Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

A comparison of albumin and saline for fluid resuscitation in the Intensive Care Unit.

The SAFE Study Investigators.

New Engl J Med 2004; 350:2247-2256 [abstract]

Reviewed by Sandrijn M. van Schaik MD, PhD, Combined Fellowship Program for Pediatric Critical Care Medicine of the Children's Hospital and Research Institute at Oakland and the University of California at San Francisco.

Review posted August 15, 2004

---

I. What is being studied?:

The study objective:

To compare the effect of fluid resuscitation with albumin or saline on mortality in ICU patients.

The study design:

Multi-center, randomized, double-blind.

The patients included:

Patients 18 year and older admitted to the multidisciplinary ICUs of 16 academic tertiary hospitals in Australia and New Zealand between November 2001 and June 2003, who required fluid resuscitation for intravascular fluid depletion that was in addition to intravenous fluid required for nutrition or to replace ongoing losses or to restore normonatremia. The need for fluid resuscitation was supported by at least one of a the following list of clinical signs:

1. Heart rate greater than 90 beats per minute
2. Systolic blood pressure (SBP) less than 100 mmHg or mean arterial pressure (MAP) less than 75 mmHg or a 40 mmHg decrease in SBP or MAP from the baseline recording or requirement for inotropes or vasopressors to maintain blood pressure at those levels.
3. Central venous pressure less than 10 mmHg
4. Pulmonary capillary wedge pressure less than 12 mmHg
5. Respiratory variation in systolic or mean arterial blood pressure of greater than 5 mmHg
6. Capillary refill time greater than one second
7. Urine output less than 0.5 mL/kg for one hour (1)

The patients excluded:

The major exclusion criteria were: patients following cardiac surgery, burns or liver transplantation, and moribund patients expected to die within 24 hours (1).

The interventions compared:

Fluid resuscitation with 4% albumin versus normal saline. The treating physician determined the amount and rate of fluid administration.

The outcomes evaluated:

1. Primary outcome: mortality at 28 days
2. Secondary outcomes:
   1. Length of stay in the ICU and in hospital.
   2. Duration of mechanical ventilation
   3. Duration of renal replacement therapy
   4. Occurrence of new organ failure (defined by an increase in the Sequential Organ-Failure Assessment Score 1 to 3 or 4 from a score of 1 or 2 at baseline)

II. Are the results of the study valid?

Primary questions:

1. Was the assignment of patients to treatments randomized?

Yes. Randomization was performed centrally for all study centers via a web site. Randomization was stratified per institution and by a diagnosis of trauma at ICU admission.

2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?

Was followup complete?

Yes, for 99% of patients follow-up data regarding primary and secondary outcomes were available. Some of the data reported in the study not pertaining to these outcomes are incomplete (e.g., CVP, albumin levels and other variables noted in table 2), presumably because these weren't measured in all patients or patient records were not complete. There is however no discrepancy in this regard between the 2 study groups; for example CVP measurements on day 1 were reported for approximately 2/3 of patients in both groups.

Were patients analyzed in the groups to which they were randomized?

Yes, data were analyzed on an intention-to-treat basis. 197 patients (2.8%, 90 in the albumin group and 107 in the saline group) did not receive any study fluid (presumably because they were deemed not to need resuscitation after all). A substantial number of patients received resuscitation fluids in addition to allocated study fluid (309 patients [8.8%] in the albumin group, 375 patients [10.7%] in the saline group). Reason for this was most commonly error (189 patients [5.4%] in the albumin group, 190 patients [5.4%] in the saline group), but clinician's preference for a specific non-study resuscitation fluid played a role also (68 patients [1.9%] in the albumin group, 103 patients [2.9%] in the saline group). Three patients were accidentally randomized twice during the 28-day study period but analyzed in the group they were first randomized to.

Secondary questions:

3. Were patients, health workers, and study personnel "blind" to treatment?

Yes. Both containers and administration sets were masked (2) and the effectiveness of blinding was reportedly confirmed prior to the start of the study.

