Andard treatment [327]. However, the cardiac index, left ventricular stroke work index, and oxygen delivery 2-(2,4-Dichloro-5-fluorophenyl)oxirane index were significantly improved. A systematic review reported in 2007 compared a PMX-DHP treatment group (978 patients) and a conventional treatment group (447 patients) for infections including those other than abdominal infections or non-gram-negative bacterial infections and found improvements in elevated mean arterial pressure, reduced dose of catecholamine, and elevated PaO2/FIO2 ratio [329]. However, the results were unreliable because the 28 articles were mostly case reports from Japan and overlapping data from the same center were included. The Early Use of Polymyxin B Hemoperfusion in Abdominal Sepsis (EUPHAS) trial published in JAMA in 2009 demonstrated significant improvement in respiratory function, SOFA score, and 28-day mortality, in addition to hemodynamic improvement by PMX-DHP in patients with severe sepsis and septic shock due to intraabdominal infections that required emergency surgery. The trial was terminated after enrolling 64 patientsOda et al. Journal of Intensive Care 2014, 2:55 http://www.jintensivecare.com/content/2/1/Page 25 ofbecause the improvement in the 28-day mortality met the criteria for trial termination through interim analysis [328]. However, Vincent, in a letter to the editor [330], pointed out that no statistically significant difference was found in the survival rate and that there were other concerns on PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/8711135 the distribution of pathogenic bacteria between groups. Thus, it is still unclear whether PMX-DHP improves prognosis. There is a report from Japan showing that the hospital mortality of patients with lower intestinal perforation requiring emergency surgery, which is almost the same indication as in the EUPHAS trial, was 17.9 without PMX-DHP treatment, which was lower than the predicted mortality rate 1-(Cyclopropylsulfonyl)-1,4-diazepane of 63.3 based on APACHE II scores [331]. The main mechanism of action of PMX-DHP is adsorption of endotoxin, although effective mechanisms other than endotoxin adsorption are reported [332]. On the basis of the understanding of the pathophysiology of sepsis, endotoxin does not play a key role but rather functions only as one of the pathogen-associated molecular patterns. Therefore, there are doubts about the efficacy against sepsis of a blood purification method that removes only endotoxin [333]. The abovementioned first investigation of the Sepsis Registry Committee [44] s12325-016-0451-1 showed that PMX-DHP was performed PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/9544797 in 43 of 266 patients (16 ). The group treated with PMX-DHP showed both significantly higher APACHE II score (23.1) and 28-day mortality (44.2 ) compared with the group without PMX-DHP. Additionally, the cases were matched by using propensity score analysis because the severity was different between the groups. The covariates included for propensity matching were four background items (age, sex, APACHE II, and SOFA score) and eight items of laboratory data before PMX treatment or interventions (lactate measurement, antibiotic administration within 1 h after admission, implementation of EGDT, etc.). Thirty-seven matched cases from each group were examined for prognosis. As a result, the 28-day mortality rate of 37 patients treated with PMX-DHP (37.8 ) was significantly lower than that of 37 patients without PMX-DHP (67.6 ) (p =0.019). In addition, the rate of implementation of intensive insulin therapy was significantly higher in the PMX-DHP group, among the items of sepsis bundle treatment.