Volume 53, Issue 8 p. 755-761
Free Access

Serum albumin and colloid osmotic pressure in survivors and nonsurvivors of prolonged critical illness

M. C. Blunt

M. C. Blunt

John Farman ICU, Box 17, Addenbrooke's Hospital, Hills Rd, Cambridge CB2 2QQ, UK

Search for more papers by this author
J. P. Nicholson

J. P. Nicholson

John Farman ICU, Box 17, Addenbrooke's Hospital, Hills Rd, Cambridge CB2 2QQ, UK

Search for more papers by this author
G. R. Park

G. R. Park

John Farman ICU, Box 17, Addenbrooke's Hospital, Hills Rd, Cambridge CB2 2QQ, UK

Search for more papers by this author
First published: 06 April 2002
Citations: 60
G. R. Park John Farman ICU, Box 17, Addenbrooke's Hospital, Hills Rd, Cambridge CB2 2QQ, UK


We retrospectively compared the changes in serum albumin concentration and colloid osmotic pressure between survivors and nonsurvivors of prolonged (≥7 days) critical illness over a 2-year period from 1 July 1995. All patients had serum albumin measured daily, and colloid osmotic pressure measured 5 days a week, throughout their ICU admission. They received crystalloid and colloid infusions as well as parenteral or enteral feeding. Infusions of albumin were not used to treat hypoalbuminaemia. One hundred and forty-five patients were included, 66 nonsurvivors and 79 survivors. Nonsurvivors were significantly older than survivors [mean (95% CI): 58 (3.8) and 49 (4.1) years, respectively] and had a greater risk of death [mean (95% CI): 0.44 (0.06) and 0.28 (0.05); p < 0.05]. There was no significant difference in gender, APACHE II score [mean (95% CI): 22 (2.7) (nonsurvivors); 18 (2.3) (survivors)] or length of stay [median (interquartile range): 14 (9–27) days (nonsurvivors); 15 (9–26) days (survivors)]. There was no difference between the two groups in the absolute minimum serum albumin concentrations reached, the time to reach that minimum or the minimum in the first 7 days. However, nonsurvivors had a significantly lower mean serum albumin concentration: [mean (95% CI): 15.7 (5.1) gl−1 compared with 18.3 (4.6) gl−1 in survivors; p < 0.05]. They also had a lower recovery mean (the weighted mean after the minimum value): [mean (95% CI): 13.3 (5.1) gl−1 (nonsurvivors) and 18.6 (5.3) gl−1 (survivors); p < 0.01]. Analysis of colloid osmotic pressure results showed no difference between the groups in mean, minimum or recovery mean. Regression analysis of mean colloid osmotic pressure and albumin revealed that albumin only contributed 17% of the colloid osmotic pressure in these patients. The similar decrease in albumin in nonsurvivors and survivors may reflect the acute inflammatory response and/or haemodilution. However, survivors showed an ability to increase serum albumin concentrations, possibly owing to resumption of synthesis. The colloid osmotic pressure varied little between or within either group of patients, possibly because of the use of artificial colloids. There was no relationship between death and colloid osmotic pressure.

Albumin has important physiological functions. In healthy humans, it is responsible for up to 80% of the intravascular colloid osmotic pressure (COP) [1–4]. It binds and transports a wide range of endogenous and exogenous substances, possibly influencing the metabolism and clearance of many drugs [5]. Albumin can inactivate certain drugs like disulfiram and it is involved in the metabolism of endogenous substances such as lipids and eicosanoids [6]. Furthermore, it may have protective functions such as the scavenging of oxygen-free radicals and the sequestering of exogenous toxins [6, 7].

The prognostic value of serum albumin has been reported in various contexts. Plasma albumin concentration decreases with age and this may be independent of underlying disease processes [8, 9]. A lower serum albumin concentration at a given age predicts an increased risk of dying within 3 years [9]. Further associations between a low serum albumin and mortality from various causes has been shown [10]. Hypoalbuminaemia in hospitalised patients is correlated with an increase in length of hospital stay, morbidity and mortality [11]. In critically ill patients a decrease in serum albumin is associated with increased morbidity and mortality [12–16]. A recent study of critically ill patients found that the serum albumin on admission to the ICU was an insensitive prognostic indicator, but when measured 24–48 h after admission a low serum albumin concentration was as reliable as the APACHE II score in predicting outcome [17].

Albumin is a negative acute-phase protein. Thus, in severe illness the plasma concentration decreases. It does not increase until the recovery phase of the illness. The longer the illness the greater the risk of death. We therefore hypothesised that either the duration of the reduction in serum albumin in critically ill patients or the absolute minimum value may be a sensitive indicator of outcome. We have retrospectively examined data in patients whose ICU stay was 7 days or more to determine whether a prolonged reduction in serum albumin was associated with nonsurvival. Since COP is measured routinely in all patients on this unit, we also decided to examine the relationship between albumin and COP in this group of patients.


