Association of Post‐Oral Glucose Tolerance Test Hypoglycaemia With Clinical Features, Incident Diabetes and Mortality in the Di@bet.es Cohort Authors: Tomás González‐Vidal, Gemma Rojo‐Martínez, Sergio Valdés, Eva García‐Escobar, Josep Franch‐Nadal, Alfonso Calle‐Pascual, Luis Castaño, Felipe Javier Chaves, Eduard Montanya, Elsa Villa‐Fernández, Ana Victoria García, Miguel García‐Villarino, Carmen Lambert, Edelmiro Menéndez‐Torre, Elías Delgado Published: 2026 ABSTRACT ABSTRACTAimsTo investigate the association of post‐oral glucose tolerance test (OGTT) hypoglycaemia with clinical features, incident diabetes and mortality.Materials and MethodsThis population‐based cohort study included 2924 adults (median age 49 years, 1680 women) without known diabetes who underwent a 75‐g OGTT as part of the Di@bet.es Study. Metabolic, demographic and lifestyle variables were collected. Serum glucose and insulin concentrations at 0, 30 and 120 min during the OGTT were obtained, allowing insulin resistance estimation using the HOMA‐IR and the Matsuda Index. Participants were grouped according to serum glucose at 120 min: < 70 mg/dL (post‐OGTT hypoglycaemia), 70 to < 140 mg/dL (normal result), 140 to < 200 mg/dL, and ≥ 200 mg/dL. The incidence of diabetes (median follow‐up, 7.5 years) and mortality (median follow‐up, 14.5 years) were assessed.ResultsParticipants with post‐OGTT hypoglycaemia (n = 327, 11.1%) were younger, leaner, less insulin resistant and more frequently active smokers. Despite having the lowest serum glucose at 0, 30 and 120 min, and the lowest serum insulin at 0 and 120 min, these participants tended to have the highest serum insulin concentrations at 30 min. Participants with post‐OGTT hypoglycaemia had the lowest incidence of diabetes and mortality, although these associations were attenuated after adjusting for confounders (such as age, sex, body mass index and physical activity) and when compared specifically with the normal result group.ConclusionsPost‐OGTT hypoglycaemia is more frequent in smokers, younger individuals, and those with a favourable metabolic profile. Post‐OGTT hypoglycaemia acts as a marker of protection for incident diabetes and mortality. Preserved early insulin secretion together with good insulin sensitivity could explain post‐OGTT hypoglycaemia. Introduction The oral glucose tolerance test (OGTT) is typically performed to detect marked hyperglycaemia following carbohydrate ingestion, given that it is a particularly sensitive test for diabetes screening [1]. In some individuals without diabetes, however, the OGTT can reveal unexpected results, such as low blood glucose concentrations measured 2 h after ingesting 75 g of glucose. These post‐OGTT hypoglycaemic episodes have no diagnostic value on their own and should not prompt suspicion of disorders such as insulinoma [2]. Nevertheless, they could reflect underlying changes in carbohydrate metabolism whose potential implications have not yet been sufficiently explored. Some authors have considered post‐OGTT hypoglycaemia a possible indicator of insulin resistance [3]. This assumption has been supported by studies reporting a high prevalence of post‐OGTT hypoglycaemia in lean women with polycystic ovary syndrome (a condition characterised by insulin resistance) [4] as well as by case reports describing resolution of post‐OGTT hypoglycaemia after treatment with insulin‐sensitising agents such as pioglitazone [5]. According to this hypothesis, insulin resistance might blunt the early insulin response to postprandial glucose elevations, leading to an exaggerated second‐phase secretion that can induce hypoglycaemia once nutrient absorption is complete [3]. Moreover, such excessive late insulin secretion would not only exhaust pancreatic beta cells but would further exacerbate insulin resistance [6]. Consequently, post‐OGTT hypoglycaemia has been regarded as a potential but unconfirmed risk factor for the development of diabetes [3]. In contrast, other studies have shown that individuals with favourable metabolic profiles and higher insulin sensitivity are more likely to experience post‐OGTT [7, 8, 9] or other forms of reactive hypoglycaemia [10]. Based on these findings, the authors have hypothesised that post‐OGTT hypoglycaemia might actually represent a protective factor for the development of diabetes [7]. Although numerous studies have examined the insulin response throughout the OGTT [8, 9, 11, 12, 13, 14, 15, 16, 17, 18, 19], investigations specifically addressing insulin secretion patterns in individuals with post‐OGTT hypoglycaemia remain limited [8, 9, 12, 13, 14, 15] and largely restricted to selected populations [9, 12, 13, 14, 15]. Similarly, few studies have examined the demographic, lifestyle and metabolic factors associated with increased post‐OGTT hypoglycaemia frequency in the general adult population [7, 8, 20], given that most available data come from studies conducted in pregnant women [21, 22, 23, 24]. To our knowledge, no study has yet evaluated the association of post‐OGTT hypoglycaemia with incident diabetes or mortality in the general population. Drawing on a representative sample of the Spanish population, our study investigated the clinical factors and insulin response related to 2‐h post‐OGTT hypoglycaemia, as well as its association with both incident diabetes and long‐term mortality. Materials and Methods Study Population and Setting The Di@bet.es Study was a national, population‐based study conducted in Spain, as previously described [25, 26]. Briefly, in this cross‐sectional [25] and cohort study [26], a cluster sampling design was used to obtain a representative random sample of the Spanish adult population by drawing participants from the national health system registry. The final baseline sample included 5072 non‐pregnant adults (≥ 18 years of age) [25]. For the present study, we selected participants with an available serum glucose determination 120 min after the start of a standard 75‐g OGTT (n = 2940). Of these, seven individuals with previously known diabetes were excluded. In addition, nine participants who did not meet diabetes criteria but were taking glucose‐lowering agents (presumably for indications other than diabetes) were excluded to ensure that the OGTT was