摘要目的：探讨不同血液灌流（HP）强度对急性百草枯中毒患者7 d和28 d预后的影响，并分析影响医师临床开具不同强度HP处方的相关因素。方法采用回顾性队列研究方法，选择2012年8月至2014年8月安徽医科大学附属省立医院危重症医学科收治的急性百草枯中毒患者，纳入年龄＞18岁、口服百草枯至入院时间＜12 h、24 h内接受HP治疗、入院后存活时间＞24 h且资料完整的患者，根据HP强度分为低强度组（LHP组，单次HP 4 h，2个灌流器）和高强度组（HHP组，单次HP 6 h、3个灌流器）。记录患者入院时相关资料及7 d、28 d预后情况，采用多元logistic回归模型评估入院时导致急性百草枯中毒患者接受HHP治疗的相关因素。建立Cox比例风险回归模型进行多因素分析，评价HHP是否可降低7 d、28 d全因病死率。结果最终入选60例百草枯中毒患者，其中LHP组28例，7 d全因病死率53.6%（15例），28 d全因病死率64.3%（18例）；HHP组32例，7 d全因病死率43.8%（14例），28 d全因病死率62.5%（20例）。与LHP组比较，HHP组患者入院时血药浓度较高，合并呼吸性碱中毒（呼碱）、急性肾损伤（AKI）者较多，乳酸（Lac）较高，但7 d全因病死率较低。多变量logistic回归模型显示，对于服毒至入院时间＞4 h〔优势比（OR）＝1.461，95%可信区间（95%CI）＝1.132～1.435，P＜0.001〕、年龄＜50岁（40～49岁的OR＝1.397，95%CI＝1.251～1.703，P＝0.002；＜40岁的OR＝1.701，95%CI＝1.253～1.836，P＜0.001），及入院时血药浓度≥2 mg/L（OR＝3.140，95%CI＝1.511～3.091，P＜0.001）、白细胞计数（WBC）＞10×109/L（OR＝1.222，95%CI＝1.032～1.275，P＝0.018）、Lac＞2.0 mmol/L（OR＝2.392，95%CI＝2.090～2.734，P＜0.001）、合并AKI（AKI 2期OR＝2.350，95%CI＝2.160～3.910，P＜0.001；AKI 3期OR＝2.821，95%CI＝1.932～3.651，P＜0.001）、合并低氧血症（OR＝2.420，95%CI＝2.131～2.662，P＝0.003）的患者，医师更倾向于采用HHP治疗。与存活组比较，死亡组患者年龄较大，入院时和HP 4 h血药浓度较高，合并AKI者较多，入院时血肌酐（SCr）、 WBC、 Lac、急性生理学与慢性健康状况评分系统Ⅱ（APACHEⅡ）评分较高，动脉血二氧化碳分压（PaCO2）、pH值较低。Cox回归模型计算风险比（HR）并校正混杂因素后显示，HHP为患者7 d生存的保护因素（HR＝0.843，95%CI＝0.732～0.971，P＝0.032），但不是28 d生存的保护因素（HR＝0.930，95%CI＝0.632～1.411，P＝0.423）；而年龄＞50岁（HR＝1.282，95%CI＝1.050～1.530，P＝0.043）、入院时血药浓度每增加1 mg/L（HR＝2.521，95%CI＝2.371～3.825，P＝0.012）及入院时合并AKI（HR＝3.850，95%CI＝2.071～5.391，P＜0.001）、WBC＞10×109/L（HR＝1.932，95%CI＝1.782～2.171， P＝0.006）、Lac＞2.0 mmol/L（HR＝2.981，95%CI＝2.210～3.792，P＝0.002）、PaCO2＜35 mmHg（HR＝1.772，95%CI＝1.483～2.516，P＝0.008；1 mmHg＝0.133 kPa）是患者28 d死亡的独立危险因素。结论 HHP虽可降低急性百草枯中毒患者7 d全因病死率，但并不能改善患者28 d全因病死率。医师在临床决策患者HP强度时仍存在可改进之处，未来或需要建立更加完善的评估体系，合理使用诸如此类较为昂贵的医疗资源。

abstractsObjectiveTo evaluate the influence of different hemoperfusion (HP) intensity on 7-day and 28-day mortality for patients with paraquat (PQ) poisoning, and examine the factors that may affect the decision of the clinicians to prescribe a high intensity HP.Methods A retrospective cohort study was conducted. The patients admitted to the department of critical care medicine of Anhui Provincial Hospital Affiliated to Anhui Medical&nbsp;University with the diagnosis of PQ poisoning from August 2012 to August 2014, fulfilling the following criteria were enrolled in the study: older than 18 years, interval from ingestion PQ to hospital admission shorter than 12 hours, and receiving HP treatment within 24 hours, and expecting surviving time exceeding 24 hours after admission, and data of the patients available for at least 28 days after admission. Depending on the intensity of HP, patients were assigned to either lower intensity HP group (LHP, defined as receiving HP for less than 4 hours, 2 columns) or higher intensity HP group (HHP, defined as receiving HP longer than 6 hours, 3 columns). Patients' data were retrieved from hospital's electronic database after hospital admission, and the results at 7th day and 28th day were recorded. Multiple logistic regression model was used to determine factors with which