Cardiovascular disease risk and glucose metabolism: agreement and discrepancies between the WHO-85 and ADA-97 classifications

Objetivo. Conocer las diferencias entre el riesgo coronario de los sujetos con glucemia basal alterada (GBA), intolerancia oral a la glucosa (ITG) y diabetes mellitus tipo 2 según las clasificaciones diagnósticas de la OMS-85 y ADA-97 en una población adulta con un riesgo alto de presentar diabetes mellitus. Diseño. Estudio descriptivo, transversal, multicéntrico. Emplazamiento. Atención primaria, 7 centros de salud. Pacientes. Un total de 970 sujetos considerados población de riesgo para diabetes mellitus tipo 2. Mediciones. Se clasificaron los sujetos según los criterios OMS-85 (normales, ITG, diabetes) y según la ADA-97 (normales, GBA y diabetes). Se recogieron las siguientes variables: edad, sexo, consumo de tabaco, índice de masa corporal, tensión arterial sistólica, tensión arterial diastólica, glucemia basal, glucemia a las 2 horas de la PTOG, HbA _1c, microalbuminuria, colesterol total, cHDL, cLDL, triglicéridos y se estimó el cálculo del riesgo coronario mediante la tabla de Wilson et al de 1998, basada en el estudio Framingham. Resultados. Se evaluó a 970 sujetos con una edad media de 58,6 ± 12,4 años, 453 varones (46,7%) y 517 mujeres (53,3%). En el análisis de la presencia de factores de riesgo cardiovascular se observa que éstos son menos frecuentes en los sujetos normales y que su prevalencia es más elevada en los diabéticos (OMS y ADA). No existen diferencias significativas entre el riesgo coronario y los diversos factores de riesgo cuando se analizan grupos homónimos OMS­ADA (normales, ITG-GBA o diabéticos). Según la clasificación de la OMS-85, el riesgo coronario en los distintos tipos de alteraciones del metabolismo de la glucosa fue un 11,3% en sujetos normales, un 14% en ITG y un 27,3% en diabéticos, y según la ADA-97 un 11,4% en sujetos normales, un 15,7% en GBA y un 29,5% en diabéticos. Conclusiones. A mayor grado de patología hidrocarbonada, mayor coexistencia de factores de riesgo y mayor estimación del riesgo coronario. No hay diferencias importantes entre los estadios de las clasificaciones OMS y ADA ni en la presencia de factores de riesgo cardiovascular ni en relación al riesgo coronario.

Aim. To identify the differences between coronary heart disease risk in patients with altered basal glucemia (ABG), oral glucose intolerance (OGI) and type II diabetes mellitus according to the WHO-85 and ADA-97 diagnostic classifications, in an adult population at high risk for diabetes mellitus. Design. Descriptive, cross-sectional, multicenter study. Setting. Seven primary health care centers in Spain. Patient. 970 persons considered the population at risk for type II diabetes mellitus. Measures. Participants were classified according to the criteria of the WHO-85 (normal, OGI, diabetes) and the ADA-97 system (normal, ABG, diabetes). The following variables were recorded: age, sex, smoking habit, body mass index, systolic blood pressure, diastolic blood pressure, basal glucemia, glucemia 2 h after an oral glucose tolerance test, HbA _1c, microalbmuniuria, total cholesterol, HDL, LDL and triglycerides. Coronary heart disease risk was calculated with the 1998 table developed by Wilson et al. on the basis of the Framingham study. Results. A total of 970 participants were studied. Mean age was 58.6 #+ 12.4 years; 453 were men (46.7%) and 517 were women (53.3%). Our analysis showed that cardiovascular disease risk factors were less frequent in normal subjects, and that their prevalence was higher in persons with diabetes (according to both WHO and ADA classifications). There were no significant differences in coronary heart disease risk or different risk factors between analogous groups in the two classification systems (normal, OGI/ABG or diabetes). Coronary heart disease risk in persons with different types of alterations in glucose metabolism was 11.3% in normal subjects, 14% in persons with OGI and 27.3% in persons with diabetes according to the WHO-85 system, and 11.4% in normal subjects, 15.7% in persons with ABG and 29.5% in persons with diabetes according to the ADA-97 system. Conclusions. The greater the alteration in carbohydrate metabolism, the greater the coexistence of risk factors and the estimated risk of coronary heart disease. There were no significant differences in the presence of cardiovascular risk factors, or in the relationship between carbohydrate metabolism and coronary heart disease risk, between analogous stages identified with one classification system or the other.

