Collaboration with the Minds and workshop activities

Occasional guidance was received from the Minds during the process of formulating these guidelines. In addition, external lecturers and librarians were invited to participate in a seminar on “Literature Acquisition Techniques for Systematic Reviews” we held independently.

Peer review

Activities were performed for various work processes while mutual peer review was conducted by team members across the region. Work products from each group were repeatedly edited and revised, with each revised draft being discussed by the Guideline Creation Committee.

Multiple rounds of public comments

CQs underwent multiple rounds of public comments generally from registered contributors: once after the initial formulation of a CQ and twice during the review of the final draft. During finalization, public commenters were requested to disclose any conflicts of interest. Opinions regarding draft CQs were also solicited over the internet.

Transparency

Although it is difficult to create guidelines that will be accepted universally, improving visibility and transparency in the development process is crucial. Members of each team created an official mailing list (ML) and discussions among team members were held using these MLs as much as possible. Core members and members of the Academic Guidelines Promotion Team joined the MLs established by each team as read-only members. Through these measures, we aimed to increase the transparency of team discussions, and by implementing the appropriate interventions, we were able to coordinate the directions taken by each team and achieve consistency throughout the entirety of the guidelines.

Vote anonymization

Votes were tallied after all 19 committee members had participated, and the rate of agreement achieved was mentioned in each recommendation. To avoid confounding from academic conflicts of interest (COIs) of committee members, committee votes concerning draft recommendations were anonymized.

Disclosure of COIs and members’ roles

Financial and academic COIs as well as the role(s) of each committee member are disclosed in the additional file. Financial COIs were disclosed in accordance with the standards used by the Japanese Association of Medical Sciences since 2013 through 2016.

Answer (recommendations)

Rationale

This clinical question (CQ) offers recommendations regarding the validity of the three biomarkers, PCT, P-SEP, and IL-6 to support the diagnosis of sepsis in two clinical settings: (1) settings with critically ill patients, such as in ICUs, where infection is suspected but difficult to confirm and (2) settings in which infection is suspected but patients are not critically ill such as the emergency room or general ward. The clinical utility of each marker was assessed individually in these two settings.

Hierarchical summary receiver operating characteristic (ROC) analysis was used during meta-analysis (data integration) of the diagnostic test accuracy for each marker, and the assessment of the quality of experience (QoE) and the recommended settings were calculated based on the estimated number of patients presenting as true positives, false positives, or false negatives determined by the “diagnostic Grading of Recommendations Assessment, Development and Evaluation (GRADE) system,” and the benefit-risk balance was assessed based on a pre-examination probability of 40%. We adopted CRP, a widely used biomarker in clinical practice, as a control. Representative meta-analyses of PCT [9], P-SEP [10], IL-6 [11], and CRP [12] were selected.

In the settings where most patients were critically ill, the benefits were evaluated to outweigh risks regarding the measurement of P-SEP or PCT, but not of IL-6 levels. As a result, we recommend the measurement of P-SEP or PCT levels as supplementary tests in the diagnosis of infection in critically ill patients when sepsis is suspected. In settings where most patients are not critically ill, significant benefit has not been established regarding the measurement of P-SEP, PCT, or IL-6 levels. Thus, we do not recommend the routine measurement of any of these biomarkers as a supplementary test in the diagnosis of infection in non-critically ill patients even when sepsis is suspected.

Access to tests for these biomarkers is variable among hospitals or facilities. Currently, only a limited number of hospitals or facilities in Japan are capable of measuring P-SEP and IL-6 values as part of routine examinations. Moreover, even in hospitals or facilities capable of performing these measurements, these tests are performed in central laboratories and may not be as useful as point-of-care-testing (POCT).

Answer (opinion)

A blood culture should be taken prior to antimicrobial administration in patients with sepsis or septic shock (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

No randomized controlled trial (RCT) was found to conform to the Patient, Intervention, Comparison, Outcome (PICO) process. The diagnosis of bacteremia fortifies the diagnostic accuracy of infection. Identifying the causative microorganisms and subsequently performing antimicrobial susceptibility tests lead to treatment optimization. However, as the volume of the blood collected increases so does the risk of iatrogenic anemia, but the benefits of taking a blood culture are considered to outweigh the potential risks in all cases of sepsis. Also, because the detection sensitivity is not expected to increase beyond three culture sets, oversampling should be avoided.

Answer (opinion)

In patients presenting with sepsis or septic shock, various culture specimens other than blood may be collected as necessary prior to administering antimicrobials (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

No RCT conforming to the PICO process has been identified. Substantial benefits can be obtained from the culture results of suspected site of infection. Such benefits are thought to outweigh the potential risks in any case of sepsis. However, as there are risks associated with this procedure, specimens should not be collected unless the collection site is suspected to be a focus of infection.

Answer (opinion)

When selecting antimicrobial agents for empiric treatment, Gram staining may be considered (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

No RCT conforming to the PICO process has been identified; thus, the risk-benefit balance is unknown. However, favorable specificity has been reported in community-acquired pneumonia, urinary tract infections, and bacterial meningitis. Considering the simplicity, rapidity, and low costs associated with this technique, the benefits may sufficiently outweigh any potential risks.

Answer (opinion)

The use of imaging examinations is recommended in the diagnosis of the foci of infection in sepsis/septic shock patients (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

There is currently no supporting RCT that conforms to the PICO process, and there is little evidence available in support of performing diagnostic imaging. The detection of infectious foci is important in sepsis and septic shock. If the diagnosis of an infectious focus can be performed accurately through imaging, the optimal treatment method can be selected, and unnecessary treatments can be avoided. However, various complications may also occur, such as allergic reaction to the iodine-based contrast agents, impaired renal function, or gadolinium-based contrast-associated nephrogenic systemic fibrosis, and caution is required when treating patients who are at risk. In addition, there are some concerns that the condition of patients with unstable hemodynamics and respiration might worsen when they are transported to the examination room. In consideration of the above, in patients with sepsis and septic shock, “performing imaging examinations for diagnosis of infectious foci is recommended (expert consensus)” while paying attention to the complications and dangers associated with patient transportation.

Answer (opinion)

Performing early (whole-body contrast enhanced) CT examination is recommended to aid in diagnosing the foci of infection in patients with sepsis/septic shock (expert consensus/no evidence) (rate of agreement, 89.5%).

Rationale

There is no RCT that conforms to the Patient, Intervention, Comparison, Outcome (PICO) process, and the evidence for performing whole-body contrast-enhanced CT at an early stage is poor. The diagnosis of the foci of infection is important for the diagnosis of sepsis/septic shock, but sometimes, it is difficult to determine the origin of infection based on simple CT examination alone. Infectious foci become apparent on contrast-enhanced CT images, which can lead to the selection of a more effective treatment for the infection. However, various complications may also occur, such as allergic reaction to the iodine-based contrast agent, impaired renal function, or gadolinium-based contrast-associated nephrogenic systemic fibrosis, and caution is required when treating at-risk patients. In sepsis and septic shock, “performing CT scans (whole body contrast enhanced) at an early stage is recommended (expert consensus).” while paying ample attention to possible complications.

Answer (opinion)

Controlling the source of infection as soon as possible is recommended in cases of sepsis arising from intraperitoneal infection (expert consensus/quality of evidence “D”) (rate of agreement, 100%).

Rationale

No RCTs conforming to the Patient, Intervention, Comparison, Outcome (PICO) process could be found, and so, a systematic review of observational studies was conducted. As a result, one observational study was extracted [48]. If sepsis arises from an intraperitoneal infection, controlling the source of infection at an early stage may improve patient outcome. Performing surgery to control the infection is invasive to the patient, but it is believed that no side effects will result if the surgery is performed early. Based on this study, it was concluded that performing an early surgery may improve patient outcomes and that the benefits to patients outweigh the potential harms.

Answer (recommendations and opinion)

Rationale

Infectious pancreatic necrosis is a disease in which the early initiation of intervention, the usual principle in controlling the source of infection, does not apply. In an RCT comparing mortality rates with regard to differences in the timing of treatment approaches to control the source of infection [60], 36 patients presenting with severe necrotizing pancreatitis were included in the early intervention group and underwent necrotic tissue resection 48–72 h after onset, while the late intervention group underwent surgery 12 days after onset. As a result of the comparison, the mortality rate was lower in the late intervention group compared to the early intervention group [60]. Two RCTs have been reported on the treatment of infected pancreatic necrosis [61, 62]. In the first RCT, the minimally invasive step-up approach to treating infectious pancreatic necrosis was compared with open necrosectomy, and no significant difference in mortality rates was observed (19 vs. 16%). However, the ICU stay times and hospitalization times associated with the minimally invasive step-up approach tended to be shorter. Regarding the frequency of complications, few incident cases of multiple organ failure or general complications, intraperitoneal bleeding requiring treatment, enterocutaneous fistula requiring treatment, or perforation into intraperitoneal organs were observed in the minimally invasive step-up approach group (a significant difference was observed with respect to the incidence of multiple organ failure and systemic complications). The second RCT was a comparison of endoscopic transgastric necrosectomy and surgical necrosectomy. As a result, it was found that the mortality rate was lower in the endoscopic transgastric necrosectomy group, and the incidence of complications such as multiple organ failure, intraperitoneal bleeding requiring treatment, enterocutaneous fistula requiring treatment, perforation into intraperitoneal organs, and pancreatic fistula was low in the endoscopic transgastric necrosectomy group. A significant difference was observed with respect to the incidence of multiple organ failure and pancreatic fistulas. Although there was no difference in survival outcomes between these two RCTs, the effectiveness of a minimally invasive approach was demonstrated by the reduction in the incidence of complications.

Based on the above observations, controlling infected lesions in patients with sepsis due to infectious pancreatic necrosis by first performing drainage (percutaneously or endoscopically) and then resecting necrotized tissue (via retroperitoneal or endoscopic approaches) is considered to be beneficial.

Answer (recommendation)

We suggest removing vascular catheters only when bloodstream infection is suspected (2D) (rate of agreement, 94.7%).

Rationale

One RCT [63] was found as the result of a comprehensive literature search. In this study of 144 patients in whom vascular catheter-related bloodstream infection was suspected, 64 patients (excluding 80 cases predicted to have been caused by vascular catheter infection) were divided into two groups (with 32 patients each). As a result, no significant difference in ICU mortality rate was observed. Accordingly, unnecessary vascular catheter removals can be reduced by restricting early withdrawals to cases when bloodstream infection is confirmed or when the patient becomes hemodynamically unstable. Such measures can be expected to lead to reductions in medical costs and risks arising from catheter reinsertion. However, it has been reported that after a catheter-related bloodstream infection is diagnosed, removal of the catheter within 24 h is associated with improved patient outcomes [64]. Based on these observations, the early removal of a vascular catheter from a patient with sepsis is considered to be beneficial only in cases where a bloodstream infection has been confirmed, or the patient has become hemodynamically unstable.

Answer (opinion)

Controlling the source of infection as quickly as possible via percutaneous nephrostomy or transurethral ureteral stent placement is recommended in cases of sepsis arising from acute pyelonephritis caused by ureteral obstruction (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

As no RCTs conforming to the PICO process could be found, this CQ referred to the American Urological Association (AUA) guidelines [53]. When considering the treatment of acute pyelonephritis caused by ureteral obstruction as well as the costs of transporting patients to specialist facilities, it is believed that the potential benefits obtained by performing a percutaneous nephrostomy or transurethral ureteral stent placement likely outweigh potential complications such as bleeding or the spreading of infection to the retroperitoneum.

Answer (opinion)

Proceeding with early surgical intervention is recommended in cases of sepsis arising from necrotizing soft tissue infection (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

No RCTs conforming to the PICO process could be found. When complicated by organ failure due to necrotizing soft tissue infection (i.e., cases of sepsis) it is likely that initiating surgical intervention including aggressive and early drainage of the infected lesion will be more beneficial to the patient. Although there is a risk of harm caused by the surgery, the benefits outweigh the risk, compared to when surgery is not performed despite the development of sepsis. Therefore, although no RCT conforming to the PICO process could be found, it was concluded that there is a strong possibility that the benefits outweigh the potential harms.

Answer (opinion)

Sepsis and septic shock patients should begin receiving an effective antimicrobial within 1 h (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Initiating antimicrobial therapy within 1 h when handling sepsis cases is now recommended in the Surviving Sepsis Campaign Guidelines based on the results of observational studies and has gained global acceptance. However, it is also true that there is no particularly strong basis for this recommendation, as no relevant RCTs have been completed to date. As such, although we have strong reservations regarding the possibility of negatively impacting patient prognosis by refraining from promptly administering antimicrobials in sepsis cases, we decided that it was necessary to offer our opinion as a target although it comes in the form of an expert consensus.

No RCTs investigating the impact of antimicrobial administration within 1 h could be found, and only results of observational studies were considered as evidence. Although the results of multiple observational studies indicate that initiation of antimicrobial therapy within 1 h or earlier reduces the risk of mortality, no significant improvement in mortality risk was observed in a systematic review of such observational studies [67].

Initiating antimicrobial therapy within 1 h after diagnosis may contribute to a lower risk of mortality, and no associated adverse effects have been reported. The increased burden placed on medical staff when antimicrobial therapy is ordered to be initiated within 1 h after diagnosis arising from the need to prioritize the corresponding preparatory tasks over others (e.g., confirmation of drugs dispensed and transportation from the hospital pharmacy) may be considered as an assumed burden. Another obstacle is the issue of space limitations for drug storage accompanying the need to routinely prepare a variety of antimicrobials for emergency outpatients. Even with the above considered, we believe that the potential benefits of this practice likely exceed any potential harms.

Answer (recommendation)

We recommend against routinely administering antimicrobial combination therapy when treating infections caused by Gram-negative bacilli (1B) (rate of agreement, 89.5%).

Rationale

In the past, there has been a view that combination therapies using antimicrobial drugs for sepsis and septic shock cases, especially in the treatment of Gram-negative bacilli, will expand the antimicrobial spectrum and that a synergistic effect should be expected. However, due to the considerable risks associated with antimicrobial combination therapies, it was important to present an opinion based on clear reasoning that was also reflective of the realities of clinical practice.

We referred to a single meta-analysis that verified the effects of using aminoglycosides in combination with β-lactam drugs [88]. No difference in mortality rate was observed for monotherapy in comparison to combination therapy, but a significant increase in the frequency of kidney injury, believed to be a side effect of aminoglycoside antimicrobials, was observed with respect to the use of combination therapy. In addition to this meta-analysis, another RCT verified the effect of using a quinolone antimicrobial (moxifloxacin) in combination with a carbapenem (meropenem), a β-lactam antibiotic [89]. This study found that while mortality rate remained unchanged as a result of using this combination therapy, the frequency of side effects associated with these drugs increased.

No significant difference in mortality rate was observed between the intervention and control groups in this study, and apart from there being no observable benefit, the frequency of kidney injury was significantly higher in patients receiving combination therapy compared with those who received monotherapy only. The development of new onset kidney injury may increase patient burden as well as medical costs as a result of the greater need for related treatment interventions. In addition, in consideration of the time and cost of prescribing, dispensing, and administering multiple antimicrobials, the potential harms associated with this practice clearly outweigh the benefits.

Answer (opinion)

The administration of anticandidal drugs in addition to general antimicrobials should be considered when treating sepsis and septic shock patients exhibiting multiple risk factors for invasive candidiasis (expert consensus/no evidence) (rate of agreement, 78.9%).

Rationale

It is known that the Candida genus of fungi is a primary cause of fungal sepsis, and also that the mortality rate associated with candidemia is higher than the rates attributed to other forms of bacteremia. Despite this, candidiasis is also frequently overlooked. As such, it is necessary to establish criteria for administering anticandidal drugs when handling cases refractory to conventional antimicrobial therapies.

No RCTs evaluating the use of anticandidal drugs in sepsis cases could be found, and the evidence considered for this CQ considered candidemia or invasive candidiasis. Multiple observational studies have been conducted with respect to the known risk factors for these conditions, and risk factors specific to ICU patients have also been reported. In addition, the sensitivity and specificity of (1-3)-β-D-glucan, a serum biomarker, have also been evaluated in the context of invasive candidiasis.

The administration of antifungal drugs following risk assessment may improve patient prognosis in invasive candidiasis or candidemia cases, but at the same time, poses a risk of adverse reactions. However, no assessment of this risk as it pertains to sepsis patients has been conducted to date. In consideration of the above, we believe that the potential benefits of anticandidal drug use likely outweigh the potential risks.

Answer (recommendation)

We suggest against administering β-lactam drugs using continuous infusion or extended infusion periods when treating sepsis and septic shock patients (2B) (rate of agreement, 100%).

Rationale

To date, intermittent administration of antimicrobial drugs has been a common practice. However, it has been found that time-dependent β-lactam drugs may be more effective in terms of pharmacokinetic characteristics when administered continuously or over an extended period. Verification of the efficacy of continuous infusion of β-lactam drugs may lead to improved patient outcomes in sepsis cases and is considered to be an important clinical issue.

We referred to four RCT reports [72, 90–92]. Among these studies, no significant differences were observed between the respective study groups with respect to 90-day mortality rate (odds ratio, 0.94; 95% confidence interval (CI) 0.69–1.28, p = 0.68), ICU mortality rate (odds ratio, 0.79; 95% CI 0.59–1.06, p = 0.11), in-hospital mortality rate (odds ratio, 0.78; 95% CI 0.59–1.03, p = 0.08), or target serum drug concentration achievement rate (odds ratio, 1.88; 95% CI 0.89–3.98, p = 0.10).

Although decreased mortality frequency is an anticipated benefit of this intervention, no significant differences were observed between the intervention and control groups in any study regarding 90-day mortality rate, in-hospital mortality rate, and ICU mortality rate. In addition, no significant differences were observed with regard to target serum drug concentration achievement rate. However, although no evaluation of side effects was conducted, because β-lactam drugs are normally administered to ICU patients via intravenous infusion, we believe that few burdens that arise from this intervention warrant consideration. We have determined accordingly that the risks and benefits associated with this intervention are comparable.

