Treatment of patients with sepsis in a closed intensive care unit is associated with improved survival: a nationwide observational study in Japan

Background The aim of this study is to investigate the association between treatment in a closed ICU and survival at discharge in patients with sepsis. Methods This is a post hoc analysis utilizing data from the Japan Septic Disseminated Intravascular Coagulation study, including data from patients with sepsis from 2011 to 2013. Multiple logistic regression analysis was used to estimate the association between ICU policy and survival at discharge, and propensity score matching analysis was performed including the same covariates as a sensitivity analysis. Multiple linear regression analysis for the length of ICU stay in surviving patients was also performed with adjustments for the same covariates. Results Two thousand four hundred ninety-five patients were analyzed. The median Acute Physiology and Chronic Health Evaluation (APACHE) II score was 22 [17–29], the median Sequential Organ Failure Assessment (SOFA) score was 9 [7–12], and the overall mortality was 33%. There were 979 patients treated in 17 open ICUs and 1516 patients in 18 closed ICUs. In comparison, the APACHE II score and SOFA scores were significantly higher in patients in closed ICUs (closed vs open = 23 [18–29] vs 21 [16–28]; p < .001, 9 [7–13] vs 9 [6–12]; p = 0.004). There was no difference in the unadjusted mortality (closed vs open; 33.1% vs 33.2%), but in multiple logistic regression analysis, treatment in a closed ICU is significantly associated with survival at discharge (odds ratio = 1.59, 95% CI [1.276–1.827], p = .001). The sensitivity analysis (702 pairs of the matching) showed a significantly higher survival rate in the closed ICU (71.8% vs 65.2%, p = 0.011). The length of ICU stay of patients in closed ICUs was significantly shorter (20% less). Conclusion This Japanese nationwide analysis of patients with sepsis shows a significant association between treatment in a closed ICU and survival at discharge, and a 20% decrease in ICU stay. Electronic supplementary material The online version of this article (10.1186/s40560-018-0322-8) contains supplementary material, which is available to authorized users.


Background
Sepsis is life-threatening organ dysfunction caused by dysregulated host responses to infection, and septic shock is a subset of sepsis in which underlying circulatory and cellular/metabolic abnormalities are sufficiently profound to substantially increase mortality [1]. According to one study, there are more than 750,000 patients annually in the USA with sepsis [2] and the incidence is rising [3]. Septic shock remains lethal even with aggressive management, with a mortality of 20 to 30% [4].
Patients with sepsis are usually treated in the intensive care unit (ICU). Sepsis results from infection, and these patients often develop multiple organ-system failure. Aggressive management, including control of the infection source and support of failing organ-systems, is needed for optimal outcomes. In recent years, guidelines for the treatment of patients with sepsis have been published [5,6], and intensivists play an important role in the care of these patients.
Intensivists improve patient outcome in the ICU [7], and the ICU organization model influences patient outcomes [8][9][10]. There are generally two staffing models for ICU treatment, including an "open organization model" and a "closed organization model." In the open model, there is no intensivist consultation or elective intensivist consultation. Mandatory intensivist consultation is conducted in closed model ICUs [8], in which intensivist directs patient care [11] regardless of the time of day.
However, focusing on sepsis, to the best of our knowledge, the effect of the ICU organization model (open/ closed) or directing of care by an intensivist on the mortality of patients is unknown and there are no studies to answer this important clinical question. We hypothesize that a closed ICU improves the outcome of patients with sepsis and septic shock, since a closed ICU is the highest-intensity physician staffing ICU model and intensivists direct care regardless of the time of day in the ICU. The aim of this study is to investigate the association between management in a closed ICU and survival at discharge of patients with sepsis. This is a nationwide study in Japan, including an analysis of clinical data regarding the severity of sepsis, pre-existing co-morbidities, the need for mechanical organ system support, and treatments given.

