Undetected dysglycaemia common in primary care patients treated for hypertension and/or dyslipidaemia: on the need for a screening strategy in clinical practice. A report from EUROASPIRE IV a registry from the EuroObservational Research Programme of the European Society of Cardiology

The primary care arm of EUROASPIRE IV was carried out in 14 European countries January 2014–April 2015 (Bosnia and Herzegovina, Bulgaria, Croatia, Kazakhstan, Lithuania, Poland, Portugal, Romania, Russian Federation, Serbia, Spain, Sweden, Ukraine and the UK). Within each country one or more geographical areas with a defined population were selected. A sample of general practices serving that population was identified according to the structure of the local health services.

Within each general practice men and women ≥ 18 to < 80 years at the time of identification, without a history of CVD, coronary or other atherosclerotic disease, who had been prescribed one or more of the following treatments: (i) blood pressure lowering drugs and/or (ii) lipid lowering drugs and/or (iii) glucose lowering (diet and/or oral drugs and/or insulin) since ≥ 6 months to < 3 years prior to the date of interview, were retrospectively identified from practice records. The total number of such patients was 4579 of whom 1325 with a history of treatment for diabetes and 42 who lacked information on diabetes were excluded. Out of the remaining 3212 patients the information on plasma glucose or HbA1c (n = 522) and the FINDRISC questionnaire (n = 295) was incomplete leaving 2395 as the study population (Fig. 1).

Venous blood was drawn in the fasting state (≥ 10 h) into a tube containing clot activator (Venosafe, Terumo Europe, Leuven, Belgium) for lipid assays and into a potassium ethylenediaminetetraacetic acid (EDTA) tube for HbA1c assay. Total and high-density lipoprotein (HDL) cholesterol and triglycerides were analysed in serum and HbA1c in whole blood on a clinical chemistry analyzer (Abbot Architect analyzer; Abbott Laboratories, Abbott Park, Illinois, USA) at a central laboratory (Disease Risk Unit, National Institute for Health and Welfare, Helsinki, Finland) accredited by Finnish Accreditation Service fulfilling the requirements of the standard SFS-EN ISO/IEC 17025:2005. Total serum cholesterol was measured using an enzymatic and high-density cholesterol (HDL-C) with a homogenous method for direct measurement while triglycerides were measured by an enzymatic glycerol phosphate oxidase method. Low-density lipoprotein (LDL-C) cholesterol was calculated according to Friedewald’s formula [17]. Glycated HbA1c was analysed with an immunoturbidimetric method and expressed in mmol/mol according to IFCC and % according to Diabetes Control and Complications Trial (DCCT). The glomerular filtration rate (eGFR) was estimated from serum creatinine by means of the Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI) [18].

The OGTT was performed using 75 g of anhydrous glucose in 200 ml of water in the morning after at least 10 h of fasting. Blood for FPG was drawn before intake of the glucose with a dip safe from the EDTA-tube in which the HbA1c was collected. Samples for 2hPG were drawn from whole venous blood using an EDTA-tube. Plasma glucose was analysed locally with a photometric point-of-care technique (Glucose 201 + , HemoCue^®, Ängelholm, Sweden). Regression analysis between the HemoCue^® instrument, and standard isotope dilution gas chromatography–mass spectrometry (IDGC–MS) showed a slope of 1.051 (95% confidence interval 1.031–1.071) an intercept of − 0.222 (95% CI − 0.016 to − 0.428; r = 0.994). The mean deviation was 0.24 mmol/l (2.0%). Values obtained with the HemoCue^® instrument were in 69% within 5%, in 91% within 10%, and always within 14.3% of the IDGC–MS method [19]. Since the HemoCue^® method is cholesterol sensitive due to the measurement in very small volumes with higher levels of glucose with low cholesterol the glucose values were corrected according to the formula: HemoCue^® glucose + 0.15 × (total cholesterol − 6). The values were converted from whole venous blood to plasma applying the formula by Carstensen et al. [19]: plasma glucose = 0.558 + 1.119 × whole blood glucose. Standardized use of the equipment was ascertained through central training of the data collectors, and retrieval of HemoCue^®-cuvette storage information, and validation sheets from a selection of the participating centres.

Hyperglycaemia defined as IFG, IGT, “high risk HbA1c” or newly detected T2DM was diagnosed according to WHO and ADA [20, 21] as outlined in Table 1.

