Characteristics of patients

The study enrolled a total of 373 patients, of whom 233 (63%) were male, and the median age was 62 years (IQR: 54–69). Among the study population, 305 patients (81.8%) underwent valve surgeries, 35 patients (9.4%) underwent aortic root surgeries, 19 patients (5.1%) underwent left ventricular outflow tract surgeries, and 14 patients (3.8%) underwent other types of surgeries. Besides, 356 (95.4%) remained on mechanical ventilation support, with a median duration of 2 days (IQR: 1–6). Postoperatively, 292 patients (78.3%) received blood transfusions. Additionally, 253 patients (67.8%) had a postoperative LVEF less than 50%, and 165 patients (44.2%) received vasopressor support, with 39 patients (10.5%) specifically receiving vasopressin support.

Patients with skin mottling

Among the total 373 patients in the dataset, 13 (3.5%) patients presented with skin mottling (Fig. 1.A). The severity of skin mottling varied among patients, with 5, 1, 2, 2, and 3 patients having SMS score ranging from 1 to 5, respectively. In comparison to patients without skin mottling, those with skin mottling demonstrated lower SBP (45 vs. 57 mmHg, p = 0.013), MAP (65 vs. 73 mmHg, p = 0.049), and UO (0.1 vs. 1.2 ml/kg/h, p < 0.001) (Table 1). They also had a higher HR (100 vs. 87, p = 0.042), lactate (6.0 vs. 2.2 mmol/L, p = 0.001), Δlactate (1.0 vs. 0 mmol/L, p = 0.001), VIS (32 vs. 7 µg/kg/min, p = 0.015), and CRT (3.1 vs. 1.2 s, p < 0.001) (Table 1). In the subpopulation of 178 cases of hypotension, the occurrence of skin mottling was higher, at 5.6% (Fig. 1.A). In addition, in the more severe population of 51 shock patients, the occurrence of skin mottling increased significantly to 15.7% in the 51 shock patients (Fig. 1.A).

Table 1 Comparison of clinical characteristics of patients with and without skin mottlingThe performance of SMS in detecting peripheral hypoperfusion

Tables 2 and Fig. 1.B demonstrated the diagnostic performance of SMS for peripheral hypoperfusion. Of the total population of 373 cases, 24 (6.4%) patients had a significant prolonged CRT. The AUROC of SMS in detecting peripheral hypoperfusion was 0.64 (95% CI: 0.59–0.69). The optimal threshold of SMS was 1, corresponding to sensitivity, specificity, PPV, and NPV of 29 (95% CI: 13–51), 98 (95% CI: 96–99), 54 (95% CI: 25–81), and 95 (95% CI: 93–97), respectively. In the hypotensive population, 13 (7.3%) patients had peripheral hypoperfusion and the AUROC was mildly elevated to 0.68 (95% CI: 0.61–0.75), corresponding to an increase in sensitivity to 38 (95% CI: 14–68). In the shock population, the proportion of patients with peripheral hypoperfusion increased to 11.8% and the corresponding AUROC and sensitivity increased substantially to 0.81 (95% CI: 0.673–0.90) and 67 (95% CI: 22–96), while the specificity remained very high at 91 (95% CI: 79–98). We also note that the NPV consistently stayed above 90, while the PPV hovered at a low level of 50, regardless of how the population was defined.

Table 2 The accuracy of SMS for detecting peripheral hypoperfusion or predicting hospital mortalityCorrelation among SMS and conventional perfusion parameters

Table 3 described the correlation coefficients between conventional perfusion parameters for SMS as well as odds ratios (OR) of conventional perfusion parameters for the presence of skin mottling. For the whole population, significant correlations were found for UO, lactate, CRT and VIS and SMS with correlation coefficients of -0.19, 0.28, 0.5 and 0.39 respectively. For subpopulations with hypotension or shock, the correlation coefficients between the most of above parameters and SMS tended to increase. According to univariate logistic regression, all parameters except Δlactate were associated with the risk of developing skin mottling. The best subset method identified UO and CRT as the perfusion parameters most closely associated with the presence of skin mottling, with adjusted ORs of 0.03 (95% CI: 0.01–0.17) and 2.15 (95% CI: 0.02–0.41), respectively. The combination of these two parameters determined the appearance of skin mottling with a C-index of 0.93 (95% CI: 0.86–1.00).

Table 3 Correlation of conventional circulatory parameters with SMSSkin mottling and mortality risk

The hospital mortality of overall patients was 14.7%, but it differed significantly between patients with and without skin mottling (84.6% vs. 12.2%, p < 0.001). In terms of survival curve, patients with skin mottling had a much worse survival (Fig. 2). In addition, the mortality risks were significantly higher in hypotension and shock populations, reaching 21.3% and 37.3%. For the overall population, the relative risk of skin mottling for mortality was 6.9 (95% CI: 4.8–9.9) and the odds ratio (OR) of SMS for mortality was 3.74 (95 CI: 1.61–8.65). The AUROCs of SMS for predicting hospital mortality were 0.60 (95 CI: 0.55–0.65), 0.63 (95 CI: 0.56–0.70) and 0.71 (95 CI: 0.57–0.83) in the overall, hypotension and shock populations, respectively (Table 1and Fig. 1.C). The optimal thresholds identified by the Youden index were all 1, corresponding to specificities of nearly 100, but a low sensitivity of 23 in the overall population, 26 in the hypotension population, and only 42 even in the shock population (Table 1).

Fig. 2

The survival curves for patients with and without skin mottling