Hydroxyethyl starch (HES) versus other fluid therapies: effects on kidney function

Abstract

Background

Hydroxyethyl starches (HES) are synthetic colloids commonly used for fluid resuscitation to replace intravascular volume, yet they have been increasingly associated with adverse effects on kidney function. This is an update of a Cochrane review first published in 2010.

Objectives

To examine the effects of HES on kidney function compared to other fluid resuscitation therapies in different patient populations.

Search methods

We searched the Cochrane Renal Group's specialised register, the Cochrane Central Register of Controlled Trials (CENTRAL, in The Cochrane Library), MEDLINE, EMBASE, MetaRegister and reference lists of articles. The most recent search was completed on November 19, 2012.

Selection criteria

Randomised controlled trials (RCTs) and quasi‐RCTs in which HES was compared to an alternate fluid therapy for the prevention or treatment of effective intravascular volume depletion. Primary outcomes were renal replacement therapy (RRT), author‐defined kidney failure and acute kidney injury (AKI) as defined by the RIFLE criteria.

Data collection and analysis

Screening, selection, data extraction and quality assessments for each retrieved article were carried out by two authors using standardised forms. All outcomes were analysed using relative risk (RR) and 95% confidence intervals (95% CI). Authors were contacted when published data were incomplete. Preplanned sensitivity and subgroup analyses were performed after data were analysed with a random‐effects model.

Main results

This review included 42 studies (11,399 patients) including 19 studies from the original review (2010), as well as 23 new studies. Fifteen studies were excluded from the original review (nine retracted from publication due to concerns about integrity of data and six lacking individual patient creatinine data for the calculation of RIFLE criteria). Overall, there was a significant increase in the need for RRT in the HES treated individuals compared to individuals treated with other fluid therapies (RR 1.31, 95% CI 1.16 to 1.49; 19 studies, 9857 patients) and the number with author‐defined kidney failure (RR 1.59, 95% CI 1.26 to 2.00; 15 studies, 1361 patients). The RR of AKI based on RIFLE‐F (failure) criteria also showed an increased risk of AKI in individuals treated with HES products (RR 1.14, 95% CI 1.01 to 1.30; 15 studies, 8402 participants). The risk of meeting urine output and creatinine based RIFLE‐R (risk) criteria for AKI was in contrast in favour of HES therapies (RR 0.95, 95% CI 0.91 to 0.99; 20 studies, 8769 patients). However, when RIFLE‐R urine output based outcomes were excluded as per study protocol, the direction of AKI risk again favoured the other fluid type, with a non‐significant RR of AKI in HES treated patients (RR 1.05, 95% CI 0.97 to 1.14; 8445 patients). A more robust effect was seen for the RIFLE‐I (injury) outcome, with a RR of AKI of 1.22 (95% CI 1.08 to 1.37; 8338 patients). No differences between subgroups for the RRT and RIFLE‐F based outcomes were seen between sepsis versus non‐sepsis patients, high molecular weight (MW) and degree of substitution (DS) versus low MW and DS (≥ 200 kDa and > 0.4 DS versus 130 kDa and 0.4 DS) HES solutions, or high versus low dose treatments (i.e. ≥ 2 L versus < 2 L). There were differences identified between sepsis versus non‐sepsis subgroups for the RIFLE‐R and RIFLE‐I based outcomes only, which may reflect the differing renal response to fluid resuscitation in pre‐renal versus sepsis‐associated AKI. Overall, methodological quality of the studies was good.

Authors' conclusions

The current evidence suggests that all HES products increase the risk in AKI and RRT in all patient populations and a safe volume of any HES solution has yet to be determined. In most clinical situations it is likely that these risks outweigh any benefits, and alternate volume replacement therapies should be used in place of HES products.

Plain language summary

Hydroxyethyl starch (HES) versus other fluid therapies: effects on kidney function

Hydroxyethyl starches (HES) are fluid products that are commonly used in clinical practice, however they have been associated with negative effects on kidney function. This review examined the effects of HES on kidney function compared to other fluid therapies in critically ill patients. Forty‐two randomised clinical trials (11,399 patients) comparing HES to another fluid therapy qualified for this review. Overall, the use of HES products was associated with a 59% increased risk of kidney failure, and a 32% increased risk of dialysis. No significant differences in effect were seen depending on the patient population studied, the type of HES solution, or the dose used. Due to the potential risks associated with HES products, alternative fluid therapies should be used.

Background

In all clinical scenarios, the main treatment for intravascular volume depletion is fluid resuscitation with either crystalloid or colloid. Hydroxyethyl starches (HES) are a group of synthetic colloids that have been commonly used for fluid resuscitation. A recent meta‐analysis restricted to critically ill patients, demonstrated increased risk of renal replacement therapy (RRT) and mortality in HES treated patients (Zarychanski 2013). However, this review excluded patients that were not deemed critically ill, such as elective surgery patients, and this study did not include data stratified by sepsis versus non‐sepsis for the largest study in the literature (Myburgh 2012). In addition, authors were not contacted for unpublished results to broaden the scope of the review. Thus, the external validity of the findings to non‐septic patients and less critically patients has remained unclear.

Description of the condition

Acute kidney injury (AKI) is a common and serious complication affecting critically ill patients, with mortality rates greater than 50% in some studies (Sear 2005; Waikar 2007). It occurs in 0.5% to 30% of patients in perioperative and critical care settings, depending on the population studied and the definition utilised (Sear 2005; Waikar 2007). 

To facilitate valid comparisons of AKI incidence a standardised classification system has been developed by consensus and given the acronym RIFLE (Bellomo 2004; Figure 1). This graded system distinguishes between three worsening levels of acute kidney dysfunction (Risk, Injury and Failure) as well as duration of requirement of RRT (Loss of kidney function and End‐stage kidney disease). The kidney dysfunction classes are defined by objective measures of glomerular filtration rate (GFR) and urine output. They have been validated to predict outcome, with incremental increases in mortality associated with worsening acute RIFLE class (RR = 2.4 Risk; RR = 4.15 Injury; RR = 6.37 Failure) (Ricci 2008). 

1.

* RIFLE (Risk of renal dysfunction, Injury to the kidney, Failure of kidney function, Loss of kidney function and End‐stage kidney disease) classification scheme. The classification system includes separate criteria for creatinine and urine output. A patient can fulfil the criteria through changes in serum creatinine or changes in urine output, or both. The criteria that leads to the worst possible classification should be used. Only GFR based criteria for Risk, Injury and Failure are utilised in this review (Bellomo 2004; with permission)

Similar to previous reviews (Ricci 2008), this review has focused on the creatinine‐based criteria in defining AKI. Compared to the criteria based on both urine output and creatinine, the creatinine only‐based criteria are superior at predicting mortality (Cruz 2007), are not confounded by diuretic use, and are more accessible when retrospectively applying the criteria to existing data (Ricci 2008).

Renal hypoperfusion, secondary to intravascular volume depletion, is commonly implicated as a cause of AKI in critically ill patients (Sear 2005; Waikar 2007). Regardless of the underlying cause of volume depletion, the mainstay of therapy is the administration of intravenous (IV) crystalloid and/or colloid to restore intravascular volume, improve renal perfusion and prevent AKI (Bagshaw 2007; Grocott 2005). 

Description of the intervention

HES are synthetic colloids used for intravascular volume expansion. Products differ by their mean molecular weight (MW), degree of hydroxyethylation (or degree of substitution), and C2:C6 hydroxyethylation ratio (Grocott 2005; Jungheinrich 2005). These features affect the pharmacokinetics of HES products (Ferber 1985; Jungheinrich 2005). The higher the MW and molar substitution, the longer it takes for alpha‐amylase to breakdown the HES molecules. Breakdown of the HES molecules produces small cleavage products that are subsequently filtered by the kidney (Ferber 1985). Therefore, the higher the molecular weight, degree of substitution and C2:C6 ratio, the faster the HES product accumulates after multiple doses and the longer they persists in the circulation. These differences were thought to affect side effect profiles including effects on kidney function (Jungheinrich 2005). However, recent evidence from a meta‐analysis of clinical studies suggests that despite faster breakdown of newer products with lower MW and degree of substitution, toxicity may be paradoxically exacerbated by increased tissue uptake, particularly in the luminal epithelial cells of the renal proximal tubules (Bellmann 2012). Despite these and other safety concerns (Barron 2004; Bork 2005; Wiedermann 2004), HES products have been popular in a wide variety of critical care settings ( McIntyre 2007 ; Miletin 2002 ; Schortgen 2004) .

How the intervention might work

Potential benefits of HES

Colloids aremore efficient intravascular volume expanders than crystalloids because more of the infused volume remains in the intravascular space (American Thoracic Society 2004, Hartog 2011). Compared to other colloids, older generation HES products had a volume efficacy and duration of action that was surpassed only by hyperoncotic albumin and some dextran products (American Thoracic Society 2004). In recent studies using lower molecular weight products with lower molar substitution, the advantage in intravascular volume expansion compared to crystalloids has been marginal (Perner 2012, Myburgh 2012). Nevertheless, in experimental models, colloids demonstrate more rapid resuscitation and improved tissue perfusion compared to crystalloids (American Thoracic Society 2004, Hartog 2011). Furthermore, complementary laboratory and clinical research has demonstrated large volume crystalloid resuscitation has multiple negative effects (Cotton 2006) that could impact kidney function indirectly. 

Mechanism of HES‐related kidney injury

In one hypothesised mechanism, hyperoncotic kidney failure, GFR is decreased secondary to a reduction in filtration fraction (Moran 1987). However, more likely HES products cause kidney injury directly as the kidney is a major target organ for HES tissue update. HES molecules have been shown to be taken up by the luminal epithelial cells in the proximal tubules via pinocytosis (Bellmann 2012). Furthermore, increased tissue uptake has been postulated to explain the toxicity of the newest HES products with lower MW and degree of substitution (Bellmann 2012). Tissue uptake does appear to be dose and time dependent (Sirtl 1999) and HES accumulates in cytoplasmic vacuoles that can persist for long periods of time even after relatively low doses for volume replacement (Metze 1997).

Pathologic evidence in support of these mechanisms is limited. One necropsy study has however demonstrated high concentrations of HES in the kidney suggesting tissue accumulation as a pathologic mechanism (Lukasewitz 1998). Osmotic nephrosis‐like lesions have also been associated with delayed graft function in kidneys from organ donors treated with HES (Cittanova 1996), however such lesions have also been seen in the absence of graft dysfunction (Legendre 1993). 

Why it is important to do this review

Given the clinical popularity of HES and the prognostic significance of AKI, a systematic review evaluating the risk of AKI in HES treated individuals is highly relevant. There have been a number of large RCT's published on this topic since the last Cochrane review on this subject (Dart 2010) in higher risk patient populations (sepsis), that have rigorously evaluated newer HES products that were anticipated to have better side effect profiles. A recent meta‐analysis has synthesized some of this new data and has shown a significant increased risk of RRT and mortality in critically ill HES treated individuals (Zarychanski 2013). This current review adds data from an additional large RCT (BaSES 2012), as well as previously unpublished subgroup data from the largest study in the literature (Myburgh 2012) and includes studies of individuals undergoing elective surgery. This is important data, because previous reviews have suggested sepsis populations were at highest risk of AKI, and non‐sepsis (mainly surgical patients) were relatively protected from renal toxicity (Dart 2010). In addition, in this review study authors were contacted to obtain additional unpublished RIFLE based outcomes. This review thus provides additional data, and a more complete evaluation of the adverse renal effects of HES products in at risk populations.

Objectives

This review aimed to:

1. Examine the risk of AKI with HES compared to other fluid therapies when used in the prevention and treatment of relative intravascular volume depletion.

2. Determine if kidney outcomes differ amongst patients treated with HES from different patient populations (sepsis and burns, trauma, cardiac and vascular surgery patients on and off cardiac bypass, non‐cardiac surgery patients and organ donors; both paediatric and adult).

3. Determine if kidney outcomes differ by HES molecular weight (MW), degree of substitution (DS), C2:C6 ratio or solvent.

4. Determine if kidney outcomes differ by administered dose of HES.

Methods

Criteria for considering studies for this review

Types of studies

All randomised controlled trials (RCTs) and quasi‐RCTs (RCTs in which allocation to treatment was obtained by alternation, use of alternate medical records, date of birth or other predictable methods) looking at HES as a therapy for the prevention or treatment of effective intravascular volume depletion were included, provided that the control group received another eligible form of fluid therapy. 

Crossover studies were excluded due to the lack of a feasible and sufficient wash‐out period in fluid management. There were no language restrictions.

Types of participants

Inclusion criteria

Human subjects of all ages who received HES as a fluid therapy for the treatment of effective intravascular volume depletion were included.   

Exclusion criteria

Healthy volunteers who received HES irrespective of volume status and euvolemic patients who received HES to decrease the viscosity of plasma in the treatment of stroke, retinal vein occlusion, angina, peripheral arterial disease, hearing loss, tinnitus or complications of pregnancy including gestational hypertension, intrauterine growth retardation or ovarian hyperstimulation syndrome were excluded. Studies of patients with kidney failure requiring RRT at the onset of the study were also excluded. 

Types of interventions

• Any HES solution (all MWs and molar substitutions) versus any other intravenous fluid therapy including:

  • crystalloid (i.e. normal saline or Ringer's lactate)

  • albumin or plasma protein fraction

  • blood or fresh frozen plasma (FFP)

  • dextran

  • gelatin

  • HES of another MW, molar substitution, C2:C6 ratio or suspended in a different solvent

• Comparisons to synthetic blood products were excluded

• Other exclusions:

  • volume replacement in preoperative autologous blood donation (remote from surgery)

  • plasmapheresis and plasma exchange

  • thrombosis prophylaxis.

Types of outcome measures

RRT was the primary outcome studied as it is a marker of severe AKI. As there are varying degrees of kidney dysfunction, and many different definitions utilised to define kidney failure, the RIFLE criteria were utilised to standardise definitions across included studies. In order to apply the criteria, authors were contacted to obtain serial creatinine measurements for each study patient. Where possible, meeting Risk, Injury and Failure were analysed as separate outcome measures, however where data from authors could not be obtained, "author defined" kidney failure was analysed as the outcome.

Primary outcomes

1. Need for RRT

2. Meeting RIFLE GFR criteria for renal risk, injury or failure (Figure 1)

3. Author defined kidney failure

Outcomes had to be assessed at least 24 hours after treatment with HES to be included. 

Search methods for identification of studies

Electronic searches

We re‐searched the Cochrane Renal Group's Specialised Register on November 19, 2012 through contact with the Trials' Search Co‐ordinator using search terms relevant to this review.

The Cochrane Renal Group’s Specialised Register contains studies identified from:

1. Quarterly searches of the Cochrane Central Register of Controlled Trials CENTRAL

2. Weekly searches of MEDLINE OVID SP

3. Handsearching of renal‐related journals and the proceedings of major renal conferences

4. Searching of the current year of EMBASE OVID SP

5. Weekly current awareness alerts for selected renal journals

6. Searches of the International Clinical Trials Register (ICTRP) Search Portal and ClinicalTrials.gov.

Studies contained in the Specialised register are identified through search strategies for CENTRAL, MEDLINE, EMBASE based on the scope of the Cochrane Renal Group. Details of these strategies as well as a list of handsearched journals, conference proceedings and current awareness alerts are available in the 'Specialised Register' section of information about the Cochrane Renal Group.

See Appendix 1 for search terms used in strategies for this review.

Searching other resources

Relevant review articles were identified and their reference lists searched for any clinical studies not previously identified. In addition, the reference lists of included studies were reviewed.

