0148-6071/80/010~-0150$02.
00/0
JOURNAL OF PARESTERAL AND ENTERAL NUTRITION
Copyright 0 I980 by the American Society of Parenteral and Enteral Nutrition
Clinical Research
Serum Albumin and Nutritional Status
R. ARMOUR
FORSE,M.D. AND
From the Department of Surgev, Royal Victoria Hospital and hlcGiIl University, hlontreal, Quebec, Canada
ABSTRACT. Serum albumin concentration is frequently used
to define nutritional status. To validate this relationship, 161
body composition studies were performed on 102 patients si-
multaneously with protein electrophoresis. The body cell mass
represented by the exchangeable potassium to total body water
ratio correlated significantly (p < 0.001) with the serum albumin
concentration (r = 0.59) and significantly (p < 0.001) to total
protein (r = 0.59). However, in both cases the 95%confidence
limits about the regression were wide. In 24 of 54 patients (44%)
with a normal nutritional state, as defined by body composition,
A knowledge of a patient's nutritional status is impor-
tant in both the application and the evaluation of nutri-
tional support, especially in evaluating the efficacy of a
nutritional support system. The earliest attempts a t eval-
uating nutritional support in a clinical setting involved
the measurement of morbidity and mortality rates.' How-
ever, many factors are responsible for the mortality and
morbidity observed in any group of patients. As a result,
the variance associated with this measurement is large.
To overcome this difficulty, nitrogen (N) balance mea-
surements are usually obtained to document the effects
of nutritional support? but such measurements are dif-
ficult to perform and are associated with a large experi-
mental error? In addition, N balance is a measure of the
effect of nutritional support and is not a measure of a
patient's nutritional state.
An association between a low serum albumin concen-
.tration and protein-calorie malnutrition has long been
suspected. A low serum albumin concentration is a com-
mon finding in protein calorie maln~trition;~ as a result,
it is commonly employed as a measure of an individual's
nutritional state. The present study was undertaken to
experimentally determine the validity of this measure-
ment.
MATERIALS AND METHODS
Data were collected in 102 patients on 161 occasions.
The majority were malnourished and the measurements
were performed just prior to or during'a course of total
p a r e n t e d nutrition (TPN). In addition, measurements
were obtained in a group of morbidly obese patients a t
Received for publication, March 29, 1980
Accepted for publication, June 2, 1980
Downloaded from pen.sagepub.com at UNIV CALIFORNIA SAN DIEGO on July 16, 2016
Val. 4, No.5
Printed in U S A .
HARRY
M. SIIIZCAL,
M.D., F.A.C.S., F.R.C.S.(C)
the serum albumin was abnormal. In 12 of 107 (11.2%)patients
with malnutrition, the serum albumin was normal. Patients
with more than one study were divided into 3 groups depending
on the changes in their nutritional state as d e f i e d by their
body composition. Serum albumin did not consistently reflect
the significant body compositional changes observed. The data
indicate that serum albumin is a valid measure of nutritional
state for epidemiological surveys. However, due to the low
sensitivity and specificity it is a poor parameter for evaluating
the individual patient's nutritional state.
varying time intervals following jejuno-ileal bypass sur-
gery performed for weight reduction. Protein malnutri-
tion was present in 25%of these patients?
Nutritional status was evaluated quantitatively by
measuring the various components of body composition.
A multiple isotope dilution technique was used to deter-
mine the body composition as previously described,6 in
which patients are injected with 500pCi of tritiated water
and 8 pCi of "Na. Blood samples are drawn before and
a t 4 and 24 hr after injection. Total exchangeable sodium
(Na,) was calculated from the 22Naspecific activity a t 24
hr. Plasma tritium concentration a t 24 hr was used to
calculate the total body water (TBW). An indirect
method was used to calculate exchangeable potassium
(K,) from TBW, Na, and the ratio in whole blood of the
sodium plus potassium content divided by the water
content? The validity of this indirect technique was
established experimentally in both experimental animals
and man.' Additional parameters of body composition
were obtained by the following calculations:
LBM = TBW/0.73
BF = Body weight - LBM
BCM = 0.00833 Ke
ECM = LBM - BCM
where LBM = lean body mass, BF = body fat, BCM =
body cell mass, and ECM = extracellular mass.
At the time of each body composition measurement, a
whole blood sample was obtained for hematocrit, hemo-