Answer (recommendation)

We suggest the use of de-escalation in conjunction with antimicrobial therapy administered to sepsis and septic shock patients (2D) (rate of agreement, 84.2%).

Rationale

Although broad-spectrum antimicrobials are frequently given at an early stage to address sepsis cases in the ICU, this practice is linked to the appearance of drug-resistant bacteria and accompanying increases in medical costs. As such, the capacity to de-escalate, or switch treatment regimens from broad-spectrum antimicrobials to drugs with narrower therapeutic indices, without risking patient safety, can be regarded as a favorable practice from the perspectives of both infection control and medical economics.

We referred to one RCT report [74]. No significant differences were observed with respect to 90-day mortality rate between the two groups, and a significant increase in the frequency of coinfection was observed in the de-escalation group.

The primary benefit expected as a result of de-escalation is the prevention of the development of drug-resistant bacteria, but this outcome could not be evaluated based on this body of evidence. Meanwhile, although de-escalation did not increase mortality rate, the results suggested that it may increase patients’ risk of contracting coinfections. However, all of the evidence considered originated from a single RCT; thus, we believe this body of evidence lacks the strength necessary to constitute a basis for overriding the notion that de-escalation can be implemented safely, which is based on the results of observational studies conducted to date. In consideration of the above, we have determined that the potential benefits of de-escalation likely exceed any potential harms.

Answer (recommendation)

We suggest using PCT values as an index when determining whether to discontinue antimicrobial therapy administered to address sepsis or septic shock (2B) (rate of agreement, 78.9%).

Rationale

The measurement of PCT levels has become feasible in routine treatment, and studies investigating the use of PCT values in cases of infection have also increased; RCTs examining the discontinuation of antimicrobial therapy in accordance with the PCT guide have been conducted. However, there is still a paucity of high-quality systematic reviews of the subgroup of these RCTs that focused on sepsis cases. To establish criteria for discontinuing antimicrobial therapy in sepsis cases, the validity of interventions calling for antimicrobial therapy discontinuation based on PCT values, particularly those accumulated by RCTs, must be evaluated.

We referred to nine RCT reports [78–86]. No significant differences were observed between the intervention and control groups with respect to ICU stay time, hospitalization period, 60-day mortality rate, and 90-day mortality rate, but significant improvement in 28-day mortality rate was observed. The number of days of antimicrobial drug use also decreased significantly (a meta-analysis was conducted only with respect to studies that clearly described the mean administration period. Significant reduction in the number of days of therapy was also observed in the studies providing median value data). In the intervention group, no significant increase in mortality rate or the period of antimicrobial drug use was observed in comparison to the control group. Potential harms could not be considered as no other side effects were assessed. PCT is measured in conjunction with other blood parameters and can be grouped among routine blood tests. As such, we believe that this intervention adds minimal additional burden, and its potential benefits likely exceed any potential harms.

Answer (opinion)

The prognostic improvement effect of IVIG administration in adult patients with sepsis is unknown based on the current randomized controlled trial (RCT) results available, and accordingly, clear recommendations pertaining to IVIG administration cannot be offered (expert consensus/quality of evidence “C”) (a rate of agreement of 67% or higher in support of its use could not be obtained).

Rationale

While formulating this clinical question, it was considered to be critical to examine the effectiveness of immunoglobulin administration to patients with sepsis while considering benefits based on reductions in all-cause mortality rate, ICU mortality rate, and ICU treatment period, and side effects caused by IVIG administration as potential harms.

In the literature search, 978 references are screened without limits on investigation period, the severity of sepsis, or IVIG dosage. Six papers were secondarily extracted via peer review of abstracts [98, 102–106]. The all-cause mortality rate in the IVIG group was significantly lower than that of the control group (n = 6, risk ratio, 0.7 [95% CI 0.56–0.95]) and the ICU mortality rate was also significantly lower (n = 1, risk ratio 0.71 [95% CI 0.60–0.84]). Reduced ICU stay time, which was the second most important and significant benefit, was also shortened significantly (n = 3, mean difference: − 3.71 [95% CI − 7.32 to − 0.09]). There was no significant increase in the risk ratio for the onset of side effects due to IVIG administration (1.63, 95% CI 0.65–4.11). The side effects were minor symptoms such as skin rash, and no serious cases or deaths were reported. IVIG administration to adult patients with sepsis resulted in improved all-cause mortality and ICU mortality rates and also significantly shortened ICU stay time without increasing the frequency of side effects in comparison to the control group.

When considering the benefit-risk balance in terms of outcomes, although there was an increase in complications, emphasis was placed on reduction in the all-cause mortality rate and ICU mortality rate, and it was determined that the potential benefits likely outweigh the potential harms. However, members of the guideline committee expressed concerns about the quality of the systematic review and the body of evidence. Another body of evidence was proposed by the Academic Guidelines Promotion Team that narrowed the subjects to cases of severe sepsis, and the internal peer review team proposed the 2013 Cochran Review [107]. The team responsible for immunoglobulin treatments suggested that “IVIGs may be administered to adult patients with sepsis (2C (weak))”, but only a 63.2% agreement was obtained in the initial vote of the committee. The reasons for this were as follows: (1) we could not evaluate the effect as there were no new studies with current sepsis definitions and standard treatment and (2) although ICU mortality rates were improved based on the three bodies of evidence presented, there was no consistency with regard to evidence for the benefit in the 28-day mortality among the three bodies of evidence. The rate of agreement was still 63.2% after the second committee vote, and the required proportion for agreement of an over two-thirds majority was not obtained. The guideline committee reached an expert consensus that “the prognostic improvement effect of IVIGs in adult patients with sepsis is unknown based on the RCT results currently available, and a clear recommendation concerning IVIG administration cannot be presented at this time.”

Answer (recommendation)

We suggest against the use of EGDT when performing initial resuscitation in patients with sepsis or septic shock (2A) (rate of agreement, 100%).

Rationale

Three RCTs [110–112] conforming to the PICO process were identified based on a search of the PubMed database and were used in the final analysis for this CQ. Regarding the 90- and 28-day mortality rates, EGDT was not effective in improving mortality rate in comparison to the standard treatment (90-day mortality rate: risk ratio 0.98 (95% confidence interval (CI) 0.88–1.10); 28-day mortality rate: risk ratio 0.98 (95% CI 0.84–1.13)). The time to shock reversal was not assessed by any RCT. Regarding intensive care unit (ICU) length of stay, the mean difference (MD) was 0.27 (95% CI − 0.33–0.87) in the comparison between the EGDT group and the standard treatment group, and no significant difference was observed.

Regarding the benefit-risk balance, no improvement in mortality rate as a result of complying with EGDT was observed in comparison to the standard treatment. In addition, no shortening of ICU length of stay as a result of complying with EGDT was observed (MD 0.27 (95% CI − 0.33–0.87)), and no benefit of EGDT over the standard treatment could be found. However, dobutamine dosages and the quantity of blood transfused increased significantly in the EGDT group [110, 111], and due to the increased frequency of arrhythmias associated with dobutamine, greater overall risk of side effects associated with transfusions, and increased time and quantity of work required of hospital staff, it is possible that compliance with EGDT may increase the risk of harm (burden) faced by patients. Based on the above, it was determined that the potential harms presented by EGDT likely outweigh its potential benefits.

Answer (opinion)

We suggest that 30 mL/kg or more of an extracellular fluid replacement solution is administrated when performing initial fluid resuscitation in patients with septic shock with a relative decrease in intravascular volume associated with vasodilatation (expert consensus/no evidence) (rate of agreement, 100%).

Comment: 30 mL/kg or more of an extracellular fluid replacement solution should be administered after assessing the decrease in intravascular volume.

Rationale

No RCTs applicable to this CQ could be found as a result of a search of the PubMed database. As such, it was concluded that an expert consensus should be offered, as the evidence for this CQ is inadequate to support a recommendation. In addition, in three large-scale RCTs evaluating the effectiveness of EGDT (ProCESS [110], ARISE [111], and ProMISe [112]), when the differences in intergroup (EGDT group vs. standard treatment group) total volume of fluid transfused prior to study protocol initiation were calculated, the following differentials were revealed: ProCESS (2.3 ± 1.5 L vs. 2.1 ± 1.4 L), ARISE (2.5 ± 1.2 L vs. 2.6 ± 1.3 L), and ProMISe (1.9 ± 1.1 L vs. 2.0 ± 1.1 L). All the subjects had already received over 30 mL/kg of crystalloid solution during the initial resuscitation prior to group assignment.

Regarding the benefit-risk balance, the concept of large-volume initial fluid resuscitation (infusion of 30 mL/kg or 2000 mL within approximately 1 h) became recognized as a common sense approach based on the conventional guidelines, and there is a possibility that the prognosis of patients with sepsis may be improved by supplementing the relative decrease in intravascular volume associated with vasodilatation and optimizing the balance of tissue oxygen supply and demand as quickly as possible.

On the other hand, excessive extracellular fluid replacement may cause a deterioration in cardiac function (heart failure) and pulmonary function (pulmonary edema). Frequent assessment of hemodynamics is necessary to avoid excessive volume loading, which may increase the burden on medical staff. The cost of extracellular fluid replacement solution may be a burden on the intervention group but is relatively low. Based on the above considerations, it was concluded that the benefits of administering 30 mL/kg or more of an extracellular fluid replacement solution during the initial resuscitation in septic shock clearly outweigh the potential risks.

Answer (opinion)

We suggest the cardiac function using echocardiography is assessed when initiating fluid resuscitation in patients with sepsis (expert consensus/no evidence) (rate of agreement, 100%).

Comment: The assessment of cardiac function using echocardiography discussed in this CQ indicates the simple evaluation of cardiac function performed at the bedside. It is focused on cardiac function (movement of the heart), and measurements related to vascular (inferior vena cava diameter, intracardiac volume) intended to afford an approximate assessment of intravascular volume prior to initiating resuscitation. It is desirable that all physicians, and not just cardiologists, involved in the resuscitation of patients with sepsis be proficient with this technique.

Rationale

Although a literature search was conducted to identify RCTs examining whether the assessment of cardiac function using echocardiography affects the prognosis of patients with sepsis undergoing initial resuscitation, no RCTs pertaining to this CQ could be found. Therefore, it was concluded that an expert consensus should be offered as the evidence for this CQ is inadequate to support a recommendation.

Regarding the benefit-risk balance, although no supporting evidence could be found, assessing cardiac function and intravascular volume using echocardiography when initiating resuscitation in patients with sepsis is useful in determining the infusion rate and in catecholamine selection. Therefore, it is believed that conducting this assessment will lead to more appropriate fluid resuscitation and drug administration. Echocardiography assessment is simple and non-invasive, and little patient burden for physicians is associated with the intervention itself. In institutions that do not routinely use echocardiography for assessment, the use of this technique will require additional time and may contribute to delays in initiating resuscitation. In addition, the price of most echocardiography devices is in the range of several million yen (approximately USD 77,000), and thus, the financial burden placed on facilities will be substantial when purchasing a new device. However, such devices have high versatility, are believed to be adequate for their desired uses, and are cost-effective. Therefore, it was concluded that the benefits of assessing cardiac function using echocardiography when initiating resuscitation in patients with sepsis clearly outweigh the potential harms.

Answer (recommendation)

We suggest against the use of an artificial colloidal solution during the initial resuscitation of patients with sepsis or septic shock (2B) (rate of agreement, 89.5%).

Answer (recommendation and opinion)

We suggest against the routine use of albumin solution during the initial resuscitation of patients with sepsis (2C).

The administration of albumin solution may be considered when large volumes of crystalloid solution are required for resuscitation, or when hypoalbuminemia is observed (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

A search of the PubMed database was conducted using the keywords “sepsis,” “septic shock,” and “albumin.” Five systematic reviews and one new RCT (CRISTAL (Colloids Versus Crystalloids for the Resuscitation of the Critically Ill) trial [138]) were extracted. The systematic reviews [139] and RCT [138] found using the most recent literature search period and that scored highly on the AMSTAR (A Measurement Tool to Assess Systematic Reviews) measurement tool (9 points) were adopted as high-quality studies for this CQ.

Among these studies, only the SAFE (Saline Versus Albumin Fluid Evaluation) 2011 study [140] was identified as an applicable RCT. No significant difference between the mortality risk ratio of 0.87 (95% CI 0.74–1.02) and ICU length of stay of 0.7 (95% CI − 0.10–1.50) was observed. No assessment was carried out regarding time to shock reversal.

Regarding the benefit-risk balance, although there was a tendency for a decrease in mortality rate, no significant difference was observed between the albumin and control groups. In addition, several complications, including infection and allergic reactions, may occur following albumin administration. Based on these findings, the potential risks for the albumin use as a standard resuscitation fluid likely outweighed the potential benefits.

Answer (opinion)

We suggest that the combination of multiple monitoring methods is used while considering the limitations of each indicator, for predicting fluid responsiveness during initial fluid resuscitation in patients with sepsis and septic shock (expert consensus/quality of evidence “C”) (rate of agreement, 94.7%).

Comment: The evidence was insufficient to support the recommendation of specific monitoring techniques to be used during initial resuscitation in patients with sepsis and septic shock.

Rationale

Two hundred seventy reports were identified as a result of a search for studies assessing survival in sepsis and septic shock patients who had undergone initial fluid resuscitation using various monitoring methods. After primary and secondary screening, five RCTs were extracted and used in the analysis [141–145]. Four RCTs included evaluation of PLR, one for transpulmonary thermodilution, and two for SVV; a meta-analysis was performed for each evaluation method.

No significant effect on the defined outcomes for analysis (mortality rate, ICU length of stay, time to shock reversal) was observed for these three evaluation methods. The monitoring methods used in the control group also varied and “performance of initial resuscitation without use of a specific monitoring method” established as the control for PICO (C) was not adopted. Therefore, it was determined that a serious issue regarding indirectness existed. Because of the difficulty in implementing study blinding, small sample size, the risk of bias, and various inaccuracies, the quality of the study was lowered, and the strength of this evidence was classified as weak (C) or very weak (D). Based on these findings, it was decided that the current evidence was not sufficient to support a recommendation and that an opinion (expert consensus) would be presented.

Regarding the benefit-risk balance, central venous catheters and arterial catheters are indwelled in most cases, and the use of some form of monitoring and optimization of infusion volume can result in improved prognosis. Therefore, it was concluded that the benefits of predicting fluid responsiveness during initial resuscitation likely outweigh the potential harms.

Answer (opinion)

We suggest that lactate levels over time are used when performing initial resuscitation in patients with sepsis (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

Primary and secondary screening of literature (174 sources) obtained after a search was conducted. As no RCTs applicable to this CQ (comparing lactate levels over time during initial resuscitation in sepsis) could be found, it was decided that an expert consensus would be presented instead of a recommendation.

Regarding the benefit-risk balance, lactate levels are associated with the patient’s prognosis in sepsis, and measuring lactate levels can aid in identifying critically ill patients. In addition, according to a report by Jansen et al., as a result of comparing patients with a lactate level of 3.0 mEq/L or higher (proportion of patients with sepsis was approximately 40% in both groups) to a comparator group that underwent initial therapy with lactate clearance as an indicator, no significant difference in in-hospital mortality rate was observed in a univariate analysis. However, in a multivariate analysis, in-hospital mortality rate improved in the group in which lactate clearance was used as an indicator [146]. Therefore, performing initial resuscitation while monitoring and assessing lactate levels over time may improve patient’s prognosis in sepsis. Arterial punctures and invasive arterial catheter insertion may also cause mechanical complications such as hematoma and embolism as well as infection, and in order to confirm lactate clearance, frequently measuring and analyzing blood gas content during initial resuscitation becomes necessary. However, as monitoring through an arterial catheter is believed to be performed in many patients, the quantity of blood collected per measurement is small, and the burden on patients appears to be minimal. Therefore, it was concluded that the potential benefits of the use of lactate levels as an indicator during initial resuscitation clearly outweigh the potential harms.

Answer (opinion)

Either ScvO₂ or lactate clearance may be used as indicators of initial resuscitation (expert consensus/quality of evidence “D”) (rate of agreement, 94.7%).

Rationale

Only one RCT (Jones et al. [118]) comparing ScvO₂ and serum lactate values was identified, and there was no significant difference in in-hospital mortality rates between using ScvO₂ and lactate clearance to guide initial resuscitation.

Regarding the benefit-risk balance, a specific central venous catheter is needed for continuous ScvO₂ monitoring. Collecting blood samples to evaluate ScvO₂ or lactate levels may increase the workload on physicians and medical staff, as well as the risk of infection. However, the potential benefits were determined to outweigh the potential harms since both measurements enable the assessment of oxygen transport capacity in tissues.

Answer (recommendation)

We recommend the use of noradrenaline as a first-line vasopressor to treat patients with septic shock that is unresponsive to initial resuscitation (1B) (rate of agreement, 100%).

Rationale

The systematic review reported by Avni et al. [147] was adopted as evidence as it had the highest quality. A total of 14 RCTs comparing noradrenaline and dopamine were identified, but these studies did not examine the time required to recover from septic shock. Noradrenaline administration significantly improved the 28-day mortality rate in comparison to dopamine (risk ratio, 0.89 [95% CI 0.81–0.98]). Regarding the ICU length of stay, the MD was 1.01 (95% CI − 0.65–2.66) as a result of the comparison between noradrenaline and other vasopressors, and no significant differences were observed. Regarding the incidence of complications, noradrenaline use resulted in a significantly lower incidence of complications compared to dopamine (risk ratio, 0.34 [95% CI 0.14–0.84]).

Regarding the benefit-risk balance, noradrenaline use resulted in a significant improvement in 28-day mortality rate in comparison to dopamine, and the frequency of harmful complications (fatal arrhythmias, myocardial/cerebral/upper, or lower limb ischemia/infarction, etc.) was significantly lower. Therefore, it was concluded that the potential benefits of noradrenaline use to treat patients with septic shock clearly outweighed the potential harms.

