetrician, endocrinologist, or family physician) and the logistical difficulty that a 2-h test poses for busy new mothers (20–22). In this context, the 1-h OGTT potentially could offer important advantages of both convenience to the patient (because of its shorter duration) and earlier detection of those experiencing glycemic deterioration. However, to our knowledge, the 1-h OGTT has not been evaluated in this setting. Thus, in this study, our objective was to directly compare 1-h and 2-h OGTT glucose measurements at 3 months postpartum as predictors of dysglycemia over the first 5 years after delivery in a cohort of women reflecting the full range of hyperglycemia in pregnancy.

This study was performed in the setting of a prospective observational cohort program at our institution in which women comprising the full spectrum of gestational glucose tolerance are recruited in pregnancy and then undergo serial evaluation in the years thereafter. The protocol has been previously described in detail (23) and has been approved by the Mount Sinai Hospital Research Ethics Board (Toronto, Canada), with all participants having provided written informed consent. In brief, all pregnant women at our institution are screened for GDM by 50-g glucose challenge test (GCT) at 24–28 weeks gestation, followed by an OGTT in those in whom the GCT is abnormal (1-h postchallenge glucose ≥7.8 mmol/L). For this program, women were recruited either before or after the GCT, with the latter recruitment serving to enrich the study population for those with varying degrees of dysglycemia in pregnancy. Women were eligible if they had no known diabetes before pregnancy or major chronic inflammatory conditions. Irrespective of the outcome of their GCT (even if it were normal), all participants completed a 3-h 100-g OGTT in pregnancy, on which National Diabetes Data Group criteria were applied for diagnosing GDM (23). As previously described (23,24), this recruitment strategy was designed to generate a cohort reflecting the full spectrum of glucose tolerance in pregnancy (from normal to mild abnormalities on GCT/OGTT to GDM) and hence a broad range of future risk of prediabetes/diabetes (24–26). After delivery, women returned for metabolic characterization, including multisample 2-h 75-g OGTT at 3 months and 1 year postpartum. At 1 year postpartum, they were recruited into the cohort analyzed herein, which involved repeat cardiometabolic characterization at 3 years and 5 years postpartum. Of 471 women recruited into this cohort at 1 year postpartum, there were 369 who have completed the visit at 5 years postpartum, thereby yielding the study population for the current analysis. Comparison of these 369 participants to those who have not completed the 5-year visit reveals that the former were slightly older (mean age 35.7 ± 4.3 vs. 34.6 ± 4.5 years, P = 0.04), with no other significant differences at 3 months postpartum in ethnicity, family history of diabetes, preceding GDM, BMI, waist, fasting glucose, 1-h glucose, or 2-h glucose (data not shown).

At each postpartum visit (at 3 months, 1 year, 3 years, and 5 years after pregnancy), participants came to the clinical investigation unit in the morning after overnight fast and underwent a 2-h 75-g OGTT. On each OGTT, venous blood samples were drawn for measurement of glucose and insulin at fasting and at 30, 60, and 120 min postchallenge, as previously described (23). These measurements enabled assessment of insulin sensitivity/resistance, β-cell function, and glucose tolerance.

Whole-body insulin sensitivity was assessed by Matsuda index (27), and insulin resistance (IR) (primarily hepatic) was measured by homeostasis model assessment (HOMA-IR) (28). The β-cell compensation was evaluated with insulin secretion-sensitivity index-2 (ISSI-2), which is an OGTT-based measure that is analogous to the disposition index from the i.v. glucose tolerance test, against which it has been directly validated (29,30). Insulinogenic index/HOMA-IR (IGI/HOMA-IR) provided a second measure of β-cell compensation (23).