the clinician decided to choose the intensity of HP for the patients, and Cox regression model was used to evaluate 7-day and 28-day mortality.Results Data of 60 patients was finally available for this study. LHP group consisted of 28 patients, with a 7-day mortality of 53.6%(15 patients) and 28-day mortality of 64.3% (28 patients); 32 patients were assigned to HHP group with 7-day mortality of 43.8% (14 patients) and 28-day mortality of 62.5% (20 patients). Twenty-eight patients constituted as the HHP group, with higher PQ concentration in plasma, higher incidence of respiratory alkalosis and acute kidney injury (AKI), and higher level of lactate (Lac) compared with LHP group. However, a lower 7-day mortality was observed in the HHP group. Multiple logistic regression model indicated that at admission, interval from ingestion PQ to hospital admission longer than 4 hours [odds ratio (OR) = 1.461, 95% confidence interval (95%CI) = 1.132-1.435,P< 0.001], younger than 50 years old (40-49 years old:OR = 1.397, 95%CI = 1.251-1.703,P = 0.002;< 40 years old:OR = 1.701, 95%CI = 1.253-1.836,P< 0.001), PQ plasma concentration≥ 2 mg/L (OR = 3.140, 95%CI = 1.511-3.091,P< 0.001), white blood cell (WBC)> 10×109/L (OR = 1.222, 95%CI = 1.032-1.275, P = 0.018), Lac> 2.0 mmol/L (OR = 2.392, 95%CI = 2.090-2.734,P< 0.001), AKI on admission (stage 2:OR = 2.350, 95%CI = 2.160-3.910,P< 0.001; stage 3:OR = 2.821, 95%CI = 1.932-3.651,P< 0.001), accompanying hypoxia (OR = 2.420, 95%CI = 2.131-2.662,P = 0.003) were more likely to receive higher intensity of HP. Furthermore when compared with patients survived for 28 days, patients who were older, with higher levels of PQ concentration at admission or after 4 hours of HP, accompanied by AKI, increased serum creatinine (SCr), WBC, Lac, and acute physiology and chronic health evaluationⅡ (APACHEⅡ) score, lower arterial partial pressure of carbon dioxide (PaCO2) and lower pH value were more likely to die. After adjusted for con-variables in COX regression model, HHP was associated with lower 7-day mortality after admission [hazard ratio (HR) = 0.843, 95%CI = 0.732-0.971, P = 0.032], but devoid of lowering effect on 28-day mortality rate (HR = 0.930, 95%CI = 0.632-1.411,P = 0.423). In addition, age> 50 years old (HR = 1.282, 95%CI = 1.050-1.530,P = 0.043), PQ concentration increased by 1 mg/L (HR = 2.521, 95%CI = 2.371-3.825,P = 0.012), AKI on admission (HR = 3.850, 95%CI = 2.071-5.391,P< 0.001), WBC>10×109/L (HR = 1.932, 95%CI = 1.782-2.171,P = 0.006), Lac> 2.0 mmol/L (HR = 2.981, 95%CI =2.210-3.792,P = 0.002), and PaCO2< 35 mmHg (HR = 1.772, 95%CI = 1.483-2.516,P = 0.008; 1 mmHg =0.133 kPa) were independent risk factors for 28-day mortality.Conclusions Though HHP was helpful in lowering mortality rate in patients with PQ poisoning within 7 days, it did not influence on 28-day mortality. Clinicians' decisions on HP intensity need further investigation, and more perfect clinical evaluation system is required for reasonable use of expensive medical resources such as HP.