Answer (opinion)

We suggest that adrenaline is used in cases in which the maintenance of hemodynamic status is insufficient despite appropriate fluid resuscitation and noradrenaline administration (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

A literature search yielded 365 reports evaluating the effects of adrenaline in septic shock when noradrenaline fails to achieve a target blood pressure, and of these, eight were identified through the primary screening. No RCTs conforming to the PICO process for this CQ were found.

Regarding the benefit-risk balance, no RCT was examined for this CQ. Adrenaline as a first-line vasopressor has not shown significant improvements in mortality rates in comparison with noradrenaline. However, 20–40% of cases of septic shock are associated with SIMD which predicts worse outcomes [148], and it has been suggested that the administration of adrenaline may improve cardiac function in cases complicated by SIMD [149]. Although adrenaline use is associated with side effects such as tachycardia, decreased tissue perfusion, and lactate acidosis, no study has shown worse outcomes following adrenaline administration. It was concluded that the potential benefits of adrenaline use when noradrenaline fails to achieve target blood pressure clearly outweighed the potential harms.

Answer (opinion)

We suggest that vasopressin is used in patients with septic shock who show evidence of persistent hypotension despite adequate fluid resuscitation and the use of noradrenaline (expert consensus/quality of evidence “B”) (rate of agreement, 100%).

Rationale

A literature search yielded 365 records, and two RCTs [150, 151] were extracted for the meta-analysis after primary and secondary screening. ICU length of stay, 28-day mortality rate, and complication rate were assessed in these two RCTs, but time to shock reversal was not assessed. The risk ratios for 28-day mortality rate and complication rate were 0.90 (95% CI 0.76–1.07) and 0.73 (95% CI 0.24–2.23), respectively. The MD in ICU length of stay was − 0.95 days (95% CI − 1.73 to − 0.17).

In the two RCTs [150, 151], noradrenaline or noradrenaline plus vasopressin was administered when vasopressors were required to maintain target blood pressure despite adequate fluid resuscitation. Compared to using noradrenaline alone, the evidence of this systematic review and meta-analysis showed that adding vasopressin decreased the ICU length of stay by 1 day, but no difference was observed regarding 28-day mortality rate. In addition, no difference was observed in complication rate between the two groups. Based on these findings, it was concluded that the benefits of adding vasopressin when using noradrenaline alone fails to achieve target blood pressure likely outweigh the harms of adding it.

Answer (opinion)

We suggest that dobutamine is used in septic shock when cardiac function remains diminished, and maintenance of hemodynamics is insufficient despite adequate fluid resuscitation and noradrenaline administration (expert consensus/quality of evidence “C”) (rate of agreement, 94.7%).

Rationale

Two RCTs [122, 123] were identified involving patients with septic shock in whom blood pressure could not be maintained with adequate fluid resuscitation and noradrenaline administration and cardiac function was normal or decreased. Adrenaline was administered to the control group. The risk ratio with respect to 28-day mortality rate was 0.88 (95% CI 0.69–1.13), and the incidence rate of complications was 0.87 (95% CI 0.62–1.22). The MD for the time to shock reversal and ICU length of stay were −1.00 day (95% CI − 1.89 to − 0.11) and 1.00 day (95% CI 0.33–1.67), respectively.

Regarding the benefit-risk balance, although the superiority of dobutamine versus adrenaline is not recognized, the 28-day mortality rate remained at approximately 40% in the RCT including patients predicted to have a very high risk of death, and as such, there appears to be some benefit in administering dobutamine. There were also no differences observed regarding the frequency of complications such as arrhythmia in comparison to the patients that received adrenaline. Based on the above observations, it was determined that the benefits of administering dobutamine likely outweigh the potential harms.

Answer (recommendation)

Corticosteroids should not be administered if patients recover from septic shock with adequate fluid resuscitation and vasopressor therapy. We suggest that a low-dose corticosteroid (HC) is administered to promote recovery from shock in adult patients with septic shock that is not responsive to the initial fluid resuscitation and vasopressors (2B) (rate of agreement, 94.7%).

Rationale

We searched for references published in 2004 or later, when the standardized care for sepsis was introduced by SSCG 2004 after being defined in 1992 as a systemic inflammatory response syndrome (SIRS) accompanying infection, with the keywords “septic shock” and “low-dose steroid,” and identified eight RCTs (Bollaert et al. [162], Briegel et al. [163], Chawla et al. [164], Annane et al. [165], Oppert et al. [166], Mussack et al. [167], Sprung et al. [156], and Arabi [168]). As these eight studies were all included in the meta-analysis by Wang et al. [160], we initially decided to use it for this CQ. However, after a second search with the same keywords at the end of December 2015, one new report by Gordon et al. [161] was identified, and we thus performed a new meta-analysis including this study.

There were nine and six RCTs studying 28-day mortality rate and the proportion of shock reversal by day 7, respectively. With regards to adverse events, six RCTs analyzed infection, six assessed gastrointestinal bleeding, and three assessed hyperglycemia. There were no problems regarding a risk of bias, inconsistency, or indirectness in the analysis of 28-day mortality rate, the proportion of shock reversal by day 7, and complication rates. However, the confidence intervals (CIs) for the two complications (infections, gastrointestinal bleeding) were wide and susceptible to inaccuracy, and their evidence levels were thus lowered by one. The risk ratios (RRs) for 28-day mortality rate and the proportion of shock reversal by day 7 were 0.96 (95% CI 0.81–1.13) and 1.32 (95% CI 1.19–1.46), respectively. The RRs for infection, gastrointestinal bleeding, and hyperglycemia were 1.09 (95% CI 0.88–1.35), 1.35 (95% CI 0.85–2.13), and 1.15 (95% CI 1.07–1.25), respectively.

Although the rates of adverse events increased or tended to increase, the proportion of shock reversal by day 7 increased significantly, and 28-day mortality rate tended to decrease; thus, we judged that the benefits likely outweighed the harms.

Answer (opinion)

We suggest that corticosteroids are administered within 6 h after the onset of septic shock to treat adult patients with septic shock that is not responsive to the initial fluid resuscitation and vasopressors (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

No RCTs comparing whether the therapeutic effects and side effects of low-dose corticosteroids in adult patients with septic shock differ depending on the timing of administration (early initiation vs. late initiation) could be found. The French study involving the administration of corticosteroids within 8 h after onset of shock [165] has demonstrated a superior 28-day mortality rate as well as improved the proportion of shock reversal compared to the CORTICUS study [156] involving corticosteroid administration within 72 h after the onset of shock. Two observational studies have recently reported efficiencies in the timing of administering corticosteroids to treat septic shock. In 2012, Park et al. [169] conducted a retrospective study using a time-dependent Cox regression model to assess the administration of corticosteroids in patients with septic shock (178 cases). As a result, it was found that the 28-day mortality rate was significantly lower in the early group that received corticosteroids within 6 h of onset of shock compared to the late group that received steroids after 6 h or more from the onset of shock (51 vs. 32%, RR 0.63; 95% CI 0.42–0.93, p = 0.002). According to the prospective study (170 cases) conducted by Katsenos et al. [170] in 2014, inotropes may be discontinued earlier in patients with early initiation of HC (< 9 h after inotropes) in comparison to patients with late initiation of HC (> 9 h after inotropes) (log-rank 18.248, p = 0.000019), and 28-day mortality rate also declined (52.2 vs. 30.6%; Fisher’s exact test, p = 0.012). Based on the above findings, the early administration of corticosteroids within 6 h of the onset of shock is recommended when steroids are to be administered for the treatment of septic shock.

Regarding the benefit-risk balance, early steroid administration for the treatment of septic shock can be expected to promote shock reversal, resulting in the prevention of irreversible organ failure caused by prolonged hemodynamic derangement and to reduce the mortality rate. There have been no reports concerning complications arising due to the timing of administration, and no increased burden on medical staff associated with the timing of corticosteroids administration (early initiation, late initiation) is anticipated. However, since no RCTs conforming to the Patients, Intervention, Comparison, Outcome (PICO) process could be found, it was determined that the benefit-risk balance is still unknown.

Answer (opinion)

We suggest that 300 mg/day or less of HC for a maximum period of approximately 7 days is used when steroids are required for patients with septic shock (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

No RCTs examining and comparing whether the therapeutic effects and side effects of corticosteroids administered to adult patients with septic shock differ depending on the dosage and administration period could be found. It was concluded that high-dose steroid administration, which had been practiced until the 1990s, was ineffective or even harmful based on the results of two RCTs and one meta-analysis [158, 159]. Low-dose, long-term administration of HC was practiced in the 2000s, and improvements in mortality rates were also reported in addition to improvements in the proportion of and time to shock reversal. In large-scale RCTs by Annane et al. [165] and Sprung et al. [156], faster shock reversal was observed in both studies as a result of administering HC 200 mg/day in four divided doses, although the 28-day mortality rate was significantly lower in the Annane study than that in the Sprung study. In a meta-analysis by Annane et al. [171] assessing 17 RCTs, HC administration methods were divided into four according to the high dose/low dose (with a 300 mg/day limit) and long term/short term (with a five-day limit) and examined. Improvements in both the proportion of shock reversal and the 28-day mortality rate were observed only in the low-dose/long-term administration groups. Also, in the meta-analysis of the latest Cochrane Review of dose and duration of treatment among the steroid administration groups by Annane et al. [172], treatment with a long course of low-dose corticosteroids (at least 5 days and 300 mg/day or less) significantly reduced 28-day mortality rate (RR 0.87, 95% CI 0.78–0.97), but did not improve in the high-dose/short-term administration groups.

While several reports recommend continuous infusion at 10 mg/h after intravenous infusion of 100 mg is given for the management of blood glucose [152], the utility of continuous intravenous infusion of steroids with long half-lives is not clear. The period of administration is not fixed at 5 days, and administration of steroids for a long time should be avoided. When discontinuing steroid therapy, gradually tapering off the dosage is safer than sudden discontinuation from the viewpoints for maintenance of hemodynamics and prevention of rebounds in immune function.

As mentioned above, administration of steroids at dosages not exceeding the 300 mg/day equivalent of HC is recommended to promote the proportion of shock reversal in patients with septic shock (over a maximum period of approximately 7 days).

Regarding the benefit-risk balance, low-dose long-term steroids that are administered until the time to shock reversal are expected to increase proportions of shock reversal and decrease the frequency of mortalities. On the other hand, high-dose short-term steroid administration is associated with a greater frequency of hyperglycemia and gastrointestinal bleeding resulting in worsened prognosis compared with the non-administration group. Although low-dose/long-term steroids did not increase the frequency of complications after an overall assessment of low-dose long-term steroids, it is necessary to pay sufficient attention to the risk of deterioration of long-term prognosis due to hyperglycemia, gastrointestinal bleeding, and increased risk of infection. However, no RCTs conforming to the PICO process could be found, and it was determined that the benefit-risk balance is still unknown.

Answer (opinion)

We suggest that HC or MPSL is administered to treat patients with septic shock (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

There were no RCTs comparing the therapeutic effects and side effects of different steroids administered in adult patients with septic shock. Glucocorticoids resulted in an improvement in the proportion of shock reversal and reduced the 28-day mortality rate in adult patients with septic shock. Although HC, a pharmacologic form of physiological cortisol, is most commonly used in large-scale RCTs, it is a short-acting steroid and has both glucocorticoid and mineralocorticoid effects. In addition, Meduri et al. [173] administered 1 mg/kg of MPSL, an intermediate-acting steroid with no mineralocorticoid effects, and then continued MPSL administration at 1 mg/kg/day for 14 days to patients with septic shock in the same way as is given to patients with acute respiratory distress syndrome (ARDS). MPSL is five times more potent than HC in terms of glucocorticoid activity, and its half-life is 1.3 times that of HC. However, the actual dosage of MPSL is approximately half of that of HC [174]. In a retrospective observational study [175] (involving HC: 21 patients: 50 mg × 4/day; MPSL: 19 patients: 20 mg × 2/day) comparing HC and MPSL with bioequivalent doses, no difference in 28-day mortality rate, the proportion of shock reversal, or incidence of complications were observed. We do not recommend combining fludrocortisone and HC. In an RCT assessing the combination of fludrocortisone and HC [165], there was no additional benefit associated with adding fludrocortisone to HC, and adding fludrocortisone increased the infection rate, especially for urinary tract infections. Dexamethasone which has greater glucocorticoid activity and a longer half-life should not be administered due to its immediate and prolonged suppressive effects on the hypothalamic-pituitary-adrenal axis [163]. Based on the above findings, we suggest using HC or methylprednisolone MPSL in patients with septic shock.

Regarding the benefit-risk balance, even if HC, or alternatively, MPSL is administered, although there was no significant difference between the groups with respect to the incidence of complications, it is necessary to pay sufficient attention to the risk of deterioration of long-term prognosis due to hyperglycemia, gastrointestinal bleeding, and increased risk of infection. However, no RCTs conforming to the PICO process could be found, and it was determined that the benefit-risk balance pertaining to patients in either group is still unknown.

Answer (recommendation)

We recommend initiating red blood cell transfusion during initial resuscitation in septic shock cases at hemoglobin levels of ≤ 7 g/dL (1B) (rate of agreement, 94.7%).

Comment: This CQ addresses red blood cell transfusions during initial resuscitation in septic shock cases, and does not address transfusions performed after the patient is hemodynamically stable.

Rationale

Although the SSCG 2012 [29] proposed performing red blood cell transfusions with a target hematocrit of 30% or higher while considering the risks of hypoxia and myocardial damage during septic shock, the debate is currently ongoing. As such, an analysis of the timing of initiation of red blood cell transfusions during initial resuscitation in septic shock cases was conducted.

Two randomized controlled trials (RCTs) [177, 178] conforming to the Patient, Intervention, Comparison, and Outcome (PICO) process were selected as final targets for analysis as a result of a search of the PubMed database. According to the results of a combined meta-analysis of the control groups (group C, 520 participants) and the intervention groups (group I, 524 participants) of these two RCTs, the risk ratio for the 28-day mortality rate in group I versus group C was 0.95 (95% confidence interval (CI) 0.80–1.11). The incidence of ischemic complications with respect to organ damage was reported for one study only, and the risk ratio for organ damage in the control group (group C, 489 participants) versus the intervention group (group I, 488 participants) was 0.9 (95% CI 0.58–1.39).

After a comparison of transfusions initiated at a hemoglobin level of ≤ 7 g/dL with transfusions initiated at ≤ 10 g/dL, no difference was observed with respect to 28-day mortality rate and incidence of ischemic complications, and no evidence supportive of initiating blood transfusions at hemoglobin values of ≤ 7 g/dL or ≤ 10 g/dL was found. However, targeting a higher hemoglobin value requires transfusion of greater volumes of red blood cells and increases the risk of adverse effects and complications associated with transfusions, such as infection and allergic reactions. In addition, in consideration of medical economics and the fact that these kinds of blood products originate from donated blood, initiating transfusions at hemoglobin values of ≤ 10 g/dL should be avoided due to the risk of various adverse events, and a threshold of ≤ 7 g/dL is recommended instead. However, threshold hemoglobin values for initiating red blood cell transfusions may change in cases involving heart failure or ischemic heart disease as underlying diseases, and further studies are required.

Answer (opinion)

We suggest against administration of fresh-frozen plasma to correct coagulopathies when patients exhibit no bleeding tendencies, and no surgical intervention is required (expert consensus/no evidence) (rate of agreement, 100%).

Comment: The use of fresh-frozen plasma should be considered in keeping with the Japanese Guidelines for Blood Product Use [176] when patients exhibit a bleeding tendency or when surgical treatment is needed.

Rationale

In Japan, fresh-frozen plasma is administered to sepsis patients who exhibit bleeding tendencies or when surgical intervention is required, but may also be given to treat coagulopathies. To date, no consensus has been reached regarding the clinical utility of administering fresh-frozen plasma to correct coagulopathies in sepsis patients. In addition, potential harms associated with fresh-frozen plasma use include the onset of transfusion-related acute lung injury (TRALI) (frequency of fatal TRALI due to fresh-frozen plasma use: 1:2–300,000 products [179]), among various other dangers. Thus, the use of fresh-frozen plasma in sepsis cases was examined for this CQ.

No RCTs assessing the clinical utility of administering fresh-frozen plasma to manage coagulopathies in sepsis patients could be found in the PubMed database. As there is currently insufficient evidence conforming to the PICO process, no recommendation could be offered for this CQ at this time, and an expert consensus is presented as an alternative.

Although there are currently no proven benefits or harm associated with the administration of fresh-frozen plasma to manage coagulopathies in patients not exhibiting bleeding tendencies or when no surgical intervention is required, administering fresh-frozen plasma increases patients’ risks of allergic reactions and infection accompanying blood transfusion, which can become a burden on circulation. Moreover, as of 2016, the cost of fresh-frozen plasma was approximately 80 United States Dollar (USD)/unit (one unit of plasma (approx. 120 mL) corresponds to 200 mL of blood). In actual practice, physicians should take note of patients’ individual views regarding donated blood products and also be aware that some patients or their family members may refuse blood transfusions for reasons such as religious beliefs.

Answer (opinion)

We suggest performing a platelet transfusion in sepsis cases when patients exhibit a bleeding tendency or when surgical treatment is needed, in keeping with the Japanese Guidelines for Blood Product Use [176] (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

In Japan, platelet preparations are often administered to sepsis patients who exhibit general bleeding tendencies or when surgical intervention is needed [176] in keeping with the Japanese Guidelines for Blood Product Use. However, there is currently no evidence upon which to base an assessment of how platelet transfusions affect the clinical course of sepsis patients. In addition, potential risks associated with platelet administration include the onset of TRALI (frequency of fatal TRALI due to platelet use: 1:3–400,000 products [179]), among various others. As such, the use of platelet preparations in sepsis patients was examined in this CQ.