not affected by medication use. After applying these inclusion and exclusion criteria, a total of 2924 adults (median age 49 years, age range 18–93 years; 1680 [57.5%] women) without known diabetes or glucose‐lowering medication who completed an OGTT between September 2007 and October 2010 were included in this prospective cohort study (Figure S1). Although none of the participants had a previous diagnosis of diabetes at study inclusion, OGTT results (including both fasting and 2‐h glucose values) identified individuals with normoglycaemia, prediabetes, and previously undiagnosed diabetes within the final cohort (Figure S1) [1, 25]. Main Determinations Oral Glucose Tolerance Test All participants underwent a standard 2‐h, 75‐g OGTT. The test was performed following the recommendations of the World Health Organisation [1]. The OGTTs were carried out in the morning after a minimum 8‐h fasting period, with no major changes in participants' diet or exercise routines in the preceding 3 days. Participants ingested within 5 min 75 g of anhydrous glucose dissolved in approximately 250 mL of water, and the following serum measurements were obtained: glucose at 0 min (n = 2923), insulin at 0 min (n = 2895); glucose at 30 min (n = 441), insulin at 30 min (n = 445); glucose at 120 min (n = 2924), insulin at 120 min (n = 2907). Participants were categorised into four groups based on serum glucose concentrations at 120 min: < 70 mg/dL (post‐OGTT hypoglycaemia), 70 to < 140 mg/dL (normal result), 140 to < 200 mg/dL (prediabetes), and ≥ 200 mg/dL (suggestive of diabetes) [1]. To assess the internal composition of the post‐OGTT hypoglycaemia group, this category was subdivided into participants with 2‐h serum glucose < 55 mg/dL and 2‐h serum glucose 55 to < 70 mg/dL for additional subgroup analyses. Fasting insulin resistance was estimated using the Homeostatic Model Assessment of Insulin Resistance (HOMA‐IR), calculated from fasting (0 min) serum glucose and insulin concentrations [27] as follows: HOMA−IR=Fasting insulinpmolL×Fasting glucosemgdL2430 Dynamic insulin sensitivity during the OGTT was assessed using the Matsuda Index, calculated from glucose and insulin concentrations at 0, 30 and 120 min (all contributing to the calculation of mean OGTT glucose and insulin; see below) [8, 12, 28]. To avoid temporal discordance between glucose and insulin measurements, the Matsuda index was calculated only in participants with at least 1 paired glucose‐insulin measurement at either 30 or 120 min [28]. Matsuda index=10000Fasting glucosemgdLxFasting insulinpmol/L6×Mean OGTTglucosemgdL×Mean OGTT insulinpmol/L6 Incidence of Diabetes A subset of participants (n = 2036) was reevaluated between January 2016 and February 2018 to determine whether they met the American Diabetes Association diagnostic criteria for diabetes [1]. Details on follow‐up availability for glycaemic status in the Di@bet.es cohort have been described elsewhere [26]. There were no clinically meaningful differences between participants in the Di@bet.es Study whose glycaemic status was reassessed and those who were not [26]. Likewise, data on the incidence of diabetes and some of its associated risk factors in the full Di@bet.es cohort have been previously reported [26]. Long‐Term Mortality Using death certificate data from the National Spanish Statistics Institute (Instituto Nacional de Estadística), survival status and causes of mortality were ascertained for 2918 participants as of December 31, 2023. All‐cause and cardiovascular mortality were evaluated, given the well‐known link between disturbances in glucose metabolism and these outcomes [29, 30]. Covariates Additional covariates were collected at the baseline evaluation (2007–2010): demographic factors (age and sex), anthropometric measures (body mass index [BMI] and waist and hip circumferences), family history of diabetes (defined as diabetes in parents, siblings, children, grandparents, or aunts/uncles), and lifestyle factors (self‐reported smoking status, alcohol consumption and physical activity). Physical activity was assessed using the short form of the International Physical Activity Questionnaire, which records the frequency and duration of vigorous, moderate and walking activities performed during the previous week [31]. This questionnaire enables the calculation of metabolic equivalent of task [32] and the classification of physical activity as low, moderate, or high [31, 33]. Ethical Issues The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the Hospital Regional Universitario de Málaga (codes CEI20070612, date 12 June 2007; CEI20110324, 24 March 2011; CEI20161026, 26 October 2016; CEI21022019, 21 February 2019), in addition to other regional ethics committees throughout Spain. All the participants provided written informed consent. Statistical Analyses We employed the chi‐square test to compare proportions (with Yates' continuity correction, when appropriate), the Mann–Whitney test to compare numerical data between two independent groups, the Kruskal‐Wallis test to compare numerical data across more than two independent groups, and Spearman's rank test to evaluate correlation. Logistic regression models were used for multivariate analyses of factors associated with post‐OGTT hypoglycaemia, as well as with incident diabetes at a given reassessment date (2016–2018). Analyses regarding incident diabetes were conducted exclusively in participants with baseline fasting serum glucose < 126 mg/dL and 2‐h post‐OGTT serum glucose < 200 mg/dL, it being assumed that the rest of the participants had newly diagnosed diabetes at baseline [1]. We employed Kaplan–Meier curves for the assessment of long‐term all‐cause mortality rates according to the four categories of serum glucose at 120 min after the start of the OGTT, and the log‐rank test for between‐group comparisons. Cox regression (proportional hazards regression) was used for the multivariate analysis of all‐cause mortality during follow‐up, after testing the proportional hazards assumption using Schoenfeld residuals. Cardiovascular mortality was analysed using competing risks methods, considering cardiovascular death as the event of interest and deaths from non‐cardiovascular or unknown causes as competing events (deaths