[Application of ROC and PR curves in the evaluation of clinical diagnostic testing].

This study reviewed the concepts and properties of the receiver operating characteristic (ROC) curve and precision recall (PR) curve, and made suggestions on the application of two curves based on the prevalence in combination with the results of simulation data. This study demonstrated that the ROC curve and PR curve had different properties, which could reflect the performance of diagnostic methods from various aspects. These two curves should be selected with a consideration of prevalence and clinical scenarios. When the prevalence was less than 20%, especially less than 5%, the PR curve could be adopted.

[Effects of nasogastric and percutaneous endoscopic gastrostomy tube feeding on the susceptibility of pulmonary infection in long-term coma patients with stroke or traumatic brain injury].

Objective: To investigate the effects of nasogastric tube (NGT) and percutaneous endoscopic gastrostomy (PEG) tube feeding on the susceptibility of pulmonary infection in long-term coma patients with stroke or traumatic brain injury. Methods: A total of 295 candidates who were in long-term coma after stroke or traumatic brain injury but without pulmonary infection and eligible for PEG catheterization were screened prospectively between January 2014 and February 2018. The patients were divided into PEG group (86 patients) and NGT group (209 patients) according to the choice of next-of-kin. Data related to the susceptibility of pulmonary infection were collected and analyzed in the two groups one month after the catheterization. Results: After follow-up for one month, compared with NGT group, patients in PEG group had a lower incidence of pulmonary infection (23.3% vs 37.8%, P=0.023), a later occurrence of pulmonary infection (average time: 21 days vs 13 days, P=0.034), and a less severe pulmonary infection (7.0% vs 12.9%, P=0.029). The following characteristics made patients more susceptible to pulmonary infection: age ≥ 70 years (HR=1.619, 95% CI 1.054-2.172), Charlson comorbidity index ≥ 2 points (HR=1.647, 95% CI 1.043-2.485), using of proton pump inhibitor ≥ 7 days (HR=1.725, 95% CI 1.214-2.738), and number of pressure ulcers ≥ 3 (HR=2.109, 95% CI 1.128-3.844). However, serum albumin concentration ≥35 g/L (HR=0.670, 95% CI 0.375-0.963) was a protective factor for pulmonary infections. The number of consistent pathogens cultivated from saliva, gastric juice and sputum simultaneously in NGT and PEG group was 35 strains (27.8%) and 8 strains (13.3%), respectively (P=0.029). The mortality of pulmonary infection was similar in the two groups (3.5% vs 4.3%, P=1.000), but the death due to pulmonary infection in the PEG group occur later (median time: 20 days vs 11 days, P=0.012). Conclusions: PEG feeding was a preferred nutrition way which could reduce the risk of pulmonary infection more effectively than NGT feeding which might favored a retrograde gastro-pulmonary route by which pathogens colonized in stomach migrated to respiratory tract. Patients with characteristics mentioned above had the susceptibility of pulmonary infection, thus risk assessment of pulmonary infection should be conducted before selecting the catheterization method.

[Significance of monitoring the gradients between transcutaneous PCO(2) and end-tidal PCO(2) in patients with septic shock].

Objective: To investigate the significance of monitoring the gradients between transcutaneous PCO(2) and end-tidal PCO(2) [P(c-et)CO(2)] in patients with septic shock. Method: Thirty-five mechanically ventilated patients with early septic shock were enrolled as the study group and 18 non-septic shock patients with stable hemodynamics as the control group between May 2014 and October 2016. The patients with septic shock were treated by early goal-directed therapy (EGDT) within 6 hours since hospitalization. The differences of baseline level of P(c-et)CO(2) and arterial lactate concentration (LAC) between the two groups and the variations of these indexes after EGDT in the study group were compared respectively. Results: The baseline levels of P(c-et)CO(2) and LAC in patients with septic shock were significantly higher than those of the control group [(26.0±16.2) mmHg vs (11.0±5.6) mmHg (1 mmHg=0.133 kPa) and (4.0±1.7) mmol/L vs (1.6±0.6)mmol/L, all P=0.000]. The area under receiver operator characteristic (ROC) curve (AUC) for baseline P(c-et)CO(2) and LAC was 0.924 (95%CI: 0.851-0.996) and 0.931 (95%CI: 0.872-1.000), respectively. P(c-et)CO(2) >12.6 mmHg and LAC >2.5 mmol/L could discriminate septic shock patients from those without shock with the same sensibility of 97% and the specificity of 83% and 78% respectively. With regard to the prognosis (Day 28) of the patients with septic shock, AUC for baseline P(c-et)CO(2) and LAC was 0.709 (95%CI: 0.533-0.886) and 0.714 (95%CI: 0.545-0.883), respectively. P(c-et)CO(2) >20.0 mmHg and LAC>3.6 mmol/L could discriminate survivors from non-survivors with the same sensibility of 92% and the same specificity of 76%. All the patients in the study group completed EGDT within 6 hours after admission, 20 (57.1%) passed EGDT and 17 (85.0%) survived, 15 (42.9%) failed EGDT and 4 (26.7%) survived, and the survival rates were significantly different (F=9.844, P=0.001). After EGDT, P(c-et)CO(2) (21.0±9.5 mmHg) and LAC(3.3±2.5 mmol/L)reduced significantly compared with the baselines (P=0.008 and P=0.046), and the associated AUC was 0.905(95%CI: 0.792-1.000) and 0.747 (95%CI: 0.576-0.917)respectively. P(c-et)CO(2) > 16.5 mmHg and LAC > 3.1 mmol/L could discriminate survivors from non-survivors with the sensibility of 97% and 91%, and the specificity of 78% and 69%, respectively. Conclusions: P(c-et)CO(2) >12.6 mmHg could play the same role as LAC in recognizing early septic shock. EGDT was an effective therapy for the septic shock and P(c-et)CO(2) reflected efficacy. P(c-et)CO(2)>20 mmHg before EGDT and >16.5 mmHg after EGDT both could predict the 28 d prognosis of patients with septic shock, and the effect of the former was equal to that of LAC, but the latter was better than LAC.