Data collection and analysis

Selection of studies

The references identified by the search strategy were divided such that each was screened independently by two of the three authors (AD, TM, CR). The titles, abstracts, and when necessary, the full text of the references were first screened to determine if they met the initial short list inclusion criteria:

• Human study

• RCT or quasi‐RCT

• HES compared to another qualifying fluid therapy

• Appropriate participant and intervention population (as defined above).

Studies which did not meet these criteria were excluded. Conflicts were resolved by consensus of all three authors.

Short listed articles were again independently examined by two authors, using the most complete publication available for the identification of relevant outcomes. Authors were contacted if data as published was incomplete. If articles mentioned serum creatinine (SCr), and were published after 1990 then the authors were also contacted for available individual patient creatinine values so that RIFLE criteria could be applied. If complete RIFLE outcomes were published in the paper, then authors were not contacted. Foreign language papers were reviewed by volunteers with medical backgrounds and proficiency in the language of publication. Translators were found for all languages required. When a study could not be located through the university library or through the Trials Search Co‐ordinator, the study was evaluated using the most complete information available. 

Data extraction and management

Data extraction was carried out independently by the same authors using standard data extraction forms. Studies reported in non‐English language journals had their data extracted onto English data extraction forms by volunteer colleagues. When more than one publication of a study existed, only the publication with the most complete data was included. Where relevant outcomes were published only in earlier versions, these data were used. Any discrepancies between published versions were to be highlighted. In order to ensure there was no duplication of data, authors examined closely author lists, time periods of enrolment, treatment protocol details, patient demographic data and any other information that may have revealed duplications. Any further information required from the original author was requested by written or electronic correspondence and any relevant information obtained in this manner was included in the review. Disagreements were resolved in consultation with ST.

Assessment of risk of bias in included studies

The following items were independently assessed by two authors using the risk of bias assessment tool (Higgins 2011) (seeAppendix 2).

• Was there adequate sequence generation (selection bias)?

• Was allocation adequately concealed (selection bias)?

• Was knowledge of the allocated interventions adequately prevented during the study (detection bias)?

  • Participants and personnel

  • Outcome assessors

• Were incomplete outcome data adequately addressed (attrition bias)?

• Are reports of the study free of suggestion of selective outcome reporting (reporting bias)?

• Was the study apparently free of other problems that could put it at a risk of bias?

Measures of treatment effect

For dichotomous outcomes (RRT, RIFLE criteria, or author‐defined kidney failure results) were expressed as risk ratio (RR) with 95% confidence intervals (CI).

Unit of analysis issues

Cross‐over studies and cluster‐RCTs were not included in this review.

Studies comparing more than one HES product with another fluid (studies with multiple intervention groups) were included in the review. In multi‐arm studies with more than one non‐HES (i.e. control) intervention, all non‐HES groups were combined into a single group as recommended (Higgins 2011). Conversely, in multi‐arm studies with more than one relevant HES group, HES groups were analysed separately and the control group divided appropriately between the HES groups for high versus low MW and high versus low volume meta‐analyses. In the latter case, the advantage of investigating sources of heterogeneity due to different MW or dose of HES were deemed to outweigh the disadvantage of incompletely resolving the unit of analysis issue by having correlated comparisons in the random effects meta‐analyses used in this study. The statistical effect of such correlations is usually trivial and this can be confirmed by a sensitivity analysis with a fixed effect model (Higgins 2011).

Dealing with missing data

Missing individual data

As the outcomes of interest in this review are adverse effects, an intention‐to‐treat analysis was not employed. Mortality data was collected but not reported because it was not consistently measured compositely with kidney outcomes. This made the mortality data difficult to interpret as it was unclear in some cases if individuals had experienced kidney dysfunction before death.

Missing summary data

Authors were contacted with a request to provide relevant outcome data which was reported as having been obtained but was not published.

Assessment of heterogeneity

Heterogeneity was analysed using a Chi² test on N‐1 degrees of freedom, with an alpha of 0.05 used for statistical significance and with the I² test (Higgins 2003). I² values of 25%, 50% and 75% correspond to low, medium and high levels of heterogeneity.

Assessment of reporting biases

Publication bias was assessed with funnel plots where appropriate.

Data synthesis

Significant clinical heterogeneity was anticipated in this review due to the different patient populations, comparison fluids, and fluid regimens studied. Realizing outcomes between studies would not be identical but could follow some distribution, data was pooled using the random‐effects model. The fixed‐effects model was also used to assess the robustness of the model chosen, and its susceptibility to outliers and correlated data from multi‐arm studies.

Subgroup analysis and investigation of heterogeneity

The following patient type subgroup analyses were planned for each identified outcome, where numbers permitted:

• Trauma versus sepsis/burns versus non‐cardiac bypass surgery versus cardiac bypass surgery versus organ donors

• Paediatric versus adult

• Type of control fluid (i.e. albumin, crystalloid, gelatin, other HES product)

• HES MW, degree of substitution (DS), C2:C6 ratio and solvent

• Dose and/or duration of therapy

• Patients with and without pre‐existing kidney dysfunction.

As there were insufficient study numbers only the following subgroup analyses were performed:

• Sepsis versus non‐sepsis patients

• High versus low MW HES (≥ 200 kDa versus 130 kDa). For the included studies in this review this was equivalent to a comparison between degree of substitution > 0.4 versus 0.4.

• High (≥ 2 L) versus low (< 2 L) cumulative HES dose.

Subgroups were explored for possible sources of heterogeneity, as well as to estimate the adverse effects unique to individual patient groups. Heterogeneity of outcomes could be related to these factors in addition to age, body mass, pre‐existing kidney pathology as well as the effect of the disease process itself on kidney function.

Sensitivity analysis

A fixed effect model was analysed, which is a measure of the 'typical intervention effect'. A priori proposed subgroup analyses explored clinical heterogeneity. Sensitivity analyses also included removing studies of poor methodologic quality as well as studies done in previous eras, to determine if differences exist compared with modern studies.  

Results

Description of studies

Results of the search

The updated search resulted in an additional 23 studies. Of the previous ongoing studies two have been published as full publications (Myburgh 2012; Magder 2010), one has since been retracted (Boldt 2007b) and one has been published in abstract form and is included in a review paper (BaSES 2012). For details of the most recent search please see Figure 2. This review now includes 42 RCTs (total 11,399 patients). 15 studies were excluded from the original review (9 retracted from publication due to concerns about integrity of data and 6 lacking individual creatinine data and thus no longer meeting criteria for inclusion).

2.

Study flow diagram.

Included studies

Of the 42 studies, 3 were published in the 1980s, 4 in the 1990s, 16 in the 2000s and 19 since 2010.

Participants

One paediatric study (Akech 2010) qualified for this review; and the remaining 41 studies involved adults in a variety of perioperative and critical care settings. The majority of adult participants were middle aged to elderly with younger patients represented in the trauma studies. Male participants were more numerous than females in the majority of studies. The median number of study participants was 60 and only 7 studies had more than 150 participants. The Myburgh study (Myburgh 2012) accounted for 61.4% of all study participants. Most studies excluded patients with pre‐existing severe kidney disease using a variety of definitions.

Interventions

The most common HES preparations investigated were 6% solutions of 130/0.4, 200/0.5, 200/0.6 or 450/0.7. Only four studies (Brunkhorst 2008; London 1989; Magder 2010; McIntyre 2008) studied hyperoncotic 10% HES (Characteristics of included studies) and only one study utilized HES mixed with hyperoncotic saline (Shmyrev 2011).

Primary outcomes

Requirement of RRT

Nineteen studies including a total of 9857 patients comparing an HES product to another fluid type reported the primary outcome of RRT (BaSES 2012; Berard 1995; Brunkhorst 2008; Cittanova 1996; Du 2011; Godet 2008; Guidet 2012; James 2011; Kumle 1999; Lee 2011; London 1989; Magder 2010; Mahmood 2007; McIntyre 2008; Mukhtar 2009; Myburgh 2012; Perner 2012; Schortgen 2001; Vlachou 2010).

Two additional studies comparing two HES products reported the outcome RRT (Kasper 2003; Ertmer 2012).

Patient populations subgroup

Eight studies included 3899 patients with sepsis or burns (BaSES 2012; Brunkhorst 2008; Guidet 2012; McIntyre 2008; Myburgh 2012; Perner 2012; Schortgen 2001; Vlachou 2010); eleven included 5911 non‐sepsis patients (Berard 1995; Godet 2008; James 2011; Kasper 2003; Kumle 1999; Lee 2011; London 1989; Magder 2010; Mahmood 2007; Mukhtar 2009; Myburgh 2012). Four studies were of cardiac surgical patients (Kasper 2003; Lee 2011; London 1989; Magder 2010); two of vascular surgery patients (Godet 2008; Mahmood 2007); four of non‐cardiac surgical patients (Berard 1995; Godet 2008; Kumle 1999; Mukhtar 2009); one of trauma patients (James 2011); one of pancreatitis patients (Du 2011); and one of intensive care patients (Myburgh 2012). One study was of deceased kidney transplant donors (Cittanova 1996) and kidney function in the kidney recipients.

Comparison fluids

The comparison fluid studied was a gelatin in six studies (Berard 1995; Cittanova 1996; Godet 2008; Kumle 1999; Mahmood 2007; Schortgen 2001). In two studies the comparison fluid was 5% albumin (London 1989; Mukhtar 2009), three studies used Ringer's lactate or acetate (Brunkhorst 2008; Du 2011; Perner 2012); one study used Hartmann's solution (Vlachou 2010), six studies used normal saline (BaSES 2012; Guidet 2012; James 2011; Magder 2010; McIntyre 2008; Myburgh 2012); and one study used a balanced plasma solution (Lee 2011).

Intervention fluids

Doses and durations of therapy ranged from 887 mL over 24 hours (Magder 2010) to 70 mL/kg with a median duration of 14 days (Brunkhorst 2008). The HES solutions studied were 10% 264/0.5 (London 1989) 10% 200/0.5 (Magder 2010;, McIntyre 2008 and Brunkhorst 2008 (200/0.5). Other fluids used were 6% 200/0.6 in Cittanova 1996, Schortgen 2001 and Vlachou 2010, 6% 200/0.5 and 70/0.5 in Kumle 1999, 6% 200/0.62 and 6% 130/0.4 in Mahmood 2007 and 6% 130/0.4 in the remaining studies.

Two studies evaluated pair‐wise comparisons of HES products and were not included in a meta‐analysis (Kasper 2003; Ertmer 2012). Kasper 2003 studied 6% 130/0.4 versus 6% 200/0.5 in cardiac surgery. In this study 2/59 patients treated with 6% 130/0.4 and 3/58 patients treated with 6% 200/0.5 required RRT. Ertmer 2012 studied 10% HES 130/0.4 versus 10% HES 200/0.5 in the perioperative management of 76 cardiac patients on bypass. RRT was not required in either group.

Author‐defined kidney failure

Seventeen studies reported kidney failure (defined by author) as an outcome measure (Abdel‐Khalek 2010; Akech 2010; Altman 1998; Brunkhorst 2008; Dehne 2001; Diehl 1982; Du 2011; Fernandez 2005; Godet 2008; Heradstveit 2010; Jungheinrich 2004; Lee 2011; Neff 2003; Schortgen 2001; Shatney 1983; Shmyrev 2011; Yang 2011). Five of these studies (Brunkhorst 2008; Du 2011; Godet 2008; Lee 2011; Schortgen 2001) also reported requirement of RRT. For nine of these studies (Akech 2010; Fernandez 2005, Godet 2008; Heradstveit 2010; Jungheinrich 2004; Lee 2011; Schortgen 2001; Shmyrev 2011; Yang 2011) data were also available for RIFLE criteria analysis. Two studies included only pair wise comparisons of HES products and therefore were not included in the meta‐analysis (Jungheinrich 2004; Neff 2003).

Patient population subgroup

Four studies included 741 patients with sepsis (Akech 2010; Brunkhorst 2008; Fernandez 2005; Schortgen 2001). The remaining studies were of non‐sepsis patients (n=620).

Comparison fluids

Comparison fluids included albumin in six of the studies (Abdel‐Khalek 2010; Altman 1998; Diehl 1982; Fernandez 2005; Neff 2003; Yang 2011), Ringer's lactate in four studies (Brunkhorst 2008, Dehne 2001, Du 2011, and Yang 2011), gelatin in two studies (Godet 2008; Schortgen 2001), plasma protein fraction (PPF) in one study (Shatney 1983), dextran in one study (Akech 2010), normal saline (Shmyrev 2011) in one study, 7.2% NS in two studies (Heradstveit 2010 and Shmyrev 2011), a plasma solution in Lee 2011 and another type of HES in Dehne 2001, Jungheinrich 2004 and Neff 2003.

Intervention fluids

The doses and durations of therapy ranged from 4 mL/kg (Shmyrev 2011) to 70 mL/kg with a median duration of 14 days (Brunkhorst 2008). The intervention fluid was 10% 200/0.5 in Brunkhorst 2008, 6.5% 200/0.62 in Altman 1998, 6% 450/0.7 in Diehl 1982 and Shatney 1983, 6% 200/0.5 in Abdel‐Khalek 2010, Fernandez 2005, Jungheinrich 2004, Heradstveit 2010, and Shmyrev 2011, 6% 200/0.6 in Schortgen 2001, and 6% 130/04 in Akech 2010,Du 2011, Godet 2008, Jungheinrich 2004, Lee 2011 and Yang 2011. Dehne 2001 evaluated 6% 200/0.5, 6% 200/0.62 and 6% 450/0.7.

Neff 2003 included 31 neurosurgical patients and treated one group with HES 130/0.4 and the other with HES 200/0.5 plus 5% albumin. Two patients in the HES 200/0.5 plus 5% albumin group developed kidney failure (no definition reported by authors) and none in the HES 130/0.4 group. This was the only study that included two HES fluids for this outcome, so it was not included in a meta‐analysis.

AKI defined by RIFLE criteria

Twenty three studies in this review, comparing an HES product to another fluid type or two HES products, either published the RIFLE criteria risk outcome or forwarded the authors individual level creatinine data to calculate this outcome (Akech 2010; Akkucuk 2012; Choi 2010; Dolecek 2009; Dubin 2010; Fernandez 2005; Gallandat 2000; Godet 2008; Guidet 2012; Heradstveit 2010; James 2011; Jungheinrich 2004; Lee 2011; Magder 2010; Myburgh 2012; Perner 2012; Protsenko 2009; Sander 2003; Schortgen 2001; Shmyrev 2011Van der Linden 2005; Yang 2011; Yassen 2011).

Nine of the studies included patients with sepsis and burns (Akech 2010; Dolecek 2009; Dubin 2010; Fernandez 2005; Guidet 2012; Myburgh 2012; Perner 2012; Protsenko 2009; Schortgen 2001). The other studies evaluated surgical or trauma patients (Akkucuk 2012; Choi 2010; Gallandat 2000; Godet 2008; James 2011; Jungheinrich 2004; Lee 2011; Magder 2010; Myburgh 2012; Sander 2003; Shmyrev 2011; Van der Linden 2005; Yang 2011; Yassen 2011) or post cardiac arrest survivors (Heradstveit 2010).

Five studies compared an HES product with albumin (Dolecek 2009; Choi 2010; Fernandez 2005; Yang 2011; Yassen 2011), four with a gelatin (Protsenko 2009; Schortgen 2001, Godet 2008; Van der Linden 2005), one with a dextran (Akech 2010), one with plasma solution (Lee 2011), six with normal saline (Dubin 2010; Guidet 2012; James 2011; Magder 2010; Myburgh 2012; Shmyrev 2011) and one with hypertonic saline (Heradstveit 2010) and two with ringer's lactate or acetate (Akkucuk 2012; Perner 2012). Three additional studies evaluated a lower versus a higher MW HES product (Gallandat 2000; Jungheinrich 2004; Sander 2003). Protsenko 2009 also included 2 different HES groups as well as a crystalloid group.