4. Were the groups similar at the start of the trial?

Baseline characteristics including age, gender, reason for admission and source of admission were similar between the 2 groups. In addition, the number of patients in predefined subgroups of patients with trauma, severe sepsis and ARDS was equally distributed over both treatment groups, although randomization was stratified for a diagnosis of trauma only. There was no difference between treatment groups in baseline APACHE II score, presence of organ failure, requirement for mechanical ventilation or renal replacement therapy. Physiological variables including heart rate, mean arterial pressure, urine output and serum albumin were similar at baseline, but there was a significant difference between central venous pressure: 9.0 ± 4.7 mmHg in the albumin group versus 8.6 ± 4.6 mmHg in the saline group at baseline (P=0.03).

This seems a negligible difference from a clinical perspective although it is notable that this is the only reported baseline physiological variable with a mean value below the preset value that served as a criterion for need for resuscitation. In other words, if low central venous pressure was the main reason for administration of resuscitation fluid, the albumin group was less in need of resuscitation than the saline group which may have led to less aggressive resuscitation.

5. Aside from the experimental intervention, were the groups treated equally?

No, but the differences in treatment are most likely a result of the experimental intervention. For example, the total volume of resuscitation fluid administered over the first 4 days to the normal saline group was 1.4 times more than the total volume administered to the albumin group. This may be because resuscitation end goals can be achieved with lesser amounts of albumin (as has been described in previous studies (3)), although these end goals were not identified in the study. As noted above (question 2) a substantial number of patients received resuscitation fluids in addition to allocated study fluid. Although the proportion of patients receiving non-study resuscitation fluid was similar between groups, it was not specified whether this was albumin, normal saline or some other fluid.

The albumin group also received more packed red cell transfusions than the saline group. The reason for this is unclear; the authors postulate that it could be due to increased hemodilution by albumin as compared to saline, or due to increased losses due to albumin-induced coagulopathy. In a randomized controlled trial (the CRIT trial) comparing a restrictive packed cell transfusion regimen with a more liberal regimen, Hebert et al. found that the liberal transfusion regimen was associated with increased mortality (4). The authors of the SAFE study refer to this observation but state that the excess of transfused packed cells in the albumin group was small (70 ml or approximately 1/4 of a unit per patient) in comparison to the excess of transfused cells in the CRIT trial (3 units/patient). They therefore state that the difference in transfused amount is unlikely to have influenced the results. In all likelihood however, the average of 70 ml of packed cells per patient reflects patients who did not receive any blood transfusion as well as patients who received 1 or more units of packed cells. Data on how many patients were transfused were not provided.

III. What were the results?

1. How large was the treatment effect?

Primary outcome: There was no difference between the 2 treatment groups in primary outcome (death at 28 days). Death rate was 20.9% in the albumin group and 21.1% in the saline group, relative risk (RR) 0.99, 95% confidence interval (CI) 0.91-1.09.

Secondary outcomes: There was no difference between treatment groups in any of the secondary outcomes:

	Albumin	Saline	Absolute difference (95% CI)	P-value
Length of stay in the ICU, days	6.5 ± 6.6	6.2 ± 6.2	0.24 (-0.06 to 0.54)	0.44
Length of stay in hospital, days	15.3 ± 9.6	15.6 ± 9.6	-0.24 (-0.70 to 0.21)	0.30
Duration of mechanical ventilation, days	4.5 ± 6.1	4.3 ± 5.7	0.19 (-0.08 to 0.47)	0.74
Duration of renal replacement therapy, days	0.48 ± 2.28	0.39 ± 2.0	0.09 (-0.0 to 0.19)	0.41
New organ failure, number (%)				0.85
No failure	1397 (52.7)	1424 (53.3)
1 organ	795 (30)	796 (29.8)
2 organs	369 (13.9)	361 (13.5)
3 organs	68 (2.6)	75 (2.8)
4 organs	18 (0.7)	17 (0.6)
5 organs	2 (0.1)	0

