Enhanced Predictive Capability of a 1-Hour Oral Glucose Tolerance Test: A Prospective Population-Based Cohort Study

Type 2 diabetes is associated with significant morbidity and mortality and represents a major burden on health care systems worldwide (1,2). Several randomized clinical trials provide evidence that type 2 diabetes can be prevented or at least postponed with lifestyle modification and drug therapy, which makes identifying high-risk individuals particularly important (3–6). Traditionally, prediabetes has been defined as impaired fasting glucose (IFG) or impaired glucose tolerance (IGT) during a 2-h oral glucose tolerance test (OGTT), and interventional studies thus far have predominantly included subjects with IGT (7,8). However, not all subjects with prediabetes develop type 2 diabetes, and conversely, a significant number without prediabetes progress to type 2 diabetes (9). Accumulating longitudinal evidence, pioneered in particular by Abdul-Ghani and colleagues (10–18), suggests that the 1-h postload glucose level during OGTT with a cutoff ≥8.6 mmol/L (155 mg/dL) may be an early marker of IGT and subsequent type 2 diabetes that is potentially more useful than either fasting or 2-h glucose levels. Furthermore, from a pathophysiological perspective, the 1-h glucose level is intriguing owing to its seemingly strong correlation with markers of both insulin secretion and sensitivity (10,19,20). However, practical clinical implications of using a 1-h glucose measurement for prediction of type 2 diabetes and its associated complications are less clear. Therefore, the aim of this study was to examine whether the 1-h blood glucose measurement would be a more suitable screening tool for risk assessment than the 2-h blood glucose alone.

All subjects were participants of the Malmö Preventive Project (1974–1992), a population-based case-finding program with the objective of identifying high-risk adults suitable for preventive measures. Inhabitants of Malmö, Sweden, belonging to prespecified birth cohorts (1921–1949) were invited for an examination of cardiovascular risk factors, alcohol abuse, and breast cancer. Progressively older men were recruited later during the course of the study. The participation rate was 71% (21–23). During the initial phase of the study (9 September 1974 to 31 May 1978 [both inclusive]), 7,200 men were consecutively included, of whom 5,364 underwent baseline OGTT. After exclusion of subjects with a missing or invalid 1-h glucose measurement (7 of 5,364) or known diabetes (100 of 5,364), and those who emigrated (71 of 5,364), 5,182 individuals were left. Of these, a subset of individuals at high risk—including those with hypertension, hyperlipidemia, diabetes, and IGT, with the latter defined per local standards as a 2-h blood glucose level ≥7.0 mmol/L (126 mg/dL) confirmed on a separate day (315 of 5,182)—underwent intervention. The intervention for IGT comprised of dietary advice and increased physical activity, most often including frequent visits at an outpatient clinic for up to 12 years after inclusion. We also excluded these subjects; thus, the final study population consisted of 4,867 men (Supplementary Fig. 1). In addition, 132 women recruited between 25 January 1977 and 6 January 1984 underwent baseline OGTT. One was excluded owing to known diabetes and 3 owing to emigration, but none of them underwent intervention, leaving a total of 128 female subjects. The Malmö Preventive Project was approved by the ethics committee of Lund University and conducted in accordance with the Declaration of Helsinki. All participants gave informed consent.

Participants used a self-administered questionnaire to provide information on lifestyle and medical history, including a history of diabetes in first-degree relatives, cardiovascular disease, and current medication. Prevalent diabetes was defined as self-reported diabetes or according to the 1985 World Health Organization criteria (24). Blood glucose and serum lipids were obtained after an overnight (≥10 h) fast. Blood glucose was analyzed using the glucose oxidase (1974–1977) or the hexokinase (1977–1992) method. Serum lipids were analyzed using the local laboratory’s standard methods. A 2-h OGTT was performed by ingestion of 30 g glucose/m² body surface area (DuBois formula), with glucose levels determined at 0, 20, 40, 60, 90, and 120 min (25). Analyses of postload glucose levels were focused on measurements at 1 h and 2 h.