from unknown cause, n = 23). Cumulative incidence functions were estimated and compared using the Fine and Gray method and Gray's test, respectively, and subdistribution hazard models were fitted using the Fine–Gray regression approach [34, 35]. Multivariate models exploring factors associated with post‐OGTT hypoglycaemia included baseline variables that were significantly associated with post‐OGTT hypoglycaemia (specifically compared with participants with normal OGTT results [70 to < 140 mg/dL]) in univariate analyses. Multivariate models assessing the association between post‐OGTT hypoglycaemia and incident diabetes and mortality included variables that are well‐known to be associated with such outcomes. Daily alcohol consumption was not included in these models because this information was missing for a substantial proportion of participants (n = 629). All multivariate analyses were adjusted for age and sex. BMI was selected as the main anthropometric variable, while other related measures (such as abdominal circumferences or insulin resistance indexes) were not included in multivariate analyses due to collinearity with BMI. All tests were two‐tailed. The sequential Bonferroni correction for multiple comparisons was applied, when needed [36]. p‐values lower than 0.05 were considered statistically significant. Data were primarily analysed using IBM SPSS Statistics version 20 (IBM Corp., USA), which was also used to construct the figures, with competing risk analyses conducted in R using the cmprsk package. Results Frequency and Determinants of Post‐Oral Glucose Tolerance Test Hypoglycaemia At 120 min of the OGTT, 327 (11.1%) participants exhibited post‐OGTT hypoglycaemia (serum glucose < 70 mg/dL). Among the remaining participants, 2088 (71.4%) had a normal OGTT result (serum glucose 70 to < 140 mg/dL), 371 (12.6%) had values consistent with prediabetes (serum glucose 140 to < 200 mg/dL), and 138 (4.7%) had values suggestive of probable diabetes (serum glucose ≥ 200 mg/dL; Table 1) [1]. TABLE 1Baseline characteristics and outcomes in relation to the categorised results of the 75‐g oral glucose tolerance test.VariableSerum glucose at 120 min during a 75‐g oral glucose tolerance test p < 70 mg/dL (n = 327)70 to < 140 mg/dL (n = 2088)140 to < 200 mg/dL (n = 371)≥ 200 mg/dL (n = 138) Baseline participants' characteristics Sex (man)140 (42.8)838 (40.1)188 (50.7)78 (56.5)< 0.001Age (years)39 (31–50)47 (35–60)* 61 (48–71)66 (54–75)< 0.001Body mass index (kg/m2) a 25.0 (22.9–28.0)27.1 (24.2–30.3)* 30.0 (26.9–33.3)30.5 (28.1–32.8)< 0.001Waist circumference (cm) b 86 (76–95)93 (83–101)* 101 (93–109)104 (97–111)< 0.001Hip circumference (cm) c 101 (96–106)104 (98–109)* 107 (101–113)106 (101–111)< 0.001Waist‐to‐hip ratio c 0.84 (0.78–0.90)0.88 (0.82–0.94)* 0.94 (0.89–0.99)0.97 (0.91–1.03)< 0.001Current smoker (yes) d 132 (40.4)527 (25.3)* 62 (16.9)25 (18.1)< 0.001Daily alcohol drinker (yes) e 65 (24.5)442 (27.1)96 (34.3)38 (32.8)0.028Physical activity f Low120 (36.8)809 (38.8)174 (47.3)58 (42.0)0.030Moderate120 (36.8)732 (35.1)124 (33.7)49 (35.5)High86 (26.4)544 (26.1)70 (19.0)31 (22.5)Family history of diabetes (yes) g 164 (52.9)1014 (50.1)175 (50.4)59 (45.7)0.583 Baseline laboratory results Fasting serum glucose (mg/dL) h 88 (81–93)92 (85–99)* 100 (91–109)116 (104–132)< 0.001Fasting hypoglycaemia (< 70 mg/dL) (yes) h 14 (4.3)70 (3.4)4 (1.1)0 (0.0)0.008Impaired fasting glucose (100 to < 126 mg/dL) (yes) i 32 (9.8)490 (23.7)* 175 (49.6)64 (70.3)< 0.001High fasting glucose (≥ 126 mg/dL) (yes) h 1 (0.3)18 (0.9)18 (4.9)47 (34.1)< 0.001Fasting serum insulin (pmol/L) j 44 (32–58)53 (38–73)* 70 (49–99)75 (53–111)< 0.001Serum glucose at 30 min (mg/dL) k 127 (111–154)140 (119–160)172 (158–203)201 (185–256)< 0.001Serum insulin at 30 min (pmol/L) l 383 (289–473)357 (248–562)346 (231–557)267 (178–348)0.036Serum glucose at 120 min (mg/dL)60 (53–66)99 (86–114)* 157 (146–174)241 (215–298)< 0.001Serum insulin at 120 min (pmol/L) m 98 (73–103)240 (153–382)* 608 (379–978)552 (368–992)< 0.001HOMA‐IR j 1.5 (1.1–2.1)1.9 (1.4–2.8)* 2.8 (2.0–4.1)3.6 (2.4–5.5)< 0.001Matsuda index n 11.7 (8.5–17.3)6.9 (4.7–9.9)* 3.3 (2.2–4.7)2.5 (1.6–3.7)< 0.001 Long term follow‐up Incident diabetes mellitus (yes) o 4 (1.7)47 (3.2)52 (22.2)NA< 0.001Fasting serum glucose at reassessment (mg/dL) p 91 (85–98)94 (88–102)* 106 (95–117)118 (108–137)< 0.001Fasting hypoglycaemia (< 70 mg/dL) at reassessment (yes) p 1 (0.7)13 (1.5)0 (0.0)0 (0.0)0.336Serum glucose at 120 min at reassessment (mg/dL) q 87 (73–103)105 (89–127)* 163 (128–209)175 (146–233)< 0.001Hypoglycaemia (< 70 mg/dL) at 120 min at reassessment (yes) q 20 (16.7)30 (4.3)* 1 (1.0)0 (0.0)< 0.001All‐cause mortality (yes) r 15 (4.6)187 (9.0)* 89 (24.1)46 (33.6)< 0.001Cardiovascular mortality (yes) r 3 (0.9)49 (2.4)20 (5.4)16 (11.7)< 0.001 Note: Categorical variables are presented as absolute numbers and percentages (in parentheses), and continuous variables are presented as medians and interquartile ranges (in parentheses). p‐values were obtained using the chi‐square test for categorical variables and the Kruskal‐Wallis test for continuous variables. p < 0.0083 were considered statistically significant after sequential Bonferroni correction.Abbreviations: HOMA‐IR, homeostatic model assessment for insulin resistance; NA, not applicable. a Data available for 2908 participants. b Data available for 2915 participants. c Data available for 2912 participants. Waist‐to‐hip ratio = Waist circumference/Hip circumference. d Data available for 2919 participants. e Data available for 2295 participants. f Data available for 2917 participants. g Data available for 2810 participants. h Data available for 2923 participants. i Data available for 2839 participants, given that those with fasting serum glucose ≥ 126 mg/dL were excluded from this analysis. j Data available for 2895 participants. k Data available for 441 participants (45 participants in the < 70 mg/dL group, 323 participants in the 70 to < 140 mg/dL group, 54 participants in the 140 to < 200 mg/dL group, and 19 participants in the ≥ 200 mg/dL group). l Data available for 445 participants (45 participants in the < 70 mg/dL group, 326 participants in the 70 to < 140 mg/dL group, 55 participants in the 140 to < 200 mg/dL group, and 19 participants in the ≥ 200 mg/dL group). m Data available for 2907 participants. n Data available for 2879 participants. o Data available for 1922 participants. p Data available for 1221 participants. q Data available for 944 participants. r Data available for 2918 participants.