Patient selection
>This is a post hoc analysis utilizing the database from the Japan Septic Disseminated Intravascular Coagulation study (JSEPTIC DIC study) (University Hospital Medical Information Network Individual Case Data Repository, UMIN000012543, http://www.umin.ac.jp/icdr/ index-j.html), which was a nationwide study in Japan [12]. The JSEPTIC DIC study retrospectively collected data from patients admitted to the ICU for the treatment of sepsis [13] from January 2011 to December 2013, excluding patients younger than 16 years or who developed sepsis after admission to the ICU (the JSEP-TIC DIC study used the definitions of sepsis, severe sepsis, and septic shock from [13] as it was performed before publication of the 2016 definitions [1]). The JSEPTIC DIC study was approved by the Institutional Review Boards of all participating hospitals. The requirement for informed consent was waived because of the retrospective nature of the study. Since this database was already anonymized for individual patient data and institutions, the Institutional Review Board waived the need for review of this post hoc study.
Patients were divided into two groups, the closed ICU group ( In Japan, the closed ICU is conventionally defined as a unit that transfers all patient care to an intensive care team that directs patient care with primary responsibility for all care and the open ICU is conventionally defined as an ICU where the intensive care team provides expertise via elective or mandatory consultation without assuming primary responsibility for patient care [14]. Patients treated in an ICU which could not clearly be classified as the closed or open were excluded from the final study population.

Exposure and outcome variables
The present study used all variables collected in the JSEPTIC DIC study. Variables for which the proportion of missing data was above 10% (fibrinogen, fibrin/fibrinogen degradation products, D-dimer, antithrombin, and lactate) and primary infection site are not included. The main exposure variable was set as the closed ICU or open ICU. The primary outcome measure was survival at discharge. Due to the healthcare system in Japan, no patients were discharged to hospice care.
Covariates for patient characteristics included age, gender, weight, pre-existing organ dysfunction and hemostatic disorders (comorbidity), Acute Physiology and Chronic Health Evaluation (APACHE) II score [15] (day 1), Sequential Organ Failure Assessment (SOFA) score [16] (day 1), systemic inflammatory response syndrome score [17] (days 1), the identification of microorganisms responsible for sepsis, blood culture results, and results of laboratory tests on day 1 including white blood cell count, platelet count, hemoglobin level, and prothrombin timeinternational normalized ratio (PT-INR). Treatment variables reviewed included administration of medications, including anti-disseminated intravascular coagulation (DIC) or anti-thrombotic drugs, immunoglobulins, low-dose steroids, and transfusion of blood products during the first week after ICU admission. Other therapeutic interventions reviewed included renal replacement therapy, renal replacement therapy for non-renal indications, plasma exchange, polymyxin B direct hemoperfusion (PMX-DHP), extracorporeal membrane oxygenation (ECMO), and use of the intra-aortic balloon pump during the first week after ICU admission.

Statistical analysis
Baseline characteristics were compared before and after the follow-up period for patients in both the closed and open ICU groups. Distributed continuous variables without a normal distribution are presented as median with interquartile range (IQR). Categorical data are summarized using numbers or percentages. The Mann-Whitney U test was used for comparing continuous variables, and the Fisher's exact test was used for categorical data. As for the main result, multiple logistic regression analysis was performed to estimate the association between the closed/open ICU and survival at discharge from the hospital, adjusted by baseline patient characteristics and treatment variables. As a sensitivity analysis, propensity score matching analysis including the same covariate was performed. For propensity score matching analysis, we use calipers of width equal to 0.1 of the standard deviation. As for sub-analysis, multiple linear regression analysis for the log-transformed values of the length of ICU stay and hospital stay in survived patients was performed with adjustments for the same covariates, and estimated the ratio of the length of stay between the closed ICU and open ICU. The level of significance was set at p < 0.05. All statistical analyses were performed with SAS (Version 9.4, SAS Institute Inc., Cary, North Carolina, USA).
In sub-analysis, multiple linear regression analysis using the data about the length of ICU stay and hospital stay as an objective variable showed that the length of ICU stay of patients treated in closed ICUs was significantly shorter (20% less) than patients in open ICUs, but this significance was not seen in the length of hospital stay (Table 4).

