healthmedicinet

A total of 3,041 patients (53.8 % male; 60.4 years, standard deviation 9.7), 35 to
74 years old, both inclusive, without a previous cardiovascular disease were included.
The variables analysed show significant differences in age, gender, years of diabetes
diagnosed, ethnicity, tobacco consumption, pulse pressure, HbA1c, single or combined
use of insulin and the prevalence of diabetic retinopathy (Table 2).

Table 2. Baseline characteristics of the cohorts

The ADVANCE equation identifies a greater number of diabetic patients with a high
cardiovascular risk ( =737, 24.2 %), while REGICOR classifies the lowest number of patients with a high
risk (?=?312, 10.2 %). The UKPDS classifies a greater number of DM2 patients with a high
cardiovascular risk than the REGICOR equation, and similar to the ADVANCE equation
(?=?691, 22.7 %).

Table 3 shows the characteristics of the DM2 patients with a high cardiovascular risk for
each equation. In the ADVANCE equation, the profile of the patients with high risk
are older (68.1 years; SD 5.9), with more years of evolution of DM2 (11.8; SD 7.1),
a smaller number of smokers, high pulse pressure (63.9 mmHg; SD 12.8), greater impaired
kidney function and retinal alterations, a higher percentage of patients treated with
single or combined insulin with oral antidiabetics (OA), and a greater number of patients
following an antihypertensive treatment. The UKPDS and ADVANCE equations show similar
results in most of the characteristics of the high risk patients, except for age,
pulse pressure, antihypertensive drug consumption, and impairment of kidney function.
REGICOR classifies as high risk patients a higher number of male smokers with elevated
levels of non-HDL cholesterol and low levels of HDL cholesterol.

Table 3. Characteristics of high-risk patients according to each risk equation

According to the real validity assessment, the correlation and agreement among the
different equations were poor, despite the statistical significance. Only the values
of correlation and agreement between the DM2 patient-specific equations (ADVANCE and
UKPDS) were significant (r?=?0.752, p??0.0001 and k?=?0.608, p??0.0001; respectively). The correlation of the REGICOR equation with UKPDS and ADVANCE
was low (r?=?0.288 and r?=?0.153, respectively; p??0.0001). The agreement scores when assigning a specific patient to the high risk
group were low for the REGICOR risk equation compared to UKPDS and ADVANCE (k?=?0.205 and k?=?0.123, respectively; p??0.0001) but good when comparing the ADVANCE equation with the UKPDS equation (Table 4).

Table 4. Agreement among the different risk engines in the classification of high risk patients

The sensitivity, specificity and positive and negative predictive values for UKPDS
and ADVANCE equations were calculated compared to REGICOR equation. For the usual
cut-off points (REGICOR ?10 %, UKPDS ?15 %, and ADVANCE ?8 %), and considering that
REGICOR is the gold standard as the only validated engine for the Spanish population,
the sensitivity was 0.518 for the UKPDS equation and 0.394 for the ADVANCE equation;
the specificity was 0.798 and 0.791, respectively. The positive predictive values
were 0.067 and 0.083, respectively, and the negative predictive values were 0.767
and 0.817, respectively. The positive likelihood ratios were 2.56 and 1.88, and the
negative likelihood ratios were 0.6 and 0.76, respectively.

The multivariate analysis (Table 5) showed that the time of evolution of diabetes, specifically for the ADVANCE equation,
the levels of HbA1c, the age of the patient, and the levels of non-HDL cholesterol
were variables associated with a greater degree of disagreement between REGICOR equation
and the DM2 patient-specific equations when classifying DM2 patients with a high cardiovascular
risk. Other variables associated with the disagreement between REGICOR and ADVANCE
equations were the presence of retinopathy, arterial hypertension with drug treatment,
tobacco consumption, pulse pressure, and urinary albumin excretion. The same multivariate
analysis model for each separate cohort showed similar results to the global results
of the three cohorts.

Table 5. Multivariate analysis of variables associated with disagreement among REGICOR (Framingham
adapted to the Spanish population), UKPDS, and ADVANCE functions