Doses ranged from 4 mL/kg (Shmyrev 2011) to 49 mL/kg (Van der Linden 2005) while duration of therapy ranged from one to six days. The HES studied was 6% 200/0.6 in Schortgen 2001 , 6% 200/0.5 in Abdel‐Khalek 2010, Fernandez 2005 and Heradstveit 2010, 10% 250/0.45 in Magder 2010, 6% 130/0.4‐0.42 in Akech 2010, Akkucuk 2012, Choi 2010, Dolecek 2009, Dubin 2010, Godet 2008, Guidet 2012, James 2011, Lee 2011, Myburgh 2012, Perner 2012, Van der Linden 2005, Yang 2011 and Yassen 2011 and 7.2%NaCl in 6% HES in Shmyrev 2011. The three studies comparing HES products used 6% solutions of 200/0.5 and 130/0.4.

Excluded studies

Due to the comprehensive nature of the search strategy employed, specific details of the excluded studies are not reported. The majority of studies not short listed did not include the identified study populations. Of those studies which met short list criteria the majority were excluded from the final review due to a lack of important reported kidney outcomes. Attempts to contact authors did increase the number of included studies.

15 studies were excluded from the original review. 7 were retracted from publication due to concerns about integrity of data Boldt 2000a; Boldt 2000b; Boldt 2003; Boldt 2006; Boldt 2007a; Boldt 2007b; Boldt 2008) and another 2 were excluded due to concerns about the credibility of the author and reliability of data (Boldt 1993; Boldt 1998); and 6 contained only mean creatinine data and thus no longer met criteria for inclusion (Allison 1999; Beyer 1997; Langeron 2001; Petrikov 2008; Sade 1985; Vogt 1999).

Two studies (Dehne 1997; Kulla 2008) were noted as significant exclusions. Kulla 2008 met all inclusion criteria but the two HES solutions used in the study were not sufficiently different in composition for comparison. Dehne 1997 studied postoperative patients in the intensive care unit. The intervention group received a 12 mL/kg/d fixed dose of 10% HES 200/0.5 as a continuous infusion for five days. The study was excluded because the control group did not receive a comparison fluid. Four of 10 patients in the HES group developed author‐defined kidney failure ("acute renal failure") compared to 5/15 control patients.

Risk of bias in included studies

Details are available for each study in Characteristics of included studies; Figure 3 and Figure 4.

3.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

4.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study

Allocation

A large number of studies did not clearly state their methods for allocation concealment. For this reason, the majority were rated as unclear.

Blinding

Only a minority of studies were blinded. This was deemed not to be a significant issue for objective outcome data. In this review, SCr based outcomes were considered objective as the lab technician running the test was unlikely to be influenced by the study arm of the patient. This includes the AKI outcome by RIFLE criteria as well as those studies where author‐defined kidney failure was based on creatinine measurements.

There is a possibility of bias for the outcome of RRT. In the majority of cases there is concomitant kidney failure evidenced by elevated SCr or electrolyte disturbance. However, RRT for other indications, such as fluid overload and the timing of initiation of RRT may be subjective in the absence of concrete pre‐defined criteria. A general weakness in the majority of studies that evaluated RRT was the absence of blinding and the absence of clear, objective criteria for RRT initiation. If an outcome was not included in the study, then a rating of 'unclear' was assigned.

Incomplete outcome data

This was not an issue in most studies. Most studies had short follow‐up periods, therefore losses to follow‐up were minimal. Two studies (Altman 1998; Gallandat 2000) had a significant amount of missing data.

Selective reporting

There was minimal evidence of selective reporting in the included studies.

Other potential sources of bias

Funnel plots for primary outcomes generally followed expected distributions (Figure 5; Figure 6; Figure 7; Figure 8; Figure 9). Small studies favouring HES were sparse for author‐defined kidney failure and small studies favouring comparator fluids were sparse for RIFLE‐Failure. However, these plots had the fewest studies with outcomes overall, and little heterogeneity in point estimates was seen in large studies. Thus, significant publication bias or unrecognised heterogeneity are not suspected for any of the main analyses.

5.

Funnel plot of comparison: 1 HES versus other fluid, outcome: 1.1 Renal replacement therapy.

6.

Funnel plot of comparison: 1 HES versus other fluid, outcome: 1.4 Kidney failure (author defined).

7.

Funnel plot of comparison: 1 HES versus other fluid, outcome: 1.5 RIFLE (Risk or worse).

8.

Funnel plot of comparison: 1 HES versus other fluid, outcome: 1.6 RIFLE (Injury or worse).

9.

Funnel plot of comparison: 1 HES versus other fluid, outcome: 1.7 RIFLE (Failure).

In three studies (Brunkhorst 2008; Godet 2008; Schortgen 2001), differences in baseline kidney function between groups could have been a source of bias. In two others (Neff 2003; Shatney 1983), there was no clear definition reported for the outcome of author‐defined kidney failure. In Mahmood 2007 it was unclear if patients who died were included in the author defined RRT counts.

Effects of interventions

HES products versus other fluids

Renal replacement therapy

Nineteen studies (9857 patients) comparing an HES product to another fluid type reported RRT.

Cittanova 1996 (47 patients) randomised deceased organ donors to HES or gelatin and recorded the need for RRT in the kidney recipients, some of whom received kidneys from the same donor. Attempts to reach the primary author to address this unit of analysis issue were unsuccessful but a sample of intraclass correlation coefficients (ICCs), ranging from ‐0.06 to 0.107 was calculated from the literature (Belli 1988; Hetzel 2002; Pfaff 1998; Suri 1999). Due to the uniqueness of this study population and the unit of analysis issue, this study was analysed as its own subgroup. As the ICCs were small, the original study data as published was entered into the meta‐analysis.

For all studies reporting RRT, there was a significant increase in the need for RRT in the HES‐treated individuals compared to those treated with other fluid therapies (Analysis 1.1 (19 studies, 9857 patients): RR 1.31, 95% CI 1.16 to 1.49; I² = 0%). For surgical and trauma patients (non‐sepsis) the RR was 1.25 (95% CI 0.96 to 1.61; Analysis 1.1.1) and the RR for septic patients was 1.32 (95% CI 1.15 to 1.53; Analysis 1.1.2). The RR for the one study of kidney transplant recipients was 6.67 (95% CI 0.92 to 48.45; Analysis 1.1.3). The results of the test for subgroup differences were: Chi² = 2.74, df = 2 (P = 0.25), I² = 27.1%.

1.1. Analysis.

Comparison 1 HES versus other fluid, Outcome 1 Renal replacement therapy.

Molecular weight/degree of substitution

There was a significant increase in the need for RRT in patients treated with either high MW HES products (Analysis 1.2.1 (9 studies, 1183 patients): RR 1.56, 95% CI 1.15 to 2.11; I² = 2%) or low MW HES products (Analysis 1.2.2 (10 studies, 8353 patients): RR 1.26, 95% CI 1.09 to 1.45; I² = 0%) when compared to another type of fluid therapy. The results of the test for subgroup differences were: Chi² = 1.59, df = 1 (P = 0.21), I² = 37.0%.

1.2. Analysis.

Comparison 1 HES versus other fluid, Outcome 2 Renal replacement therapy by MW.

Volume of HES product

There was a significant increase in the need for RRT in those treated with either high volume (≥2 L) HES product (Analysis 1.3.1 (10 studies, 2220 patients): RR of 1.43 (95% CI 1.20 to 1.71; I² = 0%) or low volume (< 2 L) HES product (Analysis 1.3.2 (7 studies, 7296 patients): RR 1.22, 95% CI 1.02 to 1.46; I² = 0%) when compared to other fluid therapies. The results of the test for subgroup differences were: Chi² = 1.52, df = 1 (P = 0.22), I² = 34.1%.

1.3. Analysis.

Comparison 1 HES versus other fluid, Outcome 3 Renal replacement therapy by volume.

Author‐defined kidney failure

Overall there was a significant increase in author‐defined kidney failure in the HES‐treated patients (Analysis 1.4 (15 studies, 1361 patients): RR 1.59, 95% CI 1.26 to 2.00; I² = 0%). There was a significant increase in author‐defined kidney failure in septic patients treated with HES products (Analysis 1.4.2 (4 studies, 741 patients): RR 1.58, 95% CI 1.24 to 2.02; I² = 0%), but no significant difference in non‐septic patients (Analysis 1.4.1 (11 studies, 620 patients): RR 1.61, 95% CI 0.79 to 3.28; I² = 0%).

1.4. Analysis.

Comparison 1 HES versus other fluid, Outcome 4 Kidney failure (author defined).

Acute kidney injury (by RIFLE criteria)

Twenty studies (8769 patients) compared an HES product with another fluid type with enough data to evaluate the outcome AKI by RIFLE criteria.

• Risk or worse: Overall, there was a significant decrease in RIFLE‐R (risk or worse) in those treated with HES products (Analysis 3.3 (20 studies, 8769 patients): RR 0.95, 95% CI 0.91 to 0.99; I² = 0%). There was a significant decrease in RIFLE‐R in non‐septic individuals treated with HES products (Analysis 1.5.1 (12 studies, 5611 patients): RR 0.90, 95% CI 0.85 to 0.94; I² = 0%) but no significant difference for septic individuals (Analysis 1.5.2 (9 studies, 3158 patients): RR 1.04, 95% CI 0.98 to 1.12; I² = 0%). The results of the test for subgroup differences were: Chi² = 12.88, df = 1 (P = 0.0003), I² = 92.2%.

• Injury or worse: Overall, there was no significant difference in risk of RIFLE ‐I (injury or worse) in those treated with HES products versus other fluid therapies (Analysis 3.4 (18 studies, 8583 patients): RR 1.01, 95% CI 0.88 to 1.17; I² =14%). There was a significant decrease in RIFLE‐I in non‐septic individuals treated with HES products versus other fluid therapies (Analysis 1.6.1 (11 studies, 5478 patients): RR 0.85, 95% CI 0.78 to 0.92; I² = 0%), but no significant difference in septic individuals (Analysis 1.6.2 (8 studies, 3105 patients): RR 1.10, 95% CI 1.00 to 1.20; I² = 0%). The results of the test for subgroup differences were: Chi² = 17.64, df = 1 (P < 0.0001), I² = 94.3%.

• Failure: Overall, there was a significant increase in RIFLE‐F (failure) in those treated with HES products (Analysis 3.5 (15 studies, 8402 patients): RR 1.14, 95% CI 1.01 to 1.30; I² = 0%). Subgroup analyses showed no significant difference in risk of RIFLE ‐F for non‐septic individuals (Analysis 1.7.1 (8 studies, 5301 patients): RR 1.04, 95% CI 0.86 to 1.27; I² = 0%) however an increased risk for RIFLE‐F was seen for septic individuals (Analysis 1.7 (8 studies, 3101 patients): RR 1.21, 95% CI 1.03 to 1.43; I² = 0%). The results of the tests for subgroup differences were: Chi² = 1.31, df = 1 (P = 0.25), I² = 23.8%.

3.3. Analysis.

Comparison 3 HES versus other fluid ‐ no subgroups, Outcome 3 RIFLE (Risk or worse).

1.5. Analysis.

Comparison 1 HES versus other fluid, Outcome 5 RIFLE (Risk or worse).

3.4. Analysis.

Comparison 3 HES versus other fluid ‐ no subgroups, Outcome 4 RIFLE (Injury or worse).

1.6. Analysis.

Comparison 1 HES versus other fluid, Outcome 6 RIFLE (Injury or worse).

3.5. Analysis.

Comparison 3 HES versus other fluid ‐ no subgroups, Outcome 5 RIFLE (Failure).

1.7. Analysis.

Comparison 1 HES versus other fluid, Outcome 7 RIFLE (Failure).

Molecular weight/degree of substitution

There was no significant difference in risk of RIFLE‐R in those treated with high MW products (Analysis 1.8.1 (6 studies, 435 patients): RR 1.26, 95% CI 0.89 to 1.79; I² = 0%) or low MW HES products versus other fluid therapies (Analysis 1.8.2 (15 studies, 7993 patients): RR 1.00, 95% CI 0.84 to 1.20; I² = 41%). The results of the test for subgroup differences were: Chi² = 1.25, df = 1 (P = 0.26), I² = 20.1%.

1.8. Analysis.

Comparison 1 HES versus other fluid, Outcome 8 RIFLE (Risk or worse) by MW.

Volume of HES product

There was no significant difference in risk of RIFLE‐R in those treated with either high volume (≥2 L) HES (Analysis 1.9.1 (10 studies, 854 patients): RR 0.90, 95% CI 0.65 to 1.26; I² = 19%) or low volume (< 2 L) HES product (Analysis 1.9.2 (9 studies, 7526 patients): RR 1.13, 95% CI 0.99 to 1.29; I² = 25%) when compared to other fluid therapies. The results of the test for subgroup differences: Chi² = 1.48, df = 1 (P = 0.22), I² = 32.5%.

1.9. Analysis.

Comparison 1 HES versus other fluid, Outcome 9 RIFLE (Risk or worse) by volume.

High versus low molecular weight/degree of substitution HES products

Acute kidney injury (by RIFLE criteria)

• Risk or worse: there was no significant difference in RIFLE‐R between those treated with high versus low MW HES products (Analysis 2.1 (3 studies, 139 patients): RR 1.15, 95% CI 0.18 to 7.8; I² = 4%).

• Injury or worse: there was no significant difference in RIFLE‐I between those treated with high versus low MW HES products (Analysis 2.2 (4 studies, 188 patients): RR 3.21, 95 % CI 0.14 to 75.68).

• Failure: due to lack of outcomes we were unable to analyse RIFLE‐F (Analysis 2.3).

2.1. Analysis.

Comparison 2 High MW/DS HES versus low MW/DS HES, Outcome 1 RIFLE (Risk or worse).

2.2. Analysis.

Comparison 2 High MW/DS HES versus low MW/DS HES, Outcome 2 RIFLE (Injury or worse).

2.3. Analysis.

Comparison 2 High MW/DS HES versus low MW/DS HES, Outcome 3 RIFLE (Failure).

Sensitivity analyses

Multiple sensitivity tests of the data were carried out for primary outcomes. If an outcome measure or either of its 95% CI crossed a RR of 1 during a sensitivity analysis, it was considered a significant change in the results. Also, if subgroup differences became or stopped being significant at the P < 0.05 level then the sensitivity analysis was considered to have changed the results significantly.

Analysis of each primary outcome using either a fixed effects model or a Peto odds ratio (OR) model did not result in any significant changes in the results.

No significant changes in results occurred when studies with baseline differences in kidney function (Godet 2008; Schortgen 2001) were removed from the RRT and author defined kidney failure analyses. Additionally, no significant changes in results occurred when the excluded study Dehne 1997 was added to the author‐defined kidney failure results.

Finally, an important sensitivity analysis was conducted to remove the urine output based RIFLE outcomes for Myburgh 2012, as only creatinine based RIFLE outcomes were utilized in the other studies. This resulted in a RR of RIFLE‐R of 1.05 for HES treated individuals compared with other fluid therapies (95% CI 0.97 to 1.14) (Analysis 4.1; 20 studies, 8445 patients), a RR of RIFLE‐I of 1.22 (95% CI 1.08 to 1.37) (Analysis 4.2; 18 studies, 838 patients) and a RR of RIFLE‐F of 1.35 (95% CI 1.15 to 1.57) (Analysis 4.3; 15 studies, 8216 patients). This analysis was not possible for the subgroups as all the data were not available.

