Analysis of 26 Studies of the Impact of Coconut Oil on Lipid Parameters: Beyond Total and LDL Cholesterol

Abstract

Coconut oil (CNO) is often characterized as an "artery-clogging fat" because it is a predominantly saturated fat that ostensibly raises total cholesterol (TChol) and LDL cholesterol (LDL-C). Whereas previous analyses assessed CNO based on the relative effects on lipid parameters against other fats and oils, this analysis focuses on the effects of CNO itself. Here, we review the literature on CNO and analyze 984 lipid profile data sets from 26 CNO studies conducted over the past 40 years. This analysis shows considerable heterogeneity among CNO studies regarding participant selection, the amount consumed, and the study duration. The analysis reveals that, overall, CNO consumption gives variable TChol and LDL-C values, but that the HDL-cholesterol (HDL-C) values increase and triglycerides (TG) decrease. This holistic lipid assessment, together with the consideration of lipid ratios, shows that CNO does not pose a health risk for heart disease. Because the predominantly medium-chain fatty acid profile of CNO is significantly different from that of lard and palm oil, studies using these as reference materials do not apply to CNO. This paper concludes that the recommendation to avoid consuming coconut oil due to the risk of heart disease is not justified.

Keywords: Friedewald formula; cardiovascular health; coconut oil; dietary fat; dietary guidelines; lipid parameters; lipid ratios; medium-chain fatty acids; saturated fat; trans-fat.

Figure 1

Comparison of fatty acid content of CNO and other so-called saturated fats, gm/100 gm. Data are from Table 1.

Figure 2

Total cholesterol levels for 14 men consuming a high-fat vs. a low-fat diet. This plot was produced from the data published in Table V of an article by Keys in 1957 including the dates [14]. The TChol results for the individual men were substantially different for January 31 and February 7, when all men were consuming an identical “house diet” with 37–42% of the total calories as fat. The results from week 1 onwards were obtained while the subjects consumed a “low-fat base diet” with 8–15% fat. Most men had a precipitous drop in TChol, reaching the lowest value between weeks 1 and 3, but it usually increased thereafter. These results show that the reliability of the data for individual prediction is low because of the large intra- and inter-individual variability.

Figure 3

Flow diagram of literature search. Note: Twenty-six studies are reported in 25 articles

Figure 4

Differences in TChol, LDL-C, HDL-C, and TG in people consuming CNO in short-, medium-, and long-duration studies. Data are from Table 5.

Figure 5

The Zararsiz (2022) study included nearly 89,000 adults and compared the LDL-C results of samples from the same person and measured using a direct colorimetric enzymatic reaction method (Roche, Siemans, or Beckman) versus each of the estimation formulas shown in the legend on the right and represented in the colored bars. None of the formulas showed concordance of more than 77% with a direct method. Of the formulas tested, the Friedewald formula had the lowest % concordance with all three methods of direct measurement, ranging from about 57 to 64%. Reprinted from ref. [94].

Figure 6

LDL-C results are shown by TG stratum for samples taken from the same person and measured using a direct colorimetric enzymatic reaction method (Roche, Siemans, or Beckman) versus each of the estimation formulas shown in the legend on the right and represented in the colored bars. None of the estimation formulas had more than 81% concordance with a direct method, and the concordance decreased with each higher TG stratum. The Friedewald formula had the lowest concordance with the Roche and Beckman direct methods for each TG stratum of 150 mg/dL or greater and had less than 10% concordance with the Beckman direct method at TG levels ≥ 400 mg/dL. The concordance of the estimation formulas with the Siemans direct method was no greater than 74% at any TG level. Reprinted from ref. [94].