Subgroup analyses for primary outcome (mortality at 28 days) revealed that there was a significant more mortality in trauma patients who received albumin than those in this subgroup who received saline (see table below). This excess mortality was exclusively due to excess mortality in patients with traumatic brain injury, since there was no difference in mortality between treatment groups if patients with traumatic injury were excluded from this subgroup analysis. In contrast, among patients with severe sepsis mortality was significantly higher in the saline group. There was no difference in mortality between treatment groups in the subgroup of patients with ARDS.

	Albumin	Saline	Relative Risk (95% CI)	P-value
Trauma	81/596 (13.6)	59/590 (10.0)	1.36 (0.99 to 1.86)	0.06
i. With brain injury	59/241 (24.6)	38/251 (15.1)	1.62 (1.12 to 2.34)	0.009
ii. Without brain injury	22/355 (6.2)	21/339 (6.2)	1.00 (0.56 to 1.79)	1.00
Severe sepsis	185/603 (30.7)	217/615 (35.3)	0.87 (0.74 to 1.02)	0.09
ARDS	24/61 (39.3)	28/66 (42.4)	0.93 (0.61 to 1.41)	0.72

2. How precise was the estimate of the treatment effect?

The 95% confidence interval for the relative risk regarding the primary outcome was narrow and crossed one, indicating a high likelihood that the observed lack of difference between treatment groups represents a true and not a chance finding. For secondary outcomes, the 95% confidence intervals for the absolute differences between groups were also relatively narrow and crossed zero, again indicating a good precision of the estimated similarity between treatment groups.

The subgroup analyses for the primary outcome only show a statistically significant difference when the trauma subgroup is split up in a "with brain injury" and a "without brain injury" group. The relative risk for death is 1.62 for patients with traumatic brain injury randomized to the albumin group compared to patients with traumatic brain injury randomized to the saline group. The 95% confidence interval is fairly wide (1.12 to 2.34) As the authors note in the discussion, caution is warranted with subgroup analyses, since the study was not designed or powered to study the differences between subgroups and it is not uncommon to find chance differences between subgroups in large trials.

Oxman and Guyatt published "A consumer's guide to subgroup analyses" (5) in which they suggest 7 guidelines for deciding whether apparent differences in subgroup response are real. These include

1. Is the magnitude of the difference clinically important?
2. Was the difference statistically significant?
3. Did the hypothesis precede rather than follow the analysis?
4. Was the subgroup analysis one of a small number of hypotheses tested?
5. Was the difference suggested by comparisons within rather than between studies?
6. Was the difference consistent across studies?
7. Is there indirect evidence that supports the hypothesized difference?

In regards to the subgroup analysis of patients with traumatic brain injury, the answer to the first 2 questions would be "yes". The hypothesis that patients with trauma would exhibit a differential effect to the intervention preceded the analysis, but it is unclear whether the same is true regarding the hypothesis that the subgroup with traumatic brain injury would have a different outcome with different resuscitation fluids. As far as can be deduced from the data presented, only a small number of subgroup analyses were performed. Related to the 6th question, the authors of the SAFE study refer to the meta-analysis published by Choi et al. in which colloid resuscitation was found to be associated with increased mortality in patients with trauma as compared to resuscitation with crystalloids (6). The analyzed trials however involved patients with abdominal trauma, not traumatic brain injury.

A search of the literature for randomized controlled trials of resuscitation fluids involving patients with traumatic brain injury did not reveal any other studies than the SAFE study. In regards to question 7, one could postulate that differences in colloid oncotic pressure are responsible for the observed difference between treatment groups in the traumatic brain injury group, with albumin perhaps leading to increased cerebral edema in the presence of a disrupted blood-brain barrier. Although 5% human albumin as available in the United States typically has a sodium concentration between 130-160 mEq/L, the sodium content of the 4% albumin solution used in the SAFE study may be lower, which would be another possible explanation for the observed difference.