Patients admitted to the general adult ICU on or after 1 July 1995, until 1 July 1997, and who stayed in ICU for 7 days or more were studied. Patients were not included if they were still inpatients on 25 July 1997.

The ICU database provided demographic (age, sex, APACHE II score, categorised admission diagnosis) and outcome data. From the latter the patients were categorised into survivors (discharged from this hospital alive) and nonsurvivors (died either in ICU or on the general ward).

Serum albumin was measured daily and colloid osmotic pressure every weekday throughout ICU admission on all patients. Serum albumin was assayed using an automated bromocresol purple (BCP) specific dye-binding method. This method is a more sensitive measure of low albumin concentrations than the older bromocresol green (BCG) method. The BCG method tended to overestimate serum albumin concentrations due to a reaction with other serum proteins, including positive acute phase proteins which are increased in critical illness [18]. COP was measured using a Wescor model 4420 Colloid Osmometer, which employs a standard membrane with a cut-off of 30 000 MW. It is calibrated using a filtered bovine serum COP standard solution, Osmocoll® II [19]. Calibrations were performed according to the manufacturer's guidelines.

During their ICU stay all patients received a combination of crystalloid and colloid infusions as well as parenteral or enteral feeding. The clinicians were aware of the albumin concentrations of all patients. A strict policy of not using albumin in critically ill patients exists in our unit. However, in total 12 patients were given human albumin solution (HAS) during their time on the ICU. Eight were hepatology patients, two were from nephrology, one had suffered a burn and one had an abdominal aortic aneurysm repaired. The total amount of 20% HAS used was 1.6 l (range 0–0.6 l) and of 4.5% HAS was 14.5 l (range 0–4 l). Eight of the patients given HAS died. Albumin was present in blood products used for the correction of specific haematological abormalities.

Statistical analysis was performed using Excel 7 (Microsoft) and Stat-100 (Biosoft, Cambridge, UK). The demographic data were compared using unpaired Student's t-tests with Bonferroni's correction (age and APACHE II score), and Chi-squared tests (for sex and diagnostic category).

The albumin and colloid osmotic pressure data gave a series of observations for each patient. As the individual observations recorded for each patient were not independent over time, analysis was performed by summarising the data from each individual patient [20].

Before the study we anticipated that mean serum albumin concentrations would decrease to a minimum and then increase as recovery progressed. Measuring this increase was difficult because the values fluctuated and assays were not always performed at consistent time intervals. Therefore, we calculated the area under the curve (AUC) of the observations after the minimum had been reached until that patient's discharge from the ICU. Dividing the AUC by the time course over which it was measured gave a weighted mean [21], which we have called the recovery mean (Rmean). For each patient we calculated the Rmean from their minimum albumin value until their discharge (Rmean.(min − disch)), and the Rmean from their minimum within the first 7 days until discharge (Rmean.(7 − disch)). The reason for this was that the minimum (and hence the Rmean.(min − disch)) could only be ascertained retrospectively. Some patients will have reached their minimum albumin concentrations towards the end of their ICU stay. In order to obtain a possible prognostic tool, we wanted to examine the trends in early critical illness, compare them with overall trends in this group of patients and use the data to test for an Rmean that may have prognostic value. For each patient we also determined the maximum serum albumin concentration achieved after the minimum, which we have called the recovery maximum (Rmax).

We also recorded the mean, the minimum, the time to reach minimum (tmin) and the minimum value for the first 7 days of ICU admission (min0–7). The derived values for the survivors and the nonsurvivors were then compared using independent two-tailed Student's t-tests for parametric data and the Mann–Whitney U-test for nonparametric data. Bonferroni's correction was applied for multiple testing. A representative set of values for one patient is shown in Fig. 1.

Details are in the caption following the image

Sample data from one patient (nonsurvivor) to illustrate the summary data. The AUC after the minimum is shaded dark grey. This is divided by its duration to provide the Rmean.(min − disch). The AUC after the minimum albumin concentration in the first 7 days is the sum of the two shaded areas. This is divided by its duration to provide the Rmean (7 − disch).

Albumin and colloid osmotic pressure measurements from the same samples were compared to assess the association between the two. Linear regression analysis was performed on this comparison.

An alpha error of less than 0.05 was considered statistically significant throughout.