For the current analysis, glucose tolerance status was evaluated with both 1) standard 2-h OGTT criteria of the World Health Organization (WHO) that are endorsed by Diabetes Canada Clinical Practice Guidelines (31) and 2) the 1-h OGTT criteria endorsed by IDF (1). The WHO/Diabetes Canada 2-h OGTT criteria (31) define dysglycemia as either prediabetes or diabetes, wherein prediabetes refers to impaired fasting glucose tolerance (IFG), impaired glucose tolerance, or both. These dysglycemic states on 2-h OGTT are defined as follows: IFG is fasting glucose between 6.1 and 6.9 mmol/L inclusive, impaired glucose tolerance is 2-h glucose between 7.8 and 11.0 mmol/L inclusive, and diabetes is fasting glucose ≥7.0 mmol/L or 2-h glucose ≥11.1 mmol/L.

The IDF 1-h OGTT criteria (1) define thresholds by which the 1-h glucose additionally defines dysglycemia as either IH or diabetes, with thresholds as follows: IH is 1-h glucose ≥8.6 mmol/L, and diabetes is 1-h glucose ≥11.6 mmol/L.

For the current analyses, the outcome of dysglycemia was defined by either WHO 2-h OGTT criteria (31) or IDF 1-h OGTT criteria (1), because application of only one of these two sets of criteria would otherwise bias the comparison of 1-h glucose and 2-h glucose in identifying this outcome. Thus, for these analyses, the outcome of dysglycemia was defined as a composite of IH, prediabetes, and diabetes, diagnosed by either 1-h or 2-h criteria.

All analyses were performed with either R 4.3.0 or Statistical Analysis System (SAS 9.4; SAS Institute, Cary NC). The study population was first stratified into tertiles of 1-h glucose on the OGTT at 3 months postpartum, with characteristics compared between the groups at each visit (Table 1). Continuous variables were compared with ANOVA (if normally distributed) or Wilcoxon rank-sum nonparametric test (if skewed). Categorical variables were compared by χ² test. A Venn diagram was constructed to show the distribution of individuals who met at least one of the two sets of criteria (2-h or 1-h) for dysglycemia at 3 months postpartum (Supplementary Fig. 1).

To assess the relationship between 1-h glucose at 3 months postpartum and subsequent time to dysglycemia (IH/prediabetes/diabetes), the cumulative incidence plot by tertiles of 1-h glucose was prepared, with curves compared by log-rank test (Fig. 1). To compare the predictive capacity of 1-h and 2-h glucose while considering time and censoring, we generated time-dependent receiver operating characteristic (ROC) curves (32) of 1-h and 2-h glucose at 3 months postpartum as predictors of dysglycemia at 1 year, 3 years, and 5 years postpartum (Fig. 2). After confirming by scatterplots that there was no extreme U-shaped relationship between the two predictors and the observed rate of dysglycemia at each postpartum visit, we performed the ROC analyses. U statistic–based test was applied to compare each pair of the area under the time-dependent ROC curves at the indicated three time points. The P values were adjusted for multiple comparisons.

To directly compare 1-h and 2-h glucose measurements on the OGTT at 3 months postpartum for predicting future risk of dysglycemia (IH/prediabetes/diabetes), we constructed Cox proportional hazard regression models with length of follow-up as the timescale, which was calculated from the date of the 3 months postpartum visit to the date at which dysglycemia first occurred in study population (n = 369) (Table 2A), women with preceding GDM (n = 135) (Table 2B), and all recruited women (n = 471) (Table 2C). Participants who had not developed dysglycemia by 5 years postpartum or had incomplete OGTT values at the latter visit were censored. For all Cox proportional regression analyses, we tested the proportional hazards assumption by generating the time-dependent covariates by creating interactions of the predictors and a function of survival time for inclusion in the model. In each panel, we performed the following three-step procedure.

1. For step 1, we constructed an unadjusted model and obtained the unadjusted hazard ratio (HR) and its 95% CI for each clinical predictor at 3 months postpartum, including fasting and 1-h and 2-h glucose.

