Comparison of different prognostic data: there was no significant difference in sex, age, time from poisoning to admission, underlying disease, systolic blood pressure, leukocyte, neutrophil, lymphocyte, hematocrit, creatinine, albumin and Na + in patients with different outcome. Compared with the group with good outcome, the levels of NSD, ONSD/ETD, NSE, CPK, lactic acid, PSS and APACHEII in patients with poor outcome were higher than those in the group with good outcome. The differences were statistically significant (P < 0.05). The comparison of clinical indexes of patients with different outcome is shown in Table 1.
Table 1
comparison of patients with different outcome
Variables | Good outcome (n = 108) | Bad outcome (n = 38) | t/Z/χ2 | P value |
Sex(Male/Female) | 62/46 | 21/17 | 0.053 | 0.818 |
Age(years) | 53.85 ± 16.62 | 54.47 ± 18.33 | 0.193 | 0.847 |
ONSD | 5.22 ± 0.47 | 5.74 ± 0.52 | 5.693 | < 0.001 |
ONSD/ETD | 22.85 ± 2.08 | 24.99 ± 2.52 | 5.161 | < 0.001 |
Poisoning until admission time | 5(4,7) | 6(4,8) | -1.279 | 0.201 |
Basis of disease | 22(20.4) | 9(23.7) | 0.185 | 0.667 |
Systolic blood pressure | 129.63 ± 16.66 | 134.55 ± 21.4 | 1.450 | 0.149 |
White blood cells | 10.79 ± 4.12 | 12.07 ± 5.21 | 1.529 | 0.128 |
Neutrophils | 79.6 ± 13.98 | 81.35 ± 13.5 | 0.671 | 0.503 |
Lymphocyte | 11.3(7.25,20.63) | 10.6(6.4,20.1) | -1.436 | 0.151 |
Hematocrit | 40.64 ± 4.51 | 42 ± 7.02 | 1.113 | 0.271 |
Serum creatinine | 55.09 ± 16.03 | 58.84 ± 17.57 | 1.210 | 0.228 |
Albumin | 38.73 ± 3.9 | 37.51 ± 3.34 | 1.712 | 0.089 |
Na | 138.53 ± 2.76 | 138.82 ± 3.73 | 0.430 | 0.669 |
CPK | 100.5(71.25,229.75) | 252.5(110.25,957) | -2.999 | 0.003 |
Lactic acid | 2.35(1.5,5.18) | 3.9(2.1,5.33) | -2.427 | 0.015 |
PSS | 2(2,3) | 4(3,6) | -3.813 | < 0.001 |
APACHEII | 6(5,9) | 10(7.75,17) | -5.878 | < 0.001 |
NSE | 17.3 ± 6.13 | 23.04 ± 6.8 | 4.821 | < 0.001 |
Analysis of related factors affecting the outcome of the patients: taking the outcome of the patients (good = 0, bad = 1) as dependent variables, sex, age, ONSD, ONSD/ETD, time from poisoning to admission, underlying disease, systolic blood pressure, white blood cells, neutrophils, lymphocytes, hematocrit, creatinine, albumin, Na+, CPK, lactic acid, PSS, APACHEII, NSE as independent variables, the binary logistic regression equation was established. The results showed that there were significant correlations between ONSD, ONSD/ETD, CPK, lactic acid, PSS, APACHEII, NSE and outcome. The results of univariate logistic regression analysis of patients with poor outcome are shown in Table 2
Table 2
logistic regression analysis of univariate factors affecting poor outcome of patients
Variables | B | SE | Waldχ2 | P | OR(95% CI) |
Sex(Male/Female) | 0.087 | 0.380 | 0.053 | 0.818 | 1.091(0.518 ~ 2.297) |
Age(years) | 0.002 | 0.011 | 0.038 | 0.846 | 1.002(0.98 ~ 1.024) |
ONSD | 2.563 | 0.564 | 20.677 | 0.000 | 12.975(4.299 ~ 39.162) |
ONSD/ETD | 0.470 | 0.109 | 18.616 | 0.000 | 1.599(1.292 ~ 1.979) |
Poisoning until admission time | 0.016 | 0.043 | 0.147 | 0.702 | 1.016(0.935 ~ 1.105) |
Basis of disease | 0.193 | 0.450 | 0.184 | 0.668 | 1.213(0.502 ~ 2.932) |
Systolic blood pressure | 0.015 | 0.010 | 2.057 | 0.152 | 1.015(0.995 ~ 1.036) |
White blood cells | 0.062 | 0.041 | 2.261 | 0.133 | 1.064(0.981 ~ 1.154) |
Neutrophils | 0.010 | 0.015 | 0.452 | 0.501 | 1.01(0.981 ~ 1.039) |
Lymphocyte | -0.008 | 0.018 | 0.179 | 0.673 | 0.993(0.959 ~ 1.028) |
Hematocrit | 0.051 | 0.037 | 1.835 | 0.176 | 1.052(0.978 ~ 1.132) |
Serum creatinine | 0.013 | 0.011 | 1.430 | 0.232 | 1.013(0.992 ~ 1.035) |
Albumin | -0.093 | 0.055 | 2.857 | 0.091 | 0.911(0.819 ~ 1.015) |
Na | 0.031 | 0.062 | 0.248 | 0.618 | 1.032(0.913 ~ 1.166) |
CPK | 0.696 | 0.282 | 6.086 | 0.014 | 2.005(1.154 ~ 3.485) |
Lactic acid | 0.222 | 0.081 | 7.467 | 0.006 | 1.249(1.065 ~ 1.465) |
PSS | 0.434 | 0.112 | 14.931 | 0.000 | 1.543(1.238 ~ 1.922) |
APACHEII | 0.325 | 0.059 | 30.355 | 0.000 | 1.384(1.233 ~ 1.553) |
NSE | 0.135 | 0.032 | 17.855 | 0.000 | 1.145(1.075 ~ 1.219) |
CPK, Phosphocreatine kinase; PSS, Poisoning severity score; APACHEII, Acute Physiology and chronic Health scoring system II; NSE, Neuron-specific enolase |
The binary logistic regression equation was established by conditional forward method with poor outcome (no = 0, is = 1) as dependent variable and ONSD, ONSD/ETD, CPK, lactic acid, PSS, APACHEII, NSE as independent variables. The results showed that ONSD, ONSD/ETD, lactic acid, APACHEII and NSE were statistically significant in the model, and all of them were risk factors for poor outcome (OR > 1, P < 0. 05). The results of logistic regression analysis of multiple factors affecting the poor outcome of the patients are shown in Table 3.