globin, and serum protein concentrations. The total se-
rum proteins were fractionated by standard serum elec-
trophoresis into the following components: albumin, al-
pha-l, alpha-2, p- and y-globulins.
The various serum protein concentrations were corre-
lated with the nutritional state as quantitatively assessed
45(1
September/October 1980 ALBUMIN AND NUTRITIONAL STATUS 451
by body composition measurements. Standard linear SERUM R L B U M I N
regression analysis was performed? A Student t test was TOTRL O R O U P
GT 4
performed to determine the significance of the two sets
of measurements, which were performed twice in the I . .:\
same individual and separated by a short interval of time.
A p < 0.05 was taken to indicate statistical significance;
all means were listed with the standard error of the mean
(SEM).
RESULTS a
m
The body cell mass (BCM) is the total mass of living,
functioning, oxygen-consuming cells of the body. Total
body potassium, which is equivalent to total K,, is lin-
.I
early related to the BCM. As a result, the size of the
BCM can be estimated from K,. The evidence which
supports the use of K, to measure the BCM has been
reviewed el~ewhere.~ In normal healthy individuals, K,
varies directly with body size? T o correct body compo- NRE/KE
sitional data for body size, data are expressed as a func- FIG. 2. Serum albumin concentration correlated with the NaJK,
tion of TBW, ie, KJTBW. It is better to use TBW for ratio; (p < 0.001),with a correlation coefficient of r = -0.47.
this, rather than body weight, since TBW is linearly
related to the lean body mass, and the resulting param- TABLE I
eter is, therefore, independent of the degree of adiposity. Correlation with bodv cell mass (K./TB It9
The serum albumin concentration was correlated with hlidpoint 95% Confidence limits
Serum protein coefficients of about the midpoint
the [BCM as estimated by the Ke/TBW (Fig. 1). The regression of regression
normal range for K,/TBW was determined in 25 normal
&
volunteers: in whom the mean K,/TBW = 80.0 +- 1.0, Total protein 0.591" 6.48 4.61-8.36
with 95% confidence limits of 69.9 to 90.1 mEq/L. In the Albumin 0.585" 3.07 1.74-4.40
present group of 161 determinations, the mean K,/TBW Alpha-1 0.025 0.36 0.16-0.56
was 62.5 mEq/L, with a range of 36.7 to 88.2 mEq/L. Alpha-2 0.241" 0.77 0.39-1.15
Therefore, a large number of individuals were malnour- fl-Protein 0.436" 0.82 0.38-1.25
y-Globulins 0.131 1.51 0.46-2.56
ished, with a depleted BCM at the time these measure- Hemoglobin 0.432" 11.74 8.26-15.23
ments were performed. Hematocrit 0.328" 35.04 24.17-45.90
The regression between serum albumin and the BCM p < 0.05.
was statistically significant (p < 0.001), with a correlation
coefficient (r) of 0.59. However, there is considerable
scatter of the data about the regression resulting in wide range for serum albumin, as performed by serum electro-
95% confidence limits (Fig. 1). Thus, for a normal K,/ phoresis, is 3.5 to 5.0 g%.
TBW of 80.0 mEq/L, the regression line indicates an Malnutrition results in a loss of body cell accompanied
albumin concentration of 3.72 g%, with 95% confidence by an expansion of the .extracellular mass: which rep-
limits of 2.4 to 5.1 g%. In our institution, the normal resents the supporting component of body composition
and is estimated by measuring total exchangeable sodium
(Na,). As a result, the Na,/K, ratio is a sensitive index of
the nutritional state, since it is a measure of the extra-
cellular mass expressed as a function of the BCM. In the
normal population, this ratio is close to unity with a
small variance. In 25 normal volunteers the mean NaJ
K, = 0.98 f 0.2, with 95% confidence limits of 0.74 to
1.22;6with malnutrition, this ratio increases significantly.
There was a significant (p < 0.001) correlation between
plasma serum albumin concentration and the Na,/K,
ratio (Fig. 2); the correlation coefficient was -0.47. Again,
there was considerable scatter in the data about the
regression, resulting in wide 95% confidence limits. With
a normal Na,/K, of 0.98, the regression indicated a serum
albumin of 3.39, with 95% Confidence of 1.9 to 4.8 g%.
A similar correlation was performed between the BCM,
as estimated by K,/TBW, and the various serum protein
m
fractions and the hemoglobin concentration and hema-
a. 20. 40. 60. 88. 180. tocrit (Table I). The .midpoint of the regression and the
KWTRU rMEa/Li
FIG. 1. Serum albumin concentration correlated with the BCM as 95% confidence limits about this midpoint are included
estimated by the K,/TBW in 161 studies; (p < 0.001) with a correlation in Table I; in each instance the 95% confidence limits
coefficient of r = 0.59. were wide.