*Asterisks denote statistically significant differences (p < 0.05 after sequential Bonferroni correction) between the < 70 mg/dL group and the 70 to < 140 mg/dL group (chi‐square test and Mann–Whitney test). Baseline characteristics and outcomes in relation to the categorised results of the 75‐g oral glucose tolerance test. Note: Categorical variables are presented as absolute numbers and percentages (in parentheses), and continuous variables are presented as medians and interquartile ranges (in parentheses). p‐values were obtained using the chi‐square test for categorical variables and the Kruskal‐Wallis test for continuous variables. p < 0.0083 were considered statistically significant after sequential Bonferroni correction. Abbreviations: HOMA‐IR, homeostatic model assessment for insulin resistance; NA, not applicable. Data available for 2908 participants. Data available for 2915 participants. Data available for 2912 participants. Waist‐to‐hip ratio = Waist circumference/Hip circumference. Data available for 2919 participants. Data available for 2295 participants. Data available for 2917 participants. Data available for 2810 participants. Data available for 2923 participants. Data available for 2839 participants, given that those with fasting serum glucose ≥ 126 mg/dL were excluded from this analysis. Data available for 2895 participants. Data available for 441 participants (45 participants in the < 70 mg/dL group, 323 participants in the 70 to < 140 mg/dL group, 54 participants in the 140 to < 200 mg/dL group, and 19 participants in the ≥ 200 mg/dL group). Data available for 445 participants (45 participants in the < 70 mg/dL group, 326 participants in the 70 to < 140 mg/dL group, 55 participants in the 140 to < 200 mg/dL group, and 19 participants in the ≥ 200 mg/dL group). Data available for 2907 participants. Data available for 2879 participants. Data available for 1922 participants. Data available for 1221 participants. Data available for 944 participants. Data available for 2918 participants. Asterisks denote statistically significant differences (p < 0.05 after sequential Bonferroni correction) between the < 70 mg/dL group and the 70 to < 140 mg/dL group (chi‐square test and Mann–Whitney test). Comparing clinical characteristics among the 4 groups, participants with post‐OGTT hypoglycaemia were the youngest and had the lowest BMI, waist and hip circumferences, waist‐to‐hip ratio, and insulin resistance (as estimated by both the HOMA‐IR and Matsuda Index; Table 1). They also had the highest frequency of current smokers and tended to have the lowest frequency of daily alcohol consumption and the highest physical activity levels. The associations between post‐OGTT hypoglycaemia and younger age, lower BMI, smaller waist and hip circumferences, lower waist‐to‐hip ratio, lower insulin resistance, and a higher frequency of current smokers remained statistically significant specifically compared with the group with normal post‐OGTT glucose levels (70 to < 140 mg/dL; Table 1). In a multivariate analysis (logistic regression), younger age, lower BMI, and current smoking status remained independently associated with increased odds of post‐OGTT hypoglycaemia specifically compared with the group exhibiting normal post‐OGTT glucose levels (Table 2). TABLE 2Multivariate analysis (logistic regression) of factors associated with hypoglycaemia (< 70 mg/dL) at 120 min during the 75‐g oral glucose tolerance test.Crude analysesAdjusted for age and sexMultiadjusted* OR (95% CI) p OR (95% CI) p OR (95% CI) p Sex (man)1.11 (0.88–1.41)0.3591.07 (0.84–1.36)0.5641.16 (0.91–1.48)0.221Age (years)0.97 (0.96–0.98)< 0.0010.97 (0.96–0.98)< 0.0010.98 (0.97–0.99)< 0.001Body mass index (kg/m2) a 0.90 (0.88–0.93)< 0.0010.92 (0.90–0.95)< 0.0010.92 (0.90–0.95)< 0.001Current smoker (yes) b 2.00 (1.57–2.55)< 0.0011.68 (1.31–2.15)< 0.0011.64 (1.28–2.11)< 0.001 Note: Analyses included only participants who did not exhibit a hyperglycaemic response during the oral glucose tolerance test (n = 2415), as those with serum glucose ≥ 140 mg/dL at 120 min during the 75‐g glucose load were excluded.Abbreviations: CI, confidence interval; OR, odds ratio. a Data available for 2405 participants. b Data available for 2414 participants.*Adjusted for sex, age, body mass index, and smoking status. This analysis included data for 2405 participants. Multivariate analysis (logistic regression) of factors associated with hypoglycaemia (< 70 mg/dL) at 120 min during the 75‐g oral glucose tolerance test. Note: Analyses included only participants who did not exhibit a hyperglycaemic response during the oral glucose tolerance test (n = 2415), as those with serum glucose ≥ 140 mg/dL at 120 min during the 75‐g glucose load were excluded. Abbreviations: CI, confidence interval; OR, odds ratio. Data available for 2405 participants. Data available for 2414 participants. Adjusted for sex, age, body mass index, and smoking status. This analysis included data for 2405 participants. When the post‐OGTT hypoglycaemia group was subdivided into participants with 2‐h serum glucose < 55 mg/dL (n = 98) and 2‐h serum glucose 55 to < 70 mg/dL (n = 229), the only significant difference in baseline clinical characteristics between these subgroups was higher physical activity among those with 2‐h serum glucose < 55 mg/dL (Table S1). Time‐Course Changes of Serum Glucose and Insulin During the Oral Glucose Tolerance Test Minute 0 (Fasting) Participants who developed post‐OGTT hypoglycaemia had the lowest fasting serum insulin and glucose concentrations (Table 1). The group with post‐OGTT serum glucose ≥ 200 mg/dL exhibited the highest fasting serum glucose and insulin concentrations (Table 1). Fasting serum glucose and fasting serum insulin concentrations were positively correlated with serum glucose concentrations at 120 min (Rho = 0.397 and Rho = 0.303, respectively; p < 0.001 in both cases). When participants with post‐OGTT hypoglycaemia were selected, fasting serum glucose