During the study period, 154 patients (mean age was 69 ± 10 years) underwent laparotomy after enhanced abdominal CT scan. Laparotomy was performed within 24 hours of the CT scan in 139 (92%) patients, one day later in 11 (7%), and two days later in 4 (3%). Overall, 110 patients (71%) underwent digestive resection. Among the remaining 44 patients, there was no appearance of transmural necrosis in 29 (19%), and extensive necrosis with open-close surgery in 15 (10%) (Figure 1). NOMI were secondary to surgery in 87/154 patients (56%) and to medical conditions in the remaining 67/154 patients (44%). Cardiac surgery was the main surgical etiology of NOMI (n=36, 23%) followed by abdominal surgery (n=27, 18%) and by aortic surgery (n=14, 9%). Sepsis was the main medical etiology of NOMI (n=23, 15%), followed by cardiorespiratory arrest (n=15, 10%), acute respiratory failure (n=14, 6%) and Hemorrhage (n=10, 6%). The mean time between ICU admission and abdominal CT was 2.7 days (range 0 – 33 days). Among the 154 patients, 87 (56%) died within 28 days after the abdominal CT.
Variables associated with 28-day mortality
Variables associated with 28-day mortality are summarized in Table 1. Regarding the variables at ICU admission, mean arterial pressure was lower in patients who died than in survivors (p< 0.001). Concerning the variable at the time of abdominal CT scan, patients who died within 28 days were more likely to have catecholamine infusion (p< 0.001). Platelet count and prothrombin rate were lower in patients who died (p=0.003 and p<0.001 respectively). Lactate levels, bilirubin and ASAT were higher in those who died (p< 0.001, p=0.008 and p<0.001 respectively).
The CT features associated with 28-day mortality are shown in Table 1. All the CT features suggestive of organ infarction were associated with death. Absence of small bowel enhancement, kidney infarction, liver and spleen infarction were more frequent in those who died (p = 0.008, p < 0.001, p = 0.002, p = 0.003 respectively). Axial diameter of the mesentery arteries was significantly lower in patients who died (p = 0.007 and p< 0.001 respectively). Portal venous gas was not associated with 28-days mortality (16 [18%] in survivors vs 7 [10%] in non survivors, P=0.253).
SMA diameter was not correlated with catecholamine infusion or noradrenaline flow (Pearson correlation coefficient, r = - 0.09, 95%CI [0.25 - 0.06] and -0.09, 95%CI [-0.25 - 0.07] respectively), whereas celiac trunk diameter was (r = - 0.27; 95%CI [-0.41 - -0.12] and -0.22, 95%CI [-0.36 - -0.06] for catecholamine infusion and noradrenaline flow respectively). Kidney infarction was not correlated with creatinine level (r = -0.01, 95%CI [-0.17 - 0.15]).
Pre-operative prognostic score for 28-day mortality
After multivariate analysis for the prediction of 28-day mortality including variables at the time of CT, a simple mortality score was established (Table 2). For each variable associated with 28-day mortality (i.e. lactates > 7 mmol/l, prothrombin rate < 60 % and kidney infarction), 1 point was attributed. Among the study population, probability of 28-day mortality was 26% (31/42), 54% (26/48), 77% (23/30) and 100% (21/21) for a mortality score of 0, 1, 2 and 3, respectively (Figure 2). The AUC for 28-day mortality prediction was 0.79, 95%CI [0.72-0.86]. The results of the Cox regression model including continuous variables (i.e. Lactates and prothrombin rate) with the nomogram built according to the four significant variables (i.e. Lactates, prothrombin rate, catecholamine infusion and kidney infarction), together are provided in Supplementary Table 1 and Supplementary Figure 1 respectively. The comparison of the AUC between the nomogram and the NOMI mortality score for 28 days mortality is available in the Supplementary Figure 2.
Influence of operative findings on 28-day mortality
The 15 patients with extensive necrosis and without bowel resection (i.e. open-close surgery) died within 24 hours. 6 of them (40%) had a mortality score of 3. Median survival in the 29 patients without surgical necrosis was 8 days, 95%CI [3 – 21 days] and 64 days, 95%CI [15 – 241 days] in the patients with bowel resection (Supplementary Figure 3). Both stomach necrosis visible at surgery, and transmural necrosis diagnosed by pathology were associated with 28-day mortality (p = 0.01, Table 1). Macroscopic small bowel necrosis appearance was not associated with 28-day mortality (40 [46%] in non survivors vs 26 [39%] in survivors, P=0.2, whereas small bowel transmural necrosis was (38 [44%] in non survivors vs 17 [25%] in survivors, P=0.002). Macroscopic jejunal necrosis appearance was associated with 28-day mortality (26 [30%] in non survivors vs 8 [12%] in survivors, P<0.001, as jejunal transmural necrosis was (25 [29%] in non survivors vs 7 [10%] in survivors, P<0.001). Macroscopic ileal necrosis appearance was not associated with 28-day mortality (32 [37%] in non survivors vs 22 [33%] in survivors, P=0.2, whereas small bowel transmural necrosis was (31 [36%] in non survivors vs 13 [19%] in survivors, P=0.002). Patients with small bowel transmural necrosis had a median survival of 2 days, 95%CI [1 - 19 days] vs 31 days, 95%CI [13 – 80 days] in patients without.
In the 110 patients with bowel resection, only small bowel transmural necrosis was associated with 28-day mortality (median survival 10 days, 95%CI [2 – 234 days] vs 80 days 95%CI [36 vs 1256 days]).
Overall multivariate analysis of prognostic factors in NOMI
The prognostic factors for 28-day mortality identified by multivariate analysis among the variables at ICU admission, at the time of the CT, and operative findings are provided in Table 3. Age was associated with an increased risk of death (HR = 1.03 CI95%[1.00 – 1.06], P = 0.025). Lactates >7 mmol/L (HR = 2.99 CI95%[1.63 – 5.47], P < 0.001), kidney infarction (HR = 1.88 CI95%[1.10 – 3.22], P = 0.021), and SMA diameter ≤ 5 mm (HR = 1.77 CI95%[1.04 – 3.01], P = 0.034) were independent risk factors for death within 28 days. Among all the operative findings associated with death by univariate analysis, only jejunal transmural necrosis was an independent risk factor for 28-day mortality (HR = 2.26 CI95%[1.14 – 4.71], P = 0.019).