Elevated LDL-C concentration was defined as a level ≥ 2.5 mmol/l.

Normal kidney function was defined as eGFR ≥ 60 ml/min/1.73 m².

Overweight was defined as a body mass index (BMI) 25.0–29.9 kg/m² and obesity as a BMI ≥ 30 kg/m². Central obesity was defined as a waist circumference of ≥ 88 cm for women and ≥ 102 cm for men [15].

Smoking was defined as self-reported smoking or an exhaled carbon monoxide > 10 ppm [22].

The EURObservational Research Programme at the European Heart House (Nice, France) was in charge of data management. All data were collected electronically through web-based data entry using a unique identification number for country, centre and individual. The data were submitted via the Internet to the data management centre where checks for completeness, internal consistency and accuracy were performed. All data were stored under the provisions of the National Data Protection Regulations.

Continuous variables were summarised according to their mean and standard deviation (SD); dichotomous variables as percentages (n). For comparing characteristics of included and excluded patients p-values were obtained from linear and logistic regression models adjusted for age and gender. All statistical analyses were undertaken using SAS statistical software (release 9.4) at the Department of Public Health, Ghent University, Belgium.

Pertinent clinical characteristics of included (n = 2395) and excluded (n = 817) patients are presented in Table 2; those excluded from the analysis were more often smokers, less physically active and had a lower educational level, higher diastolic blood pressure, FPG and HbA1c and were more frequently using statins and calcium channel blockers. Among the included patients 94% had a normal kidney function and the eGFR was unrelated to FPG, 2 h-PG and HbA1c in all patients.

The distribution of T2DM by diagnostic test is presented in Fig. 2. A FPG alone identified 80% of those with previously unknown T2DM while FPG in combination with 2hPG identified 92% and FPG together with HbA1c 90% of them.

A total of 455 patients (19%) were identified as having T2DM by the OGTT of whom 396 (87%) were detected by FPG alone. Including HbA1c ≥ 48 mmol/mol (DCCT ≥ 6.5%) as an additional diagnostic test the number of patients with newly detected T2DM increased by 13 to 468 (19.5%). Eight percent of the patients were detected by all three tests. Patients who were detected as having T2DM with HbA1c (8%) or 2hPG (10%) alone did not differ significantly from each other regarding age (mean ± SD = 59.0 ± 10.3 vs. 61.5 ± 11.9; p = 0.21), gender (males = 55.3% vs. 65.3%; p = 0.17) or FINDRISC score (mean ± SD 12.6 ± 3.5 vs. 13.9 ± 4.0; p = 0.38).

The 2hPG revealed that another 479 patients (20%) had IGT. The FPG classified 1495/2395 patients (62%) as having IFG (FPG = 5.6–6.9 mmol/l) including 395/479 of the patients with IGT. A “high risk” HbA1c [IFCC 39–47 mmol/mol (DCCT 5.7–6.4%)] was present in 72% of the total patient population.

Among the included patients 48.2–29.4% were treated with β-blockers or calcium channel blockers, respectively. The prevalence of newly detected dysglycaemia among patients on β-blockers was 41.0% (299/730) and among those without β-blockers 38.2% (634/1662) (p = 0.23 after adjustment for age and sex). About 43.4% (232/535) of the patients treated with calcium channel blockers had dysglycaemia compared to 37.8% (701/1857) in patients not treated with calcium channel blockers (p = 0.11 after adjustment for age and sex). In patients treated with statins 35.9% (256/713) had newly detected dysglycaemia compared to 40.4% (677/1678) among those who were not treated with statins (p = 0.0007 after adjustment for age and sex). Eighty-five percent of the patients were Caucasians. Among the remaining 15% 173 were Asian of whom 53 (30.6%) had newly detected T2DM. The distribution of dysglycaemia according to the different screening tests was similar to that in the Caucasian population.

The distribution of patients according to FINDRISC was: low risk to develop diabetes 10% (n = 244), slightly elevated risk 36% (n = 857), moderate risk 29% (n = 686), high risk 23% (n = 549) and very high risk 3% (n = 59). The proportion of patients with T2DM detected by FPG, 2hPG and HbA1c in relation to their FINDRISC category increased linearly (Fig. 3). The relationship between FINDRISC category and dysglycaemia (T2DM and IGT combined) diagnosed by FPG and 2hPG is shown in Fig. 4. Of patients, who according to FINDRISC had a low, moderate/slightly elevated risk 20, 34 and 41% had IGT or T2DM respectively. The corresponding proportions for the high and very high-risk category were 49 and 71%.