No RCTs examining the clinical utility of platelet administration to sepsis patients could be found in the PubMed database. As there is currently insufficient evidence that conforms to the PICO process, no recommendation can be offered for this CQ at this time, and an expert consensus is presented as an alternative.

Although there are currently no proven benefits or harm associated with the administration of platelet preparations to patients not exhibiting bleeding tendencies or when no surgical intervention is required, platelet transfusion increases patients’ risks of allergies and infection, which can increase the circulatory burden. As of 2016, the cost of platelet preparations was approximately 720 USD/10 units (200 mL), and when administering platelets, physicians should take note of how patients’ individual views regarding donated blood products and the concept of blood transfusions can differ. Also, physicians must also be aware that some patients or their family members may refuse blood transfusions for reasons such as religious beliefs.

Answer (recommendation)

We recommend setting the tidal volume to 6–8 mL/kg predicted body weight when performing mechanical ventilation in adult patients with ARDS (1B: excerpted from the ARDSGL).

Rationale

Mechanical ventilation management in ARDS patients is very important in addition to treatment of the underlying disease. In particular, mechanical ventilation settings are the highest priority for ARDS patients. Several studies have been conducted regarding ventilation strategies designed for ARDS patients that restrict tidal volume as a lung-protective measure to reduce further lung injury and limit airway pressure.

As a result of the systematic review, only six 2013 Cochrane Review [202]-adopted RCTs [191, 203–207] involving the use of lung protective ventilation methods focusing primarily on low tidal volume in adult ARDS patients were identified. Mortality statistics were reported in all six studies (n = 1305), and while there was a difference in the follow-up period, this period tended to be shorter in the low tidal volume group (risk ratio, 0.84; 95% confidence interval (CI) 0.67–1.07). Airway pressure-associated injury (pneumothorax arising from elevated airway pressure) was also reported in all six studies, but no significant decrease was observed (risk ratio 0.82; 95% CI 0.48–1.41). Although the results of three RCTs were integrated with respect to the number of ventilator-free days (VFD), the mean difference significantly increased to 2.52 days (95% CI 0.53–4.51).

VFDs increase despite the observation of high carbon dioxide levels and respiratory acidosis in conjunction with low tidal volume. Low tidal volume can be used just by adjusting the mechanical ventilator settings, and resources required remain unchanged. Therefore, it is likely that the potential benefits outweigh the potential harms.

Answer (recommendation)

We suggest a plateau pressure of ≤ 30 cmH₂O when performing mechanical ventilation in adult ARDS patients (2B: excerpted from ARDSGL).

Rationale

Mechanical ventilation-related lung injury in adult patients with ARDS is likely to occur in conjunction with decreased pulmonary compliance. There has been some concern regarding how VALI not only extends the period of mechanical ventilation but also leads to an increased risk of mortality. Increased tidal volume during mechanical ventilation and elevated airway pressure have been raised as potential causes of VALI, and it is expected that both can be controlled by limiting plateau pressure (airway pressure at the point when airflow temporarily stops at the end of inspiration). However, the results of limiting plateau pressure are not all beneficial; lower plateau pressure may also lead to adverse events such as hypercapnia. Accordingly, the optimum plateau pressure for obtaining treatment benefits without causing VALI is currently uncertain, and validation is necessary and a high priority.

As a result of the systematic review, four RCTs (1132 patients) [191, 204, 205, 207] were identified. Prolonged VFD (2.5 days on average, 95% CI: 0.5-4.45) was confirmed by setting plateau pressure to ≤30 cmH₂O for 5-7 days after the initiation of mechanical ventilation. Although the risk of death (risk ratio: 0.84, 95% CI: 0.62-1.15) and pressure-related injury (risk ratio: 0.92, 95% CI: 0.65-1.31) exhibited declining trends, these observations were not statistically significant.

VFDs are extended by limiting plateau pressure to ≤30 cmH₂O. In contrast, hypoxemia, hypercapnia, and respiratory workload are each anticipated to increase, but each has a wide range of tolerance, and the risk of iatrogenic harm is considered low. This is possible only by adjusting the mechanical ventilator settings, and resources required remain unchanged. Therefore, it is likely that the potential benefits outweigh the potential harms.

Answer (recommendation)

We suggest setting PEEP values within a range such that the plateau pressure is ≤30 cmH₂O and hemodynamics are not impacted when performing mechanical ventilation in adult patients with ARDS (2B: excerpted from ARDSGL). We suggest the use of higher PEEP values in patients with moderate or severe forms of ARDS (2B: excerpted from ARDSGL).

Rationale

Some evidence suggests that PEEP may be a factor in correcting hypoxemia by promoting the recruitment of collapsed alveoli and preventing progression of VALI, but the optimal PEEP value for this purpose remains unclear.

Seven RCTs were found as a result of the systematic review, and no significant difference was observed between the high PEEP group and the low PEEP group with respect to in-hospital mortality (risk ratio: 0.93, 95% CI: 0.83-1.04), pressure-associated injury (risk ratio: 0.97, 95% CI: 0.66-1.42), and VFD (risk ratio: 1.89, 95% CI: -3.58 - 7.36). For the analysis of in-hospital mortality, studies permitting interventions other than adjustment of PEEP values capable of influencing patient outcomes were excluded from analysis. As a result, three studies, by Brower et al. [208], Meade et al. [209], and Mercat et al. [210] were included in the final analysis. As a result of a meta-analysis involving studies in which the influence of non-PEEP interventions could not be ignored [203, 207, 211, 212], no significant difference in mortality rate was observed in the high PEEP group in comparison with the low PEEP group (risk ratio: 0.87, 95% CI: 0.74-1.02). In addition, when focusing the analysis on moderate or severe ARDS cases (PaO₂/FiO₂ ratio ≤ 200 mmHg), the results of both the sub-analyses including the studies permitting non-PEEP interventions as well as the sub-analyses excluding such interventions indicated that the high PEEP group exhibited a significantly lower mortality rate than the low PEEP group (risk ratio: 0.82, 95% CI: 0.73-0.92, risk ratio: 0.85, 95% CI: 0.75-0.96, respectively).

Setting the PEEP value is possible just by adjusting the mechanical ventilator settings, and there is no extra workload on ICU staff. However, the benefits and risks of this technique remain unclear, and therefore, the benefit-risk balance is currently uncertain.

Answer (recommendation)

We suggest fluid restriction when managing adult ARDS patients (2B; excerpted from ARDSGL).

Rationale

Pulmonary edema occurring in ARDS patients is believed to be caused by vascular endothelial damage and vascular hyperpermeability. The larger number of ARDS patients increases the mortality rate, and pulmonary extravascular fluid volume is linked to ARDS severity and frequency of mortality.

However, to date, no RCT has reported an improvement in mortality with fluid management in ARDS patients. Although attempting to optimize body fluid volume is routinely attempted in the treatment of other conditions and is regarded as an important measure, the mechanisms controlling fluid balance in ARDS patients is not well understood. As such, this is a high-priority issue, and the current recommendation is to manage fluid balance on a daily basis to avoid excessive body fluid volume as much as possible.

As a result of the systematic review, three RCTs comparing adult ARDS patients subjected to some manner of fluid restriction against those who were not were found. Studies permitting the adjustment of the infusion burden on shock patients in addition to ARDS were excluded. Although a large number of cases were included in the Fluids and Catheters Treatment Trial (FACTT 2006 study) [213], the other two studies RCTs [214, 215] included only a small number of cases. No significant differences in short-term mortality were observed, and the number of VFDs within a 28-day period increased significantly (+ 2.5 days). There were also no differences with respect to renal replacement therapy within a 60-day period. Regarding the indicators of fluid management, although two RCTs comparing pulmonary extravascular water content with pulmonary artery wedge pressure [216] and central venous pressure [143] were found, neither study demonstrated an improvement in mortality rate. The former study results demonstrated a reduction in the mechanical ventilation period, while the latter study failed to demonstrate any particular clinical utility.

Shortening of the VFD can be expected as a result of limiting fluid infusion volume. However, there is a risk of electrolyte abnormalities when diuretics are used. There are currently no clear standard indicators for assessing fluid balance, but many medical institutions use some manner of indicators to evaluate hemodynamics. The addition of new indicators for this purpose is of low necessity as treatment objectives can be achieved with general practice techniques. Thus, the benefits greatly exceed the potential risks.

Answer

Rationale

Delirium occurring in critically ill patients is an acute cerebral disorder and a form of multiple organ dysfunction involving the central nervous system. Similar to other types of vital organ dysfunction, it is believed to worsen both short-term and long-term patient prognosis. The PAD guidelines [242] also state, based on numerous high-quality observational studies, that the onset of ICU delirium increases ICU stay time as well as the overall hospitalization time and can be a cause of impaired cognitive function and neurological disorders in the long-term. The PAD guidelines further state that the onset of delirium in critically ill patients is also an independent risk factor for poor prognosis, irrespective of the underlying disease. Due to these observations, it is important for physicians attending to sepsis patients in the ICU to correctly recognize the effects of delirium, as such patients frequently have sepsis as an underlying disease.

Answer (recommendation)

Rationale

The basic principle in treating delirium is the identification and elimination of causative factors, and the first step in implementing a non-pharmacological delirium protocol is to alleviate patients’ stress through appropriate adjustments to their environment. The promotion of early rising and sleeping during the night has attracted particular attention in recent years.

Results of early rising in interventional studies targeting critically ill adult subjects [231, 232] indicated reduced delirium incidence, a reduction in excessive sedation, and significant shortening of ICU stay time as well as overall hospital admission time at moderate evidence levels. PAD guidelines [242] have also accumulated a body of supporting evidence with respect to critically ill patients. Currently, as the only effective non-pharmacological approach in the treatment of ICU delirium is early-stage rehabilitation, its implementation is strongly recommended. Regarding sleep promotion, unfortunately, at the moment there is no research with a high level of evidence, but it is easy to appreciate that improvement of sleep quality is advantageous to patients, and many authorities recommend this.

Although one randomized controlled trial (RCT) examining musical interventions for patients under mechanical ventilation has been conducted [243], as there are currently no musical therapists currently practicing at medical institutions in Japan, introducing this type of intervention at this time is believed to be difficult. Medical interventions using music require assessing together with related medical staff which types of music and sound quality are likely to produce the desired clinical effects, methods for reducing noise and other interferences, and the selection of suitable speaker devices, among various other potential factors. However, it is unlikely that interventions using music pose any harm to patients, and even if the current body of supporting evidence level is low, this approach may be considered in addition to daily therapy regimens. Although few studies regarding musical interventions are limited to sepsis patients, each of these reports is considered applicable to sepsis patients as well.

Answer (recommendation)

The use of a pharmacological delirium prevention protocol to reduce the onset and duration of delirium in adult ICU patients is not necessarily required (insufficient data) (C: excerpted from the PAD guidelines).

Rationale

While haloperidol and various atypical antipsychotic drugs have conventionally been used as drug therapies for delirium, very few reports have been published to date supporting the efficacy of these therapies with respect to ICU delirium (including postoperative delirium). The PAD guidelines [242] state that the use of these drugs cannot be recommended firmly due to a lack of supporting data. In addition, the results of a meta-analysis [229] verifying the effects of non-pharmacologic delirium prevention protocols in critically ill adult patients also indicate that the prophylactic administration of antipsychotics to surgical ICU patients and the prophylactic administration of dexmedetomidine to mechanically ventilated patients may decrease the incidence of delirium. However, others argue that these drugs are generally used to treat delirium and cannot be said to have a significant impact on clinical outcomes, including mortality rate.

Currently, while there is a possibility that antipsychotics may cause a slight decrease in the frequency of delirium, this may not necessarily lead to improvement in a patient’s final prognosis, and in consideration of the various side effects of antipsychotics (e.g., onset of extrapyramidal symptoms, torsade de pointes, ventricular arrhythmias), the use of these drugs cannot be recommended as a routine approach. As the recognition of these observations is not yet mainstream, the importance of this particular clinical question is considered to be high.

Answer (recommendation)

We recommend the routine application of a “discontinue daily sedation” protocol or an “aim for a mild depth of sedation” protocol when treating adult patients under mechanical ventilation (1B: excerpted from the PAD guidelines).

Rationale

A moderate level of evidence indicates that a protocol calling for “discontinuing daily sedation” can improve prognosis more than continuous sedation in critically ill mechanically ventilated adult patients. A moderate level of evidence similarly indicates that a protocol calling for “aiming for a mild depth of sedation” can improve patient prognosis more than maintaining a deep level of sedation. For these reasons, the PAD guidelines [242] also recommend that either of these protocols is used when determining a type and depth of sedation. However, there is currently insufficient data to support protocols calling for “discontinuing daily sedation” or “aiming for a mild depth of sedation,” and even the PAD guidelines have not yet been able to offer a consensus.

Based on the above, the current consensus is that attending physicians may select sedation protocols calling for either “discontinuing daily sedation” or “aiming for a mild depth of sedation.” In addition, although there were few studies limited to sepsis patients and most of the studies forming the basis for this consensus targeted general critically ill adult patients, each of these studies is considered also to apply to sepsis patients.

It is already clear that excessive maintenance of a sedated state unnecessarily prolongs the period of mechanical ventilation leading to the occurrence of ventilator-associated pneumonia and unsuccessful weaning trials. Critically ill patients must be kept as alert as possible to conduct routine delirium monitoring, which has high clinical importance.

Answer (recommendation)

We suggest the use of “analgesia-first sedation” when treating adult patients under mechanical ventilation (2B: excerpted from the PAD guidelines).

Rationale

Analgesia-first sedation has been demonstrated to result in greater improvement in patient prognosis in comparison to hypnosis-focused sedation in critically ill adult patients under mechanical ventilation, and the PAD guidelines [242] also recommend this approach with the support of a moderate level of evidence. In addition, although critically ill patients must be kept as alert as possible in order to routinely assess for delirium, maintaining the sedation level of these patients, who are under a great deal of stress, as low as possible must also provide “sufficient pain relief” and has high clinical importance.

Many of the studies underpinning these observations targeted general critically ill adult patients, and although there are few studies limited to sepsis patients, each of these reports is considered to apply to sepsis patients as well.

Answer (opinion)

We recommend applying of the KDIGO clinical practice guidelines for diagnosing and determining the severity of sepsis-induced AKI (expert consensus/quality of evidence “D”) (rate of agreement, 100%).

Rationale

Seven observational studies [251–257] examining patient mortality rate as a clinical outcome were extracted during an assessment comparing the KDIGO guidelines and the AKIN and RIFLE criteria, and as a result of comparing AKI diagnoses based on the KDIGO guidelines with diagnoses based on the RIFLE or AKIN criteria, diagnoses based on the AKIN criteria were found to have higher accuracy or were reflective of similar in-hospital mortality rates. Although only the study by Peng et al. [257] focused specifically on sepsis cases, the KDIGO guidelines are currently still considered to be useful in predicting prognosis in patients with sepsis-induced AKI. However, few studies regarding renal prognosis have been conducted to date, and currently little is known about this aspect. All published studies found were observational studies, and there have been no studies assessing diagnostic criteria as an interventional approach. No additional burden is placed on patients or medical personnel during diagnostic assessment. Additional medical costs are required for tests procedures such as measurement of sCre and urine volume measurement over time. Virtually, all ICUs are capable of performing such measurements, and it is likely that the potential benefits outweigh the potential harms.

Answer (recommendation)

We suggest against the initiation of blood apheresis at an early stage in sepsis-induced AKI except when emergency apheresis is necessary, such as those involving advanced metabolic acidosis, hyperkalemia, or renal overflow (2C) (rate of agreement, 100%).

Rationale

Blood apheresis in sepsis-induced AKI has come to be performed not only to supplement renal function but also to prevent or treat organ disorders through the removal and control of inflammatory mediators. Initiation of blood apheresis at an early stage is believed to improve the ease with which inflammatory mediators can be removed and controlled, and many relevant randomized controlled trials (RCTs) have been conducted in recent years. Although most of these RCTs are multiple objective studies examining both renal supplementation and inflammatory mediator removal, attempting to assess these purposes separately is not practical. A new meta-analysis was conducted after the results of two large-scale RCTs were reported in May 2016.

Three RCTs [258–260] were used to perform the new meta-analysis. In the results, the impact of early-stage initiation of blood apheresis on 28- or 30-day mortality was a risk ratio of 0.83 (95% confidence interval (CI) 0.64–1.09), and a risk ratio of 0.51 for the rate of dialysis dependence on day 60 (95% CI 0.25–1.06). No statistically significant differences and no benefits of early-stage initiation were observed. However, although there was no significant difference in the rate of dialysis dependence, early-stage initiation is expected to have some benefits regarding clinical utility. Two multicenter RCTs [261, 262] were ongoing as of June 2016, and the consensus on this subject may change based on their results.

Although decreased mortality rate is an expected benefit of this intervention, no differences between the intervention group and the control group were observed with respect to ICU stay time, 28-day mortality rate, or rate of transition to chronic dialysis. In general, hemorrhagic complications and other adverse events are recognized as potential harms, but no differences in the incidence of these events were observed between the study groups in these two RCTs. The burden on medical staff will also increase as a result of this intervention. The potential harms will outweigh the potential benefits, and associated medical costs and staff burden will also increase. Moreover, there are serious concerns over the feasibility of performing this intervention at facilities without adequate human resources or staff proficient in its implementation.

Answer (recommendation and opinion)

We suggest that attending physicians select either continuous or intermittent blood apheresis when treating sepsis-induced AKI patients exhibiting stable hemodynamics (2B) (rate of agreement, 94.7%).

We suggest selecting continuous apheresis in patients exhibiting unstable hemodynamics (expert consensus/no evidence) (rate of agreement, 84.2%).

Rationale

When deciding whether to perform continuous or intermittent blood apheresis, it is important to consider various factors such as whether the medical staff can cope with the additional workload and whether suitable equipment are available. If either option is feasible, the discretion of the attending clinician during selection is considered to be of high importance.