2. For step 2, we then constructed a core model consisting of diabetes risk factors (age, ethnicity, family history of diabetes, BMI at 3 months, duration of breastfeeding), upon which we performed further adjustments for fasting and 1-h and 2-h glucose (yielding adjusted HRs and 95% CIs).

3. For step 3, we next calculated and compared the concordance index (CCI) in CONCORDANCE package in R for each adjusted model. CCI can be interpreted as the fraction of all pairs of participants whose predicted survival times are correctly ordered among those that can actually be ordered, reflecting the probability of concordance between predicted and observed survival time (33). The larger the CCI, the greater the contribution of the predictor to the outcome. When testing the importance of each clinical predictor, a reduced model was constructed by removing the tested predictor from the core model. The percentage change in CCI from the core model and reduced model was computed. When testing the importance of each glucose measurement, we added each in turn (fasting/1-h glucose/2-h glucose) and calculated the percentage change in CCI from the additive model and core model.

Deidentified data can be made available under restricted access from the corresponding author, for academic purposes, subject to a material transfer agreement and approval of the Mount Sinai Hospital Research Ethics Board.

Table 1 shows the characteristics of the study population stratified into tertiles of 1-h glucose on the OGTT at 3 months postpartum. The highest tertile had a higher proportion of women of non-White ethnicity (P < 0.0001) and those who had GDM (P < 0.0001), compared with the middle and lower tertiles. Women in the highest tertile of 1-h glucose also had higher BMI at 3 months postpartum (P < 0.046), with no difference between the groups in waist circumference. There was a stepwise decrease in whole-body insulin sensitivity (Matsuda index) from lowest to middle to highest tertile of 1-h glucose (P < 0.0001), coupled with progressively worsening β-cell function (ISSI-2: P < 0.0001; IGI/HOMA-IR: P < 0.0001). As per definition, 1-h glucose increased from the lowest to middle to highest tertile (P < 0.0001). Of note, both fasting and 2-h glucose showed the same progression (both P < 0.0001). Moreover, rates of prediabetes and diabetes (by 2-h OGTT criteria) similarly rose from the lowest to the middle to the highest tertile (P < 0.0001).

On repeat OGTTs at 1 year, 3 years, and 5 years postpartum, this pattern of metabolic differences across the groups persisted. Specifically, on each of these assessments, there was a stepwise worsening of Matsuda index (all P < 0.0001), ISSI-2 (all P < 0.0001), and IGI/HOMA-IR (all P < 0.0001) from the lowest to the middle to the highest tertile of 1-h glucose at 3 months. Moreover, at 1 year, 3 years, and 5 years postpartum, there was a progressive increase across these groups in fasting glucose (all P ≤ 0.0002), 1-h glucose (P < 0.0001), 2-h glucose (P < 0.0001), and prevalence of prediabetes and diabetes by 2-h OGTT criteria (P < 0.0001), respectively. Thus, stratification of the study population by tertiles of 1-h glucose at 3 months postpartum was associated with metabolic risk at that visit and longitudinally thereafter at 1 year, 3 years, and 5 years.

We next evaluated the relationship between 1-h glucose at 3 months postpartum and diagnosis of dysglycemia (IH/prediabetes/diabetes). On the OGTT at 3 months postpartum, 1-h glucose identified all but 10 of the 70 women concurrently diagnosed with dysglycemia by 2-h glucose, while identifying an additional 96 women who met the 1-h diagnostic threshold but not the 2-h criterion (Supplementary Fig. 1). Thus, at 3 months, 1-h glucose identified 156 women with dysglycemia, while 2-h glucose identified 70 women (60 of whom met 1-h criterion as well). Similarly, as shown in Supplementary Table 1, the number of women diagnosed with dysglycemia by 1-h glucose exceeded the number diagnosed by 2-h glucose at 1 year, 3 years, and 5 years postpartum. Specifically, at 1 year, there were 68 women with dysglycemia by 1-h glucose alone, 13 with dysglycemia by 2-h glucose alone and 42 meeting both criteria. At 3 years, 63 women had dysglycemia by 1-h glucose, 16 had dysglycemia by 2-h glucose, and 58 women met both criteria. At 5 years, there were 48 women with dysglycemia by 1-h glucose alone, 12 with dysglycemia by 2-h glucose alone, and 76 meeting both criteria. Thus, during this time, the number of women diagnosed by only 1-h glucose decreased while the corresponding number identified by both 1-h and 2-h glucose increased, consistent with the elevation of 1-h glucose preceding that of 2-h glucose over time.