Table 3
logistic regression analysis of multiple factors affecting poor outcome of patients
Variables | Model1 | | Model2 |
OR(95% CI) | Waldχ2 | P | OR(95% CI) | Waldχ2 | P |
ONSD | 9.404(2.612 ~ 33.857) | 11.757 | 0.001 | | - | - | - |
ONSD/ETD | - | - | - | | 1.479(1.14 ~ 1.919) | 8.699 | 0.003 |
Lactic acid | 1.343(1.017 ~ 1.774) | 4.323 | 0.038 | | 1.317(1.011 ~ 1.716) | 4.163 | 0.041 |
APACHEII | 1.400(1.213 ~ 1.615) | 21.305 | 0.000 | | 1.390(1.208 ~ 1.600) | 21.117 | 0.000 |
NSE | 1.140(1.043 ~ 1.247) | 8.268 | 0.004 | | 1.154(1.056 ~ 1.26) | 10.056 | 0.002 |
Model 1 includes ONSD, ONSD/ETD, CKM, lactic acid, PSS, APACHE II, NSE; Model 2 into ONSD, ONSD/ETD, CKM, lactic acid, PSS, APACHE II, NSE. |
From the ROC curve, the areas under the curve for ONSD, ONSD/ETD, lactic acid, APACHEII, NSE and predicting severity were 0.790, 0.777, 0.633, 0.820 and 0.736 respectively, and the corresponding P values were all less than 0.05, indicating that the areas under the curve for predicting poor outcome were statistically significant (Fig. 1.1). Among them, the area under the curve of ONSD model is the largest. The sensitivity and specificity of ONSD > 5.675 mm in predicting poor outcome were 60.5% and 89.8%, respectively (P < 0.001). The sensitivity and specificity of ONSD/ETD > 0.244 in predicting poor outcome were 73.7% and 84.3%, respectively (P < 0.001); NSE > 23.1 ng/ml, the sensitivity and specificity of predicting poor outcome were 57.9% and 78.7%, respectively (P < 0.001). The sensitivity and specificity of APACHEII > 10.5 in predicting the outcome of neurological dysfunction were 65.8% and 88.8%, respectively (P < 0.001). The sensitivity and specificity of ONSD/ETD > 0.244 in predicting poor outcome were 73.7% and 84.3%, respectively (P < 0.001). Draw ROC curves according to the established model (Fig. 2.2 )the area under the ROC curve of the, ONSD prediction model (AUC) is 0.925, the best prediction probability critical value is 0.3585, the sensitivity is 89.5%, and the specificity is 94.4%. The ONSD/ETD prediction model of adverse outcome was obtained by the same method, the AUC was 0.916, the critical value of the best prediction probability was 0.3547, the sensitivity was 89.5%, and the specificity was 92.6%. ONSD has higher sensitivity and specificity for poor outcome. The results of ROC curve analysis of each index diagnosed as poor outcome are shown in Table 4.
Table 4
ROC curve of poor outcome diagnosed by each index
| AUC(95% CI) | P | Out-off value | Sensitivity | Specificity |
ONSD | 0.790(0.698 ~ 0.882) | < 0.001 | 5.675 | 60.5 | 89.8 |
ONSD/ETD | 0.777(0.680 ~ 0.875) | < 0.001 | 24.44 | 73.7 | 84.3 |
Lactic acid | 0.633(0.536 ~ 0.729) | 0.015 | 1.85 | 89.5 | 38.0 |
APACHEII | 0.820(0.730 ~ 0.910) | < 0.001 | 10.5 | 65.8 | 88.9 |
NSE | 0.736(0.639 ~ 0.832) | < 0.001 | 23.1 | 57.9 | 78.7 |
ONSD model | 0.925(0.856 ~ 0.995) | < 0.001 | 0.3585 | 89.5 | 94.4 |
ONSD/ETD model | 0.916(0.840 ~ 0.991) | < 0.001 | 0.3547 | 89.5 | 92.6 |