Downloaded from pen.sagepub.com at UNIV CALIFORNIA SAN DIEGO on July 16, 2016
452 FORSE AND SHIZGAL Val. 4, No. 5
TABLE I1
The response to changes in nirtritional state
Normal hlalnourkhed hfalnourished
deteriorated improved deteriorated

Initial Final Initial Final Initial Final

Weight (kg) 76.0 2 12.8 71.1 f 9.2, 62.2 f 3.4 62.6 f 3.1 64.4 f 4.6 67.7 f 4.6
Body fat (kg) 27.2 tfr 8.4 25.5 f 6.2 17.2 f 2.6 17.9 f 2.2 16.3 f 3.0 16.9 f 3.2
Lean body mass (kg) 48.8 tfr 5.1 45.6 f 3.8 44.9 tfr 1.8 44.7 f 1.7 48.1 f 3.6 50.8 F 2.8
Body cell mass (kg) 22.0 2 2.5 17.6 f 1.7" 14.4 tfr 0.8 17.7 f 0.9" 18.5 f 1.2 17.2 k 1.2
Extracellular mass (kg) 26.9 f 2.7 28.0 & 2.4 30.6 f 1.5 27.0 f 1.1" 29.6 f 2.5 33.6 f 2.2

KJTBW mEq/L 80.3 & 0.9 86.5 f 1.7" 95.0 f 1.9 86.7 f 1.7" 88.9 f 1.7 95.8 f 2.5"
NaJTBW mEq/L 73.4 tfr 1.0 63.6 f 2.2" 53.0 f 1.9 64.9 f 1.7" 64.2 f 1.6 56.1 f 2.9"
N&/K 1.10 tfr 0.02 1.39 f 0.08" 1.96 f 0.16 1.39 f 0.07" 1.40 f 0.05 1.59 tfr 0.16"

Albumin (g%) 3.0 tfr 0.2 3.0 f 0.2 2.6 f 0.1 2.9 f 0.1" 2.8 f 0.1 2.9 f 0.2
Alpha-1 (gW) 0.40 20.03 0.37 f 0.02 0.37 f 0.02 0.39 -C 0.02 0.38 f 0.04 0.33 f 0.02
Alpha-2 (gB) 0.85 & 0.06 0.77 f 0.07 0.77 f 0.04 0.84 f 0.03" 0.89 f 0.07 0.74 f 0.06"
Beta (g%) 0.86 tfr 0.05 0.80 f 0.06 0.74 f 0.05 0.87 f 0.03" 0.85 f 0.06 0.90 f 0.11
Gamma (g%) *
1.78 0.21 1.60 & 0.23 1.45 f 0.10 1.72 f 0.10" 1.81 f 0.25 1.58 f 0.18

Total protein 6.9 tfr 0.30 6.5 f 0.34 6.0 f 0.20 6.8 f 0.14" 6.8 f 0.3 6.4 f 0.3

Hemoglobin 11.7 2 0.4 12.2 f 0.5 11.2 f 0.4 11.5 f 0.2 12.2 f 0.5 11.4 f 0.8
Hematocrit 35.5 2 1.1 37.1 f 1.1 33.7 f 1.0 34.7 f 0.7 36.8 f 1.2 33.4 f 1.6

Patients 12 27 10
Studies 25 66 24
Time (days) 24 f 4.9 20 f 3.0 31.3 f 12.1
a p < 0.05 significantly different by paired Student t test.

NUTRITIONALLY DETERIORATED NUTRITIONALLY IMPROVED

BODY COMPOSITION ALBUM I N BODY COMPOSITION RLBUM I N

FIG.3. Body composition and serum albumin concentration in 27 FIG.4. Body composition and serum albumin concentration in 12
patients who were malnourished initially, and experienced a significant patients with an initial normal body composition who developed mal-
improvement in their nutritional state. The clear and hatched histo- nutrition. The clear and hatched histograms refer to the initial and
grams refer to the initial and final measurements, respectively. fmal measurements, respectively.