We used national and local registries to record clinical end points. Besides the diagnosis itself, the date of diagnosis is also coded in these registries. All events (type 2 diabetes, myocardial infarction, diabetic retinopathy, and diabetes with peripheral vascular complications, including ulcer) were defined according to the relevant ICD-7 to ICD-10 codes (Supplementary Table 1). Reported validities in the Swedish National Inpatient Register were high for all diagnoses (26). Mortality follow-up was based on the national registry on causes of mortality at the Swedish Central Bureau of Statistics. As follow-up was limited to 31 December 2013, the potential minimum and maximum follow-up times were 35 and 39 years, respectively.

Continuous variables are presented as mean ± SD (normally distributed variables) or median (interquartile range [IQR]) (non–normally distributed variables). Categorical variables are presented as counts and corresponding percentages. Kaplan-Meier analysis with the log-rank test and Cox proportional hazards regression with Harrell concordance index (C index), assuming an uncensored policy for handling ties, were used for assessment of discriminative ability for postload glucose measurements, including comparisons between predefined risk groups (27). Hazard ratios (HRs) were reported unadjusted and adjusted for age, BMI, IFG, triglycerides, and family history of diabetes. Furthermore, the ability of 1-h and 2-h postload glucose measurements to enhance prognostication in addition to fasting blood glucose was tested with categorical net reclassification improvement (28). All glucose measurements were assessed in a binary fashion, using the following cut points: IFG, fasting blood glucose ≥5.6 mmol/L (100 mg/dL); elevated 1-h blood glucose, ≥8.6 mmol/L (155 mg/dL); and IGT, 2-h BG ≥7.8 mmol/L (140 mg/dL) (7,10). Analyses were performed at 12 years and at maximal available follow-up, respectively. This was because of the possibility of the intervention among high-risk individuals during the first 12 years after recruitment affecting the study outcomes because even though we attempted to minimize the implications of the intervention by excluding individuals who were subjected to it, there may have been a spillover effect to the control subjects, given the population-based nature of the study. Results obtained from the women were reported separately. Therefore, unless explicitly stated otherwise, all analyses and conclusions were based on the larger male population. The significance level was 5% (two sided), and no adjustments for multiple comparisons were made, as the study was considered exploratory. All analyses were performed with IBM SPSS Statistics 23 (IBM, Armonk, NY) and Stata/IC 15 (StataCorp, College Station, TX).

Table 1 shows baseline anthropometric, clinical, and laboratory characteristics of the study participants. All 4,867 subjects were male, with a median age of 48 years (IQR 48–49; range 27–52), mean BMI 24.8 ± 3.1 kg/m², mean body surface area 1.93 ± 0.15 m², and mean total cholesterol level 5.8 ± 1.0 mmol/L (225 ± 40 mg/dL). A history of a sedentary lifestyle (leisure time mostly spent on sedentary activities, including reading, television, and cinema) was reported in 2,757 (57%), and 655 (13%) had a first-degree relative with diabetes. Normal postload blood glucose levels (NGT/1h-normal) were found in 3,139 (65%), elevated glucose at 1 h only (NGT/1h-high) was found in 1,564 (32%), and IGT only (IGT/1h-normal) was found in in 32 (1%), and 132 (3%) had both elevated 1-h glucose and IGT (IGT/1h-high). Collectively, 1,696 (35%) individuals had elevated glucose at 1 h, and 164 (3%) had IGT. Glucose levels at 1 h and 2 h were moderately correlated (Pearson r = 0.332, P < 0.001) (Supplementary Fig. 2).

Median follow-up time was 33 years (IQR 24–37). At 12 and 39 years, 65 (1%) and 636 (13%) individuals had been diagnosed with type 2 diabetes, corresponding to incidence densities of 1.2 and 4.8 per 1,000 person-years, respectively. Cumulative incidences were lowest in subjects with NGT/1h-normal, highest in subjects with IGT/1h-high, and greater in subjects with NGT/1h-high than in those with IGT/1h-normal (Table 2 and Supplementary Table 2).

