Patient population

Of the 11,440 patients included in the analysis, 31% (n = 3550) had DM and 69% (n = 7890) did not have DM. Prevalence of DM increased during the years of the study period: 28.1% between 2007 and 2010; 30.2% between 2011 and 2014; and 34.8% between 2015 and 2018 (p < 0.001). Diabetic patients were younger (80 ± 7 vs. 82 ± 7 years, p < 0.001), more frequently men (47% vs. 41%, p < 0.001) and had a higher body mass index (28 ± 5 vs. 26 ± 5 kg/m2, p < 0.001) than non-diabetic patients. Overall, they had a worse cardiovascular risk profile, with a higher prevalence of hypertension and dyslipidaemia. They more frequently reported a history of myocardial infarction, percutaneous coronary intervention, coronary artery bypass grafting, cerebrovascular events and peripheral vascular disease. Statins, aspirin and clopidogrel were more frequently used by DM patients. Diabetic patients had a higher predicted 30-day mortality according to STS-PROM (7.0%, IQR 4.1–13.5% vs. 6.1%, IQR 3.9–12.8%, p < 0.001) and EuroSCORE II (4.7%, IQR 2.7–8.2% vs. 3.7%, IQR 2.2-6.0%, p < 0.001). In contrast, the Logistic EuroSCORE, which does not include DM as a risk factor, was similar for both groups (15.5%, IQR 9.5–23.6% vs. 15.0%, IQR 9.5–22.9%, p = 0.05). Table 1 presents an overview of baseline characteristics.

Table 1 Baseline patient characteristics of the unmatched population Clinical outcomes in the unmatched population

Patients with DM and without DM had comparable rates of in hospital mortality (4.5% vs. 4.9%, RR 0.9, 95%CI 0.8–1.1, p = 0.43), stroke (1.7% vs. 2.0%, RR 0.8, 95%CI 0.6–1.1, p = 0.27), myocardial infarction (0.9% vs. 0.7%, RR 1.4, 95%CI 0.9–2.1, p = 0.18), major or life-threatening bleeding (6.1% vs. 6.7%, RR 0.9, 95%CI 0.8–1.1, p = 0.25), and permanent pacemaker implantation (13.3% vs. 13.1%, RR 1.0, 95%CI 0.9–1.1, p = 0.74). Median length of stay was 7 days in both groups (IQR 5–11, p = 0.96). Moreover, 30-day rates of all-cause mortality (5.4% vs. 6.1%, RR 0.9, 95%CI 0.8–1.1, p = 0.19) and stroke (2.3% vs. 2.6%, RR 0.9, 95%CI 0.7–1.2, p = 0.40) did not differ between diabetic and non-diabetic patients. One year mortality (17.5% vs. 17.4%, RR 1.0, 95%CI 0.9–1.1, p = 0.86) and stroke rates (5.0% vs. 5.4%, RR 0.9, 95%CI 0.8–1.2, p = 0.53) were also similar. Table 2 presents outcomes in the unmatched population.

Table 2 Clinical outcomes of patients with versus without diabetes mellitus in the unmatched population Baseline characteristics of the propensity matched population

A total of 3281 patient pairs were obtained using propensity score matching. The matched population had a mean age of 80 ± 7 years and 54% was women. GFR was lower in diabetic patients (51.4 ml/min/1.73m2, IQR 37.4–69.0, vs. 54.1 ml/min/1.73m2, IQR 40.6–71.4, p < 0.001) and aortic valve area was smaller (0.67 ± 0.19 cm2 vs. 0.69 ± 0.22 cm2, p < 0.001) than in non-diabetic patients. Baseline medical history, cardiovascular medication, echocardiographic characteristics and device types were comparable between patients with and without DM, as presented in Table 3.