The other subgroup analyses did not show statistically significant differences between treatment groups. The difference in mortality in patients with severe sepsis appeared to favor albumin for resuscitation in this subgroup. Statistically significance was not reached but the confidence interval was fairly narrow and barely crossed one. There is some evidence in the literature that colloids result in less pulmonary edema than crystalloids when administered to patients with capillary leak syndromes (7), but clinical outcome (including mortality) was not affected in these studies. Following the guidelines outlined above the observed difference in this subgroup analysis is most likely due to chance.

IV. Will the results help me in caring for my patients?

1. Can the results be applied to my patient care?

To answer this question it is important to know the characteristics of the excluded patients, information not provided by the authors of the SAFE study. In particular exclusion of moribund patients (criterion 9 (1)) may have skewed the study population away from the sickest patients.

The biggest pitfall with extrapolating the findings of the SAFE study to the PICU population is of course that pediatric patients were excluded from the study, and the etiology of pediatric diseases leading to resuscitation requirement differs to some extent from those seen in the adult population. Even with similar diseases, such as sepsis, the presentation (including hemodynamic status) can be quite different. From a biological standpoint there is no reason to believe that older pediatric patients would have different responses to albumin or saline from adults. Neonates however may be more susceptible to the calcium binding effects of albumin since the neonatal myocardium is more dependent on extracellular calcium for contractility (8).

This effect may be especially pronounced in infants after surgery for congenital heart disease, a group that in many pediatric ICU's routinely receives albumin for resuscitation in the post-operative period. Patients who underwent cardiothoracic surgery were excluded from the SAFE study and a literature search did not reveal any studies regarding choice of fluids in pediatric patients after cardiac surgery. A randomized study by Riegger et al. (9) comparing albumin with crystalloid prime solution for cardiopulmonary bypass in children did not show a clear benefit for either fluid, although the albumin group demonstrated less fluid retention and required more blood transfusion. Outcome in terms of duration of mechanical ventilation, intensive care or hospital stay and mortality was similar the groups in this study (9).

Overall, there is a striking lack of data regarding resuscitation fluids in the pediatric age group. A literature search revealed 2 randomized controlled trials comparing albumin with normal saline in (premature) neonates with hypotension (10,11). Primary outcome was resolution of hypotension in one trial (10), with no difference observed between treatment groups. The other trial studied premature infants only and outcome was defined by need for inotropic support, death or chronic lung disease, with no differences noted between treatment groups, although the albumin group was noted to have more fluid retention (11). A randomized controlled trial in children with Dengue shock syndrome compared 4 different fluids for initial resuscitation including 2 synthetic colloid solutions (dextran and gelatin) and 2 crystalloid solutions (lactated Ringer's and normal saline) (12). Primary outcome was recovery from shock as defined by restoration of normal pulse pressure. The lactated Ringer's group had a slower recovery than the other 3 groups, but despite randomization there were marked differences in severity of illness between groups at baseline. There were no deaths in the entire study cohort and an albumin group was not included.

2. Were all clinically important outcomes considered?

Other outcomes frequently considered in studies looking at resuscitation fluids include presence of pulmonary edema and coagulopathy. The former was not addressed in this study, but duration of mechanical ventilation (as an indirect, and possible clinically more relevant measure) was. In light of the increased packed cell requirement in the albumin group, it would have been interesting to know whether coagulation status changed with treatment as has been described before (13).

3. Are the likely treatment benefits worth the potential harms and costs?

As mentioned in the accompanying editorial, the SAFE study could lead people to different conclusions. On the one hand, its findings contradict the conclusion from the meta-analysis published by the Cochrane Injuries Albumin Reviewers that administration of albumin-containing fluids was associated in a 6 percent higher risk of mortality as compared to crystalloid solutions (14). The only exception may be patients with traumatic brain injury, for whom normal saline might be a better choice. On the other hand, a possibly stronger argument could be made that there is no place for albumin in fluid resuscitation since there is no benefit to albumin whereas the cost is substantial. In addition, there may be adverse effects such as coagulopathy, resulting in increased requirement for packed cell transfusion, which may by itself affect outcome (4).

