Low HDL makes Dr. Friedewald a liar

There's a $22 billion industry based on treating LDL cholesterol, a fictitious number.

LDL cholesterol is calculated from the following equation:

LDL cholesterol = Total cholesterol - HDL cholesterol - triglycerides/5

So when your doctor tells you that your LDL cholesterol is X, 99% of the time it has been calculated. This is based on the empiric calculation developed by Dr. Friedwald in the 1960s. Back then, it was a reasonable solution, just like bacon and eggs was a reasonable breakfast and a '62 Rambler was a reasonable automobile.

One of the problems with Dr. Friedewald's calculation is that the lower HDL cholesterol, the less accurate LDL cholesterol becomes. If it were just a few points, so what? But what if it were commonly 50 to 100 mg/dl inaccurate? In other words, your doctor tells you that your LDL is 120 mg/dl, but the real number is somewhere between 170 and 220 mg/dl. Does this happen?

You bet it does. In my experience, it is an everyday event. In fact, I'm actually surprised when the Friedewald calculated LDL closely approximates true LDL--it's the exception.

Dr. Friedewald would likely have explained that, when applied to a large population of, say, 10,000 people, calculated LDL is a good representation of true LDL. However, just like saying that the average weight for an American woman is 176 lbs (that's true, by the way), does that mean if you weigh 125 lbs that you are "off" by 41 lbs? No, but it shows how you cannot apply the statistical observations made in large populations to a single individual.

The lower HDL goes, the more inaccurate LDL becomes. This would be acceptable if most HDLs still permitted reasonable estimation of LDL--but it does not. LDL begins to become significantly inaccurate with HDL below 60 mg/dl.

How to get around this antiquated formula? In order of most accurate to least accurate:

--LDL particle number (NMR)--the most accurate by far.

--Apoprotein B--available in most laboratories.

--"Direct" LDL

--Non-HDL--i.e., the calculation of total cholesterol minus HDL. But it's still a calculated with built-in flaws.

--LDL by Friedewald calculation.

My personal view: you need to get an NMR if you want to know what your LDL truly is. A month of Lipitor costs around $80-120. A basic NMR costs less than $90. It's a relative bargain.

Comments (5) -

  • Mike

    3/18/2007 1:52:00 AM |

    What is shocking is that enormous prescriptions for statins are written based on the calculated LDL.

  • Dr. Davis

    3/18/2007 1:16:00 PM |

    Yes, $22 billion last year, in fact. All prescribed for a number that is a crude estimate, sometimes a complete fiction. Imagine your state trooper ticketed you because his radar device said you were doing 60 mph when you were really doing 35 mph.

  • Anonymous

    2/6/2008 1:37:00 AM |

    Why NMR over the other tests Berkeley Heart Lab or VAP?

  • Anonymous

    7/2/2008 7:02:00 PM |

    I don't understand.  If in this example, the doctor (wrongly) thinks the LDL number is 120mg/dl, how does that cause the prescription of Lipitor? Unless I'm reading it backwards, and the doctor is actually telling the patient their LDL is 170mg-220mg, but unwittingly, it's actually 120mg/dl.

    And, if a low HDL causes the LDL number to be inaccurate, does that also cause the total cholesterol number to be inaccurate too?