Only one new RCT has been conducted since the existing systematic review [263], which also yielded the same results as the systematic review, and so the assessment was made using the existing systematic review. A total of 15 studies were examined; “in-hospital mortality” was assessed based on seven RCTs [264–270] and yielded a risk ratio of 1.01 (95% CI 0.92–1.12); “transition to chronic dialysis” was examined based on three RCTs [264, 266, 269] and yielded a risk ratio of 1.01 (95% CI 0.92–1.07); no difference was observed between apheresis performed continuously and intermittently. Three RCTs [269–271] were also examined with respect to reduction in blood pressure, and no differences in the risk ratio of 0.92 (95% CI 0.72–1.16) were observed in the studies. However, the two RCTs [266, 268] extracted from the systematic review followed protocols that excluded cases involving patients with unstable hemodynamics. Continuous blood apheresis is already recognized as a standard treatment in such cases, and no RCT conducted to date has compared continuous and intermittent apheresis.

Continuous and intermittent blood apheresis are considered to be equivalent regarding potential benefits relating to in-hospital mortality rate and rate of transition to chronic dialysis. Although there is currently a lack of evidence regarding potential harms, continuous blood apheresis is associated with an increased risk of bleeding, greater medical costs arising from a longer implementation period, and increased workload on medical technicians and nurses.

Based on the above, the expert consensus reached is that continuous apheresis is preferable when patients present with unstable hemodynamics. However, because continuously and intermittently performed variants of blood apheresis have different characteristics, in cases where minimizing anticoagulant use is necessary, attending intensive care physicians or renal specialists must carefully consider these characteristics when determining whether to implement measures such as performing short-term apheresis without administering an anticoagulant.

Answer (recommendation and opinion)

We recommend against increasing the volume of blood filtered beyond the international standard volume (20–25 mL/kg/h)(1B) (rate of agreement, 89.4%). It should also be noted that the evidence supporting the blood filtration volume covered by the NHI system (10–15 mL/kg/h) is weak (expert consensus/no evidence) (rate of agreement, 73.7%).

Rationale

RCTs are actively being conducted elsewhere to assess whether increasing blood filtration volume (filtrate volume + dialysis volume) can improve patient outcomes in sepsis-induced AKI. However, the NHI system defines an upper limit for filtration volume that is coverable under the system. As such, this CQ will reassess the current body of evidence.

No new RCTs were completed since the existing systematic review [272], and the assessment for this CQ was conducted using this systematic review. Eight RCTs comparing an intervention group (high-volume, 40 mL/kg/h) and a control group (international standard quantity, 25 mL/kg/h) were assessed with respect to “28-day mortality rate.” In addition, “recovery from kidney failure” was used to evaluate “transition to chronic dialysis.” The risk ratio for “28-day mortality rate” was 0.89 (95% CI 0.76–1.04) and the risk ratio for “recovery from kidney failure” was 1.12 (95% CI 0.95–1.31). Treatment effects were evaluated in the same manner even if the blood filtration volume was increased. There is no evidence from Japan regarding a comparison of the NHI-defined filtration volume and the international standard volume, and so evaluation was not possible.

No differences between the intervention and control groups were observed with respect to 28-day mortality rate and the rate of transition to chronic dialysis. Based on these findings, the effects of blood apheresis remain unchanged even after increasing the blood filtration volume. Although there is no supporting evidence, when selecting a fluid replacement solution for blood filtration, which are widely used as dialysis fluids or fluid replenishers in Japan, electrolyte imbalances such as hypokalemia and hypophosphatemia can easily occur as a result of increasing the apheresis volume. The labor burden placed on medical technicians and nurses who replace fluid pouches will increase as a result of this intervention. In consideration of the above, the potential harms likely outweigh the potential benefits.

Answer (recommendation)

We suggest against performing PMX-DHP as a standard treatment for patients with septic shock (2C) (rate of agreement, 84.2%).

Rationale

PMX-DHP is a specialized variant of acute blood apheresis used to remove and control inflammatory mediators. This technique was developed in Japan in 1994 as an endotoxin adsorption column, is covered by the NHI, and is widely performed as a supportive therapy in patients with septic shock.

Three RCTs (Vincent [273]; Cruz [274]; and Payen [275]) were extracted as a result of the systematic review conducted for this CQ. However, the participants in each of these studies presented with septic shock arising from intraperitoneal infection requiring emergency abdominal surgery. Each of these RCTs reported increased mortality, and two reported increased mean blood pressure. In contrast, none of these RCTs reported results related to shock recovery rate. The odds ratio with regard to the impact of PMX-DHP on mortality rate was 1.1 (95% CI 0.68–1.79), and no improvements in survival rate were observed. The studies by Vincent [273] and Cruz [274] examined increases in mean blood pressure. The mean difference observed was 4.59 (95% CI − 1.71–10.90), and no significant improvements in blood pressure were observed. No differences between the intervention and control groups were observed with respect to decreases in mortality rate or increases in mean blood pressure. Moreover, these RCTs did not assess shock recovery rate, and also did not recognize thrombocytopenia as an adverse event. Meanwhile, Payen [275] reported a significant decrease in the incidence of thrombocytopenia occurring on day 3 of treatment in the PMX-DHP group, but did not quantify the reductions in platelet count. Based on the above, the potential risks associated with PMX-DHP are likely to outweigh the potential benefits.

All RCTs extracted for this CQ-targeted subjects presenting with septic shock arising from intraperitoneal infection, and no RCTs targeting other patients with septic shock could be found. For this reason, cases other than those involving intraperitoneal infections cannot be examined at this time due to a dearth of evidence. One large-scale RCT, the EUPHRATES (Evaluating the Use of Polymyxin B Hemoperfusion in a Randomized controlled trial of Adults Treated for Endotoxemia and Septic shock) study [276] is currently in progress and is scheduled to be completed in 2016. The results of the EUPHRATES study are expected to be noteworthy as it targets participants presenting with septic shock arising from diseases other than intraperitoneal infection such as pneumonia, and also focuses on severe cases.

Answer (recommendation)

We suggest against administering furosemide to prevent or treat sepsis-induced AKI (2B) (rate of agreement, 94.7%).

Rationale

Although furosemide inhibits sodium reabsorption and produces a diuretic effect, this drug is expected to prevent or promote improvement in AKI by maintaining steady urine volume, and numerous clinical studies have been conducted to assess this potential.

Two meta-analyses [277, 278] and 11 RCTs were extracted. The results of both meta-analyses indicated no effect on in-hospital mortality rate and the need for blood apheresis as a result of furosemide administration, and no new RCTs have been completed to date. Although decreased mortality and need for dialysis were the expected benefits of this intervention, no significant differences between the intervention groups that received furosemide as a preventive or treatment and the control groups were observed with regard to in-hospital mortality and necessity of dialysis. Administration of high-dose furosemide (1–3.4 g/day) was associated with temporary hearing loss and tinnitus [278] (risk ratio, 3.97; 95% CI 1.00–15.78). Based on the above findings, the potential harms likely outweigh the potential benefits.

According to the meta-analysis by Ho et al. [277], furosemide administration demonstrated no significant effect on in-hospital mortality or the need for blood apheresis. Likewise, furosemide is not recognized as effective for the prevention of AKI or for promoting recovery of impaired renal function. Neither of the two RCTs that were limited to AKI patients undergoing blood apheresis showed a significant decrease in the blood apheresis period (days) or earlier recovery from renal dysfunction in the furosemide group. In addition, the results of another meta-analysis [278] indicated that high-dose furosemide, which is often given to treat AKI, resulted in a significantly increased incidence of symptoms such as tinnitus and hearing loss in comparison with the control group. In contrast, it has also been reported in actual clinical practice that the administration of furosemide improves electrolyte imbalances such as excessive body fluid volume and hyperkalemia. However, no RCTs limited to AKI patients presenting with such clinical signs have been reported to date. It should be noted that the above recommendation does not rule out the use of furosemide to manage excess body fluid volume.

Answer (recommendation)

We recommend against adinistering dopamine to prevent or treat sepsis-induced AKI (1A) (rate of agreement, 100%).

Rationale

Although dopamine is expected to cause renal vasodilation and natriuresis at low doses (1–3 μg/kg/min) resulting in a renoprotective effect, its clinical utility has not yet been verified. In contrast, there are also concerns that dopamine administration can lead to the incidence of adverse effects such as tachycardia, myocardial ischemia, and reduced intestinal blood flow. Complications such as arrhythmia and cardiac/limb/skin ischemia have occurred clinically, but no significant increase in these events was observed in the meta-analysis targeting AKI cases. However, it has been reported that dopamine administration in the treatment of sepsis significantly increases the frequency of arrhythmias [279]. Based on the above observations, clarifying the clinical utility of low-dose dopamine in patients with sepsis-induced AKI is considered to be of importance to attending physicians.

Two meta-analyses were extracted [279, 280], and it has been shown that administering dopamine does not lead to improvement in terms of either prolonged survival time after receiving low-dose dopamine or decreased rate of dialysis initiation. No new RCTs involving patients with sepsis have been completed to date.

Although reduced mortality and need for dialysis were expected benefits of this intervention, no significant differences between the intervention and control groups in terms of in-hospital mortality rate and the need for dialysis were observed.

The risk ratio for adverse effects relating to arrhythmia and ischemic findings was 1.13 (95% CI 0.90–1.41) and was not statistically significant. Based on the above, the potential harms likely outweigh the potential benefits.

Answer (recommendation)

We suggest against administering ANP to prevent or treat sepsis-induced AKI (2B) (rate of agreement, 94.7%).

Rationale

ANP (carperitide, a synthetic analogue) is a cardiac hormone discovered in Japan together with brain natriuretic peptide (BNP) and C-type natriuretic peptide (CNP) and acts as a vasodilator, suppresses sodium reabsorption, and increases glomerular filtration rate via dilation of the afferent renal arterioles and contraction of the efferent renal arterioles. In the prevention and treatment of AKI, ANP is expected to exhibit a renoprotective effect resulting from its diuretic effect and ability to increase glomerular filtration rate, and many clinical studies have been conducted to verify this effect. However, some have reported that the administration of high doses of ANP can increase the frequency of adverse events such as hypotension and arrhythmia. Based on the above, clarifying the clinical utility of ANP in patients with sepsis-induced AKI is considered to be of importance to attending physicians.

Two meta-analyses were extracted, and neither demonstrated any reduction in the frequency of blood apheresis. Subsequently, no new RCTs on this subject have been completed to date. The incidence of complication by hypotension significantly increased (risk ratio, 1.69; 95% CI 1.29–2.22), but lower doses did not yield a significant association with hypotension (risk ratio, 1.25; 95% CI 0.87–1.81). In addition, the AKI preventive effect obtained with low doses has been verified based primarily on cases of AKI occurring after heart surgery, and the results of another meta-analysis [281] also suggest the utility of ANP. However, the level of assessment with respect to sepsis cases remains insufficient.

Although reduced mortality and need for dialysis were expected benefits of this intervention, no significant differences between the intervention and control groups were observed regarding in-hospital mortality rate and the need for dialysis.

Overall, associated risks included the significantly greater likelihood of complication by hypotension, and lower doses did not yield a significant association with hypotension. Based on the above, the potential harms likely outweigh the potential benefits.

Answer (recommendation)

We recommend that enteral nutrition rather than parenteral nutrition in sepsis patients (1B) (rate of agreement, 94.7%).

Rationale

Several studies targeting critically ill patients requiring ICU management were extracted from the JSICM guidelines. These studies included 11 RCTs concerning mortality rates [301–311] and seven RCTs concerning infection frequency [303–307, 310, 312].

Regarding potential benefits, although no improvement in mortality rate was observed in seven studies focusing on trauma, one on sepsis, one on pancreatitis, and two on surgery or serious internal illness, the benefit of enteral nutrition with respect to reduced infection frequency was confirmed in five studies focusing on trauma, one on sepsis, and one on pancreatitis. However, the study by Harvey et al. was not included in the analysis regarding the incidence of infection as it did not describe the incidence of all infections. Also, no complications impacting prognosis were observed, and as enteral nutrition was found to reduce the burden of medical costs as well [295], it was concluded that the benefits outweigh the potential risks.

Answer (recommendation)

We recommend that enteral nutrition be initiated early (within 48 h) if a patient will be unable to maintain a sufficient oral caloric intake during the first several days after the onset of sepsis (1C) (rate of agreement, 94.7%).

Rationale

Nine RCTs were extracted based on the systematic review conducted for this CQ [313–321]. Mortality rate was assessed by eight RCTs; infection frequency, the length of ICU stay, and duration of ventilation were assessed by seven RCTs, and length of hospital stay was assessed by four RCTs. The risk ratio for the impact of early enteral nutrition on mortality rate was 0.9 (95% confidence interval (CI) 0.52–1.41), the risk ratio for the impact on infection frequency was 0.7 (95% CI 0.51–1.02), the impact on length of ICU stay was − 2 days (95% CI − 5.25–1.18), no impact was observed with respect to length of hospital stay (0 days; 95% CI − 17.18–16.53), and the impact on duration of ventilation was − 1 day (95% CI − 4.82–2.49).

Regarding potential benefits, although no differences in mortality rate, duration of ventilation, or length of hospital stay were observed based on a comparison of patients that were started on enteral nutrition within 48 h of ICU admission and patients receiving enteral nutrition after 48 h, the incidence of infections was lower in the early feeding group. In the sensitivity analysis, an additional intervention group started on enteral nutrition within 24 h of ICU admission was established, and it was found that the incidence of infection and length of ICU stay were significantly lower in the early feeding group. Regarding potential harm, no clear adverse effects were observed, and as enteral nutritional solutions are inexpensive and their use does not increase medical staff workload, it was concluded that the benefits outweigh the potential risks.

Answer (recommendation and opinion)

We suggest against administering an amount of calories approximating typical caloric intake during the initial ICU period (approximately 1 week) if the patient was not malnourished before the onset of sepsis (2C) (rate of agreement, 89.5%).

We suggest against limiting caloric intake in malnutrition. In this situation, nutrition should be provided while considering the risk of developing refeeding syndrome (expert consensus/no evidence).

Rationale

Six RCTs [322–327] included in the systematic review of the JSICM guidelines were extracted. All six RCTs reported on mortality rate, three reported on infection frequency, two RCTs reported on the duration of ventilation, the length of ICU stay, and length of hospital stay; four RCTs reported on ventilator-associated pneumonia (VAP); and two RCTs reported on the frequency of continuous renal replacement therapy (CRRT) use. The risk ratio for the impact of caloric restriction on mortality rate was 0.93 (95% CI 0.83–10.7), and the risk ratio for the impact on the frequency of infection was 1.08 (95% CI 0.83–1.41). In addition, the effect of caloric restriction on duration of ventilation was − 1.04 days (95% CI − 3.29–1.20), the effect on length of ICU stay was −1.8 days (95% CI − 4.22–0.86), the impact on length of hospital stay was − 0.84 days (95% CI − 19.2–17.5), the risk ratio for the effect on VAP incidence was 0.9 (95% CI 0.68–1.17), and the risk ratio for the effect on frequency of CRRT use was 0.64 (95% CI 0.45–0.91).

The observed benefit was a lower frequency of CRRT administration in the calorie restricted feeding group. No clear adverse effects were observed. However, some previous studies have reported observing adverse outcomes (such as more hospital transfers than discharges) associated with low-volume enteral nutrition (trophic feeding). In addition, enteral nutrition is inexpensive and requires small quantities of feeding solution, and as such, its potential risks are small.

Answer (recommendation and opinion)

We suggest against that the initiation of parenteral nutrition within the first week of hospitalization if there was no malnutrition prior to the onset of sepsis/septic shock. However, enteral nutrition may be initiated within 1 week of hospitalization (2D) (rate of agreement, 84.2%).

We suggest that initiating parenteral nutritional supplementation should be considered while monitoring the risk of developing refeeding syndrome if malnutrition is observed before ICU admission, or enteral nutrition cannot be initiated within 1 week of hospitalization (expert consensus/no evidence).

Rationale

In a meta-analysis of six RCTs extracted from the JSICM guidelines [287] ([328], [296], [297], [298], [299], and [300]), all six reported on mortality rate, four reported on bloodstream infection or respiratory infection, and five reported on urinary tract infection. The risk ratio for the impact of initiating parenteral nutrition within 1 week on mortality rate was 0.95 (95% CI 0.81–1.11), the risk ratio for the impact on bloodstream infections was 1.22 (95% CI 1.02–1.46), the risk ratio for the impact on respiratory infections was 1.07 (95% CI 0.87–1.32), the risk ratio for the impact on urinary tract infections was 1.12 (95% CI 0.84–1.49), and bloodstream infections significantly increased.

Although no benefits were observed in the context of this CQ, the incidence of bloodstream infections increased in the group that was started on total parenteral nutrition within 1 week. In addition, the greater cost of parenteral nutrition could place an additional burden on these patients. Based on these observations, the potential harm was determined to outweigh the potential benefits.

Selective bias in the studies adopted for this CQ was higher in those involving patients with a higher BMI compared to the Japanese patients and the group with high caloric deficit because we excluded patients in whom malnutrition was observed prior to ICU admission. In addition, the EPaNIC (Early versus Late Parenteral Nutrition in Critically Ill Adults) study accounted for 70% of the participants in both groups, more than half of whom were post-cardiac surgery patients. The fact that many of these studies were conducted as open-label studies has been pointed out as a source of execution bias. Thus, in light of the weak body of clinical trial evidence, it was determined that the benefits and risks of this intervention should be assessed in the clinical setting.

Answer (recommendation and opinion)

We suggest that the use of parenteral nutrition in cases where enteral nutrition cannot be initiated within 1 week after the onset of sepsis/septic shock or if malnutrition is observed (expert consensus/no evidence) (rate of agreement, 94.7%).

We suggest against administering 100% of the target caloric value in such cases (2C). However, the optimal value remains uncertain (expert consensus/no evidence).

