Discussion
The purpose of this meta-analysis was to determine the effects of a low-fat diet, in comparison with the participants’ usual diet, on reducing serum lipid levels in women. There have been no prior meta-analyses focusing on the effects of a low-fat diet on women and, more specifically, on premenopausal and postmenopausal subgroups. In addition, the TC-to-HDL-C ratio was seldom assessed in previous studies.
In the present meta-analysis, the overall results suggest that a low-fat diet is efficacious in reducing the concentrations of TC, HDL-C, and LDL-C, but not in reducing TG and TC-to-HDL-C ratio in women.
The heterogeneity in the current meta-analysis was small or moderate. In postmenopausal women, only the I value for TC was 77%. All others were less than 50%. Even when we excluded Svendsen et al, the I value decreased from 77% to 36%, and the result was also not significant for postmenopausal women. As a result, the result of the current meta-analysis is robust.
High TC and LDL-C levels are strongly related to CVD risk, and reductions in LDL-C levels are associated with reduced CVD risk. The major food components that raise LDL-C are saturated fatty acids, transunsaturated fatty acids, and, to a lesser extent, cholesterol. The result of our study was similar to that of a previous meta-analysis—that low-fat diet is efficacious in reducing the concentrations of TC and LDL-C. However, the changes in TC and LDL-C observed in the current investigation seem to be greater than those in the updated Cochrane Review, which addressed the effects of reduced fat intake on men and women, yielding significant reductions in TC and LDL-C that were less than half of those observed in the current investigation.
Our result on the effect of TG is consistent with the updated Cochrane Review of 2012: reduced fat intake had no effect on TG in men and women. However, it differs from our results in that it showed no effect on HDL-C reduction. Low-fat diet is likely to result in higher TG levels and lower HDL-C levels than diets where saturated fats are wholly replaced by unsaturated fats.
In a meta-analysis of aerobic exercise and prudent diet in men and women, statistically significant interventions minus control reductions were found for TC (−15.5 mg/dL; 95% CI, −20.3 to −10.7) and LDL-C (−9.2 mg/dL; 95% CI, −12.7 to −5.8). In contrast, there were still some differences from our results, as there were no effects on HDL-C (−0.5 mg/dL; 95% CI, −4.0 to 3.1) but significant reductions in TG (−10.6 mg/dL; 95% CI, −17.2 to −4.0) and TC-to-HDL-C ratio (−0.4 mg/dL; 95% CI, −0.7 to −0.2) in the study of Kelley et al.
It is very interesting to note that, in our meta-analysis, a low-fat diet decreased HDL-C levels, differing from other systematic reviews. In the study of Hooper et al, reduced fat intake resulted in no significant reductions in HDL-C in men and women (95% CI, −0.01 to −0.02; P=0.3). In Kelley et al, aerobic exercise and prudent diet had no effect on HDL-C (−0.5 mg/dL; 95% CI, −4.0 to 3.1). Although low total and saturated fat diets have been shown not to decrease significantly (by 1.5%) after step I dietary intervention studies, but to decrease significantly (by 7%) after step II dietary interventions in the National Cholesterol Education Program; the TC-to-HDL-C ratio decreased significantly after both step I and step II dietary interventions (step I diet: ≤30% of total energy as fat, ≤10% of energy as saturated fatty acid, and ≤300 mg/d dietary cholesterol; step II diet: ≤7% of energy as saturated fatty acid and ≤200 mg/d dietary cholesterol).
The subgroup analysis in the present meta-analysis showed that reduced-fat diets are efficacious in reducing HDL-C in premenopausal women but are not effective in postmenopausal women. Moreover, we did not analyze the effects of low-fat diet as step I or step II dietary interventions in the current meta-analysis.
HDL-C concentration is an independent, inverse predictor for CVDs. HDL-C levels seem to have a stronger association with heart disease in women than in men. It is estimated that a 1-mg/dL increase in HDL-C reduces the risk of heart disease by 3% in women, in comparison with 2% in men. Despite a large body of evidence indicating that high HDL-C levels are inversely related to coronary disease risk, it has not been conclusively demonstrated that increases in HDL-C levels induced by diet and lifestyle modifications lead to reduced coronary disease risk. Thus, it remains to be determined whether increased HDL-C should be a target for dietary therapy. In certain individuals, low-fat, high-carbohydrate diets can result in reductions in HDL-C levels. Although it is not known whether diet-induced reductions in HDL-C, in conjunction with reduced TC and LDL-C, have an adverse effect on coronary disease risk, it may be prudent in those cases to couple efforts at weight management with some limitations of carbohydrate intake.