Table 3 Baseline patient characteristics of the propensity matched population Clinical outcomes in the propensity matched population

In hospital outcomes were not different between DM and non-DM patients: mortality (4.7% vs. 4.3%, RR 1.1, 95%CI 0.9–1.4, p = 0.38), stroke (1.7% vs. 2.1%, RR 0.8, 95%CI 0.6–1.2, p = 0.28), myocardial infarction (0.8% vs. 0.5%, RR 1.5, 95%CI 0.8–2.1, p = 0.21), new onset atrial fibrillation (7.8% vs. 7.1%, RR 1.1, 95%CI 0.8–1.6, p = 0.56), and permanent pacemaker implantation (13.0% vs. 13.6%, RR 1.0, 95%CI 0.8–1.1, p = 0.51). Median length of hospital stay was 7 days in both groups (IQR 5–11, p = 0.92). Thirty-day rates of mortality (5.6% vs. 5.6%, RR 1.0, 95%CI 0.8–1.2, p = 0.96) and stroke (2.3% vs. 2.6%, RR 0.9, 95%CI 0.6–1.2, p = 0.47) were comparable, as well as one-year mortality (17.3% vs. 16.2%, RR 1.1, 95%CI 0.9–1.2, p = 0.37) and stroke (4.9% vs. 5.2%, RR 1.0, 95%CI 0.7–1.2, p = 0.75) rates. Figure 1 presents time to mortality curves and Table 4 clinical outcomes in the propensity matched population. DM was not a predictor for one year mortality (HR 1.07, 95%CI 0.92–1.23, p = 0.40): neither in men (HR 0.94, 95%CI 0.76–1.17, p = 0.59) nor in women (HR 1.18, 95%CI 0.67–1.44, p = 0.10).

Fig. 1

Time-to-mortality curves of patients with and without diabetes mellitus undergoing transcatheter aortic valve implantation (matched population)

Legend: HR = Hazard Ratio; CI = Confidence interval.

Table 4 Clinical outcomes of patients with versus without diabetes mellitus in the propensity matched population Outcomes in diabetic patients

Predicted mortality with STS-PROM, Logistic EuroSCORE and EuroSCORE II was higher in diabetic patients than non-diabetic patients. In DM patients, predicted 30-day mortality with STS-PROM was 7.0%, with Logistic EuroSCORE 15.5% and with EuroSCORE II 4.7%. Observed 30-day mortality in diabetic patients was 5.4%. STS-PROM overestimated 30-day mortality with an observed-expected mortality ratio of 0.77. Logistic EuroSCORE overestimated 30-day mortality with an observed-expected mortality ratio of 0.35. EuroSCORE II underestimated 30-day mortality with a ratio of 1.15. Figure 2 presents an overview of these risk scores.

Fig. 2

Predicted versus observed mortality in diabetic and non-diabetic patients.

Legend: Comparison of predicted mortality (using STS-PROM, Logistic EuroSCORE, and EuroSCORE II) and observed 30-day mortality between diabetic and non-diabetic patients. Log ES = Logistic EuroSCORE; ES II = EuroSCORE II; STS = STS-PROM.

In the DM population, independent predictors for 30-day mortality were renal failure (OR 2.1, 95%CI 1.4-3.0, p < 0.001) and baseline atrial fibrillation (OR 2.2, 95%CI 1.6-3.0, p < 0.001). DM patients who had both conditions, were at increased risk for mortality (OR 3.6, 95%CI 2.1–6.1, p < 0.001), compared with patients who had none or only one of these conditions. Diabetes treatment data was available in 1015 of DM patients (29%). IDDM patients (n = 314) were younger, had a higher BMI, and lower GFR than NIDDM patients (n = 701) (Additional Table S2). Table 5 presents clinical outcomes in IDDM and NIDDM patients. There was a trend to higher rates of one-year mortality (15.0% vs. 10.5%, univariate HR 1.5, 95%CI 1.0-2.3, p = 0.08) and stroke (6.5% vs. 3.5%, RR 1.9, 95%CI 0.9–3.9, p = 0.07) in IDDM compared with NIDDM patients. Also in a multivariate model, there was a trend to higher mortality in IDDM, but this did not reach the threshold for statistical significance (HR 1.5, 95%CI 0.9–2.3, p = 0.08). Time to mortality curves for IDDM and NIDDM are depicted in Fig. 3. Rates of other clinical outcomes were similar.

Fig. 3

Time-to-mortality curves of patients with insulin dependent versus non-insulin dependent diabetes mellitus undergoing transcatheter aortic valve implantation.

Legend: The multivariate model included univariate predictors for mortality: age, body mass index, atrial fibrillation, hypertension, renal failure, and mean aortic valve gradient. IDDM = insulin dependent diabetes mellitus; NIDDM = non-insulin dependent diabetes mellitus; HR = Hazard Ratio; CI = Confidence interval.

Table 5 Clinical outcomes in insulin dependent versus non-insulin dependent diabetic patients