and insulin concentrations were similar between smokers and non‐smokers (Figure 1). Boxplots showing time‐course changes in serum glucose and insulin during the OGTT (0, 30 and 120 min) according to smoking status in individuals with post‐OGTT hypoglycaemia. Horizontal lines indicate the median, boxes represent the interquartile range (IQR, 25th–75th percentile), whiskers extend to values within 1.5 × IQR, white circles represent mild outliers, and asterisks (*) represent extreme outliers. Minute 30 At 30 min after the start of the OGTT, participants with post‐OGTT hypoglycaemia had the lowest serum glucose concentrations (Table 1). Participants with post‐OGTT serum glucose ≥ 200 mg/dL showed the highest serum glucose concentrations at 30 min (Table 1). Serum glucose concentrations at 30 min were positively correlated with serum glucose at 120 min (Rho = 0.409, p < 0.001). Participants with post‐OGTT hypoglycaemia tended to have the highest serum insulin levels at 30 min (Table 1). Insulin and glucose concentrations at 30 min in participants with post‐OGTT hypoglycaemia were not significantly different from those observed in the group with normal post‐OGTT glucose (70 to < 140 mg/dL; Table 1). The negative correlation between serum insulin at 30 min and serum glucose at 120 min was also not statistically significant (Rho = −0.056, p = 0.240). Nevertheless, the group with post‐OGTT serum glucose ≥ 200 mg/dL tended to have the lowest serum insulin concentrations at 30 min (Table 1). When participants with post‐OGTT hypoglycaemia were selected, smokers, compared with non‐smokers, had lower serum insulin concentrations at 30 min despite having higher serum glucose concentrations at the same time point (Figure 1). Minute 120 (End of the Test) At the end of the OGTT, and similarly to what was observed at minute 0, participants with post‐OGTT hypoglycaemia had the lowest serum insulin concentrations (Table 1). Participants with 2‐h serum glucose < 55 mg/dL had lower serum insulin concentrations at 120 min (and, consequently, a higher Matsuda Index) than those with 2‐h serum glucose 55 to < 70 mg/dL (Table S1). Participants with post‐OGTT serum glucose ≥ 140 mg/dL exhibited higher serum insulin concentrations at the end of the OGTT (Table 1). There was a strong positive correlation between serum glucose at 120 min and serum insulin at 120 min (Rho = 0.672, p < 0.001). When participants with post‐OGTT hypoglycaemia were selected, smokers, compared with non‐smokers, had lower serum insulin concentrations at 120 min, despite no statistically significant differences in serum glucose concentrations at the same time point (Figure 1). Association of Post‐Oral Glucose Tolerance Test Hypoglycaemia With Incident Diabetes Glycaemic status was reassessed a median of 7.5 years (interquartile range, 7.1–7.9 years) after the initial evaluation in 2036 participants, 1922 of whom had not been diagnosed with diabetes at the baseline evaluation. A total of 103 participants (5.3%) developed diabetes during follow‐up. None had features suggestive of typical type 1 diabetes, as no participant with incident diabetes was both diagnosed before 35 years of age and treated with insulin [1]. The time to follow‐up was similar among the 4 baseline post‐OGTT glucose groups (Figure S1, p = 0.973) and there were no differences in the availability for glycaemic status reassessment among the 4 baseline post‐OGTT glucose groups (Figure S1, p = 0.176). At reassessment, participants who had post‐OGTT hypoglycaemia at baseline had the lowest overall incidence of diabetes; however, no statistically significant differences were observed when compared specifically with participants who had normal post‐OGTT glucose (70 to < 140 mg/dL) at baseline (Table 1). Similarly, participants with baseline post‐OGTT hypoglycaemia showed the lowest fasting serum glucose levels at reassessment. In the follow‐up OGTT, the baseline post‐OGTT hypoglycaemia group again showed, at 120 min, the lowest serum glucose concentrations and the highest frequency of hypoglycaemia (Table 1). No statistically significant differences in incident diabetes rates were observed between participants with 2‐h serum glucose < 55 mg/dL and those with 55 to < 70 mg/dL (Table S1). In a multivariate analysis (logistic regression), baseline post‐OGTT hypoglycaemia was not significantly associated with the odds of incident diabetes compared with baseline normal post‐OGTT glucose (odds ratio [OR] 0.68, 95% confidence interval [CI] 0.23–1.94, p = 0.474; Table 3). As expected, participants with baseline post‐OGTT serum glucose 140 to < 200 mg/dL had increased odds of incident diabetes, as did those with higher BMI or a family history of diabetes (Table 3). Current smoking was not significantly associated with the odds of incident diabetes when smoking status was included in the logistic regression models (OR 1.34, 95% CI 0.77–2.31, p = 0.289; Table S2). TABLE 3Multivariate analyses of factors associated with incident diabetes mellitus (logistic regression) and all‐cause mortality (Cox regression) during follow‐up.Incident diabetesCrude analysesAdjusted for age and sexMultiadjusted* OR (95% CI) p OR (95% CI) p OR (95% CI) p Sex (man)1.20 (0.80–1.79)0.3571.24 (0.83–1.86)0.2791.17 (0.74–1.84)0.502Age (years)1.02 (1.01–1.03)< 0.0011.02 (1.01–1.03)< 0.0011.00 (0.99–1.02)0.323Body mass index (kg/m2) a 1.14 (1.10–1.18)< 0.0011.13 (1.09–1.18)< 0.0011.12 (1.08–1.17)< 0.001Physical activity b Low (Ref.)111Moderate0.99 (0.64–1.55)0.9910.95 (0.61–1.49)0.8481.10 (0.67–1.82)0.685High0.67 (0.39–1.17)0.1650.66 (0.38–1.16)0.1551.05 (0.56–1.95)0.870Family history of diabetes (yes) c 2.14 (1.38–3.31)< 0.0012.64 (1.68–4.14)< 0.0012.48 (1.54–3.98)< 0.001Serum glucose at 120 min during the baseline oral glucose tolerance test70 to < 140 mg/dL (Ref.)111< 70 mg/dL 0.51 (0.18–1.44) 0.2070.53 (0.19–1.51)0.2400.68 (0.23–1.94)0.474140 to < 200 mg/dL8.54 (5.59–13.04)< 0.0017.95 (5.10–12.40)< 0.0016.15 (3.84–9.85)< 0.001All‐cause mortalityCrude analysesAdjusted for age and sexMultiadjusted** HR (95% CI) p HR (95% CI) p HR (95% CI) p Sex (man)1.73 (1.39–2.14)< 0.0011.94 (1.56–2.41)< 0.0011.99 (1.59–2.50)< 0.001Age (years)1.10 (1.09–1.11)< 0.0011.10 (1.09–1.11)< 0.0011.10 (1.09–1.11)< 0.001Body mass index (kg/m2) d 1.06 (1.04–1.08)< 0.0011.03 (1.01–1.05)0.0101.02 (1.00–1.04)0.056Current smoker (yes) e 0.59 (0.45–0.79)< 0.0011.76 (1.30–2.37)< 0.0011.68 (1.24–2.27)< 0.001Physical activity f Low (Ref.)111Moderate0.95 (0.75–1.21)0.7010.83 (0.65–1.06)0.1380.90 (0.70–1.16)0.430High0.72 (0.54–0.97)0.0310.67 (0.50–0.90)0.0090.77 (0.57–1.05)0.102Serum glucose at 120 min during the baseline oral glucose tolerance test70 to < 140 mg/dL (Ref.)111< 70 mg/dL0.49 (0.29–0.84)0.0100.82 (0.48–1.40)0.4860.84 (0.49–1.42)0.518140 to < 200 mg/dL2.94 (2.28–3.78)< 0.0011.30 (1.01–1.68)0.0411.26 (0.97–1.64)0.075≥ 200 mg/dL4.24 (3.07–5.86)< 0.0011.39 (1.01–1.93)0.0461.31 (0.94–1.82)0.111 Note: Analyses regarding incident diabetes included participants with baseline fasting serum glucose < 126 mg/dL and serum glucose < 200 mg/dL at 120 min during the baseline oral glucose tolerance test (i.e., those not diagnosed with diabetes at baseline), and available data on incident diabetes (n = 1922). Analyses regarding all‐cause mortality included participants who had available data on mortality (n = 2918).Abbreviations: CI, confidence interval; HR, hazard ratio; OR, odds ratio; Ref, reference category. a Data available for 1912 participants. b Data available for 1916 participants. c Data available for 1860 participants. d Data available for 2902 participants. e Data available for 2913 participants. f Data available for 2911 participants.*Adjusted for sex, age, body mass index, physical activity, family history of diabetes, and serum glucose at 120 min during the baseline glucose tolerance test. This analysis included data for 1852 participants.**Adjusted for sex, age, body mass index, smoking status, physical activity, and serum glucose at 120 min during the baseline glucose tolerance test. This analysis included data for 2899 participants. Multivariate analyses of factors associated with incident diabetes mellitus (logistic regression) and all‐cause mortality (Cox regression) during follow‐up. 0.51 (0.18–1.44) Note: Analyses regarding incident diabetes included participants with baseline fasting serum glucose < 126 mg/dL and serum glucose < 200 mg/dL at 120 min during the baseline oral glucose tolerance test (i.e., those not diagnosed with diabetes at baseline), and available data on incident diabetes (n = 1922). Analyses regarding all‐cause mortality included participants who had available data on mortality (n = 2918). Abbreviations: CI, confidence interval; HR, hazard ratio; OR, odds ratio; Ref, reference category. Data available for 1912 participants. Data available for 1916 participants. Data available for 1860 participants. Data available for 2902 participants. Data available for 2913 participants. Data available for 2911 participants. Adjusted for sex, age, body mass index, physical activity, family history of diabetes, and serum glucose at 120 min during the baseline glucose tolerance test. This analysis included data for 1852 participants. Adjusted for sex, age, body mass index, smoking status, physical activity, and serum glucose at 120 min during the baseline glucose tolerance test. This analysis included data for 2899 participants. Association of Post‐Oral Glucose Tolerance Test Hypoglycaemia With Long‐Term Mortality Survival status was reassessed a median of 14.5 years (interquartile range, 14.0–14.8 years) after the initial evaluation in 2918 participants. At the reassessment evaluation, a total of 337 participants (11.5%) had died from any cause, and 88 (3.0%) had died from cardiovascular causes. The time to follow‐up was similar among the 4 baseline post‐OGTT glucose groups (Figure S1, p = 0.096), and there were no differences in the availability for survival status reassessment among the 4 baseline post‐OGTT glucose groups (Figure S1, p = 0.180). Participants with post‐OGTT hypoglycaemia had the lowest all‐cause mortality rates, even when specifically compared with those with normal post‐OGTT glucose (70 to < 140 mg/dL; Table 1). All‐cause mortality rates were significantly higher among participants with post‐OGTT glucose ≥ 140 mg/dL (Table 1). Kaplan–Meier curves further illustrate these associations (Figure 2). Probability of survival (Kaplan–Meier curves) according to serum glucose at 120 min during the oral glucose tolerance test. Vertical ticks represent censored data. p‐value was calculated with the log‐rank test. However, the association between post‐OGTT hypoglycaemia and lower all‐cause mortality compared with normal post‐OGTT glucose was largely attenuated after adjusting for other covariates in the multivariate Cox regression analyses (hazard ratio [HR] 0.84, 95% CI 0.49–1.42, p = 0.518; Table 3). In the multivariate analyses, male sex, advanced age and smoking were consistently associated with long‐term all‐cause mortality (Table 3). Participants with post‐OGTT hypoglycaemia also had the lowest cardiovascular mortality rates (Table 1). However, differences were not statistically significant when specifically compared with those with normal post‐OGTT glucose, in both univariate (Table 1) and multivariate competing risk models (HR 0.76, 95% CI 0.22–2.55, p = 0.650; Table S3). In these multivariate analyses, advanced age was consistently associated with cardiovascular mortality (Table S3). No statistically significant differences in all‐cause or cardiovascular mortality rates were observed between participants with 2‐h serum glucose < 55 mg/dL and those with 55 to < 70 mg/dL (Table S1). Discussion Despite frequent speculation about the metabolic implications of post‐OGTT hypoglycaemia [37], this phenomenon remains largely understudied. Using a nationally representative sample of individuals without known diabetes, the present study examined clinical features and insulin secretion patterns associated with post‐OGTT hypoglycaemia, as well as the association of post‐OGTT hypoglycaemia with incident diabetes and mortality. Our results show that post‐OGTT hypoglycaemia is more common in young individuals with a favourable metabolic profile who, unlike those with prediabetes, appear to exhibit preserved early‐phase insulin secretion. Possibly due to this, post‐OGTT hypoglycaemia was associated with a low incidence of diabetes and reduced long‐term mortality. Post‐OGTT hypoglycaemia was a