General clinical characteristics
After the exclusion of 31 patients, the study enrolled 140 patients (Figure 1), including 98 men (70.00%) and 42 women (30.00%), with an average age of 49.88 ± 13.94 years. In terms of etiology, there were seven post-endoscopic retrograde cholangio-pancreatography cases, 66 gallstone cases, 28 hyperlipidemia cases, and 39 idiopathic AP cases. Of the 140 AP patients, 60 (42.86%) underwent re-operation, including 40 cases of debridement, 13 of hemostasis, and seven of debridement combined with hemostasis. Complications occurred after debridement in 90 patients (64.29%), including sepsis in 37 patients, electrolyte disorders in 31 patients, respiratory failure in 40 patients, hemorrhage in 34 patients, shock in 36 patients and renal failure in 34 patients. Finally, 16 patients died during the perioperative period (11.43%), as shown in Table 1.
Distribution of the Cr/Alb and CRP/Alb ratios in groups of re-operation and mortality
The admission CRP/Alb was only significantly higher in the re-operation group (P < 0.05). The distributions of the preoperative CRP/Alb values showed no significant differences between both subgroups (P > 0.05). The admission Cr/Alb was significantly higher in the re-operation group (P < 0.001) and mortality group (P = 0.011). The preoperative Cr/Alb was significantly higher only in the mortality group (P < 0.01), as shown in Table 2. The APACHE II score was significantly higher both in the re-operation and mortality groups, with P < 0.05, similar to the admission Cr/Alb. As a result, admission Cr/Alb and CRP/Alb and preoperative Cr/Alb were chosen for further analysis.
Correlation analysis between the ratios and general clinical statistics of the AP patients
Table 3 shows the correlation between the three ratios and the patients’ general clinical characteristics. The admission CRP/Alb was correlated with admission Cr (r = 0.243, P = 0.013), admission CRP (r = 0.959, P < 0.001), admission Alb (r = 0.284, P = 0.022), preoperative CRP (r = 0.295, P = 0.017), admission Cr/Alb (r = 0.241, P = 0.014), and intensive care unit (ICU) duration (r = 0.294, P = 0.003). The admission Cr/Alb was correlated with the APACHE II score (r = 0.495, P < 0.001), admission blood urea nitrogen (r = 0.815, P < 0.001), admission Alb (r = -0.221, P = 0.009), admission Cr (r = 0.983, P < 0.001), preoperative Cr (r = 0.375, P = 0.002), admission CRP/Alb (r = 0.241, P = 0.014), preoperative Cr/Alb (r = 0.428, P < 0.001), and post-operative complications (r = 0.382, P = 0.002). The preoperative Cr/Alb was correlated with admission blood urea nitrogen (r = 0.326, P = 0.010), admission Cr (r = 0.412, P = 0.001), admission Cr/Alb (r = 0.428, P < 0.001), and ICU duration (r = -0.254, P = 0.040). Of those ratios, admission Cr/Alb presented the greatest application value through its correlation with most of the eight variables, including the other two ratios, simultaneously.
Predictive value of Cr/Alb for re-operation and mortality
Figures 2 and 3 show the ROC curve of the admission Cr/Alb in the prediction of re-operation and mortality. According to the ROC curve, we calculated the predictive values including cut-off values, sensitivities, specificities, and AUCs (Table 4).
The admission CRP/Alb showed the lowest efficacy, with sensitivity, specificity and AUC values of 76.1%, 58.3% and 0.691 for re-operation, and 49.5%, 75.0% and 0.635 for mortality, respectively. The admission Cr/Alb had a better predictive value, with a sensitivity of 86.3%, specificity of 61.7%, and AUC of 0.724 for re-operation, and a sensitivity of 73.4%, specificity of 81.3% and AUC of 0.794 for mortality. The predictive value of the preoperative Cr/Alb for mortality was more meaningful, with a sensitivity of 100.0%, specificity of 72.1% and AUC of 0.872, with a cut-off of 2.57. Additionally, its predictive value for re-operation was also satisfactory, with sensitivity, specificity and AUC values of 55.6%, 68.4% and 0.606 respectively. In comparison, the sensitivity, specificity and AUC values of the APACHE II score were 82.6%, 58.3 and 0.716 for re-operation, and 50.9%, 92.3% and 0.713 for mortality, respectively. In general, the predictive values of the Cr/Alb ratio and APACHE II score were basically the same.
Relationship between admission Cr/Alb and clinical characteristics
As suggested above, the cut-off admission Cr/Alb value was 3.29 for re-operation and 3.43 for mortality. The latter was chosen as the basis of grouping, as a ratio higher than 3.43 simultaneously satisfies two criteria. Of patients in the milder AP (n=94) and severe AP (n=46) group, those in the latter group with a Cr/Alb value ≥ 3.43 had a more intense heart rate and respiratory rate (P < 0.05), impaired laboratory data, including those pertaining to blood urea nitrogen, Cr, Alb, white blood cells, platelets and the APACHE II score (P < 0.05), higher morbidity values associated with post-operative complications, including electrolyte disorders, respiratory failure, hemorrhage, shock and renal failure (P < 0.05), and higher rates of re-operation and mortality (P < 0.001). Besides, the interval times between onset and debridement were shorter in the severe group but still exceeded 3-4 weeks, in keeping with the Atlanta Classification criteria for AP. These results suggest that a cut-off admission Cr/Alb value of 3.43 has the potential to distinguish severe AP from milder AP.
Logistic regression analysis of the admission Cr/Alb in AP patients
In the logistic regression analysis of the admission Cr/Alb in AP patients, the APACHE II score, sepsis, electrolyte disorder, respiratory failure, hemorrhage, shock, renal failure, re-operation and mortality were enrolled after the elimination of interference variables through colinear diagnosis. The results suggested that the APACHE II score, renal failure and re-operation were independently related to the admission Cr/Alb (P < 0.05), as shown in Table 6. This indicated that the admission Cr/Alb could predict the risk of re-operation and post-operative renal failure independently. AP patients with high admission Cr/Alb values had 4.331 times the chance of renal failure and 3.824 times the chance of re-operation. Besides, the admission Cr/Alb was correlated independently to the APACHE II score (P=0.009), suggesting that the ratio could be a simpler substitute for the APACHE II score in clinical settings.