Background: Colloid osmotic pressure (COP) plays a major role in transcapil
lary fluid shift, including in the glomerular capillary. However, COP is ge
nerally estimated by quadratic equations derived from total plasma protein
and/or albumin concentrations. The aim of this study was to assess the accu
racy of such equations, and to determine the potential role of liver-derive
d non-albumin proteins in the maintenance of COP, especially in patients pr
esenting a nephrotic syndrome. Methods: COP was directly assessed with an o
smometer in 170 patients (347 samples), and the results compared with calcu
lated COP, using 4 previously published formulas [Brenner 1972, Canaan-Kuhl
1993, Landis-Pappenheimer 1963, Navar 1977]. Results: The 4 calculated COP
values were strongly correlated with measured COP (range r = 0.88 - 0.96).
However, in absolute terms, measured COP differed significantly from each
of the 4 calculated mean values of COP (p < 0.001). Fibrinogen exerted per
se a weak oncotic effect as measured in vitro. However, fibrinogen was high
ly related to albumin and presumably reflected the oncotic effect of other
liver-derived non-albumin proteins. Inclusion of albumin and fibrinogen in
a linear model provided an excellent fit for predicted COP with a highly si
gnificant correlation (r = 0.96, p < 0.001) over a wide range of COP values
. The predicted equation was: COP(mmHg) = 6.89 x (albumin + fibrinogen) (g/
dl) - 5.68. Conclusion: None of the 4 most commonly used formulas correctly
predicted COP, and direct measurement of COP is still preferable for resea
rch studies. The introduction of fibrinogen into the formula estimating COP
leads to higher accuracy, and therefore represents a more convenient model
for routine evaluation.