  Introduction

The clinical diagnosis of diabetes mellitus is based on the determination of cut-off points for glucemia, a continuous laboratory value. In 1980 and 1985 the World Health Organization (WHO) recommended that the diagnosis be established when repeated measures of fasting glucose yielded values of 140 mg/dL or higher, or when glucemia measured after a 2-h oral glucose tolerance test (OGTT) was 200 mg/dL or higher. Values above these numbers indicated in increased risk for microvascular complications.¹ In a specific group of subjects, glucemia after a 2-h OGTT is between 140 and 200 mg/dL and their risk of cardiovascular disease (apart from microangiopathy) is high; these persons have been classified as having impaired glucose tolerance (IGT).²

Subsequently, other authors^3-6 showed that the risk of cardiovascular disease was greater in persons with glucemia levels of around 110 mg/dL or higher. In 1997, after reviewing the epidemiological studies that centered on fasting glucose, the American Diabetes Association (ADA) developed new diagnostic criteria, setting the cut-off point at 126 mg/dL for diabetes, and producing a new classification: normal (fasting glucose lower than 110 mg/dL), impaired fasting glucose (IFG, fasting glucose between 110 and 125 mg/dL), and diabetes (fasting glucose 126 mg/dL or higher.⁷ Further studies^8-12 have found differences between the WHO and ADA categories, which do not define the same groups of subjects. The DECODE study,¹³ for example, showed that agreement between the IFG and IGT categories was low (28%), and recommended that testing to detect OGT not be abandoned in view of its usefulness in predicting evolution to diabetes, and because glucemia after a 2-h OGTT is a better predictor of morbidity and mortality due to cardiovascular disease than is fasting glucose. Some studies¹⁴ considered only IGT to be a risk factor for cardiovascular disease; nonetheless, others¹⁵ have reported that mortality from cardiovascular diseases was higher among patients with IFG.

Most subjects with impaired carbohydrate metabolism (whether IGT, IFG or type 2 diabetes mellitus) also have other clinical and laboratory features (e.g., obesity, hypertension, dyslipedimia, fibrinolytic alterations) that are consistent with the so-called metabolic syndrome (Reaven´s «syndrome X»,)¹⁶ and which lead to an exponential increase in the risk of cardiovascular disease. In persons with diabetes, cardiovascular disease is the most frequent cause of death (accounting for approximately 60%-80% of all deaths in this population).^17-20 The importance of correctly classifying subjects with a moderate degree of alteration in carbohydrate metabolism (IGT or IFG) lies mainly in determining which persons are at greatest risk of developing diabetes and cardiovascular disease. The main aim of this study was therefore to identify the differences in risk of cardiovascular disease between persons with IFG, IGT and type 2 diabetes mellitus as identified by the WHO-85 and ADA-97 classification systems. Our subjects were an adult population at high risk for diabetes mellitus. Other aims of this study were to evaluate differences between different risk factors for cardiovascular disease in these patients.

Material and methods

In this descriptive, cross-sectional multicenter study the population was drawn from primary care patients followed at seven urban, semi-urban and rural health centers (total reference population approximately 150 000) in the cities of Barcelona (Raval Sud Basic Health Area), Reus (Reus 1 and Reus 2 Basic Health Areas) and Tarragona (Tarragona-Valss, Sagessa Group) in Spain. All cases were chosen from the population of patients being followed by primary health teams at these centers, and were recruited consecutively from among those who fulfilled the inclusion criteria. In all, we studied 970 persons considered the population at risk for type 2 diabetes mellitus. Sample size was calculated for a 95% confidence interval (95% CI) and an error rate of 3.1%. Inclusion criteria were based on risk factors for impaired glucose metabolism (obesity, body mass coefficient >30, antecedents of type 2 diabetes in first degree relatives, antecedents of gestational diabetes, previous disorder in glucose metabolism, and repeated use of glucemia-raising drugs such as diuretics, beta blockers, corticosteroids or estrogens).

The exclusion criteria were a prior diagnosis of diabetes mellitus, antecedents or clinical signs of any cardiovascular disease, and refusal to participate in the study.

We studied three sets of variables: sociodemographic (age, sex, risk factors for type 2 diabetes mellitus), clinical (smoking, body mass coefficient, systolic and diastolic blood pressure) and laboratory (fasting glucose, glucemia after a 2-h OGTT, HbA_1c, microalbuminuria, total cholesterol, high density lipoprotein cholesterol [HDL-C], low density lipoprotein cholesterol [LDL-C], triglycerides).

Each patient was classified according to the diagnostic criteria of the WHO-85 system, based on fasting glucose and OGTT results, as normal (fasting glucose or glucemia after a 2-h OGTT<140 mg/dL), impaired glucose tolerance (glucemia after a 2-h OGTT between 140 and 199 mg/dL) or diabetic (fasting glucose 140 mg/dL or higher, or glucemia 2 h after OGTT 200 mg/dl or higher). The same patients were each classified again on the basis of the ADA-97 criteria, based on fasting glucose alone, as normal (lower than 110 mg/dL), impaired fasting glucose (between 110 and 125 mg/dL), or diabetic (126 mg/dL or higher).