We then considered the longitudinal relationship between 1-h glucose at 3 months and dysglycemia over the subsequent years. As shown in Fig. 1, the cumulative incidence of dysglycemia up to 6 years postpartum progressively increased from the lowest to the middle to the highest tertile of 1-h glucose on the 3-month OGTT (P < 0.0001). It thus emerges that the 1-h glucose at 3 months postpartum is associated with the rate of dysglycemia over the subsequent years postpartum.

To compare the capacity of 1-h and 2-h glucose at 3 months for predicting future dysglycemia, we evaluated time-dependent ROC curves at 1 year, 3 years, and 5 years postpartum (Fig. 2). These analyses revealed no significant differences between 1-h and 2-h glucose at 3 months in identifying dysglycemia at 1 year (P = 0.31), 3 years (P = 0.94), or 5 years postpartum (P = 0.99).

Next, we constructed Cox proportional hazard regression models to directly compare the 1-h and 2-h glucose measurements at the 3-month OGTT as independent predictors of dysglycemia over the first 5 years postpartum. As anticipated, when dysglycemia is defined only by 1-h criteria, 1-h glucose at 3 months is a better predictor of this outcome than is 2-h glucose at 3 months (Supplementary Table 2), and, when dysglycemia is defined only by 2-h criteria, 2-h glucose is a better predictor than 1-h glucose (Supplementary Table 3). Thus, for unbiased comparison of 1-h and 2-glucose at 3 months as predictors of dysglycemia, the outcome was defined by either 1-h or 2-h criteria (Table 2). On the resultant unadjusted analysis, each of the three glucose values on the OGTT was significantly associated with dysglycemia (fasting glucose, 1-h glucose, 2-h glucose: all P < 0.0001). On the basis of CCI, 1-h glucose would be ranked first (CCI: 0.7003), followed by 2-h glucose (CCI: 0.6797) and fasting glucose (CCI: 0.6135). On adjusted analyses of clinical predictors at 3 months (Table 2A), the following factors emerged as significant independent determinants of future dysglycemia: each of the three glucose values on the OGTT (fasting/1-h glucose/2-h glucose: all P < 0.0001), BMI (P < 0.0001), and non-White ethnicity (Asian: P < 0.0001; other ethnicity: P = 0.004). On the basis of their impact on the CCI of the models, these predictors were ranked as follows in descending order of importance: 1-h glucose (change in CCI 16.1%), 2-h glucose (14.9%), non-White ethnicity (6.7%), fasting glucose (5.2%), and BMI (3.2%). On additional analyses restricted to only those with normal glucose tolerance at 3 months postpartum, 1-h glucose remained the highest-ranked predictor (13.0%), followed by 2-h glucose (12.0%), non-White ethnicity (5.2%), and a tie between fasting glucose and BMI (both 3.8%) (Supplementary Table 4). Moreover, upon restriction to the 135 women who had recent GDM, the relative rankings of 1-h and 2-h glucose were unchanged. Specifically, the highest-ranked predictor was 1-h glucose (13.0%), followed by 2-h glucose (12.8%), BMI (4.2%), and non-White ethnicity (3.1%) (Table 2B). Finally, performance of the analyses in all 471 recruited women identified 1-h glucose on the OGTT at 3 months postpartum as the top-ranked predictor of dysglycemia (14.7%), followed by 2-h glucose (12.1%), non-White ethnicity (6.3%), and fasting glucose (5.0%) (Table 2C).

