Abstract Background. In laboratory animals, the consumption
of soy protein, rather than animal protein, decreases
serum cholesterol concentrations, but studies in
humans have been inconclusive. In this meta-analysis of
38 controlled clinical trials, we examined the relation between
soy protein consumption and serum lipid concentrations
in humans.
Methods. We used a random-effects model to quantify
the average effects of soy protein intake on serum
lipids in the studies we examined and used hierarchical
mixed-effects regression models to predict variation as a
function of the characteristics of the studies.
Results. In most of the studies, the intake of energy,
fat, saturated fat, and cholesterol was similar when the
subjects ingested control and soy-containing diets; soy
protein intake averaged 47 g per day. Ingestion of soy
protein was associated with the following net changes
in serum lipid concentrations from the concentrations
reached with the control diet: total cholesterol, a decrease
of 23.2 mg per deciliter (0.60 mmol per liter; 95
percent confidence interval, 13.5 to 32.9 mg per deciliter
[0.35 to 0.85 mmol per liter]), or 9.3 percent; low-density
lipoprotein (LDL) cholesterol, a decrease of 21.7 mg per
deciliter (0.56 mmol per liter; 95 percent confidence interval,
11.2 to 31.7 mg per deciliter [0.30 to 0.82 mmol
per liter]), or 12.9 percent; and triglycerides, a decrease
of 13.3 mg per deciliter (0.15 mmol per liter; 95 percent
confidence interval, 0.3 to 25.7 mg per deciliter [0.003 to
0.29 mmol per liter]), or 10.5 percent. The changes in serum
cholesterol and LDL cholesterol concentrations were
directly related to the initial serum cholesterol concentration
(P0.001). The ingestion of soy protein was associated
with a nonsignificant 2.4 percent increase in serum
concentrations of high-density lipoprotein (HDL) cholesterol.
Conclusions. We found that the consumption of
soy protein rather than animal protein significantly decreased
serum concentrations of total cholesterol, LDL
cholesterol, and triglycerides. (N Engl J Med 1995;333:
276-82.)