4.1. Analysis.

Comparison 4 Sensitivity analyses, Outcome 1 RIFLE (Risk or worse) ‐ Creatinine only.

4.2. Analysis.

Comparison 4 Sensitivity analyses, Outcome 2 RIFLE (Injury or worse) ‐Creatinine only.

4.3. Analysis.

Comparison 4 Sensitivity analyses, Outcome 3 RIFLE (Failure) ‐ Creatinine only.

Discussion

Summary of main results

This is the most complete systematic review of RCTs published of the effects of HES on kidney function. Its completeness stems from its inclusion of all available studies of volume resuscitation with HES and unpublished RIFLE and subgroup data obtained from included study authors.

HES versus other fluid therapies

This updated review demonstrates an overall 31% increased risk of RRT, 59% increased risk of author‐defined kidney failure, and 14% increased risk of AKI by RIFLE‐Failure criteria in HES treated individuals versus those treated with an alternate fluid. The results of the RIFLE‐Risk and ‐Injury outcomes were more varied. When both urine output and creatinine based RIFLE criteria were included for the studies that reported them, the RIFLE‐Risk outcome showed a small but statistically significant result in the opposite direction, with a 5% reduced risk of AKI in the HES treated individuals, whereas the RIFLE‐Injury outcome showed no differences between groups.

In total, 20 studies of over 8000 patients either published RIFLE criteria or provided individual level creatinines to the authors of this review for the determination of RIFLE based outcomes. The availability of an adequately powered sample for this validated outcome measure greatly strengthens the results of this review. It was, however, surprising that the results for the RIFLE‐Risk and RIFLE‐Injury outcomes were in contrast to the results obtained for RIFLE‐Failure and RRT outcomes. The RIFLE‐Risk outcome has been attributed to pre‐renal AKI in some studies (Cerda 2011, Kellum 2008), therefore one potential explanation for these disparate findings is that HES products may, at least in the short term ameliorate renal perfusion, and thus decrease pre‐renal AKI. However, this review has demonstrated that although individuals receiving a comparator fluid are more likely to develop RIFLE‐Risk than patients receiving HES, patients receiving HES are subsequently more susceptible to developing more severe RIFLE outcomes and a need for RRT, potentially as HES by‐products accumulate over time. This finding is consistent with recent evidence that the mechanism of renal toxicity is secondary to HES particle accumulation in renal tissue (Bellmann 2012).

Further clarification of the differences in RIFLE outcomes was provided by the sensitivity analysis, in which urine output based RIFLE outcomes were excluded. When urine output based outcomes were excluded from the RIFLE‐Risk outcome as per our study protocol, the results were again in favour of non‐HES fluid therapies, however the results were not statistically significant. For the RIFLE‐Injury outcome, a significant difference between groups was not seen in the initial analysis, however when urine output based criteria were excluded, a robust difference was seen between groups, with a 22% increased risk of RIFLE ‐Injury in the HES group. The removal of urine‐output criteria from the RIFLE‐ Failure outcome also strengthened the association, and increased the risk of AKI to 35% in HES treated individuals. These findings are in keeping with the hypothesis that HES products increase the risk of established AKI, despite a relatively superior maintenance of urine output, perhaps again reflecting improved renal perfusion. Furthermore, recent literature also emphasizes the superiority of SCr based RIFLE criteria over urine output based RIFLE criteria at predicting important clinical outcomes. Wlodzimirow et al., in a cohort of critically ill adults (Wlodzimirow 2012) showed a significant difference in AKI risk when evaluating the combined urine output and creatinine based outcome (RIFLE‐Cr+UO) versus RIFLE‐Cr alone (P < 0.001). However, 83% of patients not identified with RIFLE‐Cr recovered within one or more days and those patients with AKI by RIFLE‐Cr alone had increased mortality over those with AKI diagnosed by RIFLE‐Cr+UO (38% based on RIFLE‐Cr versus 24% based on RIFLES Cr+UO (P = 0.02)). This study supports the lack of harm in excluding urine output based outcomes, as the most at risk patients are still identified without the addition of urine output to the RIFLE‐Cr outcome.

The present version of this review excluded secondary outcomes based on group mean serum creatinine (SCr) that were included in the initial version. This was due to the moderate to substantial amount of heterogeneity that existed for these outcome measures and the lack of meaningful results. SCr is influenced by many clinical factors including age, body mass and hydration status. Changes in SCr, as measured by RIFLE criteria, are more accurate measures of kidney function than static measurements. One‐time SCr measurements in acute situations cannot be reliably utilized to estimate kidney function, and therefore should not be utilised alone to evaluate kidney function in this context. This change in protocol resulted in the elimination of six studies from the original review, which are now included only as pertinent exclusions.

A common criticism of kidney outcome studies in HES treated patients is lack of sufficient follow‐up, suggested by the fact that it took 16 days before 50% of patients receiving HES developed kidney failure in Schortgen 2001. However, inspection of the published Kaplan‐Meier curve reveals subgroup differences in this study after only three days of treatment. Further, in analysing studies according to RIFLE criteria in this review, many outcomes occurred early in the course of treatment.

Sepsis versus non‐sepsis patient populations

Subgroup analyses by patient population showed statistically significant differences in RIFLE‐Risk and RIFLE‐Injury outcomes between sepsis and non‐sepsis patients. Non‐sepsis patients treated with HES had fewer adverse outcomes than those treated with a comparator fluid, whereas sepsis patients treated with HES showed worse outcomes than those receiving a comparator fluid. These results appear to be of little clinical significance given the absence of subgroup differences for the more important RIFLE‐Failure and RRT outcomes, where HES use was associated with worse outcomes in both patient population subgroups. The differences seen for RIFLE‐Risk and Injury outcomes may more likely reflect the differing initial renal response to fluid resuscitation in pre‐renal versus sepsis associated AKI.

High versus low molecular weight (MW) HES and High versus low degree of substitution (DS)

It was previously hypothesized that new HES products that were lower molecular weight and degree of substitution had better side effect profiles. Few clinical studies have directly compared 2 different HES solutions to evaluate this hypothesis. RIFLE criteria (Gallandat 2000; Jungheinrich 2004; Protsenko 2009; Sander 2003) and RRT data (Ertmer 2012, Kasper 2003Kumle 1999; Mahmood 2007; Neff 2003) were available for only a few studies with pair‐wise comparisons of different HES products. All studies were in the perioperative period except Protsenko 2009, with few adverse outcomes reported. In all but one study (Kumle 1999), 6% 130/0.4 was compared to 6% 200/0.5 or 6 % 200/0.62. When studies were analysed by comparing high versus low MW and DS HES, no significant differences were found between groups for any outcome measure. However, there were few outcomes reported and the studies were underpowered.

Subgroup differences for high versus low molecular weight and degree of substitution and volume of HES products versus other fluids for the RRT and RIFLE based outcomes did not show any evidence of differences between groups. Thus, there is insufficient evidence to support that 6% 130/0.4's favourable pharmacokinetics (Jungheinrich 2005) compared to older HES products result in improved kidney outcomes compared to higher MW and DS HES products. Newer evidence in fact has shown that the newer products actually exhibit increased tissue uptake as a potential mechanism of toxicity (Bellmann 2012).

Dose of HES product

Subgroup analyses by volume of HES product (≥ 2 L versus < 2 L) failed to show statistically significant differences between subgroups, suggesting a safe dose of HES has not been identified.

Overall completeness and applicability of evidence

Despite concerns regarding adverse kidney outcomes with HES products since the mid 1990's, hundreds of RCTs identified by this review did not evaluate kidney outcomes. This data, had it been collected, could have contributed significantly to answering this important clinical question many years ago.

The review includes data across a wide range of critically ill individuals. Due to the potential differences in patient populations and comparison fluids, where sufficient data was available, subgroup analyses were undertaken to avoid missing true effect differences where they existed. However, in this case no significant subgroup differences were identified.

The results of the review should only be applied to a patient group or intervention to the extent that it was represented in the review. The majority of the studies were of adult individuals with normal or moderately reduced kidney function, although exclusion criteria varied across studies. Organ donors and paediatric patients were under‐represented, although RCT's are not justified in these patients due to the significant potential for harm from HES products.

Quality of the evidence

A total of 42 studies were included in this review with a total of 11,399 patients. This review is significantly strengthened since the original review which included less that 3000 individuals.

This updated review has been strengthened by the addition of 23 new studies, and specifically the addition of 4 large previously ongoing studies in sepsis (BaSES 2012 (BASES), Guidet 2012 (CRYSTMAS) Perner 2012 (6S)) and critically ill patients (Myburgh 2012 (CHEST)). These four studies now contribute 76.2% of the effect size for the main outcome of interest (requirement of RRT). These studies include a robust number of patients in high risk populations for AKI, thus tremendously increasing the power of the available data to assess renal outcomes. Importantly, they were also the first studies to be double‐blinded (i.e. the study fluids were packaged in order to be indistinguishable), a very important study characteristic that eliminated the potential risk of bias driving the somewhat subjective decision to initiate RRT in some cases. Also the validated, standardized RIFLE criteria for AKI were included in all of these new studies, allowing a more valid meta‐analysis of outcomes.

The only RCT that evaluated kidney transplant donors treated with HES was Cittanova 1996. It showed that an exposure of 2100 mL of 6% 200/0.6 resulted in an increased risk of delayed graft function in the recipients, even after addressing a unit of analysis issue. This study remains the only RCT for this participant subgroup and conflicting results have been reported in non‐randomised studies (Blasco 2008; Deman 1999; Giral 2007; Legendre 1993).

Potential biases in the review process

The search strategy utilised was broad and included all potential articles that evaluated HES products in patients with intravascular volume depletion. The initial search strategy itself did not include kidney parameters such that all studies could be evaluated manually for potential kidney outcomes. All foreign language articles were reviewed and systematically evaluated. CENTRAL and the Renal Group's specialised register were included in the search to evaluate all additional studies that were picked up by handsearching. One potential limitation encountered was the inconsistent definition of kidney failure in each paper, but these findings were supported by more standardised outcomes such as RRT and the RIFLE criteria.

Agreements and disagreements with other studies or reviews

Since our Cochrane review was first published (Dart 2010), three other systematic reviews have been published on the renal effects of HES products. The first review by Hartog et al (Hartog 2011) discussed the renal adverse effects of HES in a narrative review. The authors concluded that the renal risk is increased by cumulative dose, but that no safe upper limit is known. They criticized the current literature on modern solutions because of small sample sizes, low cumulative doses, short observation periods and inadequate control fluids and suggest these solutions should not be used in at risk patients. This was the first quantitative review of the adverse kidney effects of HES to update its results since many studies by Dr J Boldt were retracted by their publisher due to concerns about the integrity of data. In 2013 there were two reviews published. The first, by Hasse et al. (Haase 2013), included only studies evaluating HES 130/0.38‐0.45 versus crystalloid or albumin in patients with sepsis. The RR of RRT for the five included studies was 1.36 (95% CI 1.08 to 1.72), a similar result to our review, despite the fact that the CHEST trial sepsis subgroup data was not available (Myburgh 2012). This review did also include the BaSES trial data. They also evaluated mortality, which was not different between groups in their main analysis, however in a post hoc subgroup analysis they did show a RR of death of 1.11 (95% CI 1.01 to 1.22) in HES treated patients in those studies with at least 28 days of follow‐up. The last review was published the same week, and included only critically ill patients (Zarychanski 2013). This review again showed similar findings with a RR of RRT of 1.32 (95% CI 1.15 to 1.50) and RR of mortality of 1.09 (95%CI 1.02 to 1.17) for HES treated patients. It also included published RIFLE criteria outcome data, however did not contact authors for additional data, and did not have access to the sepsis subgroup data from the CHEST trial or the BaSES trial.

Authors' conclusions

Implications for practice.

In this review, HES was associated with an increased risk of AKI and RRT in included study patients, with a relative risk of SCr based RIFLE‐Injury, RIFLE‐Failure, RRT and author defined kidney failure ranging from 1.22 to 1.59. There were no significant differences between septic and non‐septic patients, nor when results were stratified by molecular weight, degree of substitution and dose of HES. In all populations where HES is considered for use in volume resuscitation, these renal risks, along with other risks (Barron 2004; Bork 2005; Wiedermann 2004; Zarychanski 2013), should be weighed against any evidence of benefit from HES, and the risks and benefits of any volume replacement alternatives. In most clinical situations where HES products have been used, it is likely that these risks outweigh any benefits and other volume replacement therapies should be used instead of HES.

Implications for research.

Authors of published studies in this area are encouraged to contribute to this review's objectives by publishing their data in RIFLE format or submitting it to the authors of this review to be added directly to our analyses. Researchers contemplating RCTs involving HES for fluid resuscitation should measure individual SCr values daily for analysis by RIFLE criteria. However, due to the significant body of literature which has now shown an increased risk of AKI and RRT in HES treated individuals, further studies in this area are not warranted.

What's new

Date	Event	Description
21 May 2013	New citation required and conclusions have changed	Updated review
19 November 2012	New search has been performed	Re‐ran search and revised review

History

Protocol first published: Issue 1, 2009
 Review first published: Issue 1, 2010

Date	Event	Description
14 April 2011	Amended	Letter added to address Boldt's retracted studies

Notes

April 7, 2011

Comment regarding studies by Dr Joachim Boldt

There have recently been major concerns discovered about the work of Dr Joachim Boldt, the first author of nine studies included in this systematic review.  Four of his studies were included in the outcome analysis, need for renal replacement therapy (Boldt 2003; Boldt 2007a; Boldt 2006; Boldt 2008).  One study was included in the outcome “author defined” renal failure analysis (Boldt 1998) and seven studies were included in one of the creatinine based outcome analyses (Boldt 1993; Boldt 1998; Boldt 2000a; Boldt 2000b; Boldt 2003; Boldt 2007a; Boldt 2008). 

Landesärztekammer Rheinland‐Pfalz (“LÄK‐RLP”), the State Medical Association of Rheinland‐Pfalz, Germany, is the Institutional Review Board (IRB) responsible for clinical research performed at Klinikum Ludwigshafen.  Dr. Boldt’s most recent work was performed at this centre.  A review was conducted by the LÄK‐RLP of 102 published articles by Dr. Boldt and colleagues, submitted since 1999.  It was determined that 88 of these studies did not have IRB approval, as had been stated in the papers. Although it has not been determined at this point that the studies are fraudulent, they were deemed unethical, due to lack of IRB approval. An editors‐in‐chief joint statement was published online March 12, 2011 (Editors‐in‐Chief Statement 2011), which stated that all 88 articles would be retracted from their respective journals, including four studies included in this systematic review (Boldt 2003; Boldt 2006; Boldt 2007a; Boldt 2008)

In addition, it was confirmed that two of the studies in this systematic review did have appropriate IRB approval (Boldt 2000a; Boldt 2007a). The three additional studies were not included in their review (Boldt 1993; Boldt 1998; Boldt 2000a), therefore, it is unknown at this point whether these studies have IRB approval or not.  An investigating committee has been commissioned by Klinikum Ludwigshafen to verify the authenticity of the results in Dr Boldt’s papers. 

Due to the severe concerns regarding the validity of the results in these studies, we have run a sensitivity analysis removing Dr. Boldt’s retracted papers from our primary outcome analyses. There were no retracted papers in the renal failure categories – RIFLE or author defined. The removal of the four retracted studies from the renal replacement therapy outcome did not change our results significantly.  The RR changed from 1.38 (95% CI 0.89 to 2.16) for need for RRT to RR 1.43 (95% CI 0.89 to 2.28).

