A total of 40 patients and 14 controls were eligible and consented to the study. Eight patients were excluded (ABI > 0.9 after exercise n=2, incompressible arteries n=3, inability to perform TT because of psychological reasons or equilibrium discomfort n=3) as was 1 'healthy' control (rest ABI <0.9). Therefore, a total of 32 patients (64 legs) and 13 controls (26 legs) were used in the analyses. There were no differences in demographic data between the two groups (table 1). ABI values obtained from patients who performed a treadmill test (TT) first (n=17) were not different from data of subjects performing a bicycle test first (n=15).
Correlations in patient legs with an ABI <0.9 (n=64)
56 of 64 patient legs demonstrated an ABI<0.9 after TT. ABI values of all 64 legs obtained after both cycling tests correlated significantly with ABIs measured after
treadmill testing (TT vs SBT, r = 0.90, p<0.001; TT vs MBT, r = 0.88, p<0.001; figure 1 & 2). The average discrepancy (bias) between the TT/SBT is 0.04±0.14 (Bland-Altman plot, Figure 3). A substantial scatter is visible throughout the whole range of measurements. However, there was no converging or diverging trend visible in the scatterplot (figure 3).
Figure 4 depicts the correlation (r = 0.66, p<0.001) between the absolute fall of ABI after a treadmill exercise (?ABI, TT) compared to the drop of ABI after a SBT (?ABI, SBT). A weak correlation was also observed between these values after ?ABI, TT and MBT (r = 0.32, p=0.01, graph not shown).
Correlations in patients legs with ABI >0.9 in rest (n=22)
22 of 64 patient legs demonstrated an ABI>0.9 in rest. ABI values of these legs obtained after both cycling tests correlated significantly with ABIs measured after treadmill testing (n=22, TT vs SBT, r = 0.77, p<0.001; TT vs MBT, r = 0.51, p<0.02).
Sensitivity and specificity
ABI values of 64 patient legs were used in a sensitivity analysis whereas ABI<0.9 after TT was used as golden standard. The SBT revealed a 98% sensitivity for the diagnosis IC and a 86% specificity with an ABI cut off value of 0.89 (AUC: 0.98; figure 5). Positive predicted values (PPV) and negative predicted values (NPV) were 98% and 86%, respectively. The MBT showed a 98% sensitivity and a 43% specificity with a 0.92 ABI cut off value (AUC 0.95; graph not shown) and a PPV of 93% and NPV of 75% respectively.
ABIs of healthy controls were always >0.9 after both bicycle tests. (n=26; ABI prior to testing: 1.16±0.10, 1th ABI measured directly after SBT: 1.09±0.10, ABI measured directly after MBT: 1.02 ± 0.12).