Concerns over use of hydroxyethyl starch solutions.

Concerns about serious side effects of hydroxyethyl starch (HES) began to emerge in the 1970s, soon after it was first licensed. The concerns grew after three multicentre randomised trials found that administration of HES in patients who had sepsis or were critically ill was associated with a higher risk of kidney injury and bleeding,1 2 3 and more deaths in patients who had sepsis,1 3 when compared with crystalloids. The trials prompted the German Federal Institute for Drugs and Medical Devices to ask the European Medicines Agency (EMA) to review the use of HES in 2012. The EMA committee (of which KR was a member) recommended suspending all use of HES. However, after manufacturers requested that the decision was reviewed, the UK advised its immediate suspension triggering a second EMA review that decided to permit ongoing HES use under limited circumstances. More than two hundred million units of HES have been administered worldwide during the past 10 years (box). 4 We explain here why we believe that the decision to allow its continued use was not in the best interest of patients.

HES is a synthetic carbohydrate polymer marketed as a volume expander in different molecular weights and degrees of substitution

More than 20 different HES solutions from several manufacturers are available in the European Union and more than 30 preparations worldwide

HES has been used in a variety of clinical settings, including during surgery and to treat hypovolaemia after trauma, burns, and in sepsis, despite a lack of evidence of clinical benefit compared with other intravenous fluids such as albumin and crystalloids5

Questionable evidence

There is no question that HES can effectively expand blood volume. However, the data submitted to the US Food and Drug Authority for its licence in 1972 would today be considered inadequate to detect harms. The freedom of information requests show that the pivotal clinical trials were uncontrolled, short term studies totalling 315 patients and volunteers observed for less than 24 hours. Subsequent submissions of modified or “improved” products relied on similarly inadequate data, since submissions needed only to demonstrate non-inferiority to earlier HES preparations.6 7 Notably, a meta-analysis of clinical studies of HES up to 2011 found that many would not have been able to detect harmful effects because the studies had inadequate observation periods (median 12 hours), lacked patient relevant endpoints, and used control solutions with a risk profile similar to HES.6 Also, a substantial number of published clinical data promoting HES were produced by one of the manufacturers’ corporate funded speakers, and 88 of that author’s publications were subsequently retracted for scientific fraud in 2011.8

We examined publications evaluating HES and found over 100 almost uniformly favourable reviews during the past decade, of which more than 50 were authored or coauthored by 12 corporate funded investigators.9 The sheer number of industry sponsored publications overwhelmed the publication of carefully performed meta-analyses, 10/12 of which recommended against use of HES.9 There is also evidence of suppression of unfavourable data about HES, including non-publication of negative outcomes (selective outcomes reporting)10 11 12 13 or of entire trials with unfavourable outcomes (selective publication),14 and numerous examples of “positioning” inconclusive HES results in a positive way.9

Harms of HES

In the 1970s it was reported that HES interfered with coagulation15 and was stored in tissue,16 but not until the 1990s were there reports of renal abnormalities associated with HES administration . This may be because renal impairment can be delayed by months and may not be apparent to the surgeon or anaesthetists who performed the procedure. The FDA and BfArM say that their databases have recorded only 39 episodes of kidney failure, and reports of adverse reactions fell in the 1990s despite the concomitant increase in HES sales.

Research has found that HES has been linked with fatal intracranial haemorrhage,17 increased bleeding in patients having cardiac surgery,18 and renal failure (box).19 20 21 Between 2008 and 2012, three large randomised trials by different teams all found higher risks with HES than with other volume expanders (table 1).1 2 3 The first trial showed significantly increased rates of acute renal failure (absolute increase 12.1%) and need for renal replacement therapy (absolute increase 12.2%), increased use of blood products (7.3%), and a higher 90 day mortality (7.1%) in patients with sepsis resuscitated with HES compared with those treated with crystalloid .1 The other two found that modern tetrastarch, which was thought less toxic than earlier HES formulations, was nephrotoxic (absolute increase in number requiring renal replacement therapy 1.2% in critically ill and 5.6% in septic patients)2 3 and increased the need for blood transfusion2 3; one study also found an absolute increase in 90 day mortality of 7.5% in patients with severe sepsis (see data supplement on bmj.com for details).3