In this report, we demonstrate three key findings. First, higher 1-h glucose at 3 months postpartum is associated with poorer β-cell function and insulin sensitivity, both at the time and in the years thereafter. Second, this 1-h glucose measurement identifies dysglycemia in almost all of those so diagnosed by the 2-h glucose value, while identifying many additional women not diagnosed by the latter. Third, and most importantly, regression analyses adjusted for clinical risk factors for diabetes reveal that, on the OGTT at 3 months postpartum, the 1-h glucose value is a stronger predictor of subsequent dysglycemia over the first 5 years after pregnancy than is the 2-h glucose value, both in the full study population and in women with recent GDM. Taken together, these findings support consideration of a 1-h OGTT for postpartum reclassification of women with hyperglycemia in pregnancy.

Previous studies have shown that higher 1-h blood glucose is associated with poorer β-cell function and lower insulin sensitivity (34–36). In the current study, we first demonstrate that these pathophysiologic associations also apply to 1-h glucose at 3 months postpartum, both cross-sectionally and longitudinally at 1 year, 3 years, and 5 years after pregnancy. The significance of these findings rests in the fact that deterioration of β-cell compensation over time (often exacerbated by IR) is the key pathophysiologic determinant of postpartum progression to prediabetes and T2D in women with previous GDM (23,37). Thus, its associations with β-cell dysfunction provide a pathophysiologic basis for why 1-h glucose may predict progression to dysglycemia in this clinical setting.

In this regard, our multifaceted comparison of 1-h versus 2-h glucose then yielded very consistent findings. Notably, at 3 months postpartum, the 1-h measurement identified all but 10 of 70 women diagnosed with dysglycemia by 2-h glucose, while diagnosing an additional 96 women. This seemingly more liberal diagnosis of dysglycemia is consistent with previous studies showing that 1-h glucose exceeds 8.6 mmol/L before the 2-h measurement meets its dysglycemic threshold (≥7.8 mmol/L) (1–3,6,10). Moreover, our longitudinal analyses illustrate how the clinical significance of this capacity manifests in the years thereafter. Specifically, on direct comparison adjusted for clinical risk factors, the 1-h glucose on OGTT at 3 months postpartum is a more robust predictor of dysglycemia over 5 years both in women reflecting the full spectrum of glycemia in pregnancy and in those with the greatest severity thereof (women with GDM).

The superior performance of 1-h glucose over 2-h measurement is fully consistent with other studies outside of the postpartum context (1–7). A key consideration relevant to the current setting is that women with hyperglycemia in pregnancy are typically early in the natural history of dysglycemia wherein most will initially show normal glucose tolerance after pregnancy before progressing to prediabetes and T2D over the years thereafter. Importantly, in the years over which this progression unfolds, postchallenge glucose rises prior to fasting glucose (11,24,37). It is partly for this reason that clinical practice guidelines recommend the 2-h OGTT over measurement of either A1C or fasting glucose for postpartum reclassification following GDM (11,14). Indeed, outside of pregnancy, the most common scenario in which an OGTT is indicated in practice is postpartum reclassification. Unfortunately, its low rate of performance in this setting has remained an enduring problem in practice, driven by several factors, including the logistical challenge and inconvenience that a 2-h test poses for busy mothers caring for an infant (20–22). In this context, the comparative performance of 1-h glucose versus 2-h measurement demonstrated herein supports consideration of a 1-h OGTT as a more convenient option that potentially may increase adherence to the testing recommended for postpartum reclassification following hyperglycemia in pregnancy.