Our systematic review is due for an updated literature search in May, 2011 and for publication in January 2012.  There will certainly be more information about the validity of Dr Boldt’s publications prior to the publication of our updated review. We will thoroughly review all available information, such that our analyses and conclusions reflect only reliable data. 

Sincerely

Dr Allison Dart, Dr Thomas Mutter, Dr Chelsea Ruth and Dr Shayne Taback

University of Manitoba; Winnipeg, Manitoba; Canada 

Appendices

Appendix 1. Electronic search strategies

Database	Electronic search terms
CENTRAL	hetastarch:ti,ab,kw (hydroxyethyl next starch*):ti,ab,kw hydroxyethylstarch*:ti,ab,kw (hydroxy next ethyl next starch*):ti,ab,kw hydroxyathyl*:ti,ab,kw hydroxyethylstarke:ti,ab,kw (hydroxyethyl next starke):ti,ab,kw hydroksyetyloskrobia:ti,ab,kw hydroxyethylamidon*:ti,ab,kw hydroxyethylamylopectin:ti,ab,kw (amylopectin and hydroxyethyl):ti,ab,kw hespan:ti,ab,kw haes‐steril:ti,ab,kw haessteril:ti,ab,kw hesteril:ti,ab,kw (hes next steril):ti,ab,kw hespander:ti,ab,kw hexten:ti,ab,kw hemohes:ti,ab,kw haes:ti,ab,kw hestar:ti,ab,kw hyperhaes:ti,ab,kw heafusine:ti,ab,kw pentastarch:ti,ab,kw pentafraction:ti,ab,kw pentaspan:ti,ab,kw plasmasteril:ti,ab,kw plasmafusin:ti,ab,kw expafusin:ti,ab,kw "elo hes":ti,ab,kw elohes:ti,ab,kw elohast:ti,ab,kw elohaest:ti,ab,kw expahes:ti,ab,kw polyhydroxyamylopectin:ti,ab,kw "asl 607":ti,ab,kw asl607:ti,ab,kw ketastarch:ti,ab,kw isoh*es:ti,ab,kw onkohaes:ti,ab,kw onkohes:ti,ab,kw volex:ti,ab,kw voluven:ti,ab,kw lomol:ti,ab,kw "iles esteril":ti,ab,kw varihes:ti,ab,kw rheohes:ti,ab,kw tetrahes:ti,ab,kw tetrastarch:ti,ab,kw salinhes:ti,ab,kw isohas:ti,ab,kw venofundin:ti,ab,kw (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35 OR #36 OR #37 OR #38 OR #39 OR #40 OR #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 OR #52) (HES and not (hypereosinophil* or embryo* or HES‐1 or HES1 or HES‐6 or HES6 or (hospital next episode next stat*) or (health next examination next survey*))):ti,ab,kw (#53 OR #54)
MEDLINE	Hetastarch/ hetastarch*.tw. hydroxyethyl starch*.tw. hydroxyethylstarch*.tw. hydroxy ethyl starch*.tw. hydroxyathyl*.tw. hydroxyethylstarke.tw. hydroxyethyl starke.tw. hydroksyetyloskrobia.tw. hydroxyethylamidon*.tw. hydroxyethylamylopectin.tw. hes steril.tw. haes‐steril.tw. haessteril.tw. hespander.tw. hextend.tw. hespan.tw. hemohes.tw. haes.tw. hestar.tw. hesteril.tw. hyperhaes.tw. pentastarch.tw. pentafraction.tw. pentaspan.tw. plasmasteril.tw. (amylopectin and hydroxyethyl).tw. heafusine.tw. plasmafusin.tw. expafusin.tw. elo hes.tw. elohes.tw. elohast.tw. elohaest.tw. expahes.tw. polyhydroxyamylopectin.tw. asl 607.tw. asl607.tw. ketastarch.tw. isoh?es.tw. isohas.tw. onkohaes.tw. onkoh#s.tw. volex.tw. voluven.tw. venofundin.tw. lomol.tw. il?es esteril.tw. varihes.tw. rheohes.tw. tetrahes.tw. tetrastarch.tw. salinhes.tw. or/1‐53 (HES not (hypereosinophil* or embryo* or HES‐1 or HES1 or HES‐6 or HES6 or hospital episode stat* or health examination survey*)).tw. or/54‐55
EMBASE	Hetastarch/ hetastarch*.tw. hydroxyethyl starch*.tw. hydroxyethylstarch*.tw. hydroxy ethyl starch*.tw. hydroxyathyl*.tw. hydroxyethylstarke.tw. hydroxyethyl starke.tw. hydroksyetyloskrobia.tw. hydroxyethylamidon*.tw. hydroxyethylamylopectin.tw. haes‐steril.tw. hes steril.tw. haessteril.tw. hespander.tw. hextend.tw. hespan.tw. hemohes.tw. haes.tw. hestar.tw. hesteril.tw. hyperhaes.tw. pentastarch.tw. pentafraction.tw. pentaspan.tw. plasmasteril.tw. (amylopectin and hydroxyethyl).tw. heafusine.tw. plasmafusin.tw. expafusin.tw. elo hes.tw. elohes.tw. elohast.tw. elohaest.tw. expahes.tw. polyhydroxyamylopectin.tw. asl 607.tw. asl607.tw. ketastarch.tw. isoh?es.tw. isohas.tw. onkohaes.tw. onkoh#s.tw. volex.tw. voluven.tw. venofundin.tw. lomol.tw. il?es esteril.tw. varihes.tw. rheohes.tw. tetrahes.tw. tetrastarch.tw. salinhes.tw. or/1‐53 (HES not (hypereosinophil* or embryo* or HES‐1 or HES1 or HES‐6 or HES6 or hospital episode stat* or health examination survey*)).tw. or/54‐55
MetaRegister keywords	hetastarch* hydroxyethyl starch* hydroxyethylstarch* hydroxy ethyl starch* hydroxyathyl* hydroxyethylstarke hydroxyethyl starke hydroksyetyloskrobia hydroxyethylamidon* hydroxyethylamylopectin hes steril haes‐steril haessteril hespander hextend hespan hemohes haes hestar hesteril hyperhaes pentastarch pentafraction pentaspan plasmasteril (amylopectin and hydroxyethyl) heafusine plasmafusin expafusin elo hes elohes elohast elohaest expahes polyhydroxyamylopectin asl 607 asl607 ketastarch isohes isohas onkohaes onkohas volex voluven venofundin lomol ilaes esteril varihes rheohes tetrahes tetrastarch salinhes

Appendix 2. Risk of bias assessment tool

Potential source of bias	Assessment criteria
Random sequence generation Selection bias (biased allocation to interventions) due to inadequate generation of a randomised sequence	Low risk of bias: Random number table; computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimization (minimization may be implemented without a random element, and this is considered to be equivalent to being random).
	High risk of bias: Sequence generated by odd or even date of birth; date (or day) of admission; sequence generated by hospital or clinic record number; allocation by judgement of the clinician; by preference of the participant; based on the results of a laboratory test or a series of tests; by availability of the intervention.
	Unclear: Insufficient information about the sequence generation process to permit judgement.
Allocation concealment Selection bias (biased allocation to interventions) due to inadequate concealment of allocations prior to assignment	Low risk of bias: Randomisation method described that would not allow investigator/participant to know or influence intervention group before eligible participant entered in the study (e.g. central allocation, including telephone, web‐based, and pharmacy‐controlled, randomisation; sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes).
	High risk of bias: Using an open random allocation schedule (e.g. a list of random numbers); assignment envelopes were used without appropriate safeguards (e.g. if envelopes were unsealed or non‐opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.
	Unclear: Randomisation stated but no information on method used is available.
Blinding of participants and personnel Performance bias due to knowledge of the allocated interventions by participants and personnel during the study	Low risk of bias: No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding; blinding of participants and key study personnel ensured, and unlikely that the blinding could have been broken.
	High risk of bias: No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.
	Unclear: Insufficient information to permit judgement
Blinding of outcome assessment Detection bias due to knowledge of the allocated interventions by outcome assessors.	Low risk of bias: No blinding of outcome assessment, but the review authors judge that the outcome measurement is not likely to be influenced by lack of blinding; blinding of outcome assessment ensured, and unlikely that the blinding could have been broken.
	High risk of bias: No blinding of outcome assessment, and the outcome measurement is likely to be influenced by lack of blinding; blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement is likely to be influenced by lack of blinding.
	Unclear: Insufficient information to permit judgement
Incomplete outcome data Attrition bias due to amount, nature or handling of incomplete outcome data.	Low risk of bias: No missing outcome data; reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias); missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; missing data have been imputed using appropriate methods.
	High risk of bias: Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; ‘as‐treated’ analysis done with substantial departure of the intervention received from that assigned at randomisation; potentially inappropriate application of simple imputation.
	Unclear: Insufficient information to permit judgement
Selective reporting Reporting bias due to selective outcome reporting	Low risk of bias: The study protocol is available and all of the study’s pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way; the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre‐specified (convincing text of this nature may be uncommon).
	High risk of bias: Not all of the study’s pre‐specified primary outcomes have been reported; one or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not pre‐specified; one or more reported primary outcomes were not pre‐specified (unless clear justification for their reporting is provided, such as an unexpected adverse effect); one or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta‐analysis; the study report fails to include results for a key outcome that would be expected to have been reported for such a study.
	Unclear: Insufficient information to permit judgement
Other bias Bias due to problems not covered elsewhere in the table	Low risk of bias: The study appears to be free of other sources of bias.
	High risk of bias: Had a potential source of bias related to the specific study design used; stopped early due to some data‐dependent process (including a formal‐stopping rule); had extreme baseline imbalance; has been claimed to have been fraudulent; had some other problem.
	Unclear: Insufficient information to assess whether an important risk of bias exists; insufficient rationale or evidence that an identified problem will introduce bias.

Data and analyses

Comparison 1. HES versus other fluid.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Renal replacement therapy	19	9857	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.16, 1.49]
1.1 Non‐sepsis	11	5911	Risk Ratio (M‐H, Random, 95% CI)	1.25 [0.96, 1.61]
1.2 Sepsis	8	3899	Risk Ratio (M‐H, Random, 95% CI)	1.32 [1.15, 1.53]
1.3 Deceased organ donor	1	47	Risk Ratio (M‐H, Random, 95% CI)	6.67 [0.92, 48.45]
2 Renal replacement therapy by MW	17	9536	Risk Ratio (M‐H, Random, 95% CI)	1.32 [1.16, 1.50]
2.1 High MW	9	1183	Risk Ratio (M‐H, Random, 95% CI)	1.56 [1.15, 2.11]
2.2 Low MW	10	8353	Risk Ratio (M‐H, Random, 95% CI)	1.26 [1.09, 1.45]
3 Renal replacement therapy by volume	17	9516	Risk Ratio (M‐H, Random, 95% CI)	1.32 [1.16, 1.50]
3.1 High Volume	10	2220	Risk Ratio (M‐H, Random, 95% CI)	1.43 [1.20, 1.71]
3.2 Low Volume	7	7296	Risk Ratio (M‐H, Random, 95% CI)	1.22 [1.02, 1.46]
4 Kidney failure (author defined)	15	1361	Risk Ratio (M‐H, Random, 95% CI)	1.59 [1.26, 2.00]
4.1 Non‐sepsis	11	620	Risk Ratio (M‐H, Random, 95% CI)	1.61 [0.79, 3.28]
4.2 Sepsis	4	741	Risk Ratio (M‐H, Random, 95% CI)	1.58 [1.24, 2.02]
5 RIFLE (Risk or worse)	20	8769	Risk Ratio (M‐H, Random, 95% CI)	0.97 [0.89, 1.06]
5.1 Non‐sepsis	12	5611	Risk Ratio (M‐H, Random, 95% CI)	0.90 [0.85, 0.94]
5.2 Sepsis	9	3158	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.98, 1.11]
6 RIFLE (Injury or worse)	18	8583	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.87, 1.18]
6.1 Non‐sepsis	11	5478	Risk Ratio (M‐H, Random, 95% CI)	0.85 [0.78, 0.92]
6.2 Sepsis	8	3105	Risk Ratio (M‐H, Random, 95% CI)	1.10 [1.00, 1.20]
7 RIFLE (Failure)	15	8402	Risk Ratio (M‐H, Random, 95% CI)	1.14 [1.01, 1.29]
7.1 Non‐sepsis	8	5301	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.86, 1.27]
7.2 Sepsis	8	3101	Risk Ratio (M‐H, Random, 95% CI)	1.21 [1.03, 1.43]
8 RIFLE (Risk or worse) by MW	20	8428	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.93, 1.23]
8.1 high MW	6	435	Risk Ratio (M‐H, Random, 95% CI)	1.26 [0.89, 1.79]
8.2 low MW	15	7993	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.84, 1.20]
9 RIFLE (Risk or worse) by volume	19	8380	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.92, 1.23]
9.1 high volume	10	854	Risk Ratio (M‐H, Random, 95% CI)	0.90 [0.65, 1.26]
9.2 low volume	9	7526	Risk Ratio (M‐H, Random, 95% CI)	1.13 [0.99, 1.29]

Comparison 2. High MW/DS HES versus low MW/DS HES.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 RIFLE (Risk or worse)	3	139	Risk Ratio (M‐H, Random, 95% CI)	1.15 [0.18, 7.48]
2 RIFLE (Injury or worse)	4	188	Risk Ratio (M‐H, Random, 95% CI)	3.21 [0.14, 75.68]
3 RIFLE (Failure)	4	188	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 3. HES versus other fluid ‐ no subgroups.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Renal replacement therapy	19	9857	Risk Ratio (M‐H, Random, 95% CI)	1.32 [1.17, 1.50]
2 Kidney failure (author defined)	15	1361	Risk Ratio (M‐H, Random, 95% CI)	1.59 [1.26, 2.00]
3 RIFLE (Risk or worse)	20	8769	Risk Ratio (M‐H, Random, 95% CI)	0.95 [0.91, 0.99]
4 RIFLE (Injury or worse)	18	8583	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.88, 1.17]
5 RIFLE (Failure)	15	8402	Risk Ratio (M‐H, Random, 95% CI)	1.14 [1.01, 1.30]

3.1. Analysis.

Comparison 3 HES versus other fluid ‐ no subgroups, Outcome 1 Renal replacement therapy.

3.2. Analysis.

Comparison 3 HES versus other fluid ‐ no subgroups, Outcome 2 Kidney failure (author defined).

Comparison 4. Sensitivity analyses.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 RIFLE (Risk or worse) ‐ Creatinine only	20	8445	Risk Ratio (M‐H, Random, 95% CI)	1.05 [0.97, 1.14]
2 RIFLE (Injury or worse) ‐Creatinine only	18	8338	Risk Ratio (M‐H, Random, 95% CI)	1.22 [1.08, 1.37]
3 RIFLE (Failure) ‐ Creatinine only	15	8216	Risk Ratio (M‐H, Random, 95% CI)	1.35 [1.15, 1.57]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Abdel‐Khalek 2010.