The present data from a European primary care population consolidate that dysglycaemia is a very common and global condition in patients with one or two other risk factors for CVD. The prevalence of screen detected dysglycaemia was 37% (IGT 11%; T2DM 17%) in the Impaired Glucose Tolerance and Long-Term Outcomes Observational (IGLOO) study, based on 1377 Mediterranean participants without known CVD but with one or more cardiovascular risk factors [23]. A similar proportion, 41%, was reported by the ARIC study [24] screening 8286 middle-aged American participants with the OGTT, of whom many with hypertension, dyslipidaemia and/or central obesity. Further, 14% had screen detected T2DM in a cross-sectional population-based study from Kuwait in which more than half of the participants were overweight or obese but without any history of CVD [25].

In the Interventions in Europe and Worldwide (PREVIEW) among 424 New Zealanders with overweight and FINDRISC score of ≥ 12, about 65% had undetected prediabetes and 7% T2DM when screened with the OGTT [26]. A higher FINDRISC score was significantly associated with prediabetes and T2DM (p = 0.02). The prevalence of dysglycaemia among those with lower FINDRISC score was, however, not investigated.

It has since long been established that dysglycaemia is an important risk factor for future CVD [2]. Other important risk factors for CVD are dyslipidaemia and hypertension in the presence of which dysglycaemia causes a dramatic increase in the risk for future CVD [27, 28]. The outcome of dysglycaemia screening is therefore a prerequisite for the institution of a multifactorial risk factor management as strongly emphasized by contemporary guidelines for cardiovascular disease prevention [2, 29]. It may be argued that patients under treatment for hypertension and/or hyperlipidaemia should have received life-style advice, the corner-stone in dysglycaemia management. Still 40% had undetected T2DM or IGT indicating a need for intensified life-style advice most likely supplemented by glucose lowering drugs in patients with newly detected T2DM. As regards IGT the experiences from the Da Qing study (11) and the Finnish diabetes prevention study [30] demonstrate that an improved life-style significantly reduced the development of overt T2DM during prolonged periods of follow up. Moreover, micro- and macro-vascular complications including cardiovascular mortality was reduced in the Da Qing study.

Opportunistic screening for dysglycaemia, as applied in the present study, is not contradicted by the ongoing debate on the value of population based screening for diabetes based on the inconclusive results of The Anglo–Danish–Dutch Study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care (ADDITION) [31‐34]. A reasonable assumption is that ADDITION, testing a community based screening programme in a general population, does not reflect the benefit of screening for dysglycaemia in a population at higher risk due to already diagnosed risk factors. In such population, an early instituted and comprehensive treatment has a more obvious rationale. A remaining question is how to screen. The willingness is by apparent reasons dependent of screening tools that are easy to handle and affordable. The simplicity of the FINDRISC questionnaire makes it easy to apply for the initiation of primary prevention in high risk people [13, 35]. In addition it is valuable for the detection of IGT, relates to markers of insulin resistance [36] and predicts CVD events and mortality [37].

In the present investigation it was hypothesized that the use of FINDRISC as a first step in dysglycaemia screening would reduce the need for blood tests. This assumption could not be verified since a large proportion (20–40%) of the participants were dysglycaemic already in the low-moderate FINDRISC risk categories. The most reasonable explanation is the high overall prevalence of dysglycaemia in the present population of people treated for—hypertension and/or dyslipidaemia. FINDRISC was based on a sample derived from the general population and its discriminatory ability is reported to be less in populations at higher risk [38]. Furthermore, the different FINDRISC components may differ by race and gender as reported by a subset of the ARIC study [39]. For example, waist circumference was more predictive of incident diabetes than BMI among black men but not in white women. An alternative to the FINDRISC for the prediction of dysglycaemia may to estimate lipoprotein (a), triglyceride/HDL-C ratio and triglyceride glucose index, which has been associated with insulin resistance and the risk of future incident diabetes [40‐42].