Table 1

Selected data from three multicentre trials of HES

First author	Population studied	No enrolled	Test group	Control	No (%) with outcome (intervention v control)
					28 day mortality	90 day mortality	Requiring renal replacement therapy	Use of blood products
Brunkhorst1	Severe sepsis	600	10% HES 200/0.5	Ringer’s lactate	70/262 (26.7) v 66/274 (24.1), P=0.48	107/261 (41.0) v 93/274 (33.9), P=0.09	81/261 (31.0) v 51/272 (18.8), P=0.001*	Exposure to red cell transfusions (over 21 days): 189/275 (68.7) v199/262 (76), P=0.06
Myburgh2	Intensive care	7000	6% HES 130/0.4	0.9% NaCl	458/3313 (13.8) v 437/3331 (13.1), P=0.40	597/3315 (18.0) v 566/3336 (17.0), P=0.26	235/3352 (7.0) v 196/3375 (5.8), P=0.04†	Mean (SD) daily volume of blood products (first four days): 78 (250) ml v 60 (190) mL, P<0.001
Perner3	Severe sepsis	804	6% HES 130/0.4	Ringer’s acetate	154/398 (38.7) v 144/400 (36.0), P=0.43	201/398 (50.5) v 172/400 (43.0); P=0.03	87/398 (21.9) v 65/400 (16.3), P=0.04†	Exposure to blood product transfusion: 243/376 (64.6) v 204/380 (53.7), P=0.003
								Severe bleeding (clinical bleeding requiring ≥3 units of packed red cells within 24 hours) 38/398 (9.5) v 25/400 (6.3), P=0.09

*At 21 days.

†At 90 days.

1972: HES licensed in the US for treatment and prophylaxis of hypovolaemia based on small uncontrolled studies. Subsequent product modifications approved predominantly on the same data6 7

1970s: Reports that HES interferes with coagulation15 and that 10-30% of infused HES persists in tissue stores16

1993: Reports of renal failure first noted.19 Subsequent observations of increased kidney transplant failure when donors resuscitated with HES20

2001: French authorities add a warning label after reports of intracranial bleeding in patients with subarachnoid haemorrhage17

2001: Study finds that patients with severe sepsis resuscitated with starches developed acute kidney failure more frequently than patients receiving gelatin21

2004: FDA issues a warning label because of increased bleeding in cardiac surgery18

2008-12: Three multicentre randomised controlled studies find increased rates of renal failure 1]2 3 and bleeding 1 2 3 associated with HES in critically ill and septic patients and increased risk of death1 3 in septic patients

A subsequent meta-analysis published in JAMA reinforced these findings and increased their generalisability.22 The authors combined data from 28 clinical trials including 10 290 critically ill patients in emergency or intensive care settings (none of the trials evaluating deaths or renal failure was by the fraudulent author). HES was associated with increased mortality (1.51%; 95% confidence interval 0.02% to 3%) compared with other fluids (crystalloids, albumin, and gelatin), and increased the risk of needing renal replacement therapy by 3.12% (0.47% to 5.78%).22 These harms were not isolated to one manufacturer, molecular weight, or molar substitution of HES and probably represent a class effect of starches. The mechanisms of HES toxicity are not known, but adverse consequences may result from the rapid accumulation of HES in tissues and macrophages.16

Sales figures from market analyses estimate that from 2002 to 2012, about 240 million units of HES were sold worldwide.4 In the most recent studies using modern HES preparations (lower molecular weight and molar substitution), an average of 2.2 litres or 4.4 units were administered per patient.7 Up to 55 million patients may therefore have been exposed to HES over the past decade and, based on the risks in the JAMA meta-analysis, thousands may have died or required renal replacement therapy as a result.

Regulatory response to concerns

In 2013, after the three large trials showed harms, the FDA issued a “black box” warning about increased mortality and severe renal injury in critically ill patients, advising that it should not be used in this population.18 At about the same time, an EMA pharmacovigilance risk assessment committee (PRAC) reviewed the evidence and recommended suspension of all marketing authorisations for HES.23 Some manufacturers requested that the EMA re-examine this decision, which meant that suspension throughout Europe was delayed.24 However, the UK Medicines and Healthcare Products Regulatory Agency took independent action and withdrew HES,25 and Italy and Poland also acted to limit its use.