In total there were 976 admissions in the study period, of which 145 patients met the entry criteria. There were no significant differences in sex or APACHE II status between the two groups, but there were significant differences in age, admission diagnostic category (derived from the groups used for risk of death assessment) and the risk of death calculated according to APACHE II score and admission diagnosis (Table 1).

Table 1. Patient details and admission diagnosis in 145 patients treated in the intensive care unit between July 1995 and July 1997. Values are mean (95% CI) for age, sex and APACHE II score; median (interquartile range) for length of stay; and whole numbers. * p < 0.05. Chi-squared testing showed a difference in the distribution of admission diagnosis between survivors and nonsurvivors (p < 0.05).

The albumin concentrations since ICU admission are summarised in Table 2. There was no difference found in the minimum (min), the time taken to decrease to that minimum (tmin) or the minimum in the first week (min0–7). Eleven patients in the nonsurvivor group had minimum albumin concentrations just before death. However, the impact of this on the min and Rmean.(min − disch) appears to be negligible as the results for these indices are mimicked by the min0–7 and Rmean.(7 − disch). There was a significant difference between the two groups in the mean (p < 0.05), the Rmax (p < 0.001) as well as the Rmean.(min − disch) and Rmean.(7 − disch) (p < 0.001).

Table 2. Serum albumin concentrations (g.l−1) and colloid osmotic pressure (COP) (mmHg) in 145 patients treated in the intensive care unit. Values are mean (95% CI) except for tmin reported as median (interquartile range). min = minimum serum albumin value; min0–7= minimum value in first week; tmin= time taken to reach minimum albumin value; Rmean.(min − disch)= recovery mean from minimum albumin value until discharge; Rmean.(7 − disch)= recovery mean from minimum albumin value in first 7 days until discharge; Rmax= maximum serum albumin concentration achieved after the minimum value. * p < 0.05; ** p < 0.001.

Figure 2 shows the mean daily albumin concentration over time from ICU admission. It is important to point out that the numbers of patients contributing to the daily mean albumin concentrations decreased with time. Thus the significance of the results changes with the length of ICU stay. After 90 days there was a very small number of patients in each group.

Details are in the caption following the image

Mean daily albumin concentrations for survivors and nonsurvivors. Values are mean (95% confidence intervals). The numbers in rectangles along the x-axis refer to the mean number of patients in each group (S:NS) in the 10-day period shown.

Analysis of the colloid osmotic pressure measurements revealed no significant differences in the values (Table 2 and Fig. 3[link]). Albumin made only a minimal contribution to the colloid osmotic pressure. Linear regression of the measurements lead to the equation

Details are in the caption following the image

Comparison of mean daily colloid osmotic pressure values between survivors and nonsurvivors (error bars — 95% CI).

inline image

The small value of the albumin coefficient illustrates the minimal effect albumin has on COP in these patients.


This study confirmed our clinical suspicion that prolonged hypoalbuminaemia in critically ill patients is associated with a poor outcome. Survivors had significantly higher overall mean serum albumin concentrations and were able to recover to a higher mean serum albumin than nonsurvivors.

There were no significant differences between the two groups in the absolute minimum values, the minimum in the first week or the time to reach the minimum. Serum albumin concentrations decrease during critical illness for several reasons. Increased capillary permeability causes a redistribution of albumin from intravascular to extravascular compartments [22]. This is cytokine-mediated as part of a systemic acute-phase response [6, 23]. Haemodilution with resuscitation fluids will reduce serum albumin [12]. There may also be loss of albumin in haemorrhage, burns and into large surgical wounds [5]. Redistribution and dilution play by far the most important part in early stages of critical illness.

Changes in the synthesis and catabolism of albumin during illness will reduce serum albumin, but not acutely. Synthesis has been shown to decrease in animal studies of injury and inflammation [24–26]. Interleukin-6 and TNF-α both depress albumin gene transcription, and thus albumin synthesis [27, 28]. Human in vitro studies confirm the depression of albumin synthesis as an acute-phase response [27]. The in vivo situation, however, is not simple. In patients with septic shock, rates of albumin synthesis can vary enormously, from very low to significantly increased [29].

Albumin catabolism is enhanced by the increased corticosteroid concentration during the stress response, which leads to a general increase in protein breakdown [6]. Studies have shown that the rate of degradation seems to decrease with the plasma concentration [6]. However, while the absolute rate of degradation may be reduced in response to a low plasma albumin concentration, the fractional degradation rate (the percentage of the total exchangeable albumin pool, intravascular and extravascular, broken down each day) may be normal or even increased [30].