Several studies added biochemical measures, in particular FPG to improve the predictive capacity of risk scoring models [43]. Thus, screening was based on the IFG cut point (≥ 6.1 mmol/l) in the ARIC study to be followed by a clinical detection rule based on the risk factors of the metabolic syndrome and CVD for those below this value. With this strategy 86% of those with T2DM and 66% of all hyperglycaemia cases were detected, identifying 42% of the study population as screen positive [24]. In the IGLOO study, addressing opportunistic screening strategies based on the FINDRISC in individuals with one or more cardiovascular risk factors, the best screening strategy seemed to be based on FPG measurement in all participants and the performance of OGTT only in those with a FPG ≥ 5.6 mmol/l [23]. Based on this strategy, in which an OGTT was required for 56% of the studied population, 97% of the participants with T2DM and 78% of those with IGT was identified.

In the present study, the combination of FPG and 2hPG identified 92% of those with T2DM and the 2hPG another 20% with IGT. The superiority of this combination corresponds to the outcome of a previous study from EUROASPIRE IV on glycaemic screening of patients with established coronary artery disease [44] and in a recent population based study population consisting of overweight and obese subjects free from cardiovascular disease [45]. “High risk” HbA1c [DCCT 5.7–6.4% (IFCC 39–47 mmol/mol)] classified a majority (72%) of the present population as having glucose perturbations. This proportion is reasonably too high. Furthermore “high risk” HbA1c is less sensitive than IFG and IGT to detect individuals with insulin resistance and ß-cell dysfunction [46]. According to a WHO expert group, “high risk” HbA1c is a less effective diagnostic tool than FPG and 2hPG and current evidence is insufficient to make any formal recommendation on the interpretation of HbA1c levels < 6.5% [20, 47].

An alternative, limiting OGTT to 62% of the population (those with IFG), would be to start screening with a FPG continuing with an OGTT only in patients not fulfilling the criteria for T2DM. With such strategy 18% of patients with IGT would be missed. It has to be debated whether this is acceptable.

A major strength of EUROASPIRE IV is that data are based on interviews and standardised examinations of a large cross-sectional European population of well-characterised high-risk patients treated for hypertension and/or dyslipidaemia. Thus, the studied population is representative for a common at-risk group which, so far free from CVD, should be of particular interest for targeted life-style based interventions [4, 5, 48]. All tests (FPG, 2hPG and HbA1c) were uniformly undertaken by centrally trained staff members using the same methodology including a central laboratory. A limitation is that, as in almost all similar studies, dysglycaemia was based on a single glucose recording rather than, as recommended by WHO for a clinical diagnosis, repeated measurements. Moreover, it may be argued that patients who, due to missing data, were excluded had a somewhat higher risk profile (Table 2). This may, if of any importance, contribute to an underestimation of the true prevalence of dysglycaemia. Markers of dysglycaemia may vary by race and ethnicity [49, 50]. The vast majority of the present population was of Caucasian origin limiting the generalisability of the present findings to such populations.

Statin use may increase the risk for T2DM as well as β-blockers and possibly calcium channel blockers [51]. In the present study patients on statins had a lower proportion of dysglycaemia and there were no significant differences for β-blockers or calcium channel blockers. These data, derived from a cross-sectional investigation, should be looked at as descriptive for the present patient population. Accurate information on drug induced dysglycaemia can only be derived from longitudinal (cohort) studies. In contrast, the aim of the present study was to screen for undetected dysglycaemia in the presence of other risk factors and not to outline the risk for pharmacologically induced dysglycaemia.

It has been claimed that an OGTT has a limited reproducibility in particular for establishing the presence of IGT [52]. In a study by Wallander et al. [53] patients were screened with OGTT 5 days, 3 and 12 months after an AMI. Of those who were identified to have dysglycaemia at hospital discharge 93% were still classified as such, either having diabetes (64%) or IGT (29%) after 12 months. Of patients with IGT at hospital discharge 71% hade IGT or diabetes 12 months later and 29% normal glucose metabolism while 60% of patients classified with normal glucose metabolism at discharge remained normal after 12 months and 40% had developed IGT or type 2 diabetes. Thus, an OGTT seems to be reliable enough for screening purpose. Still, a second test should be performed to confirm the diagnosis, as recommended by WHO, before the institution of treatment.