Making Dr. Friedewald an honest man

Colleen started with the usual discrepancy between conventional calculated LDL cholesterol of 121 mg/dl and the far more accurate LDL particle number (NMR) of 1927 nmol/L.

Those of you following this conversation or our many conversations on the Track Your Plaque Forum know that a useful and highly reliable rule-of-thumb for converting NMR LDL particle number to LDL is to drop the last digit: 1927 nmol/L becomes 192 mg/dl. (This is, admitttedly, arrived at empirically, not by design. However, it has held up through thousands of NMR analyses and plays out reasonably when you compare distributions of Friedewald LDL and LDL particle number on a population basis.)

In other words, by this simple manipulation, Colleen's Friedewald calculated LDL is off by 58%. This is very common, a phenomenon I witness several times every day.

By LDL particle size, 75% of all Colleen's LDL particle were abnormally small (small LDL particle number 1440 nmol/L). This is a moderately severe small LDL tendency.

So we took all the steps for reduction of small LDL/LDL, including elimination of wheat and cornstarch, exercise, weight loss (which happens inevitably when wheat and cornstarch are eliminated), fish oil, vitamin D, etc.

Another NMR lipoprotein panel showed an LDL particle number of 882 nmol/L and a Friedewald calculated LDL of 87 mg/dl. Using our rule-of-thumb, LDL by particle number is virtually the same as the calculated LDL. This time, small LDL numbered only 237 nmol/L, or 26.8% of the total, a marked reduction.

Isn't that interesting? As small LDL is corrected, the crude Friedwald calculated LDL approximates the more accurate LDL particle number.

It assumes that accuracy of the Friedewald calculation may be more likely to occur as LDL size approaches normal. However, when LDL size is abnormally small--a condition shared by at least 70% of people with coronary heart disease--then the Friedewald LDL becomes increasingly inaccurate.

The opposite can also happen: When all or nearly all LDL particles are large, Friedewald calculated LDL can markedly overestimate LDL particle number. Yesterday, for instance, a patient had a Friedewald calculated LDL of 183 mg/dl, but an NMR particle number of 1110 nmol/L--drop the zero . . . LDL 110 mg/dl. This woman was advised to take a statin drug by her primary care physician, based on the Friedewald LDL. Instead, she proved to have a far lower LDL. She would not have benefitted from taking a statin drug.

As I've warned many times before: Beware the Friedewald calculated LDL.

Comments (3) -

  • Nancy LC

    8/27/2008 9:55:00 PM |

    Oooh, thanks for posting this!  On low carb diets we see people getting alarmed all the time because their LDL goes up (calculated, of course).  However their triglycerides are very, very low and their HDL usually is good, sometimes extremely high.

    But their doctors are all alarmed over their LDL and keep pushing them into using statins.  I'll let folks know that they should ask for an NMR test that directly counts their LDL particles before jumping onto statins.

    BTW: Have you ever run into people getting higher fasting blood glucose readings taking fish oil?  It seems to happen to me.  I was taking a high dose (8g) for my arthritis and my FBG went to over 100 at times, despite a low carb, no grain diet.

  • Jonathan

    8/31/2008 11:10:00 AM |

    "She would not have benefitted from taking a statin drug."  Isn't it true that women have never been shown to benefit from statins?  If I understand correctly, statins have only been shown to reduce total mortality for middle-aged (not elderly) men with full-blown heart disease.

  • Isaac

    11/11/2008 3:07:00 PM |

    Is it really that the Friedewald calculation is wrong, or just that the two tests are measuring different things? Standard panels report cholesterol components in mg/dl, a mass per volume. The NMR test reports nmol/L, a particle count per volume. If the particles are large then there are fewer particles but perhaps more mass.

    This doesn't say anything about which one is more accurately describing cardiovascular risk, but I don't see why the Friedewald calculations are necessarily incorrect; they're just measuring things differently than a NMR test does.