The effect of soybean-based foods on plasma lipid and lipoprotein concentrations

Tufts University
711 Washington Street
Boston, MA 02111
E-mail: alice.lichtenstein{at}tufts.edu

Dear Sir:

We would like to clarify statements printed in the editorial by Steinberg (1) accompanying our recent publication about the effect of different forms of soybean-based foods on plasma lipid and lipoprotein concentration and other cardiovascular disease risk factors (2). It was written that "...when the change from baseline was examined, all groups had a substantial improvement in lipid values." There appears to have been a misunderstanding in the nature of the data presented and the statistical analysis. We did not examine a change from baseline, nor, given the study design, would it have been appropriate to do so. We reported baseline characteristics of the study volunteers at the time of screening (Table 1 from reference 2). These data were collected to verify that the study volunteers met the predetermined inclusion criteria for recruitment and were collected during a period when they consumed their habitual diet. This habitual diet likely differed among individuals; hence, it is inappropriate to compare those data with the data that were collected during the controlled intervention period (Table 3 from reference 2) and draw conclusions. Regardless of the intervention, as we have seen in the past, plasma lipid and lipoprotein values observed during the screening period were frequently less favorable with respect to cardiovascular disease risk profiles than were those observed during any of the intervention periods (3–6). This misunderstanding also led to a second inaccurate statement in the editorial: "It also provides good evidence that soy protein can elicit a hypocholesterolemic effect similar to that of animal protein when the dietary components are held constant." Differences between data collected at the end of the intervention periods and those collected at baseline are likely attributable to multiple factors, such as the saturated fat, cholesterol, and fiber contents of the diets, and not to a single dietary component. However, this hypothesis cannot be tested from the available data.

ACKNOWLEDGMENTS

The authors had no conflicts of interest to declare.

REFERENCES