Clusters
The analysis resulted in 7 significant clusters, 4 in the case of
patients with T2DM, and 3 in the non-diabetic patients. Dendrograms
portraying patients with and without T2DM are shown in supplementary
figure 1.
The resulting clusters are shown in table 1 and figure 3 (above).
Clusters 1, 2, 3 and 4 pertain to patients with T2DM. Cluster number 1
(n: 201) included predominantly female patients with high cardiovascular
risk without arrhythmia (only 1% had AF/flutter). They had high
systolic blood pressure (SBP) and high prevalence of both dyslipidemia
and atherothrombotic stroke. Cluster number 2 (n: 303) included older
patients, also predominantly women, with no dyslipidemia, lower SBP, and
a prevalence of AF/flutter greater than 50%. Cluster number 3 (n: 140)
included mostly male patients with COPD, dyslipidemia, and liver
disease. Although the presence of AF/flutter was 58.6%, this cluster
showed the lowest rate of cerebrovascular disease (92.1% patients
without any stroke). Cluster number 4 (n: 263) was similar to number 1.
However, the prevalence of AF/flutter (93.9%), TIA (7.2%), and
cardioembolic stroke (3.4%) was much higher. Clusters 5, 6 and 7
pertain to patients without T2DM. On average, these clusters portrayed
older patients, with lower SBP and BMI and better eGFR than those with
T2DM. Overall, the associations among the variables included are less
significant (figure 2). However, variables such as hemoglobin and eGFR
are more significant in defining the clusters, whereas the presence of
arrhythmia is not. Cluster 5 contains predominantly men (66.5%) with
COPD and a high prevalence of AF/flutter (64.8%). They also have the
highest levels of eGFR (62.3 +/- 32.7ml/min, p=0.0001) and hemoglobin
(12.8 +/- 2.9 gr/dl, p=0.01). Cluster 6 again contains mainly women
(69.6%) with excess weight and a high prevalence of dyslipidemia
(97.1%) and stroke, both hemorrhagic and cardioembolic or
atherothrombotic (p=0.003). Approximately half have AF/flutter (55.1%)
and of the three clusters without T2DM, patients in cluster 6 have the
lowest levels of eGFR (p=0.0001) and hemoglobin (p=0.01). Finally,
cluster 7, with a 67% predominance of women, is quite undifferentiated,
with a slightly higher SBP than the other clusters without T2DM, and a
notable prevalence of AF/flutter (63.2%) and TIA (4%).
Some additional findings by clusters are shown in Table 2. Patients with
T2DM had more history of hypertension, mainly clusters 3 and 4. Charlson
comorbidity index is similarly higher in these patients, the highest
levels being found in cluster 3. Regarding etiology of HF, ischemic
cardiomyopathy was significantly prevalent in patients with T2DM
(p=0.001), particularly in cluster 3 (40%). Finally, significant
differences in treatment were detected among the clusters. Overall,
patients with T2DM were more often treated with angiotensin-converting
enzyme inhibitors (ACEi)/angiotensin receptor blockers (ARB) (p=0.004),
beta-blockers (p=0.003), loop diuretics (p=0.007), and thiazides
(p=0.009).