References:

1. Supplementary Appendix to The SAFE Study Investigators. A Comparison of Albumin and Saline for Fluid Resuscitation in the Intensive Care Unit. N Engl J Med 2004;350:2247-2256, accessed on-line via http://content.nejm.org/cgi/content/full/350/22/2247/DC1
2. ANZICS Clinical Trials Group and Institute for International Health SAFE Study Investigators. The Saline vs. Albumin Fluid Evaluation (SAFE) Study (ISRCTN76588266): Design and conduct of a multi-centre, blinded randomized controlled trial of intravenous fluid resuscitation in critically ill patients. Accessed on-line via http://bmj.bmjjournals.com/cgi/content/full/326/7389/559/DC1
3. Haupt MT, Rackow EC. Colloid osmotic pressure and fluid resuscitation with hetastarch, albumin and saline solutions. Crit Care Med 1982;10(3):159-62. [abstract]
4. Hebert PC, Wells G, Blajchman MA, Marshall J, Martin C, Pagliarello G, Tweeddale M, Schweitzer I and Yetisir E. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group N Engl J Med. 1999;340(6):409-17 [abstract] [PedsCCM EB Journal Club Review]
5. Oxman AD, Guyatt GH. A consumer's guide to subgroup analyses. Ann Int Med 1992;116:78-84. [abstract]
6. Choi PT, Yip G, Quinonez LG and Cook DJ. Crystalloids vs. colloids in fluid resuscitation: a systematic review. Crit Care Med. 1999;27(1):200-10 [abstract] [PedsCCM EB Journal Club Review]
7. Haupt MT, Kaufman BS and Carlson RW. Fluid resuscitation in patients with increased vascular permeability. Crit Care Clin 1992;8(2):341-353 [abstract]
8. Schwartz SM, Duffy JY, Pearl JM and Nelson DP. Cellular and molecular aspects of myocardial dysfunction. Critical Care Medicine 2001; 29(10 Suppl): S214-9 [abstract]
9. Riegger LQ, Voepel-Lewis T, Kulik TJ, Malviya S, Tait AR, Mosca RS and Bove EL. Albumin versus crystalloid prime solution for cardiopulmonary bypass in young children. Crit Care Med. 2002;30(12):2649-54. [abstract]
10. Oca MJ, Nelson M and Donn MS. Randomized Trial of Normal Saline versus 5% Albumin for the Treatment of Neonatal Hypotension. J Perinatol. 2003;23(6):473-6 [abstract]
11. So KW, Fok TF, Ng PC, Wong WW and Cheung KL. Randomised controlled trial of colloid or crystalloid in hypotensive preterm infants. Arch Dis Child Fetal Neonatal Ed. 1997;76(1):F43-6 [abstract]
12. Ngo NT, Cao XT, Kneen R, Wills B, Nguyen VM, Nguyen TQ, Chu VT, Nguyen TT, Simpson JA, Solomon T, White NJ and Farrar J. Acute management of dengue shock syndrome: a randomized double-blind comparison of 4 intravenous fluid regimens in the first hour. Clin Infect Dis. 2001;32(2):204-13 [abstract]
13. Johnson SD, Lucas CE, Gerrick SJ, Ledgerwood AM and Higgins RF. Altered coagulation after albumin supplements for treatment of oligemic shock. Arch Surg. 1979;114(4):379-83. [abstract]
14. Cochrane Injuries Group Albumin Reviewers. Human albumin administration in critically ill patients: systematic review of randomised controlled trials. BMJ. 1998;317(7153):235-40 [abstract] [PedsCCM EB Journal Club Review]

---

Back to the EB Journal Club Index