common finding in our study, occurring in more than 10% of the tests performed. This high prevalence underscores that post‐OGTT hypoglycaemia lacks sufficient specificity to diagnose rare disorders such as endogenous hyperinsulinism, as previously noted [2]. Although it has been hypothesised that post‐OGTT hypoglycaemia might be more frequent in individuals with early signs of insulin resistance [3], our results show that this phenomenon was more common in younger individuals with healthier metabolic profiles, as determined by anthropometric, laboratory and insulin resistance measures. These findings are in line with previous studies [7, 8, 9, 10], including those conducted in pregnant women [21, 22]. It is well known that individuals with prediabetes exhibit impaired early‐phase postprandial insulin secretion [11, 17, 18, 19]. This leads to an exaggerated late‐phase insulin response [11, 16, 17, 18], as confirmed by our findings in participants with post‐OGTT glucose ≥ 140 mg/dL. Individuals with post‐OGTT hypoglycaemia in our study displayed the opposite insulin secretion pattern: in response to rising glucose, they exhibited preserved early‐phase insulin secretion (at 30 min), followed by a suppression of late‐phase insulin secretion as glucose levels declined (at 120 min). This early insulin peak has also been reported in previous studies of populations with post‐OGTT hypoglycaemia [8, 12, 13, 14]. It remains unclear whether this preserved early‐phase insulin secretion is physiologically appropriate or potentially excessive. On one hand, it could possibly be considered excessive, given that the early insulin peak occurred in response to glucose concentrations that were the lowest in the post‐OGTT hypoglycaemia group at 30 min. However, when compared specifically with individuals who had a normal post‐OGTT result (70 to < 140 mg/dL), glucose and insulin concentrations at 30 min were similar. This might suggest that the early insulin response itself is not excessive; rather, greater insulin sensitivity in the post‐OGTT hypoglycaemia group could explain why a similar insulin peak led to hypoglycaemia only in this group. Another factor associated with post‐OGTT hypoglycaemia was active smoking. In a previous study analysing factors associated with fasting hypoglycaemia, the association between smoking and hypoglycaemia did not remain significant after adjusting for BMI, which was lower in smokers [38]. Thus, the independent association between smoking and post‐OGTT hypoglycaemia observed in our study is intriguing, particularly considering that smoking is known to be associated with both insulin resistance and impaired insulin secretion [39], traits opposite to those exhibited by participants with post‐OGTT hypoglycaemia in our cohort. This suggests that the mechanisms underlying post‐OGTT hypoglycaemia could differ between smokers and non‐smokers. Interestingly, when we selected participants with post‐OGTT hypoglycaemia and compared insulin secretion during the test between smokers and non‐smokers, we found that smokers had lower insulin concentrations at 30 min despite higher glucose levels at that time. Therefore, our hypothesis that post‐OGTT hypoglycaemia results from a combination of preserved early insulin secretion and high insulin sensitivity might not fully apply to smokers. The specific mechanisms involved in post‐OGTT hypoglycaemia among smokers are unclear and need further investigation. To our knowledge, this is the first population‐based study to assess the association of post‐OGTT hypoglycaemia with incident diabetes and mortality. Our findings are consistent with those reported in pregnant women [23, 24] and in individuals with cystic fibrosis [40, 41], in whom post‐OGTT hypoglycaemia was not associated with the subsequent development of various forms of diabetes. Similarly, a prior study followed 29 women with post‐OGTT hypoglycaemia for a mean of 6.4 years, none of whom experienced new‐onset diabetes [42]. In addition, a study conducted in Japanese Americans reported a lower risk of incident diabetes among participants with preserved early‐phase insulin secretion (although this was not typically followed by post‐OGTT hypoglycaemia) [18]. In our study, the association between post‐OGTT hypoglycaemia and incident diabetes was not statistically significant compared specifically with the normoglycaemic group and after adjusting for confounders. Mortality rates, including cardiovascular mortality, were particularly low in the post‐OGTT hypoglycaemia group. All‐cause mortality rates in people with post‐OGTT hypoglycaemia were even lower than those observed in participants with a normal OGTT result. This observation is noteworthy given the high frequency of smokers among participants with post‐OGTT hypoglycaemia; a factor that, as expected, emerged as a strong risk factor for mortality in the multivariate analyses. In these multivariate analyses, the association between post‐OGTT hypoglycaemia and mortality was also attenuated. This lack of statistically significant associations with incident diabetes and mortality after adjusting for confounders suggests that post‐OGTT hypoglycaemia likely represents a marker of younger age and a favourable metabolic profile rather than a true protective factor against incident diabetes or mortality. From a broader perspective, our findings may have implications for research, clinical practice and public health. In addition to the previously mentioned need for studies to explore the mechanisms underlying post‐OGTT hypoglycaemia in smokers, this study opens other relevant avenues for research. The meaning of post‐OGTT hypoglycaemia needs to be further explored. Gold‐standard methodologies, such as hyperinsulinaemic‐euglycaemic and hyperglycaemic clamp techniques, are warranted to disentangle the relative contributions of insulin sensitivity and insulin secretion to post‐OGTT hypoglycaemia. Similarly, further mechanistic studies are needed to elucidate the contribution of counterregulatory hormones and incretin pathways to the physiology of post‐OGTT hypoglycaemia. Continuous glucose monitoring‐based studies could also assess whether individuals with post‐OGTT hypoglycaemia develop postprandial hypoglycaemia under real‐life conditions. The