Global risk of coronary heart disease was calculated for each subject with data from the table of Wilson et al. (1998), which are based on the Framingham study. This table records the following variables: sex, age, diastolic and systolic blood pressure, cigarette smoking, diabetes, total cholesterol and HDL-C.²¹ In the present study diabetes was defined according to WHO criteria, and the risk of coronary heart disease was calculated according to the WHO system. Then each patient was classified again with the ADA system, and coronary heart disease risk was again calculated with the latter system. Scores were assigned only for those patients with a diagnosis of diabetes. The final score was the sum of the scores obtained from the table for each variable. These scores were transferred to a new table to predict the percentage likelihood of a new adverse cardiovascular event in the next 10 years. The 10-year risk of coronary heart disease was considered high when the likelihood was >20%, and very high when it was >30%.^22-24

Descriptive statistics were based on calculation of the mean and standard deviation for quantitative variables, and calculation of the proportions and 95% CI for qualitative variables. Normal distribution was verified with the Kolmogorov-Smirnoff method. Analytical statistics were based on Student´s t test and analysis of variance (ANOVA) for quantitative variables, and on the chi-squared test for qualitative variables. The differences were considered significant when P<.05. Agreement between the two measurements of the same phenomenon was measured as the kappa coefficient, and was considered very good at >0.75, acceptable at 0.40-0.75, and poor at <0.40.²⁵

Results

The study sample consisted of 970 persons (453 men [46.7%], 517 women [53.3%]) with a mean age of 58.6±12.4 years and a mean body mass coefficient of 30.6±5.34 kg/m². Table 1 summarizes the findings for the sociodemographic, clinical and laboratory variables. When we compared different risk factors for cardiovascular disease in subjects with equivalent diagnoses according to the WHO and ADA systems (normal, IGT or IFG, type 2 diabetes mellitus), we found few statistically significant differences. In comparison with the WHO classification, persons considered normal in the ADA system had lower fasting glucose values, higher glucemia values after a 2-h OGTT, and higher 24-h microalbuminuria values. Subjects with IFG had higher fasting glucose and glucemia after a 2-h OGTT than did those diagnosed as having IGT. Subjects considered to have diabetes in the ADA system had lower glucose values after a 2-h OGTT than those found to have diabetes with the WHO system (Table 1).

 

When we compared the two systems for agreement in diagnosis (Table 2), we found that agreement was acceptable for type 2 diabetes mellitus (kappa coefficient 0.69) and normal status (kappa coefficient 0.54). However, agreement was poor for the OGT and IFG categories (kappa coefficient 0.19), and the two diagnoses matched for only 20.7% of the subjects.

 

The risk of coronary heart disease, as calculated from the table of Wilson et al. based on the Framingham study (1998), is shown in Table 3 for each category of diagnosis. The values obtained for each risk factor were used to calculate the mean score for each group, expressed as the percentage risk of having an adverse cardiovascular event in the next 10 years. We found that the WHO criteria led to risks of 11.3% in normal subjects, 14.0% in subjects with IGT and 27.3% in subjects with type 2 diabetes mellitus, whereas the ADA criteria yielded values of 11.4% (normal), 15.7% (IFG) and 29.5% (type 2 diabetes mellitus). Thus the diagnosis of IFG and type 2 diabetes mellitus (ADA system) implied a greater risk of coronary heart disease than did the analogous diagnoses with the WHO system (although the difference was statistically significant only for the diagnosis of diabetes). Regardless of which system was used, the risk of coronary heart disease was greater in subjects with IGT or IFG, or diabetes, than in those with normal carbohydrate metabolism (P<.05). When we analyzed different degrees of coronary heart disease risk, we found that the risk was high (>20%) in 27% of the subjects with IGT, 29.2% of those with IFG, 72.1% of those with type 2 diabetes mellitus according to the WHO system, and 76.2% of those with type 2 diabetes mellitus according to the ADA system, whereas this level of risk was found in only 18.1% (WHO system) and 18.8% (ADA system) of the normal subjects (Table 4).

 

Discussion

Diabetes is associated with a 2-fold to 4-fold greater risk of coronary heart disease than that observed in the population of persons without diabetes; the greater risk, obviously, implies greater mortality.^17-20,26 The risk of atheromatous lesions is also increased in the group of subjects with a moderate degree of alteration in glucose metabolism, e.g., IGT or IFG,^14,15 and even in persons with whose glucemia values are at the upper limit of normality.²⁷ Therefore, changes in the diagnostic criteria for diabetes can lead to changes in how the risk of cardiovascular disease is assessed. The aim of our study was to compare the influence of the new ADA-97 criteria with that of the older WHO-87 criteria on the assessment of coronary heart disease risk, and to identify the differences between the groups of patients identified as having IGT or IFG.

