By this time, I figure most everyone knows what LDL cholesterol is and those over age 30 have undoubtedly had their LDL checked numerous times, with many of them put on a statin medication because their “LDL is too high.”
Setting aside for a moment the role that LDL plays in heart disease (which, as you can read in my book, is a concept that has been dangerously oversimplified—it’s really the number of LDL particles, not total LDL amount that’s important in assessing heart disease risk), what you probably don’t know is the “standard” cholesterol test that your doctor orders is not really a measurement of your LDL—it’s just an estimate. And it’s an estimate that can be off by a frightening amount, depending upon a variety of factors.
When routine testing of LDL, HDL, and triglycerides, was first started back in the early 1970’s, it was relatively easy to measure HDL and triglycerides but measuring LDL was a bit more difficult and therefore a bit more expensive. To solve this problem, testing labs estimated LDL based upon a formula derived in 1972 by Dr. William Friedewald from the National Institute of Health. This “Friedewald Equation” calculates LDL (also called LDL-C) using the formula
LDL = Total cholesterol – (VLDL-C) – HDL.
“VLDL-C” stands for “very low density lipoprotein cholesterol” and in order to calculate the value of VLDL-C, the formula uses the triglyceride levels to approximate the levels of VLDL. (That’s right, the estimation formula itself uses an estimated value for one of its components!). In this case, the Friedewald formula assumes that (triglycerides/5) equals VLDL-C. Keep this assumption in mind, it will be important later.
Thus, in replacing VLDL-C with (triglycerides/5), the form of the Friedewald equation becomes:
LDL = Total cholesterol – HDL – (triglycerides/5).
Applying this formula, if a person has a total cholesterol value of 200, HDL of 60, and triglycerides of 150, their LDL would calculate out to be 200 – 60 - (150/5) or 110 mg/dL. This hypothetical person would probably be told they needed to be on a statin. But, how useful is this formula? How much error is inherent in this method of “estimating” one’s LDL? Let’s see…
A Flawed Model
Despite its widespread, almost universal use, the Friedewald equation is very far from perfect and rarely reflects the true amount of LDL cholesterol present.
Here are but a few of the problems with using the Friedewald equation to calculate LDL:
Chylomicrons are particles that transport dietary (ingested in foods) lipids from the intestines to other locations in the body. Because chylomicrons contain lipid particles that would distort the Friedewald equation, the researchers deriving the Friedewald equation ensured that no chylomicrons were present in their research samples. They did this by storing the samples at 4ºC for 18 hours and then discarding any sample that showed evidence of chylomicrons.
It’s not feasible, however, for your local medical lab to go to this trouble just to ensure that there are no chylomicrons in your blood sample, so they do the next best thing: they tell you to fast for 12 to 14 hours before the blood draw. According to the journal Clinical Biochemistry, however, the assumption that fasting for this amount of time ensures the absence of chylomicrons “may frequently be wrong” and “the error that inevitably results is undetected by the laboratory and undetectable by the clinician.”
The first problem with using triglycerides in the Friedewald equation is the assumption that triglycerides are an accurate tool for estimating VLDL-C levels. This is flawed reasoning because the ratio of triglycerides to VLDL is not constant, a fact that the Friedewald researchers admitted in their own research paper proposing the Friedewald equation. In disclosing this fact, the Friedewald researchers argued that the error inherent in using triglycerides to estimate VLDL-C was not significant because there would always be considerably more LDL than VLDL in a blood sample. In the words of the journal Clinical Biochemistry, “…it is disconcerting that the assumption on which the [Friedewald] calculation is based was disproved in the paper that proposed it.”
Not only is the ratio of triglycerides to VLDL-C not constant, it varies significantly depending on the level of triglycerides in the sample. Clinical guidelines warn against using the Friedewald equation if a patient’s triglycerides are over 400 mg/dL, but problems occur at levels far below that; levels that are common in today’s population. A study comparing true (directly measured) LDL levels to those estimated by the Friedewald equation found that if triglycerides were below 200 mg/dL, the estimated LDL was within 10% of the directly measured for 85% of the samples. If, however, triglycerides were 201-300 mg/dL, only 77% of the sample estimates were within 10% of the directly measured value. The success rate dropped even further with samples having triglycerides in the 301-400 mg/dL range, with only 59% of the samples falling having values within 10% of the directly measured value. This increasing failure rate happens because the ratio of triglycerides to VLDL-C changes as triglyceride levels increase.
To show how this changing ratio will distort the LDL estimate, consider this example cited by The American Journal of Cardiology in which the levels of HDL and total cholesterol were kept constant across three samples. Changing only the triglyceride levels from 350 mg/dL down to 50 mg/dL will increase the estimated LDL value from 90 mg/dL to 150 mg/dL.
In other words, even though the LDL in the samples did not change, by decreasing the triglyceride levels from 350 mg/dL to 50 mg/dL, the resulting LDL estimate increases by 67%, from 90 mg/dL to 150 mg/dL.
No lab test is perfect. And when a formula is based upon a series of tests, the result is an accumulated error that grows larger with the number of components in the formula. Because the Friedewald equation is based upon tests for triglycerides and HDL, errors in those two measurements will compound to greatly affect the calculate LDL even further. In the words of Clinical Biochemistry, “…even if total cholesterol, HDL cholesterol, and triglycerides are measured satisfactorily, the cumulative error involved can fatally sabotage the calculation of LDL-C.”
Summarizing the Shortcomings of Friedewald
I think the Clinical Biochemistry authors of “Triglycerides and small dense LDL: the twin Achilles heels of the Friedewald formula” summarize Friedewald’s shortcomings quite well:
“On its own, there are numerous pitfalls and inaccuracies in the Friedewald approach. Samples must be taken after a prolonged fast. Chylomicrons should void the calculation, but we no longer check whether they are present. Triglycerides greater than [400 mg/dL] make calculation of LDL-C inaccurate, but so do those from [220 mg/dL to 400 mg/dL]. With therapy that lowers triglycerides, changes in VLDL composition make changes in LDL-C concentrations impossible to interpret. Moreover, the method may also break down at lower levels of LDL-C. Worse yet, even if LDL-C were measured accurately, it is really the number of LDL particles and their composition that are critical, not just concentration of LDL-C.”
It’s obvious to me that today’s standard clinical practice of using the Friedewald calculation to estimate LDL is decades behind the times. And when I think about this deeply flawed estimate being used as justification for placing countless millions of people on statins, I see red.
How You Should Test Your LDL
If you really want to know your true LDL value, have it directly measured, not estimated. The test is called “Low Density Lipoprotein Cholesterol (Direct)” or sometimes just “LDL Direct.”
What’s better than a Friedewald-approximated LDL or even a directly measured LDL value? An “LDL-P” that measures the number of LDL particles is a far more relevant indicator of cardiac risk, but that’s a subject that my book covers in great detail!
“Triglycerides and small dense LDL: the twin Achilles heels of the Friedewald formula,” Allan D. Sniderman, David Blank, Robert Zakarian, Jean Bergeron, Jiri Frohlich; Clinical Biochemistry 36 (2003) 499-504.
“Calculated low-density lipoprotein cholesterol level: time for a change,” DM Lane; The American Journal of Cardiology 1997;80:823
“Calculated values for low-density lipoprotein cholesterol in the assessment of lipid abnormalities and coronary disease risk,” McNamara JR, Cohn JS, Wilson PW, Schaefer EJ; Clinical Chemistry 1990:36:36-42.