In accordance with EU legislation, the action in the UK triggered further urgent review. A second assessment committee, with different leaders and external advisers, looked at the available data. Although the second committee agreed that HES solutions must not be used in critically ill patients or those with sepsis and burn injuries, it allowed their continued use in patients with hypovolaemia due to acute blood loss within the first 24 hours after elective surgery or trauma, and limited the maximum dose. There was, however, considerable dissent about this decision, with only 19 of the 33 eligible members voting in favour.26 27 The committee also requested large randomised post-marketing clinical trials to establish the efficacy and safety of HES in perioperative and trauma populations.28 This recommendation was ratified by the European Commission in December 2013 and became a legally binding regulation in all member states.29

The 2013 EU regulation created a dilemma for the countries that had banned HES but were now required by EU law to reinstate its use. Although alternative treatments with equivalent efficacy and superior safety profiles are available,5 countries can now be pressured by manufacturers to return HES to pharmacy shelves across Europe. Three of us (KR, CN, CSH) were signatories to an open letter protesting the revised decision, which was sent to the EMA at the end of 2013 by more than 80 clinicians and leaders of medical societies worldwide.30

The case for keeping HES

According to the EMA, important new information from three sources led to the revised decision.28 These were a small unpublished single centre study of HES in sepsis,31 unpublished data from a German intensive care registry sponsored by HES manufacturers and conducted by a marketing and communications agency,32 and data from a new trial of colloids versus crystalloids in critically ill hypovolaemic patients.33 The primary endpoint in this trial, 28 day mortality, was not significantly different in the two arms of this trial. It did find that 90 day mortality was significantly lower in the colloid group, although only before correction for multiple comparisons. Furthermore, the 90 day endpoint was added midway through the trial and patients in the colloid arm received three different colloids, not just HES.33 34 Since all the above new data apply to patients with sepsis or in intensive care, it is puzzling that the committee used them to support its decision to retain HES use in surgical and trauma patients.

According to the detailed assessment report, the second PRAC review of numerous published and unpublished studies provided “some reassurance that the risks of mortality and renal injury in surgical and trauma patients may be lower than those in critically ill and [septic] patients.”27 We analysed the strength of data from these studies based on generally accepted criteria of rigour for clinical trials (table 2). Our analysis suggests that only the three multicentre randomised trials in patients with sepsis or in intensive care1 2 3 had the needed sample size, adequate control fluids, and length of follow-up to evaluate HES safety. The fact that no harmful effects have been identified in surgical populations may therefore be the result of poor study quality rather than an absence of side effects in these patients.

Table 2

Methodological rigour of original research investigations examined by two EMA committees reviewing use of HES (reference details are available in the data supplement on bmj.com)