On the basis of their body composition, patients were
divided into normal and malnourished groups. Malnutri-
tion was defined by an NaJK, ratio that exceeded 1.22.
The upper 95% confidence limit of this ratio in our group
of normal volunteers was 1.22. PreGous work in our
laboratory has validated the use of this ratio to define
the presence of malnutrition." Fifty-four patients were
nutritionally normal with an NaJK, 4.1.22; of these, 24
(44%) had an abnormal serum albumin. Malnutrition was
present in 107 instances, in which the serum albumin was
normal in 12 (11.2%).
In 32 individuals, measurements were performed more
than once, separated by a significant time interval. This
permitted a comparison of changes in nutritional state
with changes in serum protein concentrations. T h e pa-
tients were divided into three groups according to the
observed change in nutritional state (Table 11).
The nutritional state deteriorated in 12, with a signifi-
cant decrease in the BCM from 22.0 f 2.5 to 17.6 f 1.7
kg which was also reflected by a significant decrease in
K,/TBW (p < 0.01). This was accompanied by an expan-
sion of the extracellular mass. The significant increase in
NaJK, from 1.10 k 0.02 to 1.39 0.08 (p < 0.01) reflected
this deterioration. In spite of these significant changes in
body composition, the serum total protein did not change
significantly and albumin remained unchanged at 3.0 A
0.2 (Fig. 3).
-In the second group of 27 patients there was a signifi-

Downloaded from pen.sagepub.com at UNIV CALIFORNIA SAN DIEGO on July 16, 2016
September/October 1980 ALBUMIN AND NUTRITIONAL STATUS 453
cant improvement in body composition, as indicated by regression are large. These data, therefore, indicate that
a 22.9% increase in the BCM and a 22.5% increase in Ke/ serum albumin is useful in epidemiological surveys to
TBW. There was a concomitant contraction of the extra- evaluate a population's nutritional state. The data also
cellular mass. This improvement in nutritional state was support previous observations of a low serum albumin in
reflected by a decrease in Nae/Ke from 1.96 k 0.16 to 1.39 the presence of malnutrition. The presence of large 95%
f 0.07 (p < 0.001). The serum albumin increased signifi- confidence limits indicates that serum protein and albu-
cantly (p < 0.001) by 11.5% from 2.6 f0.1 to 2.9 k 0.1 g% min concentrations both lack sufficient sensitivity and
(Table 11,Fig. 4). Significant increases were also recorded specificity to accurately assess an individual's nutritional
in many other serum protein fractions. state, as indicated by the sequential measurements in the
T h e third group of 10 patients were initially malnour- same patients. In spite of significant changes in nutri-
ished and remained malnourished. In this group, the tional state, there were minimal changes in the serum
nutritional state deteriorated further during the period protein concentrations. This is because albumin metab-
of observation. There was a further significant decrease olism is affected by a number of factors such as malnu-
in the Ke/TBW accompanied by a significant increase in trition, malabsorption, liver failure, renal failure, surgical
the NaJTBW. The mean NaJK, increased significantly stress, infection, wounds and burns, carcinoma, hor-
(p < 0.01) from 1.40 +. 0.05 to 1.79 +. 0.16, an indication mones, osmotic equilibrium, and environment.I2 In ad-
of significant deterioration of body composition. The dition, the serum albumin concentration is related to the
albumin concentration remained essentially unchanged, total albumin mass and to the size of the albumin pool.
with an initial mean of 2.8 f 0.1 and a final mean of 2.9 Thus, overhydration results in a decreased serum albu-
f 0.2 g%. min concentration. The present study has demonstrated
In all three groups the hemoglobin and hematocrit did that the nutritional state is an important determinant of
not change significantly. There was a significant (p < the serum albumin concentration, but it is not the only
0.001) correlation between body weight and the BCM as determinant, which accounts for the large variance as-
determined by Ke/TBW the correlation coefficient was sociated with this parameter. Similar problems arise
0.41. However, the confidence limits were large. In addi- when relating the patient's nutritional state to morbidity
tion, the body weight did not reflect the observed changes and mortality. In the hospitalized patient, malnutrition
in nutritional state in the three groups of patients (Table is associated with an increased morbidity and mortality.
11). However, the latter are a function of numerous factors in
addition to the nutritional state. As a result, there is a
similar large variance associated with the relationship
DISCUSSION between the nutritional state and the clinical outcome of
Nutritional assessment is essential for the proper man- the hospitalized patient, ie, the morbidity and mortality.
agement of nutritional support in the clinical environ- Malnutrition has been defined as a serum albumin of
ment. It is important to identify the patients who require < 3.2 g%, a weight loss of more than 20 lb, and a decreased
specialized nutritional support and to determine the type food intake for more than 2 weeks.I3In the present study,
of support required. The calorie and protein require- the serum albumin was < 3.2 g% on 101 occasions, of
ments of a patient with pre-existing malnutrition is dif- which 19% had a normal body composition. In contrast,
ferent from one who is not malnourished but unable to serum albumin was > 3.2 g% in 60 patients, of which 43%
meet his requirements by the usual oral route." Nutri- were malnourished according to body composition data.
tional assessment is also important in following the prog- The data indicate that the serum protein concentra-
ress of nutritional support, especially with TPN, which tions are useful indicators of the nutritional state of a
is an expensive form of nutritional therapy. With maras- population and are, therefore, valid for epidemiological
mus, or protein-calorie malnutrition, there is a loss of surveys. However, because of the poor specificity and
both body fat and the BCM. On the other hand, kwash- sensitivity of the measurement, the serum proteins are
iorkor, or protein malnutrition, results in a depletion of of little value in an individual patient.
tKe BCM while body fat remains normal. Thus, with
both forms of malnutrition, there is a catabolic break- REFERENCES
down of body protein resulting in a loss of cellular mass,
since a labile pool of body protein does not exist. With 1. Sheldon EF, Gardiner BM, Way LW, et a 1 hlanagement of gas-
both forms of malnutrition, this loss of BCM is accom- trointestinal fEtulas. Surg Gynecol Obstet 133:385-390,1971
panied by an expansion of the extracellular mas^.^.^' 2. Dudrick SJ,W h o r e DW, Vars HM, et al: Long-term total par-
enteral nutrition with growth, development, and positive nitrogen
Body composition measurements, therefore, provide an balance. Surgery 64:134-142,1.968
accurate assessment of the nutritional state. These mea- 3. Vinnars, E Effects of intravenous amino acid administration on
surements, although accurate and precise, are complex nitrogen retention. Scand J Clin Lab Invest 27 (Suppl 117):l-55,
and, therefore, difficult to perform in the average hospital 1971
4. Weinsier RL, Hunker EM, Krumdieck CL, et al: Hospital malnu-
environment. However, serum protein determinations trition. A prospective evaluation of general medical patients during
are simple to perform and are available'in almost all the course of hospitalization. Am J Clin Nutr 32418423,1979
hospitals. The present study was performed to evaluate 5. Shizgal HM, Forse RA, Spanier AH, et al: Protein malnutrition
the use of serum proteins, especially albumin, to quanti- following intestinal bypass for morbid obesity. Surgery 86:60-69,
tatively assess the nutritional state. 1979
6. Shizgal Hhl, Mane CA, Spanier A H L The effect of nitrogen-
A highly significant correlation exists between the sparing, intravenously administered fluids on postoperative body
BCM and both total serum protein and serum albumin composition. Surgery 85496-503,1979
concentrations; however, the confidence limits about the 7. Shizgal HM, Spanier AH, Humes J, et al: Indirect measurement of