Rationale

No RCTs conforming to the PICO process could be found. Evidence for this CQ was gathered from three essential papers [297, 298, 328]. Differences in the nutritional management discussed in these three papers were examined, and the results were used to formulate the opinions serving as a basis for the recommendation offered by this CQ (expert consensus). One more RCT [329] was found as in an additional search. This was a small-scale study examining 50 cases involving gastrointestinal disease. The incidence rates of sepsis in 50 consecutive cases requiring nutritional therapy for 5 days or more were compared between a group receiving 60% of the typical caloric value based on Schofield’s estimation formula and a group receiving 100% of the typical caloric value. The incidence of sepsis (3 cases vs. 12 cases; p = 0.003), the incidence of systemic inflammatory response syndrome (SIRS) (9 cases vs. 16 cases; p = 0.017), and the frequency of nutrition-related complications (2 cases vs. 9 cases; p = 0.016) were significantly lower in the 60% administration group. Based on these results, it is recommended that 100% of the target caloric value should not be given at least during the acute phase.

Regarding potential harm, early parenteral nutrition is expected to prevent various complications arising from an increased caloric deficit, particularly in malnutrition. However, the possibility that administering 100% of the target caloric value during parenteral nutrition may increase the risk of complication by infection could not be ruled out.

Answer (recommendation)

We suggest an optimal target blood glucose range of 144–180 mg/dL in patients with sepsis (2C) (rate of agreement, 100%).

Rationale

Our analysis included 14,495 patients from 27 studies [128, 330–332, 337–359]. None of the studies were especially prone to bias. We performed an NMA in addition to a direct comparison because there were no studies that directly compared patients with a target blood glucose level > 180 mg/dL to those with a target blood glucose level of 144–180 mg/dL.

The direct comparison revealed that compared with target blood glucose levels of 144–180 and > 180 mg/dL, target blood glucose levels of 110–144 mg/dL were associated with a significantly higher risk of hypoglycemia; however, there were no significant differences in the risk of mortality and infection.

The NMA showed that target blood glucose levels of 144–180 mg/dL were associated with a significantly lower risk of hospital mortality and sepsis or bloodstream infection compared with blood glucose levels > 180 mg/dL, and there were no significant differences in the risk of hypoglycemia between the target blood glucose levels of 144–180 and > 180 mg/dL.

The quality of evidence in the direct comparison was B (moderate) to C (weak), and that in the NMA was C (weak) to D (very weak). Finally, we decided that the quality of evidence was C (weak).

Reduction of hospital mortality and infection is considered a benefit of blood glucose management. Our NMA showed that compared with the target blood glucose levels > 180 mg/dL, levels of 144–180 mg/dL were associated with a significantly lower risk of hospital mortality and sepsis. Furthermore, there were no significant differences between the target blood glucose levels < 110 and 110–144 mg/dL with regard to these outcomes. The increase in the incidence of hypoglycemia is considered an adverse effect of blood glucose management.

Therefore, we assessed the balance between the risk and benefit of blood glucose management as follows: (1) target blood glucose levels < 110 and 110–144 mg/dL—equilibrium or uncertain, (2) target blood glucose levels of 144–180 mg/dL —benefit may exceed risk, (3) target blood glucose level > 180 mg/dL—risk may exceed benefit.

Answer (recommendation)

We recommend against the use of a glucometer with capillary blood in patients with sepsis (1B) (rate of agreement, 94.7%).

We suggest the use of a glucometer with arterial/venous blood (2B) and recommend the use of an arterial blood gas analyzer (1C) in patients with sepsis (rate of agreement, 94.7%).

Rationale

The recommendation presented for this CQ was based on the results of a systematic review conducted by Inoue et al. [336] In a comparison of measurements taken using a glucometer with capillary blood or a blood gas analyzer using arterial blood, measurements taken using the arterial blood gas analyzer were less likely to include measurement errors falling significantly outside the acceptable range (odds ratio, 0.04; 95% confidence interval (CI) 0.01–0.14) [336]. In addition, in a comparison of glucometer measurements using capillary blood or arterial blood, the risk of measurement error was significantly lower when arterial blood was used (odds ratio, 0.36; 95% CI 0.25–0.52). No significant difference was observed between glucometer measurements using arterial blood and blood gas analyzer (arterial blood), but measurement errors tended to be lower when the blood gas analyzer (arterial blood) was used (odds ratio, 0.17; 95% CI 0.01–2.46) [336].

Regarding the benefit-risk balance, it was determined that (1) the potential risks of glucometer (capillary blood) use clearly outweighed the benefits, (2) the benefits of glucometer (arterial/venous blood) use likely outweighed the potential risks, and (3) the benefits of using a blood gas analyzer (arterial blood) clearly outweighed the potential risks.

Answer (recommendation)

We suggest against the routine use of antipyretic therapy for septic patients presenting with fever (2C) (rate of agreement, 94.7%).

Comment: Administering antipyretic therapy may be considered as a method of alleviating physiological responses accompanying fever, such as tachycardia, tachypnea, and discomfort.

Rationale

Six randomized controlled trials (RCTs) were extracted as a result of the systematic review conducted for this CQ [374–379]. All six RCTs assessed mortality rates, one assessed ICU-free survival days, four examined the length of ICU stay, and two examined the rates of complication from infection.

The risk ratio for the impact of antipyretic therapy on mortality rate was 1.12 (95% confidence interval (CI) 0.83–1.51), the impact on the number of ICU-free survival days was + 1 day (95% CI − 0.38–2.38), and the risk ratio for the impact on the length of ICU stay was − 0.04 days (95% CI − 0.76–0.68). Regarding the incidence of infection, one RCT reported the frequency of infection and another RCT reported the number of patients who developed infections; thus, it was not possible to compare the two.

In this CQ, mortality rate, ICU-free survival days, and length of ICU stay were assessed as benefits of antipyretic therapy, and the incidence of infectious complications was assessed as a potential harm associated with antipyretic therapy. The strength of the evidence concerning the primary outcome of mortality rate was C (weak). While the strength of the evidence regarding ICU-free survival days and length of ICU stay was B (moderate strength), there was no consistent assessment method for the potential harm of complication by infection, and evaluating the strength of this evidence was not possible. The strength of the evidence regarding all outcomes was C (weak) in keeping with the strength of the evidence on mortality rate as the primary outcome.

The risk ratio for the impact of antipyretic therapy on mortality rate was 1.12 (95% CI 0.83–1.51), but the extent to which it reduces mortality is unclear. The results of one RCT suggested that ICU-free survival days may increase by approximately 1 day. No significant shortening in length of ICU stay was observed. The results of two other RCTs indicated the clear potential for an increased risk of complication by infection as a result of antipyretic therapy. Accordingly, the balance of risk and benefit of administering antipyretic therapy for septic patients with fever is still uncertain. In addition, the use of antipyretic drug therapy or external cooling can be expected to increase the workload of medical staff.

Answer (opinion)

We suggest that the correction of body temperature while carefully considering hemodynamic stability for hypothermic sepsis patients with hypothermia-related complications such as diminished cardiac contractility/dilatability or coagulation abnormalities (expert consensus) (rate of agreement, 100%).

Rationale

There were no RCTs conforming to the Patient, Intervention, Comparison, Outcome (PICO) process. Attempts to gradually correct patient body temperature in cases where diminished cardiac contractility/dilatability or coagulation abnormalities occur in hypothermia [380, 381] are believed to offer potential benefits when such symptoms are determined to be hypothermia-related. However, carefully monitoring for possible hemodynamic instability due to factors such as decreased blood pressure or the relative decrease in circulating blood volume is necessary while patients recover from hypothermia.

The benefit-risk balance can differ depending on the condition of the patient. Attending physicians should determine whether to act while considering both the adverse effects of hypothermia itself as well as the risk of attempting to correct body temperature and if treatment is necessary, body temperature should be corrected gradually through methods such as extracorporeal circulation, passive incubation, or covering the patient with a blanket.

Coagulation and fibrinolytic changes in sepsis

In the clinical course of sepsis, coagulation/fibrinolytic abnormalities are identified early on, and the risk of death due to multiple organ failure increases markedly when complicated by disseminated intravascular coagulation (DIC) [382, 383]. DIC in sepsis manifests primarily as a state of severe systemic activation of the blood coagulation cascade, and it is believed that microcirculatory damage caused by intravascular coagulation can become a cause of organ damage [384]. In DIC, although fibrinolytic function also increases in response to the activation of the coagulation cascade, the extent of this increase varies depending on the underlying disease, and thus, DIC can occur with suppressed fibrinolysis or with enhanced fibrinolysis. Of these, DIC occurring in sepsis exhibits typical fibrinolytic suppression patterns in which fibrinolytic function is suppressed relative to the enhancement of coagulation [385]. In addition, DIC with suppressed fibrinolysis is considered to have a poor prognosis, particularly when complicated by multiple organ disorders [385].

Necessity of diagnosing DIC in sepsis

There are two factors underlying the significance of evaluating the coagulation/fibrinolytic status of patients with sepsis: (1) gaining an accurate grasp of the patient’s disease state and (2) determining the need for therapeutic intervention [386]. The results of numerous studies have indicated that DIC is linked with poor prognosis in sepsis [387], and diagnosing DIC is necessary to predict outcomes and to determine the timing of interventions. As anticoagulant therapy carries a risk of excessive bleeding, selecting appropriate cases and applying rationally planned timing are essential [382]. Administering anticoagulants inappropriately will not only fail to produce the desired results but can also increase patients’ risk of adverse events [388]. Therefore, when treating patients with sepsis, attending physicians should acquire a sufficient grasp of the state of coagulation/fibrinolytic functions in real-time and perform suitable interventions after the diagnosis of DIC.

Usefulness of anticoagulant therapy for sepsis-associated DIC

Many anticoagulants have been evaluated to date in connection with sepsis-associated DIC based on the understanding that excessive coagulation activity can cause microcirculatory damage leading to organ failure [389]. However, there is currently no unified view regarding the efficacy of these drugs. One reason for this is that in the USA and European countries, the clinical utility of various anticoagulants has been studied primarily in large-scale randomized controlled trials (RCTs) involving patients with severe sepsis [390, 391]. These studies did not focus on sepsis-associated DIC, and the anticoagulant therapies administered to the participants of these studies clearly differed from those used in Japan. According to the results of a recent meta-analysis, anticoagulants cannot be expected to be effective in general sepsis cases, and their effectiveness is reported to be limited to the treatment of sepsis-associated DIC [387]. This guideline was formulated based on the entirety of available data, although the current body of evidence regarding anticoagulant therapy for sepsis-associated DIC is still limited in both quality and quantity.

Answer (opinion)

The Japanese Association for Acute Medicine (JAAM) DIC diagnostic criteria have been recognized as valid both as treatment initiation criteria and useful as an index of severity and are also considered useful in diagnosing sepsis-associated DIC (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

The question of what diagnostic criteria should be used for the diagnosis of sepsis-associated DIC is frequently encountered in routine practice, and the current state of this issue on the background of multiple diagnostic criteria in use is addressed by this CQ. Given the fact that the JAAM DIC diagnostic criteria (hereinafter referred to as “acute-phase criteria”) created in 2005 by the Research Committee on DIC of the JAAM have gained widespread acceptance in Japan, this CQ focuses the discussion on the clinical utility of the acute-phase criteria.

The main diagnostic criteria currently in use are the Japanese Ministry of Health, Labour and Welfare DIC Diagnostic Criteria [392], the International Society on Thrombosis and Hemostasis (ISTH) overt DIC Diagnostic Criteria [393], and the acute-phase criteria [382] created by the JAAM.

It is impossible to discuss the superiority or inferiority of these diagnostic criteria, as there is no “correct diagnosis” for DIC. The acute-phase criteria was formulated for the purpose of facilitating early-stage DIC diagnoses, of the three diagnostic criteria, DIC should be diagnosed for the most widespread coagulopathies [382, 394]. This criterion has become the most widely used diagnostic criterion for sepsis-associated DIC in Japan because of the simplicity of its associated diagnostic procedures.

The acute-phase criteria are considered to be useful for diagnosing sepsis-associated DIC because they have been recognized as valid both as treatment initiation criteria [395] and also suitable as an index of disease severity [396]. The acute-phase criteria are considered to be useful for diagnosing sepsis-associated DIC and appear to present minimal risks. As such, it is believed that the potential benefits likely outweigh the potential risks.

Answer (opinion)

No clear recommendation can be offered at this time concerning the use of recombinant thrombomodulin preparations in sepsis-associated DIC (expert consensus/quality of evidence “B”) (a rate of agreement of 67% or higher in support of its use could not be obtained).

Rationale

Recombinant thrombomodulin, which was first launched in 2008, is one of the several anticoagulants that have come into wide use in treating sepsis-associated DIC in Japan. However, to date, no consensus has been reached regarding its usefulness. Although the bulk of clinical evidence regarding this drug has been the product of a Japanese phase III trial [397] and a multiregional Phase II trial [398], the scales of these studies were determined to have been inadequate, and no recommendation was offered. Another multiregional phase III trial is currently in progress with results expected in 2018, and this clinical question (CQ) was formulated based on the currently available body of evidence in view of the importance of clarifying the potential benefits and risks of recombinant thrombomodulin use in sepsis-associated DIC treatment. An existing systematic review [399] was used to formulate this recommendation.

The systematic review for this CQ was conducted using three RCTs [398, 400, 401] as mentioned previously. The quality of evidence was “B” (moderate), and it was determined that the potential benefits likely outweigh the potential risks. The estimated effect of this treatment intervention on mortality rate was represented by a risk ratio (RR) of 0.81 (95% confidence interval (CI) 0.62–1.06), and the point estimate for the number needed to treat was 15, with moderate benefits expected. The estimated effect on hemorrhagic complications was represented by an RR of 0.83 (95% CI 0.22–3.11), and it was determined that there was a low risk that the treatment intervention increases the frequency of hemorrhagic complications. Meanwhile, a study by Aikawa [400] comparing thrombomodulin therapy against heparin as a control was excluded from analysis due to the possibility that its evaluation of hemorrhagic complications may not have been appropriate. As a result, the RR was 1.11 (95% CI 0.59–2.11), and the possibility that thrombomodulin may cause a slight increase in the frequency of hemorrhagic complications could not be eliminated.

The potential benefits of treatment intervention in this CQ were evaluated in terms of its capacity to improve mortality rate. Although the possibility of increased frequency of hemorrhagic complications as a result of thrombomodulin therapy could not be ruled out, it was determined to be highly likely that the potential benefits of its use outweigh the potential risks. However, there are different viewpoints regarding how to interpret the benefit-risk balance, and no consensus could be reached as some committee members emphasized that no significant differences in beneficial outcomes were observed. It was decided that no recommendation would be offered at this time in anticipation of the results of the aforementioned ongoing phase III trial in 2018.

Answer (recommendation)

We suggest the use of antithrombin replacement therapy in patients with sepsis-associated DIC whose antithrombin activity has decreased to ≤ 70% (2B) (rate of agreement, 68.4%).

Rationale

Guidelines for use in various countries do not recommend the use of antithrombin based on the results of a large-scale clinical trial (the KyberSept trial) [390]. In contrast, antithrombin replacement therapy is often used in patients with sepsis-associated DIC in Japan. On this background, this guideline discusses the clinical utility of antithrombin replacement therapy based on a repeat analysis of patients with sepsis-associated DIC.

This analysis targeted studies limited to patients with sepsis-associated DIC. The KyberSept trial [390] assessed severe sepsis rather than sepsis-associated DIC, and so could not be adopted as evidence for this CQ, as it does not consider treatments for sepsis patients without DIC. Thus, a post hoc analysis in a study by Kienast et al. [402], which was limited to sepsis-associated DIC, was adopted. Meanwhile, the efficacy and risks of the dosage covered by the National Health Insurance for the intervention itself (1500 units/day for patients with sepsis-associated DIC with antithrombin activity levels of 70% or lower, and 40–60 units/kg in surgical cases) were also assessed.

In addition, four Japanese and overseas studies [395, 402–404] were analyzed. The treatment intervention was expected to be beneficial in terms of reduced mortality rate based on an RR of 0.68 (95% CI 0.49–0.93), but bias risk/indirectness may lower the reliability of the effect estimate. Although the estimated effect on hemorrhagic complications was represented by an RR of 1.17 (95% CI 0.45–3.01) and the possibility of harm could not be eliminated, the confidence interval was wide, and thus, the reliability of this finding was low.

The benefit of this treatment intervention was evaluated in terms of its capacity to reduce mortality. Although the possibility of increased frequency of hemorrhagic complications as a result of antithrombin replacement therapy could not be ruled out, it was determined to be highly likely that the potential benefits of use outweigh the potential risks.

Answer (opinion)

We suggest against the use of protease inhibitors as standard treatment in sepsis-associated DIC (expert consensus/quality of evidence “D”) (rate of agreement, 89.5%).

Rationale

The body of evidence concerning the usefulness of protease inhibitors is poor, and only two RCTs [404, 405] have been conducted to date. However, the results of these studies are currently referred to frequently in Japan. On this background, this guideline offers a discussion of the results of an analysis of the clinical utility of protease inhibitors in patients with sepsis-associated DIC.

There have been two RCTs that have evaluated the utility of protease inhibitors in sepsis-associated DIC; Hsu et al. [405] reported a lack of utility, while Nishiyama et al. [404] reported a potential utility. However, both reports were based on small-scale studies that were not double-blinded. Although these studies also assessed 28-day mortality rate, they did not assess the frequency of hemorrhagic complications or the DIC recovery rate. No significant improvements in mortality rates were observed in these studies, and it was concluded that no prognostic improvement effect attributable to protease inhibitors could be determined, and at present, the available evidence is insufficient to support a recommendation. In addition, although the RCT conducted by Nishiyama et al. [406] included trauma patients, it was ultimately decided that the results of this study were not suitable for adoption as evidence for this CQ due to a lack of sub-analysis results.