Methods	Design: RCT Time frame: April 2008 to February 2009
Participants	Country: Egypt Setting: multi‐centre Adult patients undergoing paracentesis with volume replacement Number (treatment/control): 67/68 Age (mean ± SD) years: treatment group (46 ± 2.96); control group (47 ± 3.54) Sex (males): treatment group (54%); control group (52%) Exclusions: kidney disease (creatinine > 2 mg/dL)
Interventions	Treatment group Poly 0‐2‐HES Hemohes 6% 0.45 to 0.55, 8 g/L of ascites removed, half after the procedure and the remainder over 6 to8 hours No volume reported Control group Albumin 20%, 8 g/L of ascites removed, half after the procedure and the remainder over 6 to8 hours No volume reported
Outcomes	Kidney failure (Cr > 50% increase from baseline and over 1.5 mg/dL)
Notes	Published data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'random numbers generated by SAS'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data; death only losses
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Akech 2010.

Methods	Design: RCT Time frame: June 2006 to December 2008
Participants	Country: Kenya Setting: single centre Paediatric patients with severe malaria Number (treatment/control): 40/39 Age: groups combined 40 months (IQR 28‐53) Sex: NS Exclusions: established kidney failure
Interventions	Treatment group HES 130/0.4 Voluven, up to 2 doses of 20 mL/kg 50 (42 to 58) mL/kg Control group Dextran 6%, up to 2 doses of 20 mL/kg 49 (41 to 57) mL/kg
Outcomes	Author defined kidney failure 'evidence of renal impairment' RIFLE
Notes	Published and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'an independent person not involved in the recruitment'
Allocation concealment (selection bias)	Low risk	'opaque envelopes'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	'no evidence of renal impairment'
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Akkucuk 2012.

Methods	Design: RCT Time frame: NS
Participants	Country: Turkey Setting: single centre Children 2‐16 years undergoing cardiac surgery ‐ pump Number (treatment/control): 12/12 Age (mean ± SD) years: treatment group (5.1 ± 3.7); control group (3.9 ± 1.7) Sex (M/F): treatment group (7/5); control group (6/6) Exclusions: Kidney disease (Cr > 1.5 mg/dL)
Interventions	Treatment group HES 130/0.4% Volume (mean ± SD): 1,538.5 ± 999.2 mL over 2 to 4 days Control group Ringers Lactate Volume (mean ± SD): 387.9 ± 186.7 mL over 2 to 4 days
Outcomes	RIFLE
Notes	Published study of abstract found in review
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Divided into two randomised groups
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Altman 1998.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: France Setting: Multicentre Adult patients with cirrhotic ascites requiring paracentesis, repeated daily until ascites drained Number (treatment/control): 27/33 Age (mean ± SD) years: treatment group (56.3 ± 11.2); control group (55.9 ± 11.2) Sex (M/F): treatment group (21/6); control group (23/10) Exclusions: kidney disease (Cr > 1.4 mg/dL)
Interventions	Treatment group Elohes 6.5% Volume (mean ± SD): 1538.5 ± 999.2 mL over 2 to 4 days Control group 20% albumin Volume (mean ± SD): 387.9 ± 186.7 mL over 2 to 4 days
Outcomes	Kidney failure (50% increase from baseline or > 1.4 mg/dL)
Notes	Published data only ‐ unsuccessful attempt to contact author
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'randomly allocated (random number table)'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective criteria
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	Excluded patients after randomisation
Selective reporting (reporting bias)	Low risk	Subgroup analysis but compared as subset of whole
Other bias	Low risk	The study appears to be free of other sources of bias

BaSES 2012.

Methods	Design: RCT Time frame: 2006‐2011
Participants	Country: Switzerland Setting: single centre Adult ICU patients with sepsis Number : 241
Interventions	Treatment group 6% HES Voluven Volume median: 3775 mL (IQR 2019 to 6347) Control group Isotonic saline
Outcomes	RRT
Notes	Published data only available in systematic review published by the authors Haase 2013
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unable to assess as not published
Allocation concealment (selection bias)	Unclear risk	Unable to assess as not published
Blinding (performance bias and detection bias) RRT	Unclear risk	Unable to assess as not published
Blinding (performance bias and detection bias) Creatinine based	Unclear risk	Unable to assess as not published
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Unable to assess as not published
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Unable to assess as not published
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Unable to assess as not published
Selective reporting (reporting bias)	Unclear risk	Unable to assess as not published
Other bias	Unclear risk	Unable to assess as not published

Berard 1995.

Methods	Design: RCT Time Frame: May 1992 to March 1993
Participants	Country: France Setting: NS Adult patients with GI bleeding Number (treatment/control): unclear Age (mean ± SD) years: treatment group (58.5 ± 16/6); control group (62.1±15.8) Sex (males): treatment group (65.75%); control group (64%)
Interventions	Treatment group HES: hydroxyethylamidons Control group gelatin Volumes: NS
Outcomes	RRT only
Notes	French abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'tires au sort'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Insufficient information to permit judgement
Blinding (performance bias and detection bias) Creatinine based	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Insufficient information to permit judgement
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Selective reporting (reporting bias)	Unclear risk	Insufficient information to permit judgement
Other bias	Unclear risk	Insufficient information to permit judgement

Brunkhorst 2008.

Methods	Design: parallel RCT Time frame: April 2003 to June 2005
Participants	Country: Germany Setting: multicentre (18 academic tertiary hospitals) Adult sepsis patients with intravascular volume depletion Number (treatment/control): 262/275 Age (mean ± SD) years: treatment group (64.4 ± 13.3); control group (64.9 ± 14.1) Sex (males): treatment group (60.3%); control group (59.6%) Exclusions: kidney disease (dialysis dependent or Cr > 320 µmol/L)
Interventions	Treatment group 10% Hemohes 200/0.5 for 21 days, death or ICU discharge Volume: Median 70.4 mL/kg (IQR 33.4 to144.2) Control group Ringer's lactate for 21 days, death or ICU discharge Volume: 1.32 x HES group
Outcomes	RRT (published) Kidney failure (2x baseline Cr or need for RRT (published) 2x baseline Cr (unpublished subgroup data) Death (published)
Notes	Could receive up to 1000 mL HES prior to entry, 18/275 in Ringer's lactate group received H Published data only ‐ no further data provided by author
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Stratified by hospital and using random permuted blocks with variable size (supplementary online appendix)
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	High risk	No criteria and no blinding
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	Subjective need for RRT part of author defined kidney failure
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	All patients accounted for and reported, study stopped early after safety analysis
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study was stopped after the first stage of a two‐stage adaptive study design for glucose arm only

Choi 2010.

Methods	Design: RCT Time frame: January to November 2008
Participants	Country: South Korea Setting: single centre Adult cardiac surgery patient on bypass Number (treatment/control): 18/18 Age (mean ± SD) years: treatment group (54 ± 12); control group (55 ± 14) Sex (M/F): treatment group (5/13); control group (6/12) Exclusions: kidney disease (Cr > 1.4 mg/dL pre op)
Interventions	Treatment group 6% 130/0.4 HES for pump prime Volume: 500 mL for pump prime plus mean 921 ± 361 mL 8 hours postoperatively and 15 ± 45 mL 16 hours postoperatively Control group 5% Albumin 500 mL for pump prime Volume: 500 mL albumin plus HES mean 799 ± 241 mL 8 hours postoperatively and 14 ± 42 mL 16 hours postoperatively
Outcomes	RIFLE (unpublished data)
Notes	Both groups received HES postoperatively
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computerized randomisation table'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Cittanova 1996.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: France Setting: single centre Deceased organ donors Number (treatment/control): 21/21 Age (mean ± SD) years: Treatment group (45.6 ± 10.8); control group (45.7 ± 11.1) Sex (M/F): treatment group (12/8); control group (12/9) Exclusions: none
Interventions	Treatment group 6% HES 130/0.4 Voluven (Fresenius Kabi, Germany) Control group Gelofusine ® (B/Braun, Germany) 4% succinylated gelatin     Volumes not clearly reported
Outcomes	RRT Author defined kidney failure
Notes	Published data only ‐ attempted to contact author, unsuccessful
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	NS
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Low risk	Outcome assessors blinded to fluid type
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome criteria
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Insufficient information to permit judgement
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not included
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Unit of analysis issue: analysis of recipients without adjustment for multiple kidneys from same donor

Dehne 2001.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Germany Setting: single centre ASA I/II middle ear surgery patients with expected surgery duration > 3 hours Number (group 1/group 2/group 3/control): 15/15/15/15 Age (mean ± SD) years: group 1 (50.4 ± 13.5); group 2 (53.2 ± 0.6); group 3 (50.2 ± 10.6); control group (51.9 ± 7.9) Sex (M/F): NS Exclusions: kidney disease ('absence of renal dysfunction')
Interventions	Treatment group 1 6% HES 200/0.5 Volume: 15 mL/kg over 1 day (actual not reported) Treatment group 2 6% HES 200/0.62 Volume: 15 mL/kg over 1 day (actual not reported) Treatment group 3 6% HES 450/0.7 Volume: 15 mL/kg over 1 day (actual not reported) Control group Ringer's lactate Volume: 60 mL/kg over 1 day (actual not reported)
Outcomes	Kidney failure ('evidence of renal dysfunction') death (all graph values)
Notes	Published data only ‐ author contacted, no further data available Only 6% 200/0.5 versus Ringer's lactate presented in Analysis 1.4
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'prospectively randomised'
Allocation concealment (selection bias)	Low risk	'via sealed envelope assignment'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	No specified threshold, markers used of uncertain significance, no blinding
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Diehl 1982.

Methods	Design: Quasi‐RCT Time frame: NS
Participants	Country: USA Setting: single centre Adult cardiac surgery patients on cardiac bypass requiring fluids for intravascular volume depletion; kidney disease included Number (treatment/control): 27/33 Age (mean ± SD) years: treatment group (58.0 ± 8.0); control group (56.6 ± 8.1) Sex (M/F): treatment group (29/4); control group (20/7) Exclusions: NS
Interventions	Treatment group 6% HES 450 70‐90% substitution Volume (mean ± SD): 1210 ± 273 mL over 1 day Control group 5% albumin Volume (mean ± SD): 1241 ± 481 mL over 1 day
Outcomes	Kidney failure (Cr > 1.5 mg/dL) Death
Notes	Published data only ‐ author not contacted, study too remote
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	'preoperatively randomly divided...according to hospital identification number'
Allocation concealment (selection bias)	High risk	'preoperatively randomly divided...according to hospital identification number'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective outcome measure (Cr > 1.5 mg/dL and Cr > 2.0 mg/dL)
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes reported
Other bias	Low risk	The study appears to be free of other sources of bias

Dolecek 2009.

Methods	Design: RCT Time frame: May 2005 to 2008
Participants	Country: Czech Republic Setting:single centre trauma hospital Adult patients with severe sepsis requiring artificial ventilation Number (treatment/control): 26/30 Age (median, range) years: treatment group (47, 19‐81); control group (43, 23‐67) Sex (M/F): treatment group (22/4); control group (26/4) Exclusions: 'acute renal failure'
Interventions	Treatment group 6% HES 130/0.4 Volume 250ml every 6 hours for 72 hours (3000 mL) Control group 20% albumin Volume 100ml every 12 hours for 72 hours (600 mL)
Outcomes	RIFLE only
Notes	Unpublished data only
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computer generated randomisation list'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Du 2011.

Methods	Design: RCT Time frame: 2008 to 2009
Participants	Country: China Setting: single centre Severe acute pancreatitis Number (treatment/control): 21/21 Age (mean ± SD) years Treatment group: donor (38 ± 12); recipient (44 ± 1) Control group: donor (40 ± 17); recipient (44 ± 11) Sex (M/F) Treatment group: donor (11/4); recipient (20/7) Control group: donor (8/4); recipient (9/11) Exclusions: renal insufficiency
Interventions	Treatment group Elohes 6% 200/0.6 (max 33 mg/kg) Donor volume (mean ± SD): HES 2100 ± 660 mL over 1 day Control group Ringer's lactate Donor volumes (mean ± SD): gelatin 2875 ± 1384 mL over 1 day
Outcomes	RRT
Notes	Published data only ‐ attempted to contact author, unsuccessful
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computer derived random number sequence'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	High risk	'not blinded'
Blinding (performance bias and detection bias) Creatinine based	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	Patients who died at less than 72 hours were excluded
Selective reporting (reporting bias)	High risk	Patients who died at less than 72 hours were excluded
Other bias	Low risk	The study appears to be free of other sources of bias

Dubin 2010.

Methods	Design: pilot RCT Time frame: NS
Participants	Country: Argentina Setting: multicentre Adult sepsis/burns with intravascular volume depletion Number (treatment/control): 12/13 Age (mean ± SD) years: treatment group (62 ± 21); control group (65 ± 12) Sex (males): treatment group (67%); control group (55%) Exclusions: NS
Interventions	Treatment group Voluven 130/0.4 6% for volume expansion for 24 hours according to protocol then for 5 days ongoing hydration Volume (median, 90% CI): 2610 ± 885 mL then 2071 ± 486 mL Control group Normal saline for volume expansion for 24 hours according to protocol then for 5 days ongoing hydration Volume (median, 90% CI): 6254 ± 2603 mL then 2114 ± 726 mL
Outcomes	RIFLE criteria (unpublished) Death
Notes	Unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'simple randomisation'
Allocation concealment (selection bias)	Low risk	'sealed envelopes'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	High risk	4 excluded after randomisation due to loss of follow‐up, and one excluded as not meeting criteria
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Small study with difference in group creatinines at baseline

Ertmer 2012.

Methods	Design: RCT (post‐hoc analysis) Time frame: November 2004 to July 2005
Participants	Country: Germany Setting: multicentre Adult cardiac surgery patients on cardiac bypass requiring fluids for intravascular volume depletion; kidney disease included Number (treatment/control): 37/39 Age (mean ± SD) years: treatment group (61 ± 11); control group 62 ± 11) Sex (M/F): treatment group (31/6); control group (33/6) Exclusions: pre‐operative creatinine > 2mg/dL or oliguria (urine excretion <500 mL/d)
Interventions	Treatment group 10% HES 130/0.4 for perioperative replacement Volume:1577 mL Control group 10% HES 200/0.5 for perioperative replacement Volume: 1540 mL
Outcomes	RRT
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Permutated blocks of six
Allocation concealment (selection bias)	Low risk	By patient number offsite
Blinding (performance bias and detection bias) RRT	Low risk	Identical bottles
Blinding (performance bias and detection bias) Creatinine based	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	One study centre excluded from results

Fernandez 2005.

Methods	Design: parallel RCT Time frame: January 2002 to February 2003
Participants	Country: Spain Setting: single centre Adult patients with cirrhosis and spontaneous bacterial peritonitis and intravascular volume depletion Number (treatment/control): 10/10 Age (mean ± SD) years: treatment group (60 ± 10); control group (62 ± 9) Sex (M/F): treatment group (5/5); control group (6/4) Exclusions: 'renal parenchymal disease'
Interventions	Treatment group 6% Hesteril (HES) 200/0.5 for 3 days Volume Baseline: 25 mL/kg Day 3: 16.6 mL/kg (either fluid) Control group 20% albumin for 3 days Volume Baseline: 7.5 mL/kg Day 3: 5 mL/kg
Outcomes	Author defined kidney failure (Cr > 1.5 mg/dL or BUN > 25 mg/dL) (published) RIFLE (unpublished)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'random numbers generated by SAS'
Allocation concealment (selection bias)	Low risk	'sealed envelopes'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome criteria
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective outcome criteria, Cr > 1.5 mg/dL or BUN > 25 mg/dL
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome criteria
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Gallandat 2000.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Netherlands Setting: multicentre Adult cardiac surgery patients on cardiac bypass for pump priming and intravascular volume depletion Number (group 1/group 2): 30/29 Age (mean ± SD) years: group 1 (63.5 ± 9.0); group 2 (61.0 ± 10.3) Sex (M/F): group 1 (25/5); group 2 (25/5) Exclusions: kidney disease ('renal disorders')
Interventions	Treatment group 1 Low MW HES: Voluven 6% 130/0.4 from intraoperatively until 16 hours postoperatively Volume (mean ± SD): 2550 ± 561 mL Treatment group 2 High MW HES: Pentaspan 6% 200/0.5 from intraoperatively until 16 hours postoperatively Volume (mean ± SD): 2466 ± 516 mL
Outcomes	RIFLE (unpublished)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'randomised'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Ojective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	High risk	Only two patients from Voluven group with no Cr values at postoperative day 1 (unpublished data), multiple missing values sufficient to effect change in outcomes from postoperative day 2 (unpublished data)
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Godet 2008.