The Kaplan-Meier plot shows the unadjusted type 2 diabetes–free survival for risk categories based on postload blood glucose measurements (Fig. 1A and Supplementary Figs. 3 and 4). Congruent with the above, both crude and adjusted HRs for development of type 2 diabetes were significantly greater in risk categories including subjects with elevated 1-h blood glucose levels (Table 3 and Supplementary Table 3).

The presence of an elevated 1-h blood glucose level was associated with significantly greater discriminative ability than IGT based on 2-h blood glucose at both 12 years (C index 0.698 vs. 0.553, P < 0.001) and 39 years (C index 0.637 vs. 0.511, P < 0.001). Additionally, the presence of IFG or elevated 1-h blood glucose was associated with a greater risk of type 2 diabetes than IFG or IGT at 12 years (HR 7.22 [95% CI 3.93–13.26], P < 0.001, C index 0.720, vs. HR 4.64 [95% CI 2.69–7.99], P < 0.001, C index 0.600; P for difference <0.001) and 39 years (HR 3.31 [95% CI 2.82–3.88], P < 0.001, C index 0.636, vs. HR 2.28 [95% CI 1.83–2.84], P < 0.001, C index 0.537; P for difference < 0.001).

The abilities of elevated 1-h and 2-h blood glucose levels to reclassify risk in addition to fasting blood glucose measurements were tested in separate models. Two risk categories were applied, and reclassification could only be present if postload glucose levels were elevated in subjects with normal fasting glucose. As such, the addition of 1-h glucose was associated with greater net reclassification improvement than the addition of 2-h glucose at both 12 years (0.308 vs. 0.066) and 39 years (0.214 vs. 0.016) (Supplementary Table 4).

Restricting analyses to subjects in whom contemporary guidelines suggest screening for diabetes (8), i.e., subjects aged ≥45 years or subjects with a BMI ≥25 kg/m² and at least one additional risk factor (> 99% of our study population met the criteria for screening) did not alter the results.

Similar overall patterns were observed when we used all-cause mortality as the outcome, i.e., subjects with elevated 1-h blood glucose levels had a significantly higher mortality risk than those with NGT/1h-normal, whereas those with IGT/1h-normal did not (Table 3, Supplementary Table 5, and Fig. 1B). The risk of myocardial infarction or fatal ischemic heart disease was also greater among subjects with elevated 1-h glucose levels versus NGT/1h-normal. Finally, elevated 1-h glucose levels were associated with greater risks of retinopathy and peripheral vascular complications. No differences were detected between IGT/1h-normal and NGT/1h-normal (Supplementary Tables 6 and 7).

To highlight the importance of 1-h blood glucose measurements, we performed supplemental analyses comparing glucose levels obtained at 20 and 40 min with those obtained at fasting, 1 h, and 2 h in predicting incident type 2 diabetes and all-cause mortality (Supplementary Tables 8 and 9). Standardized HRs for glucose measurements obtained at time points earlier than 2 h were generally greater among the younger and thinner halves of the study population, but only the interaction terms between BMI and 40-min and 1-h glucose, and between age and 40-min glucose, were statistically significant (Supplementary Table 10).

Median age at baseline was 39 years (IQR 38–40; range 29–56). Median follow-up was 36 years (IQR 35–36). Mean blood glucose concentrations were 4.7 ± 0.5 mmol/L (84 ± 8 mg/dL) at fasting, 7.5 ± 1.9 mmol/L (135 ± 34 mg/dL) at 1 h, and 5.8 ± 1.2 mmol/L (104 ± 22 mg/dL) at 2 h. Four (3%) had an elevated 2-h glucose concentration, and 41 (32%) had an elevated level at 1 h. Accordingly, 86 (67%) were categorized as NGT/1h-normal, 38 (30%) were categorized as NGT/1h-high, 1 (1%) was categorized as IGT/1h-normal, and 3 (2%) were categorized as IGT/1h-high. None had developed type 2 diabetes at 12 years. At complete follow-up, 13 (10%) had been diagnosed with type 2 diabetes, corresponding to an incidence density of 3.2/1,000 person-years. Cumulative incidences (and incidence densities per 1,000 person-years) in the four categories were 8% (2.5), 13% (4.2), 0% (0), and 33% (13.6), respectively. Compared with the NGT/1h-normal group, HRs were 1.72 (95 CI 0.55–5.43), P = 0.35, for the NGT/1h-high group and 6.68 (95% CI 0.82–54.52), P = 0.08, for the IGT/1h-high group.

