Albumin or plasma protein fraction

Of all intravenous fluids, albumin has received the most attention, with over 50 randomised controlled trials published as of December 2000. Six meta-analyses have reviewed the evidence from randomised controlled trials of albumin use in fluid resuscitation with regard to all cause mortality15-19 or postoperative bleeding.20 The first systematic review on albumin was published concurrently in the British Medical Journal16 and the Cochrane Database of Systematic Review. The Cochrane review was subsequently updated.18 Three systematic reviews compared albumin with crystalloids;15,17,18 two compared albumin with other colloids.19,20 Table 7.2 summarises the evidence on albumin.

Wilkes and Navickis identified 42 randomised controlled trials that compared albumin with isotonic or hypertonic crystalloids with no albumin or lower doses of albumin with regard to all cause

Table 7.2 Meta-analyses of randomised controlled trials comparing albumin or plasma protein fraction with crystalloids or other colloids with all cause mortality or postoperative bleeding as an outcome

No of patients Effect size

Reference Control group Population (studies; years) (95% CI)*

Mortality

Table 7.2 Meta-analyses of randomised controlled trials comparing albumin or plasma protein fraction with crystalloids or other colloids with all cause mortality or postoperative bleeding as an outcome

No of patients Effect size

Reference Control group Population (studies; years) (95% CI)*

Mortality

Wilkes ef al. 200117

Crystalloids with no or

All patients

2958 (42; 1962-2000)

RR

1-11 (0-95, 1-28)

lower albumin dose

Surgery or trauma

1339 (20; 1977-2000)

RR

1-12 (0-85, 1-46)

Burns

197 (4; 1975-95)

RR

1-76 (0-97, 3-17)

Hypoalbuminaemia

357 (4; 1988-97)

RR

1-59 (0-91, 2-78)

High-risk neonates

304 (6; 1973-99)

RR

1-19 (0-78, 1-81)

Ascites

373 (4; 1962-99)

RR

0-93 (0-67, 1-28)

Alderson et al. 200215

Isotonic crystalloids

All critically ill

641 (18; 1977-97)

RR

1-52 (1-08, 2-13)

Alderson et al. 200218

Isotonic crystalloids or

All critically ill

1519 (31; 1973-97)

RR

1-52 (1-17, 1-99)

no albumin

Hypovolemia

719 (18; 1977-97)

RR

1-46 (0-97, 2-22)

Burns

163 (3; 1978-95)

RR

2-40 (1-11, 5-19)

Hypoalbuminaemia

637 (10; 1973-97)

RR

1-38 (0-94, 2-03)

Bunn et al. 200219

Hydroxyethylstarch

All critically ill

1029 (20; 1982-98)

RR

1-17 (0-91, 1-50)

Bunn et al. 200219

Gelatin

All critically ill

542 (4; 1987-96)

RR

0-99 (0-69, 1-42)

Hydroxyethylstarch

Cardiopulmonary bypass 653 (16; 1982-98)

Postoperative bleeding Wilkes et at. 200120

Hydroxyethylstarch

Cardiopulmonary bypass 653 (16; 1982-98)

CI = confidence interval; RR = relative risk; SMD = standardised mean difference

"For relative risks, RR < 1 favours the albumin group; RR > 1 favours the control group. For standardised mean differences, SMD < 0 favours the albumin group; SMD > 0 favours the control group. Results are statistically significant (P < 0-05) if the 95% CI do not include 1 for relative risks or 0 for standardised mean differences mortality.17 From their meta-analysis, they concluded that there was no statistically significant difference in mortality between the two groups. However, the point estimates for all study populations, with the exception of ascites, favoured the crystalloid group.

By contrast, the Cochrane Injuries Group Albumin Reviewers completed two systematic reviews comparing albumin to crystalloids.15,18 In their first meta-analysis, Alderson et al. pooled the results from 18 randomised controlled trials that compared albumin with isotonic crystalloids for fluid resuscitation in critically ill patients. They found a statistically significant increase in mortality in the albumin group (RR 152; 95% CI 108-213). These results were confirmed in their second meta-analysis, which pooled results from 31 randomised controlled trials that compared albumin with isotonic crystalloids, no albumin, or lower doses of albumin. The relative risk for mortality remained unchanged and the 95% CI narrowed (117-199).

Alderson et al. also found one trial that compared albuminhypertonic crystalloid with isotonic crystalloid.15 The trial enrolled only 14 patients and did not find any difference in mortality (050; 0-06-4-33).25

The results of the three meta-analyses have generated much discussion and some controversy. Alderson and colleagues concluded that the "use of human albumin in critically ill patients should be urgently reviewed and that it should not be used outside the context of a rigorously conducted randomised controlled trial".18 By contrast, Wilkes and Navickis stated that their "findings should allay concerns about the safety of albumin".17 The discrepancy in conclusions relate, in part, to the differences in selection criteria for the non-albumin comparative group. Wilkes and Navickis included studies with hypertonic crystalloids; Alderson and colleagues did not. In spite of this difference, concerns about the safety of albumin cannot be dismissed. In all three meta-analyses, the point estimates have favoured the non-albumin group. At this time, one can conclude that there is no benefit to mortality with the use of albumin compared with crystalloids in fluid resuscitation. Given the possibility of discrepant results between meta-analyses of small studies (as in this case) and large randomised controlled trials, whether albumin increases mortality or not remains to be determined.

The efficacy of albumin compared with other colloids has been reviewed in two meta-analyses of randomised controlled trials (Table 7.2). Bunn et al. identified 25 studies that compared albumin to hydroxyethylstarch (20 studies), dextrans (three studies), or gelatins (four studies).19 There was no statistically significant difference between albumin and hydroxyethylstarch with regard to mortality, but the point estimate favoured the latter (117; 0-91-1-50).19 No deaths were observed in the three studies that compared albumin with dextrans. Three of the four studies that compared albumin with gelatins did not find any deaths. In the one study with deaths, albumin was not significantly better than gelatins in reducing mortality in fluid resuscitation, although the point estimate favoured albumin (0 99; 0-69-1-42).26

Wilkes et al. compared albumin with hydroxyethylstarch with regard to the amount of postoperative bleeding in patients undergoing cardiopulmonary bypass (Table 7.2).20 They identified 16 randomised controlled trials and found a statistically significant reduction in postoperative bleeding with albumin (standardised mean difference 0 24; 0 40-0 08).20 The clinical significance is unclear. In adults (14 trials), the pooled mean blood loss was 693 ± 350 ml in the albumin group compared with 789 ± 487 ml in the hydroxyethylstarch group.20 No differences between the two colloids were seen in terms of the duration of postoperative ventilatory support or intensive care unit stay, although the standardised mean differences favoured the albumin group in both outcomes.20

Was this article helpful?

0 0

Post a comment