Discussion
In this analysis of a large nationwide Japanese cohort of patients with sepsis and septic shock, patient management in a closed ICU is significantly associated with improved rate of survival at discharge and a decrease in the length of ICU stay.

Efficacy of the treatment of sepsis in a closed ICU
According to multiple logistic regression analysis adjusted for baseline characteristics and treatment variables, treatment in a closed ICU had a significant association with improved survival at discharge (odds ratio = 1.59, 95% CI [1.276-1.827], p = 0.001) ( Table 2). This result suggests that the efficacy of care in a closed ICU may depend on the management of sepsis conducted in a closed ICU. However, according to the crude comparison (Table 1), the patient cohorts were different in the closed and open ICUs. We additionally performed propensity score matching analysis as a sensitivity analysis, and the closed ICU group has a significantly higher survival rate, compared to open ICU (Table 3, Additional file 1: Table S1). The result of the main analysis is fully supported by the sensitivity analysis. There is a significant association between treatment in a closed ICU and improved survival at discharge. A significant difference in the rate of using various treatment modalities in closed and open ICUs is identified in this study (Table 1). Although these treatments are adjuvant therapies for sepsis and their efficacy is controversial, the difference in the rate of using these treatments performed in each ICU may be partly responsible for the observed significant association between treatment in a closed ICU and survival at discharge. The present study demonstrates a significant association between prognosis and each specific treatment for sepsis including use of recombinant human soluble thrombomodulin (rhsTM), low-dose steroid therapy, or PMX-DHP.
Some DIC treatment guidelines recommend the use of rhsTM rather than heparin for patients with DIC due to sepsis [18], and both positive and negative evidence regarding the survival benefit of rhsTM have been reported [19,20]. The result of a phase 3 trial of rhsTM [21] is awaited. The JSEPTIC DIC study using propensity score analysis reported a survival benefit with the use of rhsTM [12]. The present study, which uses the same data as the JSEPTIC DIC study, also shows a significant association between survival at discharge and rhsTM, using a different statistical analysis, and the use of rhsTM was more frequent in closed ICUs and the use of heparin was more frequent in open ICUs. This   difference may be partly responsible for the association between improved survival at discharge and treatment in a closed ICU. The efficacy of steroid therapy to reduce mortality in patients with sepsis has been controversial [22,23]. In the present study, the use of low-dose steroids (which is more frequent in a closed ICU) was significantly associated with in-hospital mortality, but the route of administration, timing, and presence of side effects were not reviewed. A significant association between the use of PMX-DHP and survival at discharge was also shown, but there was no significant difference of the using rate of PMX-DHP between the closed and open ICUs. Some trials showed a survival benefit with the use of PMX-DHP [24], but others have not [25]. The effect of PMX-DHP remains controversial, and the results of another prospective multicenter randomized controlled trial is awaited, including a phase 3 trial of PMX-DHP [26].
We consider that a significant association between treatment in a closed ICU and survival at discharge can be related to the high-quality care provided in a closed ICU. A high rate of compliance with guidelines is one element of "high quality" intensive care. This element affects the patients' prognosis, and compliance is different in open and closed ICUs. Some reports show that patients with acute respiratory distress syndrome (ARDS) cared for in a closed ICU had lower hospital mortality due to the increased rate of compliance with guidelinerecommended lung-protective ventilation [27] (a safety management was given for the patients with ARDS in the closed ICU). Considering the present study, a significantly lower rate of obtaining blood cultures (blood cultures are strongly recommended in the sepsis guideline [5,6]) in the open ICU (Table 1) suggests that compliance with the guidelines in an open ICU may be lower than that in closed ICUs and the lower compliance may contribute to the higher mortality in open ICUs. Patient with sepsis commonly need ICU care, and in this decade, guidelines for the treatment of sepsis have been published from international societies [5]. The management of patients with sepsis has been qualified and standardized throughout the world, and the closed ICU can provide standardized intensive care for patients with sepsis based on the guidelines, which can improve the outcomes [28].
The safety management, prevention, or rapid and optimal response to complications is an element of "high quality" intensive care. Another report from the Leapfrog Group reported that applying ICU physician safety staffing standards could save more than 54,000 lives in the USA each year [29]. Intensivists can be an expert in safety management and complication management (prevention or appropriate response) in the ICU. Although the present study did not investigate the differences in incidence of complications or other safety management issues between closed and open ICUs, this potential difference may contribute to the significant association between treatment in a closed ICU and a higher survival rate at discharge. Future study should be focused on the differences in these factors.