In fact, the TC-to-HDL-C ratio is considered more important than TC or lipoprotein cholesterol concentrations in estimating the risk of coronary artery disease. Among the eight RCTs included, only two trials reported on the TC- to-HDL-C ratio. Even though we obtained six comparison groups after we had calculated the TC-to-HDL-C ratio from the data that two authors provided, low-fat diet still did not significantly reduce the serum TC-to-HDL-C ratio in women. Based on this observation, more trials on the TC-to-HDL-C ratio need to be included when reporting serum lipid outcomes if we are to determine the effects of low-fat diet on the TC-to-HDL-C ratio in women. In addition, previous meta-analyses could predict the effects of fats on plasma lipids but they could not determine whether fat will cause CVDs. We also need to analyze the classes of fatty acids and their effects on serum lipids to determine whether or not the risk is reduced most effectively when trans-fatty acids and saturated fatty acids are replaced with unsaturated fatty acids.
Unfortunately, we also found that the effects of low-fat diets on TC, HDL-C, and LDL-C only showed significant reductions in premenopausal women, and not in postmenopausal women, within the subgroup analysis. However, we cannot compare our results with other meta-analyses because there were no prior analyses of premenopausal and postmenopausal subgroups. We did not evaluate age, exercise, and other factors in premenopausal and postmenopausal women in the eight studies included, so we cannot conclude that a low-fat diet alone has no effect on postmenopausal women. Age plays an important role in serum lipid levels in women. Postmenopausal women are older than premenopausal women; therefore, they gain weight easier and exercise less than younger women.
Strengths and Limitations
The strengths of this review are twofold: (1) there was a predetermined strategy for study selection and (2) there was quality assessment of the studies included by two independent assessors.
Our meta-analysis has some limitations. First, one of the major elements limited the inclusion criteria to studies published in English or Chinese. As a result, selection bias could not be avoided. Some important studies published in languages other than English or Chinese were excluded. In addition, our review included studies with small sample sizes and low statistical power, which increased the probabilities of obtaining nonsignificant results and of reducing the probability of their publication. However, previous studies have proven that meta-analyses of small trials yield results that are very similar to those of meta-analyses of large trials of the same intervention. The results were based on only eight trials even though they were all RCTs. For TG, HDL-C, and LDL-C, 1,536 women (900 in the intervention group and 636 in the control group) met the inclusion criteria. For TC, only 542 women (285 in the intervention group and 257 in the control group) were included in one study. Bhargava suggested that we report LDL-C and HDL-C separately and drop TC. Third, our analysis did not evaluate the influence of the duration of the study on the results. Only two trials reported separately for three durations: Chen et al reported outcomes 3, 6, and 12 months after intervention, and Clifton et al reported outcomes at 4, 8, and 12 weeks. To avoid duplication of data derived from the same participants from studies with multiple time points, we used only the endpoints for the longest duration in our study. Additional studies are needed to further address the effects of the duration of low-fat diet on women. Fourth, lipid profile is the main outcome and a surrogate endpoint for CVDs in our study. We have not assessed the effect of low-fat diet on cardiovascular morbidity, mortality, and individual outcomes, including myocardial infarction or stroke. Lastly, some included articles did not mention dietary fat as either saturated or unsaturated. In the updated Cochrane Review, it was not clear whether replacing saturated fat with polyunsaturated or monounsaturated fat (either or both) was beneficial for cardiovascular events. Thus, we need to analyze the effects of unsaturated fatty acids on serum lipids in women to determine the effects of the classes of fatty acids on serum lipids. In brief, it is important to observe the weaknesses of language limits, sample size, and lack of analysis of duration in our analysis, which limit the interpretation of the results. In future studies, we need to search more published studies that are not limited to English or Chinese or to modify the inclusion criteria to include more studies. As well, we need to address a more definitive hard endpoint for CVDs.
Implications for Practice
The present findings seem to be important from a practical perspective, especially for premenopausal women. Based on our results, providers should suggest that women adopt a low-fat diet before menopause to improve serum lipid levels.