The benefits of the treatment intervention were considered in terms of its ability to improve mortality rate, but the confidence intervals in the results were determined to contain major inaccuracies. Hemorrhagic complications were raised as a potential harm associated with this intervention and a major outcome, but the two RCTs extracted did not assess this outcome. As such, it was determined that the risks of this intervention could not be assessed, and thus, its benefit-risk balance is uncertain.

Answer (opinion)

We suggest against the use of heparin or heparin analogs as a standard treatment in sepsis-associated DIC (expert consensus/quality of evidence “D”) (rate of agreement, 84.2%).

Rationale

As mentioned previously, no conclusions have been reached regarding the utility of anticoagulant therapy for sepsis-associated DIC. This assessment is particularly difficult because heparin and heparin analogs are often administered to sepsis patients to prevent deep vein thrombosis regardless of the presence of DIC. Thus, to verify the clinical efficacy of heparin/heparin analogs as treatments for sepsis-associated DIC, this CQ discusses the results of an assessment limited to patients with sepsis-associated DIC.

28-day mortality rate, hemorrhagic complications, and DIC recovery rate were adopted as outcomes. Three RCTs [400, 407, 408] conforming to the Patient, Intervention, Comparison, Outcome (PICO) process were selected to be included in the analysis. Of these three RCTs, unfractionated heparin was used in the study by Aikawa et al. [400] as a control for recombinant thrombomodulin, while in the study by Aoki et al. [407], unfractionated heparin was used as a control for activated protein C concentrate. A report by Liu et al. [408] focused on patients with sepsis-associated pre-DIC as defined in the Chinese diagnostic criteria. The 28-day mortality rate statistics were reported by Aikawa et al. [400] and Liu et al. [408], and the quality of evidence was determined to be “D” (very weak) as a result of the meta-analysis.

Statistics pertaining to hemorrhagic complications were reported by Aikawa et al. [400] and Aoki et al. [407], and the quality of evidence was determined to be “D” (very weak) as a result of the meta-analysis. Only Aikawa [400] reported on DIC recovery rate, and the quality of this evidence was also determined to be “D” (very weak) as a result of the meta-analysis.

It was determined that the treatment intervention had the beneficial effect of improving the mortality rate, but the supporting evidence was found to be inadequate. Hemorrhagic complications were cited as potential serious adverse outcomes of this intervention, and it was concluded that the available evidence was insufficient to support a recommendation regarding hemorrhagic complications. In addition, although beyond the scope of the meta-analysis, heparin-induced thrombocytopenia (HIT) is a known complication associated with heparin use, and caution is required when administering heparin. Accordingly, it was concluded that the benefit-risk balance for this intervention is currently uncertain.

Answer (opinion)

We suggest the use of anticoagulant therapy, compression stockings, and/or intermittent pneumatic compression for the prevention of DVT in accordance with a patient’s risk level (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

VTE affecting inpatients and postoperative patients is widely recognized as a complication requiring preventative measures. In Japan, the Guidelines for the Diagnosis, Treatment and Prevention of Pulmonary Thromboembolism and Deep Vein Thrombosis (Revised 2009) [409] and the Guidelines for Preventing Pulmonary Thromboembolism/Deep Vein Thrombosis (Venous Thromboembolism) [415], offer a classification of various risk factors for developing VTE as well as corresponding prophylactic measures. Among them, severe infection is classified as a moderate risk factor for VTE and establishing a patient’s risk level based on the underlying disease and clinical history are recommended. SSCG 2012 [23] recommends the administration of low molecular weight heparin (grade 1B) or unfractionated heparin (grade 2C) in addition to intermittent air compression of the lower extremities (grade 2C) for prophylaxis against DVT. However, the recommendations offered by these guidelines are based on studies not limited to sepsis patients, as well as data from postoperative or critically ill patients admitted to the ICU. The analysis conducted for this guideline was limited to sepsis patients to allow for the assessment of DVT prevention measures in such patients.

While no studies limited only to sepsis patients were found as a result of the literature search, DVT prophylactic measures are expected to prevent both pulmonary thromboembolism and mortality in sepsis patients, similar to other critically ill patients. However, there is currently no evidence limited only to sepsis patients, and the frequency and severity of adverse effects are unknown. The potential risks associated with anticoagulants include hemorrhagic complications and heparin-induced thrombocytopenia (HIT), and care must be taken with their use. Compression stockings and intermittent pneumatic compression should be used with caution in patients with arterial blood flow disorders, such as diabetes, due to the associate risk of exacerbation of circulatory disorders. However, because the interventions addressed in this CQ include intravenously administered pharmacologic therapies, compression stocking use, and intermittent pneumatic compression, the intervention itself creates little physical burden on patients. Based on the above, the potential benefits of VTE prevention measures clearly outweigh the risks.

Answer (opinion)

We suggest diagnosing DVT using techniques such as clinical symptoms, D-dimer fluctuations, venous compression ultrasonography findings, and contrast-enhanced computed tomography scan in assessing risk factors and adverse effects (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Sepsis patients often present with high D-dimer values accompanying disseminated intravascular coagulation, and ruling out DVT based only on D-dimer values can be difficult. However, actively monitoring for DVT onset in cases involving prolonged elevation or repeated spikes in D-dimer values during treatment is considered to be good practice.

Patients with sepsis are at a high risk of developing VTE, but it should be noted that their clinical symptoms are easily masked by factors such as sedation. As a result, attending physicians should check for associated features such as elevated D-dimer values. The early diagnosis and treatment of DVT is considered to have a high potential for benefiting patients. In contrast, the use of contrast agents is contraindicated in sepsis cases accompanied by cardiac and renal dysfunction, and the burden on the patient should be carefully evaluated in cases involving artificial respiration or continuous apheresis due to the risks involved in transporting patients. Accordingly, the benefit-risk balance may also vary depending on the condition of the patient.

Answer (recommendation)

We suggest against performing electrical muscle stimulation as an ICU-AW preventative measure when handling sepsis or ICU patients (2C) (rate of agreement, 100%).

Rationale

Although artificial respiration period, ICU stay time, and hospitalization period may each increase as a result of ICU-AW, no consensus has been reached regarding effective therapies for ICU-AW, and preventative measures are expected. Electrical muscular stimulation induces muscle contractions by percutaneously channeling low-frequency electrical currents. In some cases, sufficient rehabilitative therapy may not be implemented for patients with chronic heart failure and chronic obstructive pulmonary disease arising from exertional dyspnea, and electrical muscle stimulation, which can be performed even in a resting state is used as an alternative therapy [431, 432]. Although improvements in muscular strength and exertion capacity have been reported as results [433], the effectiveness of this intervention in critically ill patients or sepsis patients is currently unknown. As such, this CQ examines the capacity of electrical muscle stimulation to prevent the onset of ICU-AW.

Two single-center RCTs have reported on the efficacy of electrical muscle stimulation as a measure to prevent ICU-AW [434, 435]. According to the results of an intent-to-treat analysis [436] of the findings obtained by Routesi et al. [434] and Kho et al. [435], there were no significant differences in the incidence of ICU-AW in comparison with the control group. This body of evidence can be said to be inadequate, in consideration of the low number of subjects in the electric muscle stimulation group and the associated risk of bias, in addition to the lack of a high-quality systematic review or meta-analysis at this time. Three single-center RCTs examining whether muscle mass increases as a result of electrical muscle stimulation [437–439] were identified and subjected to a meta-analysis [440]. Although subject muscle mass increased significantly, the total number of subjects in the electric muscle stimulation group was low (72 subjects) and the bias risk was high, and accordingly, this evidence may be considered to be of poor quality. No studies analyzing the period of artificial respiration and ICU stay time have been conducted. Although the incidence of ICU-AW is the most important outcome considered by this CQ, due to the low statistical power of the two studies examined, this body of evidence was rated C (weak).

To perform this intervention, patients must undergo electrical muscle stimulation in the lower limbs for approximately 1 hour each day with time allotted for rest periods. Some pain may occur during this intervention. Although the labor burden placed on nurses, attending physicians and physical therapists in connection with this intervention is not anticipated to be great, medical facilities must possess an electrical stimulation apparatus to perform the intervention, and equipping all facilities for this intervention is considered to be impracticable.

Answer (recommendation)

We suggest implementing early-stage rehabilitation as a PICS preventative measure for sepsis or ICU patients (2C) (rate of agreement, 100%).

Rationale

PICS, in which the functional prognosis of the body’s physical, cognitive, and psychological capacities deteriorate as a result of staying in the ICU, has increasingly been cited in recent years as a concern facing ICU patients, including sepsis patients. The epidemiology, prevention, and treatment of PICS are also gaining greater research attention. Early-stage rehabilitation interventions are conducted as preventive measures. No RCTs assessing such interventions in sepsis cases only has been conducted to date. However, several RCTs targeting intensive care patients have been completed, and it was determined that this evidence could be extended to sepsis cases as well to assess validity.

We evaluated the effect of early-stage rehabilitation on PICS-related outcomes used in our meta-analysis (eight studies: [441], [442], [443], [444], [445], [446], [447], and [231]) and two meta-analyses [436, 448]. These results revealed that implementation of early-stage rehabilitation interventions resulted in significantly improved motor function, 6-min walk distance (6MWD), and shortened artificial respiration periods. Although no significant differences were observed regarding the incidence of PICS, because intervention subjects’ exhibited significant improvements in MRC score (an ICU-AW assessment tool), 6MWD, and artificial respiration period, this intervention is expected to be more beneficial than harmful. However, patients targeted by the studies examined were ICU patients rather than sepsis patients, and moreover, the scale of the studies for each PICS-related outcome was small. As such, in considering the influence of the median/interquartile range to mean/standard deviation conversion applied during the meta-analyses, the level of this body of evidence cannot be said to be high. In addition, the lack of analysis of side effects also increased the difficulty of the assessment.

To administer this intervention, patients must participate in a rehabilitation program scheduled on a daily basis. This intervention will increase the workloads of nurses, physical therapists, and attending physicians. Administering the intervention with great care and under adequate observation is necessary in serious disease cases, and this intervention should be considered to be highly technical. As such, serious concern is warranted regarding the feasibility of performing this intervention at facilities without sufficient medical personnel or other appropriately trained staff.

Answer (opinion)

Pediatric sepsis is currently defined as “SIRS arising from infection” according to the criteria and definitions proposed by Goldstein et al. in 2005 [455]. Although “severe sepsis” was defined as “a condition accompanying organ dysfunction” in the definition of Goldstein et al., this term has now been replaced with “sepsis,” in line with the terminology revision in the Sepsis-3 [5] definition. When using this criterion, refer to the proposed changes presented in CQ19-2 below pertaining to the number of inspirations, hypotension, and creatinine value. However, because the concept of SIRS was excluded from the adult Sepsis-3 definition [5], equating similar expressions, specifically “organ dysfunction accompanying infection,” to “sepsis” should not be dismissed (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Meanwhile, a new definition for adult sepsis (Sepsis-3) was proposed in 2016 [5]. Under this new definition, sepsis refers to infection (including suspected infection) accompanying organ dysfunction with a SOFA score of two or higher [454]. However, no pediatric SOFA score system currently exists. The definition of pediatric sepsis is expected to be revised internationally in the future.

Answer (opinion)

A clear recommendation regarding the threshold value for respiratory rate in the diagnosis of pediatric SIRS cannot be offered at this time. As an example, refer to the criteria proposed by Nakagawa & Shime [460] (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Answer (opinion)

A clear recommendation concerning a threshold value for systolic blood pressure as a diagnostic criterion for pediatric septic shock cannot be offered at this time. As an example, refer to the hypotension criteria used in the SPROUT (Sepsis Prevalence, Outcomes, and Therapies) study [463] (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Therefore, although septic shock can be said to be a relatively broad concept under Goldstein’s definition [455], under the Sepsis-3 definition [5], severe septic shock is defined in the group of patients presenting with shock facing a particularly high degree of mortality risk (approximately 40% mortality in adults). That is, the Goldstein definition [455] anticipates an intention to administer early-stage treatment, while the Sepsis-3 definition [5] is considered to isolate a specific group of patients with a relatively high mortality risk. The proportion of pediatric patients satisfying the criteria for septic shock under both the Goldstein definition [455] and the Sepsis-3 definition [5] is currently unknown.

Currently, the Goldstein definition [455] is used as a standard for pediatric septic shock, and the low blood pressure thresholds to be used in such cases are described in Table 4.

Answer (opinion)

We suggest establishing respective reference creatinine values for pediatric patients in different age groups (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Answer (opinion)

We suggest the use of pediatric blood culture bottles for pediatric patients (until approximately school-age) (expert consensus/no evidence). We also suggest the use of adult bottles even in pediatric patients if their physique is similar to an adult’s and the patient can safely sustain blood collection (approximately ≥ 36 kg) (expert consent/no evidence) (rate of agreement, 100%).

Rationale

Blood culture testing is an essential technique for the optimization of antibiotic therapy in the treatment of infections/sepsis. In adults, typically as much blood is collected as possible to improve test accuracy. However, when handling pediatric patients, the collection of large volumes of blood is more difficult due to issues related to circulating blood volume, and collection methods similar to those employed in adult patients are generally not used [466]. Blood culture test procedures suitable for children, including the use of pediatric blood culture bottles, are therefore necessary.

No randomized controlled trials (RCTs) conforming to the Patient, Intervention, Comparison, Outcome (PICO) process could be found. In observational studies, the comparison of aliquots (up to 5 ml) of blood from both adult and pediatric bottles revealed a high rate of positivity in the pediatric bottles and a short time to positive test results [467]. In essence, this suggests that test results are more likely to be positive when a pediatric bottle is used even if the volume of blood collected is smaller. The target blood sample volume for neonates is 1–2 mL, 2–3 mL for infants, 3–5 mL for infants/school-age children, and 10–20 mL for pubescent children [468]. The use of pediatric bottles is generally desirable with pediatric patients. The use of adult bottles is appropriate with older children of school-age (approximately ≥ 36 kg) who are able to provide a sufficient volume of blood.

Because the positivity rate for blood volumes lower than 1 mL is low even when a pediatric bottle is used [466], pediatric bottles should contain at least 1 mL of blood. As the same amount of blood is collected, the frequency of adverse effects is not likely to increase. Therefore, the potential benefits likely outweigh the potential risks.

Answer (opinion)

Adrenaline is the first-line vasopressor drug for use in treating septic shock in pediatric patients (expert consensus/quality of evidence: “C”).

Noradrenaline should not be used as the first-line treatment drug in pediatric patients with septic shock (expert consensus/no evidence).

Vasopressin should not be used to treat vasodilatory septic shock in pediatric patients (expert consensus/quality of evidence: “C”).

Dobutamine or milrinone may be used to treat pediatric septic shock as appropriate given the patient’s condition (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

As with adults, the infusion of appropriate volumes of fluid often does not improve the hemodynamics of pediatric septic shock patients, and the use of circulatory inotropes often becomes necessary. Although the consensus statement for the management of pediatric patients with severe sepsis mentions that “positive inotropes/vasoconstrictors should be administered as soon as possible to pediatric patients who are unresponsive to transfusion loading,” no specific drugs were recommended [452].

SSCG 2012 also does not offer strong recommendations regarding the use of any specific circulatory inotrope in children [23]. Noradrenaline is recommended for use in adults, but there have been no RCTs upon which to validate its use in children. The selection of circulatory inotropes for use in children presenting with septic shock is a routinely encountered issue in practice, and it is important to maintain current knowledge about which drugs are effective under different circumstances.

One RCT [469] reported that adrenaline use in pediatric patients with sepsis was associated with a lower mortality rate in comparison to dopamine [469] and recommended adrenaline as a first-line drug therapy. However, this was a single-center study with 120 participants, and it is necessary to recognize that the design of this study weakens it as an evidentiary basis for the use of adrenaline as a first-line drug.

While there have been no RCTs conducted to date comparing noradrenaline with other drugs or placebo, noradrenaline may also be considered for first-line use in patients exhibiting high cardiac output and peripheral vascular dilatation. Another RCT investigating children presenting with vasodilatory shock reported that low-dose vasopressin temporarily increased the blood pressure compared with placebo, but worsened survival and prognosis [470]. It is important to note, however, that this study did not focus only on sepsis patients. In the management of pediatric patients with septic shock, vasodilators may be considered in cases characterized by peripheral vasoconstriction and blood pressure is maintained, but there is currently no basis for recommending the use of milrinone. Moreover, there have been no RCTs conducted to date comparing dobutamine with other drugs or placebo.

In another study [469] comparing adrenaline and dopamine, adrenaline was associated with a lower rate of mortality than dopamine. Potential risks include a tendency for dopamine to exacerbate inflammation, and adrenaline was linked to hyperglycemia and persistent hyperlactatemia. Based on these observations, the potential benefits of adrenaline use are believed to outweigh the risks. According to the results of a comparison of vasopressin and placebo [470], vasopressin was linked to a higher mortality rate and incidence of adverse events, and the risks likely exceed the potential benefits.

Answer (opinion)

CRT should be used in conjunction with other hemodynamic indicators to monitor the state of circulatory management in pediatric sepsis patients (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

According to SSCG 2012 [29], when performing initial resuscitation, ideal conditions include a CRT within 2 s, normal blood pressure for the age group, normal heart rate, no difference in between the central and peripheral pulses, warmness of the extremities, 1 mL/kg/h urine volume, normal consciousness, central venous oxygen saturation (ScvO₂) of 70% or higher, and a cardiac index of 3.3–6.0 L/min/m². However, the measurement of ScvO₂ requires the insertion of a central venous line, and accurately measuring the cardiac index in children can be difficult. A CRT of ≤ 2 s in children admitted to the PICU has been reported to be correlated with an ScvO₂ ≥ 70% [471]. A meta-analysis also found that abnormal CRT was linked to a higher risk of mortality [472].

It is important to verify the significance of CRT as a noninvasive and continuous circulatory management index that can easily be used by clinicians in the initial diagnosis of pediatric patients with sepsis, particularly in Japan, which has underdeveloped intensive care treatment systems.