Study limitations
This is a sub-analysis of the JSEPTIC DIC study, which used retrospective data. The 205 patients excluded because of data insufficiency may have an impact on the results although these excluded patients represent less than 10% of the study population. The J-SEPTIC DIC study did not clearly define open and closed ICUs, and each ICU had reported subjective information about their ICU model (open/closed/unclassified). This study comparing the outcomes of patients with sepsis in open and closed ICUs was conducted based on a definition that was subjectively reported from each institution. Although the present study demonstrates the potential of improved efficacy of the treatment of sepsis in a closed ICU, this result may be unreliable because of an unclear definition of the types of ICUs. Further study should clearly define open and closed ICUs to investigate the association between the outcome of patients with sepsis and ICU organization models.
This study focuses only on the association between outcomes and the type of ICU model. Details of differences in ICU treatment and quality of care in each institution have not been analyzed. Although many kinds of ICU staffing components influence ICU outcomes, the information such as hospital characteristics, general/university; the staffing numbers for open or closed ICU's in this study, nursing/junior doctors/fellows/certificated specialist intensivists/non-certificated intensivist; or nighttime coverage by certificated specialist intensivists was not available in this post hoc analysis. These staffing patterns must influence the outcomes of patients in the ICU, but this information was unavailable in this study. This is a severe limitation of the present study. Future studies should include staffing information to investigate a relationship between the outcomes of patients with sepsis and treatment using each ICU model (open/closed).
Although many kinds of guidelines or safety management protocols in the ICU such as hand hygiene protocols, ventilation-associated pneumonia bundles, and catheter-related blood stream infection management impact outcomes in the care of patients with sepsis, compliance with these care bundles was not evaluated in this study. Future studies should focus on compliance with sepsis and other care guidelines, treatment differences, and quality of care in the ICU. To further define the benefit of a closed ICU, additional prospective multicenter studies are warranted.

Conclusions
This nationwide retrospective post hoc analysis of patients with sepsis in Japan shows a significant association between treatment in a closed ICU and improved survival at discharge, although there are acknowledged study limitations. Future prospective trials are indispensable to evaluate the efficacy of treatment in a closed ICU for the treatment of patients with sepsis. These studies should define the ICU models (closed/open) clearly and focus on staffing patterns, compliance with care guidelines, treatment differences, and quality of care in ICUs with both closed and open ICU models.

Additional file
Additional file 1: Table S1. Comparison for patient characteristics and treatments after matching. (DOCX 25 kb) Abbreviations APACHE II score: Acute Physiology and Chronic Health Evaluation II score; ARDS: Acute respiratory distress syndrome; DIC: Disseminated intravascular coagulation; ECMO: Extracorporeal membrane oxygenation; ICU: Intensive care unit; JSEPTIC DIC study: Japan Septic Disseminated Intravascular Coagulation study; PMX-DHP: Polymyxin B direct hemoperfusion; PT-INR: Prothrombin time-international normalized ratio; SOFA score: Sequential Organ Failure Assessment score