Methods	Design: parallel RCT Time frame: April 2002 to June 2004
Participants	Country: France Setting: multicentre Adult vascular surgery patients, not on cardiac bypass with intravascular volume depletion; kidney disease (Cr > 250 µmol/L or CrCl < 80 mL/min) Number (treatment/control): 32/33 Age (mean (range)) years: treatment group (72.9 (57 to 89)); control group (73 (55 to 86)) Sex (M/F): treatment group (28/4); control group (28/5) Exclusions: NS
Interventions	Treatment group Voluven 6% 130/0.4 Volume (mean ± SD): 2350 ± 1355 mL over 6 days Control group 3% Plasmion (gelatin) up to 50 mL/kg Volume (mean ± SD): 2136 ± 1174 mL over 6 days
Outcomes	RRT Death Kidney failure (Cr > normal or increase by 44.2 µmol/L from baseline) all published RIFLE (unpublished)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'the randomisation list was generated by DATAMAP using balanced blocks'
Allocation concealment (selection bias)	Low risk	'envelopes identified by the randomisation number'
Blinding (performance bias and detection bias) RRT	High risk	Open study, no objective criteria
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Differences in kidney function between groups at baseline

Guidet 2012.

Methods	Design: RCT Time frame: 2012
Participants	Country: France/Germany Setting: multicentre Sepsis patients with intravascular volume depletion; Number (treatment/control): 100/96 Age (mean ± SD) years: treatment group (65.8 ± 15.4); control group (65.9 ± 14.7) Sex (M/F): treatment group (28/4); control group (28/5) Exclusions: creatinine >3.39 mg/dL or anuria > 8 hours
Interventions	Treatment group 6% HES 130/0.4 for 4 days Volume 50 mL/kg first day, 25 mL/kg for days 2 to 4 (2615 ± 1499 mL) Control group Normal saline Volume no maximum (2788 ± 1799 mL)
Outcomes	RRT Death RIFLE All published outcomes
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	NS
Allocation concealment (selection bias)	Low risk	'double blind'
Blinding (performance bias and detection bias) RRT	Low risk	'double blind'
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Selective reporting (reporting bias)	Unclear risk	Insufficient information to permit judgement
Other bias	High risk	Fresenius‐Kabi involved in study design, analysis and preparation of report

Heradstveit 2010.

Methods	Design: RCT Time frame: September 2005‐March 2007
Participants	Country: Norway Setting: single centre Adult patients with witnessed out of hospital cardiac arrest in ICU Number (treatment/control): 10/9 Age (median (range)) years: treatment group (60 (48 to 74)); control group (60 (22 to 75)) Sex (M/F): treatment group (8/2); control group (8/1) Exclusions: NS
Interventions	Treatment group HyperHAES 6% 200/0.5 Volume median (range): 4750 mL (3150 to 9075) 500 mL HyperHAES then Ringer's lactate/normal saline Control group 7.2% NaCl Volume median (range): 8010 mL (5515 to 12,908)
Outcomes	Author defined kidney failure RIFLE (unpublished)
Notes	Published and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'stratified randomisation'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	Not blinded and no definition for 'renal failure'
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Deferred consent, 5 excluded after randomisation which met exclusion criteria (low risk). Differences at baseline (3 in HES group had balloon pumps at baseline)

James 2011.

Methods	Design: RCT Time frame: NS
Participants	Country: South Africa Setting: single Centre Severe trauma requiring greater than 3L volume, subgroup by penetrating versus blunt trauma Number (treatment/control) penetrating: 36/31, blunt: 20/22 Age (median (range)) years penetrating treatment: 27.6 (18 to 49) penetrating control: 32.6 (21 to 56) blunt treatment: 33.0 (18 to 50) blunt control: 35.7 (20 to 58) Sex (M/F) penetrating treatment: 33/3 penetrating control: 27/4 blunt treatment: 15/5 blunt control: 15/7 Exclusions: excluded pre‐existing kidney failure with oliguria or anuria
Interventions	Treatment group HES 130/0.4 Volume mean (SD) penetrating treatment: 5093 ± 2733 mL blunt treatment: 6113 ± 1919 mL Control group 0.9% NS Volume mean (SD) penetrating control: 7473 ± 4321 mL blunt control: 6295 ± 2197 mL
Outcomes	Need for dialysis RIFLE
Notes	Published data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'random numbers grouped in blocks of 8'
Allocation concealment (selection bias)	Low risk	'pre packed numbered boxes containing study fluid labelled'
Blinding (performance bias and detection bias) RRT	Low risk	'identical bags'
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	High risk	Not true intention to treat,excluded some patients from analysis
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Jungheinrich 2004.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Germany Setting: multicentre Adult orthopaedic surgery patients with expected blood loss > 2000 mL Number: 26/26 Age (mean ± SD) years: group 1 (63 ± 12); group 2 (61 ± 8) Gender (M/F): group 1 (16/10); group 2 (16/10) Exclusions: 'renal dysfunction'
Interventions	Treatment group 1 Low MW HES: 6% Voluven 130/0.56% from induction to 5 hours postoperatively Volume (mean ± SD): 1602 ± 569 mL at end of surgery; 2035 ± 446mL postoperative day 1 Treatment group 2 High MW HES: HAES‐steril 6% 200/0.5 from induction to 5 hours postoperatively Volume (mean ± SD): 1635 ± 567 mL at end surgery; 2000 ± 424 mL postoperative day 1
Outcomes	RIFLE injury/failure Author defined kidney failure (double baseline Cr)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'performed using a method of randomly permuted blocks of six'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure, 'solutions were blinded by the manufacturer and indistinguishable from each other'
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Funded by Fresenius Kabi, Bad Homburg, Germany.

Kasper 2003.

Methods	Design: parallel RCT Time frame: August 2000 to April 2001
Participants	Country: Germany Setting: single centre Adult cardiac surgery patients on cardiac bypass with intravascular volume depletion Number (group 1/group 2): 59/58 Age (mean ± SD) years: group 1 (63 ± 8); group 2 (64 ± 7) Sex (M/F): group 1 (47/12); group 2 (46/12) Exclusions: kidney disease (Cr > 1.5 mg/dL)
Interventions	Treatment group 1 Low MW HES: Voluven 6% 130/0.4 intraoperatively until 24 hours postoperatively Volume (mean (range)): 3500 mL (2000 to 4500) Treatment group 2 High MW HES: Haes‐steril 6% 200/0.5 intraoperatively until 24 hours postoperatively Volume (mean (range)): 2500 mL (1850 to 3250)
Outcomes	Cr (mg/dL) at 24 hours (ICU) and postoperative day 7 RRT Death
Notes	Published data only ‐ author contacted, no further data available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computer generated code that was prepared at a remote site'
Allocation concealment (selection bias)	Low risk	'sealed in sequentially numbered opaque envelopes'
Blinding (performance bias and detection bias) RRT	Low risk	'HES solutions supplied in identical looking sequentially numbered plastic bags' 'patients were kept blinded throughout the study'
Blinding (performance bias and detection bias) Creatinine based	Low risk	'HES solutions supplied in identical looking sequentially numbered plastic bags' 'patients were kept blinded throughout the study'
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	Pre‐specified outcomes were reported
Selective reporting (reporting bias)	Low risk	Dropped 1 (Voluven) and 2 (Haes‐steril) from analysis, re‐explored for bleeding and found to have non‐generalized bleeding, therefore unlikely to be related to primary outcome
Other bias	Low risk	The study appears to be free of other sources of bias

Kumle 1999.

Methods	Design: parallel RCT, stratified by age (< 65 and 65 to 85) Time frame: NS
Participants	Country: Germany Setting: single centre Adult surgical (non‐cardiac) patients, intravascular volume depletion Number (group 1/group 2/control): 20/20/20 Age (mean ± SD < 65; mean ± SD > 65) years Group 1: 52.6 ± 8.8; 73.9 ± 5.7 Group 2: 53 ± 9; 72 ± 6 Control: 52.9 ± 4.4; 71 ± 3.1 Sex (M/F < 65; > 65) Group 1: 6/4; 4/6 Group 2: 6/4; 6/4 Control: 3/7; 5/5 Exclusions: kidney disease (Cr > 180 µmol/L)
Interventions	Treatment group 1 High MW HES: Hemohes 6% 200/0.5 Volume (mean ± SD < 65; > 65): 1690 ± 560 mL; 1600 ± 570 mL over 3 days Treatment group 2 Low MW HES: Rheohes 6% 70/0.5 Volume (mean ± SD < 65; > 65): 2150 ± 470 mL; 1964 ± 520 mL over 3 days Control group 35 kD Gelafundin (gelatin) Volume over 3 days (mean ± SD < 65; > 65): 2100 ± 460 mL; 2180 ± 510 mL
Outcomes	RRT Death
Notes	Published data only ‐ author contacted, no further data available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'the random number table used to randomise patients was generated using FILEMAKER PRO 4.0'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	High risk	No blinding, no criteria
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Lee 2011.

Methods	Design: RCT Time frame: NS
Participants	Country: Korea Setting: single centre Adult surgical (cardiac) patients, intravascular volume depletion Number: 53/53 Age (mean ± SD) years: treatment group (63 ± 8); control group (65 ± 9) Sex (M/F): treatment group (43/10); control group (44/9) Exclusions: kidney disease
Interventions	Treatment group HES: HES 130/0.4 Volume (mean ± SD): 1458 ± 465 mL + crystalloid Control group Crystalloid Volume (mean ± SD): 8342 ± 1794 mL
Outcomes	RRT (published) RIFLE (unpublished) Author defined kidney failure (published)
Notes	Published data only ‐ author contacted, no further data available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'randomly allocated'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Open label trial, 'endpoints assessed blindly'
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	'endpoints assessed blindly'
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

London 1989.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: USA Setting: single centre Adult cardiac surgery patients, on cardiac bypass, postoperative intravascular volume depletion; Cr > 2.0 mg/dL Number: 50/44 Age (mean ± SD) years: treatment group (63 ± 7); control group (64 ± 7) Sex (M/F): 89/1
Interventions	Treatment group 10% Pentastarch 264/0.45 for 24 hours postoperatively Volume (mean ± SD): 1706 ± 393 mL Control group 5% albumin for 24 hours postoperatively Volume (mean ± SD): 1794 ± 341 mL
Outcomes	RRT
Notes	Published data only ‐ author contacted, no further data available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'randomised'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Insufficient information to permit judgement
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data (deaths/causes accounted for)
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Magder 2010.

Methods	Design: RCT Time frame: NS
Participants	Country: Canada Setting: single centre Adult patients on bypass cardiac surgery Number (treatment/control): 119/118 Age (mean ± SD) years: treatment group (65.5 ± 10.6); control group (65.9 ± 10.6) Sex (males): treatment group (74.8%); control group (69.5%) Exclusions: NS
Interventions	Treatment group 250 MW 10% Pentastarch Volume as boluses (mean ± SD): 887 ± 546 mL Control group Normal saline Volume as boluses (mean ± SD): 1397 ± 1041 mL
Outcomes	RRT RIFLE
Notes	Published data and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computerized random generator by central pharmacy'
Allocation concealment (selection bias)	Low risk	'unmarked bags in numbered boxes'
Blinding (performance bias and detection bias) RRT	Low risk	'unmarked bags'
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Only patients admitted prior to 14:30 to ICU were randomised, all patients received 750ml HES in pump prime

Mahmood 2007.

Methods	Design: parallel RCT Time frame: April 2001 to January 2003
Participants	Country: UK Setting: single centre Adult vascular surgery patients off cardiac bypass for intravascular volume depletion Number (group 1/group 2/control): 21/21/20 Age (mean ± SD) years: group 1 (72 ± 7); group 2 (72 ± 7); control group (73 ± 8) Sex (M/F): group 1 (16/5); group 2 (19/2); control group (15/2) Exclusions: kidney disease (Cr > 177 µmol/L)
Interventions	Treatment group 1 Elohes (HES) 6% 200/0.62 Volume (mean ± SD): 3443 ± 1769 mL Treatment group 2 Voluven (HES) 6% 130/0.4 Volume (mean ± SD): 3911 ± 1783 mL Control group 4% Gelofusin (gelatin) Volume (mean ± SD): 4490 ± 1499 mL
Outcomes	RRT Death
Notes	Published data only ‐ author not contacted
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation by blocks using random number table
Allocation concealment (selection bias)	Low risk	'sealed envelopes'
Blinding (performance bias and detection bias) RRT	High risk	No criteria, no blinding
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data; death only losses
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	"The study was funded by Fresenius Kabi, including the salary of the research fellow (A. Mahmood). Dr P. Gosling and Mr R. Vohra are full‐time employees of University Hospital Birmingham NHS Trust with no conflicts of interest. The sponsor had no role in study assimilation, design, analysis or in the writing of the manuscript at any stage. The data and final manuscript were the authors' responsibility."

McIntyre 2008.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Canada, New Zealand Setting: International multicentre Adult patients in early septic shock; patients with renal insufficiency (Cr > 300 µmol/L) Number (treatment/control): 21/19 Age (mean ± SD) years: treatment group (63.1 ± 13.1); control group: (63.6 ± 16.3) Sex (M/F): treatment group (13/8); control group (11/8) Exclusions: 'chronic renal failure requiring dialysis'
Interventions	Treatment group Pentastarch 10% 200/0.5 HES Volume (mean ± SD): 1.9 ± 2.1 L (maximum 3 L or 28 mL/kg) Control group Normal saline for first 12 hours after randomisation Volume (mean ± SD): 5.2 ± 1.91 L (maximum 3 L or 28 mL/kg)
Outcomes	7 day organ failure (RRT) Death
Notes	Published data only ‐ author contacted, no further data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'central computerized permuted 4‐block randomisation scheme'
Allocation concealment (selection bias)	Low risk	'only the designated pharmacist at each institution was aware of the treatment allocation for individual patients'
Blinding (performance bias and detection bias) RRT	Low risk	'only the designated pharmacist at each institution was aware of the treatment allocation for individual patients', 'study fluids blinded'
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Mukhtar 2009.

Methods	Design: RCT Time frame: NS
Participants	Country: Egypt Setting: NS Adult patients recipients of living donor liver transplant Number (treatment/control): 20/20 Age (mean ± SD) years: treatment group (55 ± 5.8); control group (51 ± 6) Sex (M/F): treatment group (19/1); control group (16/4) Exclusions: 'primary renal dysfunction'
Interventions	Treatment group Voluven 6% HES 130/0.4 Volume (mean ± SD): 3080 ± 417 mL intraoperative, 6229 ± 1140 mL postoperative Control group Albumin 5% Volume (mean ± SD): 3500 ± 1000 mL intraoperative, 4636 ± 1140 mL postoperative
Outcomes	RRT
Notes	Published data only ‐ author contacted, no further data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'randomly allocated'
Allocation concealment (selection bias)	Low risk	'sealed envelope'
Blinding (performance bias and detection bias) RRT	High risk	Not blinded and criteria not listed
Blinding (performance bias and detection bias) Creatinine based	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Unclear risk	All 3 patients with severe renal impairment randomised to HES group

Myburgh 2012.