Downloaded from pen.sagepub.com at UNIV CALIFORNIA SAN DIEGO on July 16, 2016
454 FORSE AND SHIZGAL Vol.4, No. 5
total exchangeable potassium. Am J Physiol. 233F253-259,1977 1980 (in press)
8. Armitage P Statistical Methods in Medical Research. New York, 11. Shizgal HM: Protein requirements with total p a r e n t e d nutrition.
John Wiley & Sons, 1971 Surg Forum 2960-62, 1978
9. Moore FD, Olesen KH, hlcMurrey JD, et al: The Body Cell Mass 12. Rothschild hlA, Oratz hl, Schreider SS: Albumin synthesis. N Engl
and Its Supporting Environment. Body Composition in Health and J Med 286749-757; 816-821, 1972
Disease. Philadelphia, W.B. Saunders Co, 1963 13. Law DK, Dudrick SJ, Abdou NI: Irnrnunocompetence of patients
10. Shizgal HM: Body composition by multiple isotope dilution. Pro- with protein-calorie malnutrition. T h e effects of nutritional reple-
ceedings of the 2nd Ross Conference on Nutritional Assessment, tion. Ann Int Med 79:545-550,1973

Downloaded from pen.sagepub.com at UNIV CALIFORNIA SAN DIEGO on July 16, 2016