Our results demonstrate that in middle-aged men, for whom screening for type 2 diabetes would be recommended, the 1-h blood glucose level is a significant predictor of future type 2 diabetes, regardless of the 2-h blood glucose level. In addition, the 1-h blood glucose level has greater detection rates (higher sensitivity) than the 2-h blood glucose level and can correctly reclassify subjects with traditionally defined prediabetes, especially with longer duration of follow-up (high specificity). Importantly, the 1-h blood glucose level is also associated with diabetes complications and mortality.

Prior studies have focused mainly on the ability of 1-h glucose measurements to correctly subdivide subjects with NGT into those with low(er) and high(er) risk of future type 2 diabetes and focused less on the direct comparison between 1-h and 2-h OGTT from a clinical standpoint (10,12–15). Previous results from the Malmö Preventive Project have further suggested that the 1-h blood glucose level also adds significant predictive value to clinical risk factors (16,17). However, such strategies do not allow for the omission of the 2-h OGTT or preclude the use of complex multivariable clinical prediction models, for which clear cutoff values and benefits upon intervention have not been established. Our study therefore extends previous findings by using a simple and clinically applicable approach, directly comparing risks in groups stratified according to postload blood glucose levels, in a larger population sample with longer follow-up duration. An elevated 1-h blood glucose level was associated with a significantly increased risk of incident type 2 diabetes at both short-term and long-term follow-up and further provided significant net reclassification improvement in subjects categorized per fasting blood glucose. Consistent with prior findings, subjects in the IGT/1h-normal group constituted a minority (18), and, importantly, few individuals categorized as such progressed to diabetes, without any significant difference from persons in the NGT/1h-normal group. Conversely, the IGT/1h-high subgroup contained individuals at very high risk of future type 2 diabetes. Therefore, virtually all individuals with true IGT at 2 h who progressed to manifest type 2 diabetes were captured by an elevated 1-h glucose level.

The ability of 1-h blood glucose levels to predict diabetes complications and mortality is less well studied. Elevated glucose levels at 1 h are associated with adverse metabolic and cardiovascular changes, reflected by body composition, cholesterol levels, and subclinical target organ damage, including arterial stiffness, carotid intima-media thickness, and left ventricular hypertrophy (19,20,29–31). However, very few reports of the clinical consequences have been published (32–35), and although associations with macrovascular events and mortality have been shown, only one study of 1,945 subjects included glucose levels at 2 h and indicated that 1-h glucose levels predicted all-cause mortality among subjects with NGT (35). Our study demonstrates for the first time an association between elevated 1-h blood glucose levels and adverse cardiovascular prognosis, including microvascular complications, in subjects with NGT, while at the same time showing the lack of such an association in subjects with IGT only. Furthermore, the 1-h blood glucose level predicted mortality, whereas isolated IGT did not.