A strength of this study is the systematic evaluation of a cohort of women reflecting the full spectrum of glycemia in pregnancy with multisample OGTTs at 3 months, 1 year, 3 years, and 5 years postpartum. The serial assessment of this population made it possible to compare the clinical implications of 1-h and 2-h glucose measurements over a critical window after pregnancy during which many of these women will develop dysglycemia. Although GDM was diagnosed by National Diabetes Data Group criteria rather than International Association of Diabetes in Pregnancy Study Groups (IADPSG) criteria, the anticipated applicability of the observed findings to those meeting IADPSG thresholds is supported by the fact that the recruitment of the cohort was specifically designed to reflect the full range of glycemia in pregnancy, as previously described (23,24). An additional limitation is the absence of data on lipids and metabolic syndrome and data evaluating 1-h glucose in different ethnic groups. Another limitation is that the observational design of the study precludes definitive conclusion that implementation of a 1-h OGTT within the first 6 months after delivery would necessarily increase rates of testing. Rather, the data reported herein strongly support the need for a clinical trial to test this assertion, wherein women with hyperglycemia in pregnancy would be randomly assigned to either 1-h OGTT or the standard 2-h OGTT for postpartum reclassification. Such a trial could provide evidence for a solution to address the current missed opportunity resulting from low rates of the recommended postpartum testing in practice.

Of note, the impact of this missed opportunity may be amplified in the coming years. Specifically, the opportunity cost associated with this problem is likely to increase because of the confluence of two factors. First, the prevalence of hyperglycemia in pregnancy has been rising, with the IDF estimating that one in six pregnancies globally are now affected (38). Second, we are currently entering a new era in medicine in which pharmacotherapeutic intervention can induce weight loss far greater than that which could ever be achieved in the past (39,40). As weight loss is a primary focus of management for reducing the likelihood of progression to prediabetes/T2D in at-risk individuals, the cost of continued low rates of postpartum testing will include the missed opportunity to intervene early in the growing number of women with hyperglycemia in pregnancy. These considerations thus further underscore the need for new strategies, such as the 1-h OGTT, that may increase adherence with postpartum testing in practice. At the same time, we recognize that the 1-h OGTT is not going to fully solve this problem because it does require a 1-h test after overnight fasting that may still be problematic for some, particularly in the primary care setting. Moreover, recognizing that the 2-h duration of the OGTT is not the sole factor contributing to poor adherence, it is necessary to continue evaluating other strategies such as dedicated coordinators for arranging testing and early postpartum assessment during the hospital admission for delivery (20,21).

In conclusion, a higher 1-h blood glucose measurement on OGTT at 3 months postpartum is associated with poorer β-cell function and insulin sensitivity, providing a pathophysiologic basis for its capacity to predict risk of dysglycemia. Indeed, this 1-h glucose measurement identifies dysglycemia in almost all of those so diagnosed by the 2-h glucose value, while expanding the number of women identified as being at risk. The significance of its capacity to identify risk at 3 months postpartum is highlighted by the demonstration that 1-h glucose is a more robust independent predictor of subsequent dysglycemia over the first 5 years after pregnancy than is the 2-h glucose value, both in women comprising the full spectrum of gestational glycemia and in those with recent GDM. Thus, the findings reported herein support consideration of a 1-h OGTT as a potential solution to increase the uptake of postpartum reclassification following hyperglycemia in pregnancy and thereby address a missed opportunity in current clinical practice.

Funding. This study was supported by operating grants from the Canadian Institutes of Health Research (CIHR) (PJT-156286, PJ9-192251 and PJT-195810). R.R. holds the Boehringer Ingelheim Chair in Beta-cell Preservation, Function and Regeneration at Mount Sinai Hospital, and his research program is supported by the Sun Life Financial Program to Prevent Diabetes in Women.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.R., C.K.K., A.J.H., P.W.C., M.S., and B.Z. designed and implemented the study. C.Y. performed the statistical analyses. R.R. wrote the manuscript. R.R. and C.Y. verified the data. All authors critically revised the manuscript for important intellectual content. All authors approved the final manuscript. R.R. is guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were John B. Buse and Adrian Vella.