First author	Population studied	Study discussed in first EMA decision	Study discussed in second EMA decision	Double blind	Multicentre	>500 patients enrolled*	Crystalloid or albumin control *	90 day follow-up *	Published in peer review journal
Perner3	Sepsis	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Myburgh2	Intensive care	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Brunkhorst1	Sepsis	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes
Guidetw14	Sepsis	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes
Annanew3	Intensive care	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes
Feldheiserw10	Surgery	Yes	Yes	Yes	No	No	Yes	Yes	Yes
Siegemundw45	Sepsis	Yes	Yes	Yes	No	No	Yes	Yes	No
Jamesw24 w25	Trauma	Yes	Yes	Yes	No	No	Yes	No	Yes
NCT01117649	Surgery	No	Yes	Yes	Yes	No	Yes	No	No
Gondosw15	Fluid bolus†	No	Yes	No	Yes	No	Yes	No	Yes
Hamajiw22	Surgery	No	Yes	Yes	No	No	Yes	No	Yes
Magderw30	Fluid bolus§	Yes	Yes	Yes	No	No	Yes	No	Yes
Gallandet Huetw13	Surgery	Yes	No	Yes	Yes	No	No	No	Yes
Hanartw20	Pediatric surgery	Yes	No	Yes	No	No	Yes	No	Yes
Standlw48	Pediatric surgery	Yes	No	No	Yes	No	Yes	No	Yes
Godetw14	Surgery	Yes	No	No	Yes	No	No	Yes	Yes
Frenchw12	Preload‡	No	Yes	Yes	No	No	Yes	No	Yes
Alaviw2	Surgery	No	Yes	Yes	No	No	Yes	No	Yes
Martin 2002w31	Surgery	No	Yes	No	No	No	Yes	No	Yes
Yangw55	Surgery	No	Yes	No	No	No	Yes	No	Yes
Leew28	Surgery	No	Yes	No	No	No	Yes	No	Yes
Verheijw52	Fluid bolus†	No	Yes	No	No	No	Yes	No	Yes
Rackoww41	Fluid bolus†	No	Yes	No	No	No	Yes	No	Yes
Trofw49	Fluid bolus†	No	Yes	No	No	No	Yes	No	Yes
Hasanw21	Preload‡	No	Yes	No	No	No	Yes	No	Yes
Rileyw42	Preload‡	No	Yes	No	No	No	Yes	No	Yes
Ueyamaw50	Preload‡	No	Yes	No	No	No	Yes	No	Yes
Siddikw44	Preload‡	No	Yes	No	No	No	Yes	No	Yes
MadiNoJebaraw29	Preload‡	Yes	Yes	No	No	No	Yes	No	Yes
Ickxw23	Surgery	Yes	No	Yes	No	No	No	No	Yes
Lindroosw27	Neurosurgery	No	Yes	No	No	No	Yes	No	Yes
FengerNoEriksenw11	Surgery	No	Yes	No	No	No	Yes	No	Yes
Younesw56	Trauma	Yes	No	No	No	No	Yes	No	Yes
Akkucukw1	Pediatric surgery	No	Yes	No	No	No	Yes	NR	Yes
Nagyw36	Trauma	Yes	No	No	No	No	Yes	No	Yes
Vlachouw53	Burns	No	Yes	No	No	No	Yes	No	Yes
Kasperw26	Surgery	Yes	No	No	No	No	No	No	Yes
Mercier§	Preload‡	Yes	No	NR	NR	No	Yes	No	No

NR=not reported.

*Factors essential to detect harm from HES.

†Fluid bolus studies with up to two hours duration performed in patients with hypovolaemia after surgery or in the intensive care unit.

‡Preloading of fluid bolus to prevent hypotension after spinal block for caesarean section.

§Same study as Martin 2002.w31

The way forward

The EMA has requested further research but is this feasible? Member states may find it difficult to comply with the EMA’s request for post-marketing studies in perioperative and trauma patients. If trials are designed, ethical review boards, clinical investigators, and patients may question whether participants should consent to receiving a product that has no benefit over alternatives and for which there is evidence for harm in critically ill patients. There is little reason to expect different results in elective surgery or trauma. Notably, a recent meta-analysis of studies investigating use of HES in surgical settings confirmed an increased risk of requiring renal replacement therapy (relative risk=1.44, 95% confidence interval 1.04 to 2.01).35 Replacing acute blood loss in surgical and trauma patients with a product that induces coagulopathy may also be difficult to justify when safer alternatives are available.18 36 37 Postmarketing studies must be rigorously designed with patient relevant outcomes, suitable control fluids, and follow-up of at least 90 days.

While awaiting further studies, we believe that the market authorisation for HES should be suspended for all clinical conditions. We advise doctors to avoid using HES for resuscitation when other fluids are available. Withdrawing marketing authorisation now would not prevent future studies under current regulatory oversight. We believe that manufacturers should use preclinical studies to investigate new formulations of HES designed to eliminate the recognised harms before embarking on further clinical trials.

The data used to approve HES in the 1970s would be insufficient by modern regulatory standards

Three randomised controlled trials found that critically ill recipients are at greater risk of kidney injury requiring dialysis and, in two studies, of death

The FDA has issued a black box warning for HES products in all critically ill adult patients, and it has been withdrawn in the UK and some other European countries

EMA revised its initial decision to withdraw HES and allowed limited use for acute blood loss in trauma and after elective surgery

Studies in such patients are of poor quality and may not have been able to detect harms