Although the 2-h OGTT has received less emphasis by the American Diabetes Association (8), the majority of evidence for intervention among subjects with prediabetes comes from studies of overweight or obese individuals with IGT (3–6). Furthermore, both fasting glucose and HbA_1c have limited sensitivity and specificity for detecting subjects at risk, especially because of high false-negative rates, and the current study provides evidence that the 1-h OGTT is superior to fasting glucose alone for identifying high-risk subjects (7,10,12,13,15,16,36–38). Hence, a more sensitive strategy is needed, which at the same time is simple to use in a primary care setting and less time consuming and more convenient than the 2-h OGTT. The current study extends the results from a previous report from the Malmö Preventive Project and shows that the 1-h time limit provides not only the best compromise in terms of time consumption (performs equivalent to or better than glucose measurement at 90 min) but also prognostic ability (performing better than glucose measurements at 20 and 40 min) (17). The proposed cut point for 1-h blood glucose identifies a substantially larger proportion of subjects at high risk compared with conventional IGT (10,12,13), and its use could lead to more widespread preventive efforts. However, since subjects for whom screening for type 2 diabetes is recommended already have an adverse risk profile (7,8), using a 1-h OGTT should lead to a reduced burden of both diabetes and its complications, without an excess risk of harm.

The pathophysiological significance of elevated 1-h blood glucose levels is not fully understood. Proposedly, these individuals may be at an intermediate stage between NGT and IGT or represent a phenotype distinct from that of individuals with IGT (10,19,20). In subjects with NGT, an elevated 1-h glucose level is associated with insulin resistance to a degree similar to that seen among individuals with IGT. Insulin secretion is also affected, albeit to a lesser extent (19,20). Furthermore, the risk of future type 2 diabetes associated with elevated 1-h blood glucose levels was temporally persistent but greater in the IGT/1h-high group than in the NGT/1h-normal group, particularly at shorter follow-up. These findings, in addition to those stated above, largely support the first theory and further support the concept that worsening β-cell function, not insulin resistance, is the main culprit involved in progressive glucometabolic deterioration (13).

Our results highlight the potential benefit of targeting individuals with an elevated 1-h blood glucose level and provide a strong rationale for an interventional study in which subjects are selected based on this abnormality. Should preventive efforts prove beneficial in this relatively large group, it would become prudent to consider 1-h blood glucose as a replacement for 2-h blood glucose.

A few limitations deserve mention. The definition of type 2 diabetes has changed over the past decades, notably with the lowering of the fasting glucose threshold in 1997 and the introduction of HbA_1c-defined diabetes in 2011 (7). However, despite there being only a partial overlap between subjects with prediabetes and diabetes defined according to these different measures, the predictive capability of clinical risk factors remains comparable (7,39). The 30 g/m² glucose load was the standard procedure at the initiation of the study, and based on estimates of body surface area, this resulted in an average glucose load 23% lower than with the use of 75-g glucose. Prior studies have suggested that a larger glucose load results in greater differences in glucose concentrations at 2 h than at 1 h. In individuals without IGT or diabetes, the glucose levels at 1 h are virtually identical when comparing a 50-g with a 100-g glucose load. Even at 2 h, the difference appears to be small, with highly correlated values (40–43). Discrepancies between glucose levels are larger among subjects with IGT at both 1 h and 2 h (42). Theoretically, the lower glucose dose might have had a slightly greater impact on the 2-h level than the 1-h level, improving the relative sensitivity of the 1-h measurement, but significant differences have not been shown for 75 g versus 100 g, and it is not likely that important differences would exist for the glucose load used in our study compared with the standard 75-g load (43). In contrast with the well-established partitions for fasting glucose and 2-h OGTT, the cut point of 8.6 mmol/L for 1-h OGTT has been less thoroughly investigated and was derived from the San Antonio Heart Study (10). However, use of a cut point derived from our own cohort would have inflated its utility compared with 2-h OGTT. An additional limitation might be attributed to the use of two different approaches for measurement of glucose levels, although the glucose oxidase and hexokinase methods have been shown to deliver well-correlated results (44). The generalizability of our results beyond white men may be limited, but there is no evidence to suggest sex- or race-related differences in the prognostic ability of 1-h versus 2-h OGTT. However, women may have slightly lower mean concentrations of 1-h glucose, with a slower return to baseline levels (45). This fits well with the observation that traditionally defined IGT is more prevalent in females than in males (9). Most women were recruited later during the Malmö Preventive Project, when routine performance of the comprehensive OGTT had been omitted owing to financial restrictions. Thus, very few women underwent such testing and the results obtained from this subgroup should be interpreted very cautiously, given the minuscule number of events. The isolated IGT group was small, and although results were consistent, CIs were wide. Still, this finding by itself is valuable, as it indicates a minimal loss by not capturing these individuals. Finally, although we excluded the individuals who underwent lifestyle intervention to enhance the interpretability of our study, this might have resulted in removal of the subjects who had the highest a priori risk for type 2 diabetes and associated complications, underestimating the true incidence of outcomes, particularly in the subgroups with abnormal baseline glycemic status.