clinical implications of this study are clear: post‐OGTT hypoglycaemia in individuals without known diabetes should not be interpreted as suggestive of insulin resistance or as a state of prediabetes, but rather as a marker of good metabolic health. Consequently, given that post‐OGTT hypoglycaemia identifies young, insulin‐sensitive individuals, it also serves as a surrogate marker of favourable long‐term prognosis. In general, our findings indicate that the OGTT provides an excellent characterisation of metabolic status, performing particularly well as a predictor of diabetes in individuals with post‐OGTT glucose concentrations of 140 to < 200 mg/dL. While other functional tests might offer greater precision, the OGTT could serve as an inexpensive test to indirectly assess insulin resistance, which can be relevant from a public health resource perspective. Despite strengths such as the population‐based design and long‐term follow‐up, our study has limitations that should be acknowledged. Hypoglycaemia‐related symptoms are an important aspect of the evaluation of non‐diabetic hypoglycaemia [2, 38, 43]. However, information on symptoms during the OGTT was not systematically collected in our study. Incident diabetes was assessed in a subset of participants approximately 7 years after the baseline evaluation, and the exact date of diabetes diagnosis could not be ascertained. Therefore, after confirming that median follow‐up duration was comparable across the four groups, diabetes incidence was evaluated based on glycaemic status at follow‐up, treating the analysis as a comparison between two cross‐sectional assessments rather than a time‐to‐event analysis. The majority of incident cases were likely type 2 diabetes [26], but the specific type was not accurately determined. Although none of the participants with incident diabetes had features suggestive of typical type 1 diabetes, more atypical presentations of type 1 diabetes (such as slowly progressive type 1 diabetes) or other forms of diabetes could not be excluded [1]. Blood glucose and insulin concentrations at 30 min were measured in only a small subset of participants from a specific geographical area; thus, results based on these measurements should be interpreted with caution. The Spanish population is predominantly White, which could limit the generalisability of our findings to other populations. Our study excluded individuals with previously diagnosed diabetes and therefore included participants with prediabetes. Although information on previously known prediabetes was not available, OGTT results (including fasting and 2‐h glucose values) allowed us to estimate that 29.9% (n = 875) of the study participants had prediabetes (Figure S1). The threshold for defining non‐diabetic hypoglycaemia, including post‐OGTT hypoglycaemia, is controversial [43]. We employed a threshold of < 70 mg/dL to define post‐OGTT hypoglycaemia, although other studies have used lower cut‐offs [7, 8], such as < 55 mg/dL. To explore the internal composition of the < 70 mg/dL group, we compared participants with post‐OGTT serum glucose < 55 mg/dL and those with 55 to < 70 mg/dL. No statistically significant differences were observed in clinical characteristics or outcomes between the groups, except for baseline physical activity levels and, as expected, insulin concentrations at 120 min (Table S1). These results support that the < 70 mg/dL group was homogeneous and appropriate for the analyses performed. Conclusions Post‐OGTT hypoglycaemia is a common finding that identifies younger individuals with a favourable metabolic profile, who show a low incidence of diabetes and reduced long‐term mortality. Preserved early‐phase insulin secretion together with good insulin sensitivity might explain post‐OGTT hypoglycaemia, particularly in non‐smokers. When detected incidentally, post‐OGTT hypoglycaemia should not be interpreted as a pathological finding, but rather as a marker of adequate metabolic health and, consequently, a favourable prognosis. Author Contributions T.G.‐V. was responsible for the study design, data curation, data analysis, interpretation of the results, drafting of the manuscript, and preparation of its graphical content. E.G.‐E., E.V.‐F., A.V.G., M.G.‐V. and C.L. contributed to the curation and interpretation of the data. G.R.‐M., S.V., J.F.‐N., A.C.‐P., L.C., F.J.C., E.M., E.M.‐T. and E.D. were responsible for the conceptualisation of the Di@bet.es Study, as well as for the data acquisition, data interpretation, data curation, funding acquisition, project administration and supervision. All authors reviewed the manuscript, provided critical comments, and approved the final version. Funding This investigation has been supported by CIBERDEM (Ministerio de Economía, Industria y Competitividad‐ISCIII), Ministerio de Sanidad, Servicios Sociales e Igualdad‐ISCIII, Instituto de Salud Carlos III (research grants PI17/02136, PI14/00710), and was co‐funded by the European Union and the Consejería de Salud de la Junta de Andalucía (PI‐0337‐2016). This study has also been co‐funded by the European Regional Development Fund (ERDF) “A way to build Europe”. Tomás González‐Vidal was supported by a Río Hortega research contract (CM24/00080) from the Instituto de Salud Carlos III. Gemma Rojo‐Martínez belongs to the Nicolás Monardes research program of the Consejería de Salud (C‐0060‐2012; Junta de Andalucía, Spain). Elsa Villa‐Fernández is a recipient of a Formación de Profesorado Universitario (FPU) grant from the Ministry of Education of Spain (MCIU‐24‐FPU23‐03765). Carmen Lambert is a recipient of a Sara Borrell grant from the Instituto de Salud Carlos III (CD23/00037). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Flowchart of the study. Table S1: Comparison of clinical characteristics and outcomes of participants with 2‐h serum glucose < 55 mg/dL and 2‐h serum glucose 55 to < 70 mg/dL. Table S2: Multivariate analysis (logistic regression) of factors associated with incident diabetes mellitus during follow‐up. Table S3: Multivariate analysis (competing risk model) of factors associated with cardiovascular mortality during follow‐up. Table S4: STROBE statement. Checklist of items that should be included in reports of observational studies.