Methods	Design: RCT Time frame: December 2009 to January 2012
Participants	Country: Australia and New Zealand Setting: multi centre Adult patients admitted to ICU Number (treatment/control): 3500/3500 Age (mean ± SD) years: treatment group (63.1 ± 17.0); control group (62.9 ± 16.9) Sex (males): treatment group (60.5%); control group (60.3%) Exclusions: please see article for complete list. Excluded patients who had already received RRT
Interventions	Treatment group Voluven 6% HES 130/0.4 up to a daily maximum of 50 mL/kg Volume (daily mean ± SD): 526 ± 425 mL Control group 0.9% saline Volume (daily mean ± SD): 616 ± 488 mL
Outcomes	RRT RIFLE
Notes	Unpublished data provided by author included RRT and RIFLE outcomes for patients with sepsis. Non‐sepsis outcomes were calculated by subtraction from totals published
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'web based randomisation system'
Allocation concealment (selection bias)	Low risk	'encrypted web based randomisation system'
Blinding (performance bias and detection bias) RRT	Low risk	Indistinguishable bags
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	0.1% lost to follow‐up, similar in both groups
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	Fresenius Kabi unrestricted grant

Neff 2003.

Methods	Design: parallel RCT Time frame: April 1999 to May 2000
Participants	Country: Switzerland Setting: single centre Adult head injury patients Number (group 1/group 2): 16/15 Age (mean ± SD) years: group 1 (36 ± 13); group 2 (39 ± 12) Sex (M/F): group 1 (14/2); group 2 (13/2) Exclusions: chronic renal insufficiency
Interventions	Treatment group 1 Voluven (HES) 6% 130/0.4 for up to 28 days Volume (mean ± SD): 19,000 ± 16,000 mL Treatment group 2 HES 6% 200/0.5 + 5% albumin for up to 28 days Volume (mean ± SD): 22,000 ± 11,000 mL
Outcomes	Kidney failure ('renal failure in multiorgan failure') Death Cr (statement)
Notes	Exclusions for kidney disease unknown Published data only ‐ author contacted, no further data available
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'randomised in block size 6 performed by DATA‐MAP Gmb H'
Allocation concealment (selection bias)	Low risk	'opaque sealed envelopes'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	No definition of kidney failure given, not blinded
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	CrCl has significant losses to follow‐up not accounted for in day 8 figures, numbers lower in earlier data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	No definition for kidney failure given; supported, in part, by a grant from Fresenius Kabi, Bad Homburg, Germany

Perner 2012.

Methods	Design: RCT Time frame: December 2009 to November 2011
Participants	Countries: Denmark, Norway, Finland, Iceland Setting: multicentre adult patients in ICU with severe sepsis Number (treatment/control): 398/400 Age (median (IQR)) years: treatment group (66 (56 to 75)); control group (67 (56 to 76)) Sex (male): treatment group (60%); control group (61%) Exclusions: need for RRT, see complete list in article
Interventions	Treatment group Voluven 6% HES 130/0.42 to maximum 33 mL/kg/d Volume (median (IQR)): 3000 mL (1507 to 5100) Control group Ringer's acetate Volume (median (IQR)): 3000 mL (2000 to 5750)
Outcomes	RRT RIFLE
Notes	Published data only , no further data required
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computer generated allocation sequence'
Allocation concealment (selection bias)	Low risk	'treatment assignments were concealed' at all levels of trial including analysis
Blinding (performance bias and detection bias) RRT	Low risk	'treatment assignments were concealed' at all levels of trial including analysis
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	'treatment assignments were concealed' at all levels of trial including analysis
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Protsenko 2009.

Methods	Design: RCT, triple arm Time frame: NS
Participants	Country: Russia Setting: multicentre Abdominal sepsis Number: group 1(15); group 2 (11); control group (10) Age: 32 to 72 years Sex: NS Exclusions: 'severe chronic renal failure'
Interventions	Treatment group 1 Hemohes® 6% (B/Braun, Germany) 200/0.5 Volume: 11.3 mL/kg Treatment group 2 Venofundin® 6% (B/Braun, Germany) 130/0.4 Volume: 12.9 mL/kg Treatment group 3 Gelofusine® (B/Braun, Germany) Gelatin 4% succinylated gelatin Volume: 12.7 mL/kg Duration: NS (likely single infusion)
Outcomes	RIFLE
Notes	Published and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	NS
Allocation concealment (selection bias)	Unclear risk	Envelopes
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Sander 2003.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Germany Setting: single centre Adult major gynaecological surgery patients Number (group 1/group 2): 27/29 Age (mean ± SD) years: group 1 (45 ± 15); group 2 (47 ± 14) Sex: All female Exclusions: kidney disease (Cr > 120 µmol/L)
Interventions	Treatment group 1 6% HES 200/0.5 Volume (mean ± SD): 1389 ± 610 mL Treatment group 2 6% HES 130/0.4 Volumes (mean ± SD): 1224 ± 544 mL
Outcomes	RIFLE
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'block randomisation computerized using a RANCODE'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data insufficient to affect outcomes
Selective reporting (reporting bias)	Low risk	Missing data insufficient to affect outcomes
Other bias	High risk	The study was supported by grants of B. Braun Melsungen AG, D‐34212 Melsungen, and Serum‐Werk Bernburg AG, D‐06406 Bernburg, Germany

Schortgen 2001.

Methods	Design: parallel RCT Time frame: April 1998 to September 1999
Participants	Country: France Setting: multicentre (3) Adult sepsis patients with intravascular volume depletion Number (treatment/control): 65/64 Age (median (range)) years: treatment group (60 (47 to 74)); control group (56 (44 to 71)) Sex (M/F): treatment group (43/22); control group (46/18) Exclusions: kidney disease (previous dialysis or Cr > 320 µmol/L)
Interventions	Treatment group Elohes (HES) 6% 200/0.6 for up to 4 days Volume (median (IQR)): 31 mL/kg (19 to 51) Control group 3% gelatin (35 kDa) for up to 28 days Volume (median (IQR)): 43 mL/kg (19 to 60)
Outcomes	RRT Kidney failure (2x increase in serum Cr from baseline or need for RRT, composite outcome) Death (all published) Subgroup of 2x increased Cr (unpublished), RIFLE (unpublished)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'centralized randomisation, blocks prepared for each treatment centre by statisticians not involved in executing treatment assignments'
Allocation concealment (selection bias)	Low risk	'sealed opaque envelopes serially numbered and used in sequence'
Blinding (performance bias and detection bias) RRT	High risk	Not blinded, no objective criteria
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Cr based or need for RRT
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data (death only losses)
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Difference in Cr at baseline between groups. The study was supported by grant CRC 97151 from the Assistance Publique‐Hôpitaux de Paris.

Shatney 1983.

Methods	Design: Quasi‐RCT Time frame: NS
Participants	Country: USA Setting: single centre Adult trauma patients Number (group 1/group 2): 16/16 Age (mean (range)) years: group 1 (30.4 (16 to 55)); group 2 (30.6 (18 to 66) Sex (M/F): group 1 (12/4); group 2 (13/3) Exclusions: NS
Interventions	Treatment group 1 6% Hetastarch for 5 days Volume (mean): 3900 mL (extrapolated from graph) Treatment group 2 Plasma protein fraction for 5 days Volume (mean): 3900 mL (extrapolated from graph)
Outcomes	Kidney failure ('renal failure')
Notes	Published data only ‐ author not contacted, study too old
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	'alternation'
Allocation concealment (selection bias)	High risk	'alternation'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	No evidence of blinding
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	Lack of definition for kidney failure

Shmyrev 2011.

Methods	Design: RCT Time frame: NS
Participants	Country: Russia Setting: NS Adult cardiac surgery patients on bypass Number (treatment/control): 8/7 Age (mean ± SD) years: treatment group (48 ± 13.8); control group (46.2 ± 11.1) Sex (M/F): treatment group (2/6); control group (2/5) Exclusions: pre‐existing kidney disease
Interventions	Treatment group 7.2% NaCl in hyperHAES 6% 200/0.5 Volume (mean): 4 mL/kg Control group Normal saline Volume (mean): 4 mL/kg
Outcomes	Author defined kidney failure RIFLE (from unpublished data)
Notes	Published and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Blind randomisation protocol
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome criteria
Blinding (performance bias and detection bias) Author defined kidney failure	Low risk	Aiken criteria
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome criteria
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data reported; no losses to follow‐up
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Van der Linden 2005.

Methods	Design: parallel RCT Time frame: NS
Participants	Country: Belgium Setting: single centre Adult cardiac surgery patients on cardiac bypass for pump priming and intravascular volume depletion Number (treatment/control): 65/68 Age (mean ± SD) years: treatment group (67 ± 11); control group (66 ± 8) Sex (M/F): treatment group (50/15); control group (48/20) Exclusions: kidney disease (Cr > 1.3 mg/dL)
Interventions	Treatment group 6% HES 130/0.4 intraoperatively until 20 hours postoperative Volume (mean ± SD): 48.9 ± 17.2 mL/kg Control group 3% gelatin intraoperatively until 20 hours postoperative Volume (mean ± SD): 48.9 ± 14.6 mL/kg
Outcomes	Death (published) RIFLE (unpublished)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'randomly allocated (computer generated)'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data reported; death only losses
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Vlachou 2010.

Methods	Design: RCT Time frame: 2004 to 2006
Participants	Country: Birmingham Setting: single centre Adult patients with burns (15% to 80% BSA) Number (treatment/control): 12/14 Age (mean ± SD) years: treatment group (40.8 ± 20.1); control group (42.4 ± 23.5) Sex (M/F) treatment group (7/5); control group (10/4) Exclusions: Cr > 130 µmol/L on admission
Interventions	Treatment group 6% HES 200/0.6 Elo‐Haes Volume: 1585 mL colloid at 24 hours Control group Hartmann's solution Volume: 8450 mL at 24 hours
Outcomes	RRT
Notes	Published data only (unpublished sought)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Block randomisation
Allocation concealment (selection bias)	Low risk	Sealed envelope system
Blinding (performance bias and detection bias) RRT	Unclear risk	Criteria (NS); blinding (NS)
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Unclear risk	Outcome not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	3 patients not analysed, not intention to treat
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	High risk	High number of eligible patients not enrolled

Yang 2011.

Methods	Design: RCT Time frame: NS
Participants	Country: China Setting: NS Adult surgical (non cardiac, hepatocellular carcinoma) Number (treatment/control 1/control 2): 26/30/25 Age (mean ± SD) years: treatment group (49.7 ± 1.8); control group 1 (47.8 ± 11.9); control group 2 (51.2 ± 12.9) Sex: (M/F): treatment group (21/5); control group 1 (25/5); control group 2 (22/3) Exclusions: kidney failure requiring dialysis
Interventions	Treatment group Voluven 6% 130/0.4 for 5 days Volume (mean ± SD): 3484 ± 1072.5 mL Control group 1 20% albumin for 5 days Volumes (mean ± SD): 3163 ± 999.5 mL Control group 2 Ringer's lactate for 5 days Volumes (mean ± SD): 3372 ± 965 mL
Outcomes	Author defined kidney failure RIFLE
Notes	Published and unpublished data
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'computerized random number generator'
Allocation concealment (selection bias)	Unclear risk	NS
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	High risk	Not blinded and not defined
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	High risk	4 patients in HES and 5 in RL required albumin and were excluded
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

Yassen 2011.

Methods	Design: RCT Time frame: NS
Participants	Country: NS Setting: NS Adult patients recipients of living donor liver transplant Number (treatment 1/treatment 2/control): 15/15/15 Age: NS Sex: NS Exclusions: NS
Interventions	Treatment group 1 Voluven 6% HES 130/0.4 low dose 30 mL/kg/d Treatment group 2 Voluven 6% HES 130/0.4 high dose 50 mL/kg/d Control group Albumin 4%
Outcomes	RIFLE
Notes	Published and unpublished data, abstract only
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	'randomly stratified'
Allocation concealment (selection bias)	Unclear risk	'blind'
Blinding (performance bias and detection bias) RRT	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) Creatinine based	Low risk	Objective outcome measure
Blinding (performance bias and detection bias) Author defined kidney failure	Unclear risk	Outcome not reported
Blinding (performance bias and detection bias) RIFLE	Low risk	Objective outcome measure
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data reported
Selective reporting (reporting bias)	Low risk	Pre‐specified outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias

BUN ‐ blood urea nitrogen; Cr ‐ creatinine; CrCl ‐ creatinine clearance; IQR ‐ interquartile range; NS ‐ not stated; RRT ‐ renal replacement therapy

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aksun 2009	No individual creatinines available
Allison 1999	No individual creatinines available
Ando 2008	No individual creatinines available
Beyer 1997	No individual creatinines available
Boldt 1993	Excluded due to concern regarding reliability of data
Boldt 1998	Excluded due to concern regarding reliability of data
Boldt 2000a	Retracted
Boldt 2000b	Retracted
Boldt 2003	Retracted
Boldt 2006	Retracted
Boldt 2007a	Retracted
Boldt 2007b	Retracted
Boldt 2008	Retracted
Boldt 2009	Retracted
Boldt 2010	Retracted
Chen 2006	No individual creatinines available
Dehne 1997	The intervention group received a 12 mL/kg/d fixed dose of 10% HES 200/0.5 as a continuous infusion for 5 days. The study was excluded because the control group did not receive a comparison fluid.
Hanart 2009	No individual creatinines available
Kalayanarooj 2008	10% Haes‐steril versus Dextran in Dengue Fever patients, mean creatinines reported only, unable to contact author for individual data
Kulla 2008	Double intervention, 6% 130/0.42 in acetate + balanced electrolyte crystalloid versus 6% 130/0.42 in NS + less balanced electrolyte crystalloid.
Kvalheim 2010	No individual creatinines available
Langeron 2001	No individual creatinines available
Liet 2003	Volume not used in management of hypovolemia
NCT00576849	Questionable data quality
Ooi 2009	No individual creatinines available
Petrikov 2008	No individual creatinines available
Sade 1985	No individual creatinines available
Schewior 2008	No individual creatinines available
Shahbazi 2011	No individual creatinines available
Tiryakioglu 2008	Voluven versus Ringer's lactate in cardiac surgery patients on bypass, mean creatinines reported only, unable to contact author for individual patient data for RIFLE analysis
Vernetta 2012	Inappropriate study group ‐ patients undergoing nephrectomy
Vogt 1999	No individual creatinines available
Wu 2010	No individual creatinines available

Differences between protocol and review

Normovolaemic and hypervolaemic haemodilution studies were removed because they were not treatments of intravascular volume depletion. Due to insufficient data, the patient population subgroup analysis was modified after studies were selected but before any analysis was undertaken.

Contributions of authors

1. Draft the protocol: AD, TM, CR

2. Study selection: AD, TM, CR

3. Extract data from studies: AD, TM, CR

4. Enter data into RevMan: AD, TM, CR

5. Carry out the analysis: AD, TM, CR

6. Interpret the analysis: AD, TM, CR

7. Draft the final review: AD, TM, CR

8. Disagreement resolution: All authors

9. Update the review: AD, TM, CR

Sources of support

Internal sources

• Department of Anesthesia, Faculty of Medicine, University of Manitoba, Canada.

  Funds for statistical consultation were provided through a department grant.

External sources

• No sources of support supplied

Declarations of interest

None declared.

New search for studies and content updated (conclusions changed)