| Part A:Estabilishment of early warning model for severe H7N9avian influenza patientsBackground:The high mortality of avian influenza H7N9in humans has caused a great concern in China. The novel H7N9virus caused acute onset and rapid progression. Within a short time, acute respiratory distress syndrome (ARDS), shock and even multiple organ dysfunction syndrome (MODS) appears, which cause high rate of death. According to published researches, the median time from onset of illness to ARDS were7days; from the onset to death was14day. The mortality was up to39.2%. Thus, how to select patients who will develop to sever cases and make right medical decision is a challenge for frontline clinicians.ObjectTo identify the risk factors which may precipitate patients to ARDS or even death, we collected and analyzed the epidemiological characteristics, clinical features, pathological and physiological indicators of H7N9avian influenza patients. Then we attempt to establish early warning model, which is especially suitable for clinician to select critically ill patients and fatal cases from avian influenza H7N9patients and make right medical decision. Furthermore, we also want to evaluate the value of APACHE II score in prediction of the prognosis of patients with avian influenza A H7N9. Methods:A case-control retrospective study of165laboratory confirmation cases with avian influenza A H7N9was conducted. Univariate analysis and multivariate Logistic regression analysis were used to identify independent predictors of ARDS and death. Chi-squared Automatic Interaction Detection was used to weight the risk factors. A H7N9score system of early warning for ARDS was formulated, ROC analysis was also performed to evaluate the accuracy of this early warning score system. Meanwhile, we use acute physiology and chronic health evaluation Ⅱ (APACHE Ⅱ) score to assess the degree of the illness and the prognosis for H7N9patients. ROC analysis was also performed to evaluate the accuracy of APACHE Ⅱ score. All analyses were performed with the use of SPSS software for Windows (version18.0).Results:Univariate analysis showed that risk factors included age, comorbidity, the median time from onset to admission, the median time from onset to confirmation, the median time from onset to antiviral therapy, dyspnea, pneumonia, lesion involved two lungs, Laboratory tests on admission (including absolute neutrophil count and lymphocyte count, Hemoglobin level, HCT level, AST level, SCr level, CK level, LDH level, CRP level). Multivariate Logistic regression showed increased age (OR1.083,95%CI1.041-1.127, p=0.000), absolute neutrophil count (OR1.417,95%CI1.052-1.908,p=0.022), plasma level of SCr (OR1.038,95%CI1.007-1.071, p=0.016) and LDH (OR1.005,95%CI1.002-1.007,p=0.000) were independent predictors for development to ARDS, but Hemoglobin (OR0.968,95%CI0.940-0.997,p=0.028) was protective factors for ARDS. H7N9-ARDS prediction model (including LDH level, age, SCr level), established by decision tree technology of CHAID method, showed a high accuracy and can be used for prediction higher risk of ARDS in avian influenza A H7N9patients(with a AUC value of0.892, sensitivity value of96.1%, specificity value of55.9%).In analysis of risk factors for death, we found age, comorbidity, the median time from onset to confirmation, the median time from onset to antiviral therapy, dyspnea, ARDS, rhabdomyolysis, shock, acute kidney injury and laboratory tests on admission (including absolute lymphocyte count, HCT level, ALT level, AST level, SCr level, CK level, LDH level, CRP level) were risk factors for death. Multivariate Logistic regression also showed increased age (OR1.057,95%CI1.023-1.093, p=0.001), the median time from onset to confirmation (OR1.067,95%CI1.013-1.124, p=0.014), plasma level of LDH (OR1.002,95%CI1.001-1.004, p=0.018), shock (OR5.516,95%CI2.024-15.031,p=0.001)and acute kidney injury(OR3.583,95%CI1.419-9.045, p=0.007) are independent predictors of development to death in these patients.Using APACHE Ⅱ score to predict the death risk of avian influenza A H7N9patients had an AUC of0.837(p=0.000,95%CI:0.774-0.930), for a cut-off of20point or more, the sensitivity and specificity to predict were92.3%and68.4%, respectively.Conclusions:Increased age, absolute neutrophil count, SCr level and LDH level were independent risk factors for development to ARDS in avian influenza A H7N9patients, but hemoglobin was protective factor for ARDS. Increased age, the median time from onset to confirmation, plasma level of LDH, shock and acute kidney injury were independent predictor for development to death. APACHE II score can be used to predict death risk of these patients. H7N9-ARDS prediction model (including LDH level, age, SCr level) had a high sensitivity and was easy to use. It is suitable for clinician to practice. Part B:Clinical study of Bacteria Co-infection and The Value of Procalcitonin for Early Diagnosis in Avian Influenza A (H7N9) Patients.Background:Bacterial co-infection in people suffering from influenza is a key element that promotes severe disease and mortality. Patients contracting influenza A (H7N9) often develop severe disease. However, information on the contribution of bacterial co-infection to the severity of avian influenza A (H7N9) is limited.Objectives:We aimed to investigate the incidence of bacterial co-infection, the commonest pathogen, and an accurate diagnosis marker before obtaining positive cultures for this potential fatal disease and to determine which patients should receive antibiotic therapy at the time of admission. Meanwhile, we analyzed PCT and CRP levels for early diagnosis of bacterial co-infection in influenza A(H7N9) infection patients.Methods:A retrospective study was conducted in83patients with confirmed avian influenza A (H7N9) infection from April2013to February2014in our hosipital. Data on bacterial co-infection within the first72hours of admission were collected. The severity of patients with bacterial co-infection and those without was compared. Meanwhile we analyzed markers for early diagnosis of bacterial co-infection in influenza A (H7N9) infection patients.Results:1, Bacterial co-infection was confirmed in16of83patients (19.3%) based on positive cultures, and Staphylococcus aureus (4/16,25%) was the most prevalent pathogen and75%was MRSA(3/4).2, Higher Acute Physiology and Chronic Health Evaluation Ⅱ score (25.63±5.30vs18.57±8.27, p=0.002), shock (62.5%vs28.4%, p=0.010), renal replacement treatment (81.2%vs23.9%, p=0.000), mechanical ventilation (93.8%vs43.3%, p=0.000), and extracorporal membrane oxygenation treatment (50.0%vs20.9%, p=0.018) were more frequently observed in patients with bacterial co-infection. High mortality was also observed (62.5%vs23.9%, p=0.003)3, Elevated procalcitonin was an independent marker to predict bacterial co-infection within72hours after admission.4, Procalcitonin at a cut-off of0.81μg/1had an area under the receiver operating characteristic curve of0.96(compare to0.68of CRP). When PCT≥0.81μg/1, the sensitivity was91.7%and specificity was90.2%for diagnosis of bacterial co-infection.Conclusions:Influenza A (H7N9) infection patients with bacterial co-infection had a more severe condition. Elevated procalcitonin is an accurate marker for early diagnosing bacterial co-infection in influenza A (H7N9) patients, thus enabling earlier antibiotic therapy. |
PDF Full Text Request