In conclusion, the 1-h blood glucose level is a powerful predictor of future type 2 diabetes, with net reclassification improvement and detection rates significantly greater than 2-h blood glucose measurements, especially with longer duration of follow-up. The 1-h blood glucose level is further associated with diabetes complications and mortality. Finally, simple logistics favor the use of a 1-h versus a 2-h OGTT. Therefore, 1-h blood glucose should be considered as a replacement for 2-h blood glucose as the preferred marker of IGT.

Acknowledgments. The authors sincerely thank data managers Anders Dahlin and Håkan Andersson at the Department of Clinical Sciences, Lund University, Skåne University Hospital, for help with retrieving data.

Funding and Duality of Interest. This study was funded by the Danish Diabetes Academy, supported by the Novo Nordisk Foundation. D.L.B. discloses the following relationships: Advisory board, Cardax, Elsevier Practice Update Cardiology, Medscape Cardiology, and Regado Biosciences; Board of directors, Boston VA Research Institute and Society of Cardiovascular Patient Care; Chair, American Heart Association Quality Oversight Committee; Data-monitoring committees, Cleveland Clinic, Duke Clinical Research Institute, Harvard Clinical Research Institute, Mayo Clinic, Mount Sinai School of Medicine, and Population Health Research Institute; honoraria, American College of Cardiology (Senior Associate Editor, Clinical Trials and News, acc.org), Belvoir Publications (Editor in Chief, Harvard Heart Letter), Duke Clinical Research Institute (clinical trial steering committees), Harvard Clinical Research Institute (clinical trial steering committee), HMP Communications (Editor in Chief, Journal of Invasive Cardiology), Journal of the American College of Cardiology (Guest Editor, Associate Editor), Population Health Research Institute (clinical trial steering committee), Slack Publications (Chief Medical Editor, Cardiology Today’s Intervention), Society of Cardiovascular Patient Care (Secretary/Treasurer), and WebMD (CME steering committees); other, Clinical Cardiology (Deputy Editor), NCDR-ACTION Registry Steering Committee (Chair), and VA CART Research and Publications Committee (Chair); research funding, Amarin, Amgen, AstraZeneca, Bristol-Myers Squibb, Chiesi, Eisai, Ethicon, Forest Laboratories, Ironwood, Ischemix, Lilly, Medtronic, Pfizer, Roche, Sanofi, and The Medicines Company; royalties, Elsevier (Editor, Cardiovascular Intervention: A Companion to Braunwald’s Heart Disease); site co-investigator, Biotronik, Boston Scientific, and St. Jude Medical (now Abbott); trustee, American College of Cardiology; and unfunded research, FlowCo, Merck, PLx Pharma, and Takeda. M.H.O. discloses receipt of a part-time clinical research grant from the Novo Nordisk Foundation. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.P. conceptualized the hypothesis, designed the work, analyzed and interpreted data, drafted the manuscript, and revised the manuscript critically for important intellectual content. D.L.B. and M.H.O. conceptualized the hypothesis, designed the work, interpreted data, and revised the manuscript critically for important intellectual content. M.L.N., R.J., K.-F.E., P.M.N., and M.B. conceptualized the hypothesis, interpreted data, and revised the manuscript critically for important intellectual content. All authors read and approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. M.P. and M.H.O. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 52nd Annual Meeting of the European Association for the Study of Diabetes, Munich, Germany, 12–16 September 2016.