With regard to our methods, the fact that ours was a multicenter study based on the population seen at primary care centers helped to minimize some of the biases the appear most often in studies of a single, highly selected population, generally in hospital settings (i.e., populations which are less than ideally representative of the general population). To obtain an acceptable number of subjects with impaired carbohydrate metabolism, we worked with a population of subjects who had risk factors for type 2 diabetes mellitus. This may have resulted in selection bias, as our study population might also have had more risk factors for cardiovascular disease in association with diabetes even though their glucose values at the time of the study were normal. This would imply that the differences we observed between the group with a metabolic disorder and the group of normal subjects were smaller than those that would be found in the general population. The strict application of the WHO-87 criteria involves the use of OGTT for most subjects, in contrast with usual clinical practice, in which fasting glucose is the predominant measure (ADA criteria).

To calculate the risk of cardiovascular disease we used the Framingham table, because it is well known in our setting.²⁸

With regard to the agreement in diagnosis obtained with the WHO and ADA classifications, the ADA system is relatively useful for diagnosing normal status (k=0.54) and type 2 diabetes (k=0.69). However, agreement for the diagnosis of IGT or IFG was poor (k=0.19), as various authors have shown previously^8-15. This may be explainable because of differences in physiopathological mechanisms and the stage of glucemia in the two entities. What appears clear is that IGT and IFG do not identify the same subjects or define the same disorder, although both lie in the range between normality and type 2 diabetes mellitus.

Subjects classified as normal with both systems had lower fasting glucose levels and fewer risk factors for cardiovascular disease than subjects with IGT or IFG; in turn, there were fewer risk factors in these intermediate groups than in subjects with diabetes. This finding is significant in that it translates as a parallel relationship between the intensity of the carbohydrate disorder and the risk of coronary heart disease the subjects are exposed to.

In the present study we found no significant differences between subjects who were classified as normal with one system or the other. The only finding of note was the lower fasting glucose value in subjects considered normal according to the ADA criterion; this was expected in light of how this diagnosis is defined. There were no differences in any of the other risk factors for cardiovascular disease.

In subjects classified as having IGT or IFG, we found differences only in fasting glucose and glucemia after a 2-h OGTT. In subjects with IFG (ADA-97) the diagnosis is arrived at later (fasting glucose); this probably explains the greater number of risk factors in these persons. Mean 10-year risk of coronary heart disease for subjects with IGT (WHO) was 14.0% for subjects with IGT (WHO) and 15.7% for those with IFG (ADA); this difference was not statistically significant. Although these intermediate stages between normality and diabetes also showed an intermediate degree of risk for cardiovascular disease, the analysis of individuals who satisfied the criteria for both IGT and IFG (about one in every five) is of particular interest. In an earlier study the lipid profile of this subgroup was shown to be much more atherogenic (higher concentration of total and LDL cholesterol) than in subjects with either IGT or IFG.²⁹ This finding supports the recommendation that the primary prevention of diabetes and its concomitant risk of cardiovascular disease should be aimed specifically at individuals with one or both disorders.

In persons diagnosed according to one criterion or the other as having type 2 diabetes, we found no significant differences in the presence of cardiovascular risk factors, although the global risk of coronary heart disease was 27.3% in patients diagnosed with the WHO criterion, versus 29.5% with the ADA criterion (P<.05).

When we compared those subjects most likely to develop atheromatous lesions (with a 10-year risk of coronary heart disease >20%), we noted that this risk was present in 27% of those with IGT and 29.2% of those with IFG. This situation was much more frequent in patients with type 2 diabetes mellitus.

In conclusion, use of the new ADA criteria do not, in global terms, involve significant changes with respect to determining the risk of cardiovascular disease in patients with any of the three grades of glucose metabolism disorder. The Framingham Heart Study used a population residing near Boston, and the use of these data probably overestimated the risk of coronary heart disease in the current Spanish population, as the mortality rate for cardiovascular disease in the USA (close to 500 per 100 000) is higher than that for Spain (lower than 300 per 100 000).³⁰ Although there are other recognized methods for calculating the risk of coronary heart disease,^31-33 we believe prospective studies in our setting are fundamental, especially for population groups at greatest risk of atherogenesis. We are currently following our sample to determine the actual appearance of cardiovascular events over the next 10 years, in order to determine the validity of predictions based on standard tables.

Correspondence: Isabel Otzet Gramunt. CAP Raval Sud, Avda. Drassanes, 17-21 7.ª planta, 08001 Barcelona, España.E-mail: [e-mail] Manuscript accepted for publication 5 September 2001.