No RCTs conforming to the PICO process could be found. CRT is an index of circulatory management that can be evaluated noninvasively and repeatedly over time. Correlations between CRT and ScvO₂ as well as between abnormal CRT and mortality have also been reported [472], and the use of CRT as an indicator of hemodynamic status is believed to offer substantial benefits to patients.

In contrast, the use of CRT alone to gauge the state of circulatory management may lead to excessive treatment intervention. CRT is also influenced by a variety of factors such as measurement site, compression time, and temperature [473], and it must be recognized that inconsistencies in the method of evaluation may lead to erroneous interpretation [474]. However, there are no RCTs conforming to the PICO process, and therefore, the benefit-risk balance is currently unknown but is believed to differ depending on the unique circumstances of individual patients.

Answer (opinion)

Both ScvO₂ and serum lactate values may be used as indicators of circulatory management in pediatric sepsis cases. However, use of comprehensive circulatory evaluation together with other hemodynamic indicators is required (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

ScvO₂ and serum lactate value are both indicators of the tissue oxygen supply/demand balance, and numerous studies have been conducted to investigate the positive and negative aspects of their use as indicators of hemodynamic status in patients with sepsis. ScvO₂, in addition to central venous pressure, mean arterial pressure, and urine volume, have been recommended as hemodynamic indicators when performing initial resuscitation in adult sepsis patients [29]. However, the clinical utility of quantitative protocols for circulatory management applying these metrics has been called into question in recent years. In children, the measurement of ScvO₂ has been recommended in conjunction with the use of indices measurable through noninvasive methods, such as vital signs, peripheral circulation, and urine volume [29, 453]. However, the invasiveness and costs accompanying this technique warrant its reconsideration.

SSCG 2012 recommends the use of normalization of serum lactate values as an indicator of initial resuscitation in adult sepsis patients presenting with hyperlactatemia [29]. Also, according to Sepsis-3 [5], elevated serum lactate level is used to define septic shock. The utility of the absolute or chronological lactate value in predicting mortality or organ dysfunction in pediatric patients with severe sepsis has also been reported based on an observational study [452]. An evaluation of the utility of these metrics in sepsis cases as well must be conducted in the future.

One RCT [475] conforming to the PICO process was found. Application of the American College of Critical Care Medicine (ACCM)/Pediatric Advanced Life Support (PALS) guideline [453], which calls for the continuous monitoring of ScvO₂, was linked to a decrease in 28-day mortality rate and a lower incidence of novel organ disorders in comparison to cases where ACCM/PALS was not applied. However, this was an unblended overseas study, and it must also be recognized that baseline characteristics of the two study groups were different.

No RCTs investigating the use of serum lactate value in managing circulation could be found.

ScvO₂ and serum lactate value are both indicators of the tissue oxygen supply/demand balance, and implementing circulatory management aiming for normalization of these indices is considered valid. However, the placement of a central venous catheter or an arterial catheter is necessary. While both are useful in the management of pediatric patients with sepsis, adverse events such as mechanical complications during placement, catheter-related bloodstream infection (CRBSI) following placement, thrombosis, or peripheral blood flow disorders may occur. Also, sedation, as well as tracheal intubation, may be necessary when placing the central venous catheter in pediatric patients, and the workload of attending medical staff can also be expected to increase. However, the frequency of complications will vary depending on the physician/technician, the facility environment, and individual patient characteristics, and clinical benefits obtained will also vary depending on the patient’s condition. Therefore, the benefit-risk balance is expected to vary by case.

Answer (opinion)

Hgb > 7 g/dL may be used as a target value as appropriate based on the patient’s condition after shock and hypoxemia have been corrected, although each patient’s course will be different. (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Anemia develops readily in sepsis as a result of multiple concurrent events, such as bleeding tendency due to disseminated intravascular coagulation and frequent blood sampling for diagnostic testing. Correcting anemia is important to normalize the supply of oxygen to tissues and to ease the burden on the heart. Contrastively, cardiac overflow due to excessive blood transfusion may worsen respiratory condition and hemodynamics. Performing blood sampling and transfusions only when necessary is also advantageous from the perspective of infection control and prevention.

The Transfusion Requirements in the Pediatric Intensive Care Unit (TRIPICU) study [476] investigated target Hgb values for pediatric intensive care patients. Slightly less than 40% of the patients targeted were sepsis patients, and the study compared two groups that underwent “blood transfusion with Hgb < 7 g/dL ; target range: 8.5-9.5g/dL” and “blood transfusion with Hgb < 9.5 g/dL; target range: 11-12g/dL”. As a result, the average Hgb values were 8.7 ± 0.4 g/dL vs. 10.8 ± 0.5 g/dL, respectively, and the mortality rate, incidence of multiple organ failure, and hospitalization time did not differ between the two groups. When determining the target Hgb levels in pediatric sepsis patients, a literature review of relevant studies was considered to be an important clinical task.

No RCTs conforming to the PICO process could be found. By establishing Hgb > 7 g/dL as the target value in septic shock cases or patients presenting with hypoxemia despite stable hemodynamics, blood transfusion frequency can be reduced, medical resources can be utilized more effectively, medical costs can be reduced, and the frequency of complications associated with blood transfusion may be lowered.

In contrast, gaining a sufficient grasp of the patient’s circulatory condition is necessary if Hgb > 7 g/dL is to be established as the target Hgb value after hemodynamics have stabilized. Also, when maintaining a lower Hgb value, responses to acute hemorrhaging may narrow the scope of options for managing shock. As no RCTs conforming to the PICO process were found, the benefit-risk balance is currently unclear. However, this balance is believed to differ depending on the condition and circumstances of individual patients.

Answer (recommendation)

We suggest against the administration of steroids as a standard treatment in pediatric septic shock cases (2D) (rate of agreement, 100%).

Rationale

Administering steroid therapy in sepsis cases had long been considered to be a vital clinical task. However, according to SSCG 2012 [29], the routine use in adult septic shock patients is not recommended. The use of steroids in pediatric patients with severe sepsis has not been linked with survival prognosis in the overseas or Japanese literature [451, 477].

In contrast, steroid therapy is recommended in the pediatric section of SSCG 2012 [29]. This recommendation is based on the observation that pediatric septic shock cases in Europe and the USA are often caused by meningococcal bacteria, and the effect of the rapid development of acute adrenal dysfunction and the high rate of mortality in the pediatric population must also be taken into account. Japan has a low incidence of this condition, and its treatment warrants a different approach than in the western nations [478]. In addition, dengue fever patients with shock have been disproportionately represented in other clinical studies assessing the significance of steroids in the treatment of pediatric sepsis, which is far from the situation in Japan. As such, further investigations not influenced by the results of studies on dengue heat shock are believed to be necessary. Although many clinical studies have assessed neurological prognosis by administering multiple steroid drugs from the start of meningitis treatment, this CQ only addresses the use of steroids to support recovery from shock.

Administering steroids will not lead to a reduced risk of mortality or shortened shock recovery period in pediatric septic shock patients who are unresponsive to fluid resuscitation and dependent on circulatory inotropes. However, the incidences of complications such as bleeding and secondary infections do not increase. The evidence upon which this recommendation is based was limited to a single RCT conducted in a developing country [479]. The most important outcome considered in this CQ is mortality rate, and shock recovery rate and the incidence of complications are somewhat less important. As the evidence pertaining to mortality rate is rated “D” (very weak), the overall evidence regarding this outcome is also rated “D” (very weak).

Decreased mortality rate and increased shock recovery rate/faster shock recovery are considered to be benefits of this intervention. However, no significant differences with respect to mortality rate and shock recovery period were observed in the only RCT adopted for this CQ. Meanwhile, potential harm associated with this intervention include an increased incidence of complications (e.g., bleeding and secondary infection), but no significant difference was observed. Accordingly, the benefit-risk balance for this intervention is currently suspected to be unfavorable, or uncertain.

Answer (opinion)

No recommendation regarding the use of blood apheresis as a treatment in pediatric septic shock cases can be offered at this time due to insufficient evidence (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

SSCG 2012 [29] does not address the use of blood apheresis to treat septic shock in adult or pediatric patients. However, as several new RCTs [274, 275, 480] to investigate the utility of blood apheresis in the treatment of sepsis in adult patients have been announced successively in Europe and the USA, assessing the significance of this intervention for pediatric patients is believed to be valuable.

Only a single study [481] conforming to the PICO process was found. However, this study enrolled few subjects and was terminated early; thus, the body of evidence for this CQ is currently considered to be inadequate. Accordingly, no recommendation could be offered, and an expert consensus is presented instead.

Patient condition may be improved as a result of performing blood apheresis to aid in regulating immunological response through the removal of inflammatory cytokines and mediators. However, no meta-analysis or RCT has been conducted to date assessing renal replacement therapy and direct hemoperfusion with polymyxin B-immobilized fiber in children, and a study on plasma filtration [481] found no improvement in mortality rate even after adjusting for disease severity. Adverse events associated with blood apheresis include bleeding while securing venous access, drops in blood pressure after initiation of blood apheresis, electrolyte anomalies, hypothermia, bleeding due to anticoagulant use, among others.

Mechanical complications occurring while securing blood access are expected to be particularly high in pediatric patients. When handling pediatric patients, the experience level of the attending medical personnel can affect the frequency of adverse events. A decision to perform blood apheresis may greatly increase the workload of medical personnel, and this is considered to have a major impact. Although the evidence obtained from the literature search was inadequate to form a basis for any conclusion, the severity of the potential risks associated with this intervention must be considered fully.

Answer (opinion)

We recommend against the administration of immunoglobulin therapy as a standard treatment in pediatric sepsis cases (expert consensus/no evidence) (rate of agreement, 94.7%).

Rationale

The use of immunoglobulin therapy to address severe infections is covered under the Japanese National Health Insurance program and is in wide use despite unclear evidence regarding its capacity to improve clinical prognosis. Meanwhile, although numerous studies have been conducted overseas to assess a technique known as immunomodulation, which involves administering large doses of immunoglobulin preparations, the results of these studies have not been consistent. Also, with the exception of neonates, high-quality RCTs targeting pediatric patients are currently lacking. Immunoglobulin preparations are expensive, and further clarification of their clinical efficacy will have great significance.

The adult patient-oriented sections of SSCG 2012 [29] do not support the administration of intravenous immunoglobulin (IVIG) preparations, and while the accompanying commentary touches on the significance of the International Neonatal Immunotherapy Study (INIS) trial [482], a multicenter RCT that verified the efficacy of IVIG in neonates, SSCG 2012 does not contain a pediatric patient-oriented chapter on IVIG. The INIS trial is the largest multicenter RCT on this subject, and the fact that IVIG’s effectiveness was not observed in adult or pediatric patients cannot be ignored. The fact that IVIG was found to be ineffective despite how many subjects were premature infants with a history of hypogammaglobulinemia is particularly significant.

No studies conforming to the PICO process were found. According to the results of a systematic review or meta-analysis [107] limited to neonates only, no improvement in mortality rate as a result of polyclonal IVIG use was observed, and in the INIS trial [482] (n = 3493) as well, which was also adopted for this systematic review, no significant differences between the intervention and control groups were observed with respect to mortality rate and incidence of severe sequelae. Although adverse effects such as hyperviscosity syndrome and acute renal failure have been associated with this intervention, these were not among the frequently occurring adverse events reported for the intervention group in the INIS trial [482] (intervention group: 12/1759; control group: 10/1734). The INIS trial [482] was not adopted as a direct basis for the recommendation decision for this CQ because the majority of subjects were premature infants and the targets differed. It can, however, be inferred from the fact that the administration of IVIG preparations to adult sepsis patients leads to neither benefits nor risks that the same is likely true for pediatric patients.

Answer (recommendation)

We do recommend against applying strict glycemic management in pediatric sepsis cases (1B) (rate of agreement, 100%).

Rationale

As with adult patients, numerous reports have suggested relationships between hyperglycemia and high mortality rate, as well as hyperglycemia and hospitalization period in pediatric patients as well. Although not limited to sepsis cases, several successively announced large-scale RCTs [483–486] have assessed the significance of implementing strict glycemic management in critically ill children. Thus, investigating the importance of this type of intervention in patients with severe sepsis is considered to be an important task.

While four RCTs [483–486] targeting critically ill pediatric patients were adopted for a meta-analysis of strict glycemic management (Srinivasan 2014) [487], all four studies involved intensive care unit (ICU) patients other than sepsis patients, and no subgroup analysis limited to sepsis patients was reported.

While the incidence of complication by secondary infection declines as a result of implementing strict glycemic management in pediatric sepsis or PICU patients, hypoglycemia occurs more frequently, and no significant improvement in mortality rate can be expected. Hypoglycemia is a serious complication that can lead to severe neurological sequelae over the longer term in children, particularly infants. Therefore, it was concluded that the potential risks associated with this intervention likely outweigh the benefits, and so it is not recommended, regardless of cost or feasibility.

Although lowered mortality is an expected benefit of this intervention, no significant difference was observed between the intervention and control groups, and the corresponding odds ratio was 0.79 (95% confidence interval (CI) 0.55–1.15). Meanwhile, the decline in complications by secondary infections is also considered to be a clinical benefit, but one of less overall importance. The corresponding odds ratio was 0.76 (95% CI 0.59–0.99), and a statistically significant decrease was observed in the intervention group. While increased incidence of hypoglycemia is a harm caused by this intervention, the odds ratio for this adverse effect was 6.14 (95% CI 2.74–13.78) and was significantly higher in the intervention group. Given that severe hypoglycemia is a serious complication that gives rise to concerns regarding its long-term impact on neurological development, it was concluded that the potential harms associated with this intervention likely outweigh its potential benefits.

Answer (opinion)

The ACCM-PALS initial treatment algorithm may be used as necessary in consideration of patient condition and needs of the clinical environment (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

Performing resuscitative measures without delay and facilitating recovery from the shock state as soon as possible is desirable when treating pediatric septic shock. As such, the application of the ACCM-PALS algorithm for pediatric septic shock has been globally adopted, and use of its translated version has entered the mainstream in Japan [29]. At the same time, it is also important to carefully consider the reliability and validity of other algorithms and their translations and to confirm the utility of existing algorithms.

No RCTs applicable to the subject of this CQ could be found during a literature search of the PubMed database. Although no systematic reviews or RCTs evaluating the validity and utility of the ACCM-PALS algorithm itself currently exist, several observational studies were found [488–490]. However, it was determined that summarizing these observational studies (conducting a meta-analysis) would be difficult due to the low quality study design, variation in outcome indicators, and the possibility for high heterogeneity, and no repeat systematic review and meta-analysis was conducted during the formulation of this guideline. In line with the above, the supporting evidence for this CQ is currently insufficient, and no recommendation can be presented at this time. However, as the expert consensus reached by the Guideline Creation Committee, it was determined that the ACCM-PALS initial treatment algorithm might be used as necessary in consideration of patient condition and needs of the clinical environment during the treatment of pediatric sepsis. The usefulness and validity of the algorithm itself will need to be verified in the future.

By following this algorithm, various pediatric sepsis treatments can be administered as appropriate and without omission. Contrastively, by adhering to the algorithm, excessive treatments may be offered while treatments not covered by the algorithm may be ignored or their implementation delayed. In addition, some increase in workload can be expected to accompany the work of monitoring and confirming adherence to the algorithm, but this additional burden is considered to be minor. No RCTs conforming to the PICO process were found and as such the benefit-risk balance for this technique is currently unclear and is believed to vary depending on patient condition. Because the ACCM-PALS algorithm is freely accessible, no additional medical costs will be incurred, and the drugs and medical devices covered by the algorithm are available in many intensive care units.

Answer (opinion)

The intraosseous route may be used as a temporary route of administration for fluid resuscitation and circulatory inotropes in pediatric septic shock cases, in consideration of patient condition and needs of the clinical environment (expert consensus/no evidence) (rate of agreement, 100%).

Rationale

The use of the intraosseous route as a temporary delivery route for the administration of fluids and circulatory inotropes is well recognized as a pediatric resuscitative technique. However, the use of the intraosseous route accelerates the start of initial resuscitation and may influence outcomes even in pediatric patients with septic shock requiring rapid infusion and the use of circulatory inotropes, and as such the clinical utility of this technique is worth considering.

No RCTs examining the utility of the intraosseous route in pediatric sepsis cases have been conducted to date, but according to the results of one RCT [491], the intraosseous route is as useful as the peripheral venous route in pediatric patients with severe dehydration. In this RCT, all bone marrow needles were placed within 5 min (100%), while the venous route was successfully secured 67% of the time. In addition, successfully securing a venous route required much more time in comparison to the time needed when using bone marrow needles (venous route: 129 ± 13 s, 95% CI 103–156 s vs. intraosseous route: 67 ± 7 s, 95% CI 55–80 s).

Patients requiring resuscitation treatment due to shock or similar conditions exhibit collapsed peripheral blood vessels and securing a venous delivery route frequently presents difficulties. At the same time, resuscitation treatments such as fluid transfusion and drug therapy are also impeded as they often cannot be initiated without access to a venous delivery route. In situations such as these, the intraosseous route can allow for rapid transfusion and drug delivery and is believed to have a high potential for being beneficial to patients. The use of the intraosseous route is also believed to cause little additional increase in workload in comparison to central venous puncture in cases where securing peripheral venous access during resuscitation is difficult. However, careful attention should be paid to the potential for complications when using bone marrow needles, such as malpositioning, hemorrhage, osteomyelitis, compartment syndrome, fat embolism, and tibial fractures.

Although there is currently insufficient evidence to support a recommendation, based on the available evidence pertaining to use of the intraosseous route in pediatric patients with severe dehydration in addition to the accepted understanding of the difficulty in securing peripheral and central venous access in such patients compared to adults, it was determined by expert opinion that both the intraosseous route and the peripheral venous access may be used to facilitate initial fluid resuscitation and administration of circulatory inotropes as treatments in pediatric patients with sepsis.