Fasting vs. Nonfasting and LDL-C Accuracy

Study Questions:

How accurate are methods for estimation of low-density lipoprotein cholesterol (LDL-C) in the nonfasting state?


Data from the VLDL (Very Large Database of Lipids) study, a cross-section of 1,545,634 patients (959,153 fasting ≥10-12 hours; 586,481 nonfasting) were used for the present analysis. Rapid ultracentrifugation was used to directly measure LDL-C content (direct-LDL). Accuracy was defined as the percentage of direct-LDL falling within an estimated LDL-C (either a novel method for estimating LDL or the Friedewald method for estimating LDL) category by clinical cut points. The novel estimation of LDL-C was calculated as total cholesterol–high-density lipoprotein (HDL-C)–triglyceride (TG)/adjustable factor. The adjustable factors were derived from the authors’ previously reported method, whereby TG and non–HDL-C were used to assign 1 of 180 different patient-specific factors to estimate VLDL-C based on ratios of TG and non–HDL-C across a wide distribution of values. For low estimated LDL-C (<70 mg/dl), accuracy was evaluated by TG levels. The magnitude of absolute and percent differences between direct-LDL and estimated LDL-C (novel method estimated LDL or Friedewald method estimated LDL) was stratified by LDL-C and TG categories.


In both fasting and nonfasting samples, accuracy was higher with the novel method across all clinical LDL-C categories (range: 87-94%) compared to Friedewald estimation (range: 71-93%) (p ≤ 0.001). With LDL-C <70 mg/dl, nonfasting LDL-C accuracy using the novel method was superior to estimations using the Friedewald method (92% vs. 71%, p < 0.001). In this LDL-C range, 19% of fasting and 30% of nonfasting patients had differences ≥10 mg/dl between the Friedewald method for estimating LDL-C and direct-LDL, whereas only 2% and 3% of patients, respectively, had similar differences with novel estimation. Accuracy of LDL-C <70 mg/dl further decreased as TG increased, particularly for the Friedewald estimation (range: 37-96%) versus the novel method (range: 82-94%). With TG 200-399 mg/dl in nonfasting patients, the novel method for estimating LDL-C <70 mg/dl accuracy was superior to the Friedewald method for estimating LDL-C (82% vs. 37%, p < 0.001). In this TG range, 73% of fasting and 81% of nonfasting patients had ≥10 mg/dl differences between LDL-C for the Friedewald method compared to the direct-LDL measures. In the same TG range, 25% of fasting and 20% of nonfasting patients had ≥10 mg/dl differences between LDL for the novel method compared to the direct-LDL measures.


The investigators concluded that a novel adaptable LDL-C estimation performs better in nonfasting samples than the fixed Friedewald estimation, with a particular accuracy advantage in settings of low LDL-C and high TG.


The method for estimating LDL levels in the nonfasting (and fasting) state appears to improve accuracy and thus clinical decision making. Implementation into systems that make it easy for providers to use, and related research on practice changes and implications for management of lipids, are warranted.

< Back to Listings