We failed to prove our hypothesis that administration of psyllium in the morning would have a greater cholesterol-lowering effect than it would in the evening. Not only was there no observable difference in lipid levels between the crossover periods but the daily ingestion of a greater daily dose than the 10.2 g of psyllium for which the FDA allows health claims to be made  had no effect on lipid levels in our study group. No change in any lipid parameter, including total and LDL cholesterol was observed. No difference was found when subgroup analysis was undertaken for the sex of the patients, the time of day they took their psyllium, or whether they were recruited from the gastroenterology clinic or the lipid clinic.
We used a crossover design since this was the most appropriate one for the primary question being addressed; accordingly, our study did not include a control group. However, the nature of the study should not have provided any motivation for study subjects to adopt any new lifestyle or dietary changes beyond those implemented well before the introduction of psyllium. Observational data has been shown to provide valid information, which is consistent with that observed in randomized, controlled trials [17, 18]. The nature of the intervention was in keeping with those undertaken in day-to-day clinical practice and the protocol used should, therefore, have high "clinical relevance".
Failure of lipid-lowering by psyllium has also been demonstrated in twenty hypercholesterolemic children , in twenty-four hyperlipidemic adults  and in a large observational study of elderly patients taking psyllium . A report of lipid-lowering therapies in hypercholesterolemic veterans showed only a 2% reduction in LDL cholesterol and a small increase in LDL/HDL ratio in patients taking psyllium, but does not provide a measure of statistical significance . One study revealed no difference in total cholesterol-lowering compared to placebo, but a reduction of LDL cholesterol resulted from psyllium treatment . Another demonstrated no difference in total cholesterol-lowering compared to placebo, a reduction of LDL in 11 "responders" and no change in 9 "nonresponders" . A reduction of HDL cholesterol has been noted in some studies [25–27] and was associated with changes in LDL/HDL ratios similar to placebo treatment [25, 26].
Published studies include few normocholesterolemic subjects. Cholesterol reduction was observed in 7 normal men  and in 5 of 9 subjects , in both studies after 3 weeks of treatment. A reduction of cholesterol levels was also observed in 12 elderly patients given psyllium for 4 months , while 5 normocholesterolemic subjects in another study showed no reduction after 2 to 7 months of treatment .
A meta-analysis of 17 studies of patients with hypercholesterolemia has suggested a small but significant cholesterol-lowering effect of psyllium . All of these investigations were associated in one way or another with the product manufacturer. Additional studies have also indicated some cholesterol-lowering by psyllium in hypercholesterolemic individuals [32–37] or in diabetics [38–40]; however, much of this work is uncontrolled and some protocols have specifically excluded premenopausal women [33, 38]. The association of cholesterol-lowering effects with psyllium may be weakened in some studies by the use of a supplement containing additional forms of soluble fibre  or by apparent differences in intake of calories [43–46], soluble fibre  or cholesterol  in control and treatment groups or periods. Several reports include only small numbers of patients and/or are of short duration. There is a strong predominance of male subjects in these publications and some protocols incorporate additional treatment interventions [20, 48].
Several factors may contribute to the difference between our observations and those of others. A meta-analysis has demonstrated that the initial level of cholesterol was highly predictive of the subsequent reduction of cholesterol by oat bran . A greater effect of psyllium in men compared to women has been suggested [23, 46] and a diet high in soluble fibre produced less cholesterol-lowering in post menopausal women than in men . Soluble fibre has a lesser effect on lipid metabolism in female than in male guinea pigs  and there is a sex-based difference in mechanism of action in this animal . Oat bran fails to lower cholesterol in young women, in contrast to men and older women . The dominance of women in our study, the "normal", or only slightly abnormal cholesterol states of our subjects and the relatively young ages of some of them may, accordingly, account for some of the variance of our observations with some of those previously reported.
The small increase in the weight of subjects is believed to be have resulted from reduced physical activity. In a meta-analysis of the effect of weight reduction on lipids, predominantly through dietary change, a reduction in total cholesterol of 0.05 mmol/L and of 0.02 mmol/L in LDL cholesterol per kilogram of weight lost was identified . Dietary intakes were stable throughout our study and the average weight gain of less than one kilogram is very unlikely to have raised cholesterol levels to a degree sufficient to offset a significant cholesterol lowering effect of psyllium.
A small cholesterol-lowering effect of psyllium appears to occur in hypercholesterolemic individuals, at least in men and possibly postmenopausal women. The notion of a benefit accruing to the general population requires additional study. The promotion of foods containing psyllium as reducing the risk of heart disease for the population at large  may be premature. Additional study is required and this should be undertaken in a manner that is free from concern regarding the possibility of publication bias which Brown L, Rosner B, Willett WW and Sacks FM have raised .