Serum albumin had no prognostic value in our patients early in the course of critical illness. This contrasts with a recent study by McCluskey et al. [17], who showed that nonsurvivors had lower albumin concentrations on admission to ICU and that their albumin concentrations decreased more rapidly in the first 24–48 h. The admission albumin was not a sensitive indicator of outcome, but the value at 24–48 h was as accurate as APACHE II in predicting mortality. We did not confirm this in our group of patients. While our nonsurvivors were older and had a higher risk of death, they did not stay on the ICU longer than survivors and their APACHE II scores were not significantly different. Our findings may be related to the fact that we selected patients with prolonged critical illness.

The value of serum albumin as a prognostic indicator has been studied elsewhere. In patients with renal failure on dialysis [31], in stroke rehabilitation patients [32] and in critical illness [13, 16], low serum albumin is associated with increased length of stay, higher complication rates and higher mortality. Large community-based studies have shown a link between low serum albumin and higher morbidity and mortality [9, 10]. Reinhardt et al. [11] found that in hospitalised patients a serum albumin concentration of less than 3.4 g.dl−1 was associated with a 30-day mortality rate of 24.6%. This increased to 62% if the serum albumin was 2.0 g.dl−1 or less.

Data in patients with critical illness are conflicting. The retrospective study of McCluskey et al. [17] has been mentioned above. Bradley et al. [14] studied serum albumin concentrations in 44 critically ill surgical patients. Admission concentrations were low in both groups and could not be used as a prognostic indicator. Repeat measurements before discharge from the ICU or before death still showed that both groups had serum albumin concentrations lower than normal. Significantly, nonsurvivors had lower mean serum albumin concentrations than survivors, but for individual patients the difference was too small to be of any prognostic value. This contrasts with the findings of Ching et al. [33], who showed in a small study of 34 critically ill surgical patients on equivalent nutritional support, survivors were able to increase their serum albumin to normal while nonsurvivors could not.

It is unknown if the degree of hypoalbuminaemia correlates with the degree of critical illness. A greater inflammatory response would cause more distribution of albumin to the extravascular compartment and more extensive fluid resuscitation would be necessary in the acute phase of critical illness. Ongoing inflammation may cause a reduction in synthesis and an increase in catabolism of albumin [5, 23]. A patient's ability to recover to a higher albumin level could then conceivably be an indicator of recovery from systemic insult and thus be used to predict long-term outcome. Sapijaszco et al. [15] have examined the role of serum albumin in predicting successful weaning of ventilated patients and concluded that when measured daily, the trend of albumin concentrations could be used to predict outcome, though mortality was not looked at. We have shown a highly significant difference in the recovery means, both Rmean.(min − disch) and Rmean.(7 − disch) between survivors and nonsurvivors.

The role of albumin in the maintenance of COP in healthy subjects is significant. In critical illness, however, the relationship may be altered. A low COP has previously been shown to be associated with increased morbidity and mortality [34, 35] and a level of 15 mmHg is associated with a 50% survival rate [36]. Our patients had mean minimum values of greater than 15 mmHg. We have confirmed that albumin contributed very little to the COP in this group of patients. If all mean COPs and serum albumin concentrations are examined, albumin made up only 17% of the colloid osmotic pressure. In the acute stages of critical illness, our patients were resuscitated with plasma expanders, such as gelatin (Gelofusine) and starch (Haes-steril), both of which effectively maintain COP and these would have formed the major part of the COP. Since there was no significant difference between the COPs of survivors and nonsurvivors, we can assume that there is no justification in artificially increasing plasma albumin merely to improve the COP. A growing body of evidence shows that there is no benefit in using albumin to correct hypoalbuminaemia [3, 35, 37–39].

We have shown that all our patients with prolonged critical illness had a similar minimum albumin concentration, but that nonsurvivors were less able to recover from this minimum than survivors. The inability of nonsurvivors to recover from a low serum albumin may signify an ongoing inflammatory reaction, leading to decreased albumin synthesis and increased breakdown. We found no difference between the colloid osmotic pressures of the two groups and an insignificant association between serum albumin and COP.


Calculation of weighted mean from AUC

The total AUC of variable y is calculated by adding the AUC between each pair of consecutive measurements. This may be calculated easily as it is a trapezium and accordingly has an area equal to (y1 + y2)(t2 − t1)/2. Thus the total AUC for n + 1 measurements recorded for times ti (where i = 0, … , n) is the sum and the weighted mean may be determined by dividing this by the time course (tn − t0), thus:

inline image

Although this appears complicated it can be calculated using a Visual Basic macro within Excel that iterates through the values in two ranges (time and observation), performs the individual area calculations, compounds the result then divides by the time span.


We thank Mr Graham Lawrence, Department of Pathology, Addenbrooke's Hospital, for kindly supplying us with the information from the biochemistry and haematology databases.