In 1948 an ambitious, long-term prospective study started in the small town of Framingham, Massachusetts1. The goal was to follow a large group of adults over many years to study the evolution of hypertension and cardiovascular disease and in so doing, identify risk factors and prognostic markers in subjects who went on to develop either of these conditions.

The subjects had several annual tests performed, but by the 1960s the correlation of total serum cholesterol with cardiovascular disease was evident. Soon the roles of low-density lipoprotein (LDL) and high-density lipoprotein (HDL) were discovered.

More than 50 years later, virtually all adults in the United States have heard of “good cholesterol” (HDL) and “bad cholesterol” (LDL), and many know their own levels of each. Clinicians rely heavily on the values reported by the laboratory to guide recommended lifestyle changes versus more aggressive therapy. Lipoprotein analysis is a high-volume, commonly performed test, yet for a majority of specimens, laboratories still use a formula first published in Clinical Chemistry in 19722.

Early analysis of lipoprotein fractions was performed by electrophoresis or ultracentrifugation. The terms LDL and HDL come from their relative density in the ultracentrifuge. Performing this type of analysis on the limited number of specimens from the Framingham study was possible, but neither electrophoresis nor ultracentrifugation was suitable for mass screening.

Enzymatic analysis of serum total cholesterol, triglyceride and HDL were comparatively easy, so the team of Friedewald, Levy, and Fredrickson developed a simple formula to estimate LDL using this data. The three lipid particles that carry virtually all the cholesterol in blood are LDL, HDL, and VLDL, so subtracting HDL and VLDL cholesterol from total cholesterol should yield LDL cholesterol. VLDL is decently estimated as 1/5 of triglyceride in mg/dL, so the Friedewald equation was born (LDL = Total cholesterol – HDL cholesterol – Triglyceride/5).

The Friedewald equation has been a cost-effective and valuable tool over many decades to estimate LDL. That said, it is an estimate and not a direct measurement, so there are many situations in which the estimation is inaccurate. Of particular importance is inaccuracy due to extremes in triglyceride measurement. For example, the higher the triglyceride, the more likely that LDL will be underestimated.

It has long been the consensus that the LDL estimate is inaccurate when triglyceride exceeds 400 mg/dL, but the underestimation of LDL begins at triglyceride levels as low as 150 mg/dL3. Thus, the Friedewald equation works best in relatively normolipidemic patients, not hyperlipidemic patients who are most likely to need therapy monitoring.

Why does an estimate that worked well in 1972 and for many years thereafter need to be replaced? The short answer is statin and PCSK-9 inhibitor therapies for hypercholesterolemia. These drugs are extremely effective at lowering LDL cholesterol, and clinicians are heavily influenced by LDL measurements when initiating and adjusting drug therapy. Since the drugs have undesirable side effects, inappropriate use of them is to be avoided, and the need for the extremely accurate LDL measurement is acute. Direct LDL measurements offer better accuracy but are costlier than a “free” calculation.

Enter the Martin-Hopkins formula.4 Dr. Seth Martin, based at Johns Hopkins, is first author on a paper detailing analysis of lipid specimens from 2,129 children; 8,165 adolescents; and 1,340,614 adults. Cholesterol content of lipid fractions was determined after ultracentrifugation, and triglycerides were directly measured. These values were considered “true,” and the team subsequently developed a modification of the Friedewald equation that provides a more accurate estimate of LDL.

The Martin team found that the level of VLDL cholesterol varies primarily with triglyceride and non-HDL levels. Although Friedewald’s original VLDL estimate of Triglyceride/5 is still valid when the total population is considered, the estimate is better if the denominator is customized for each patient’s levels.

The team constructed a matrix that allows the user to select a denominator which more accurately calculates LDL. Instead of dividing all triglyceride values by 5, one of 180 different options is selected based on total triglyceride and non-HDL levels. Online you can access the matrix and a free LDL Calculator.

The new formula is also more reliable in non-fasting patients because it adjusts for triglyceride level, mitigating effects from patient non-compliance. Implementing the new formula on a large scale will initially be costly for the re-programming of laboratory information systems. After this initial expense, however, the LDL cost will again be “free,” minimizing the need for direct LDL measurements.

The time has come to modernize the “free” LDL estimate and improve prevention and care for cardiovascular and cerebrovascular diseases, the leading cause of death in the United States.


  1. www.framinghamheartstudy.org, accessed 22 November 2017.
  2. Friedewald, W.T., Levy, R.I., and Frederickson, D.S. 1972. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clinical Chemistry 18: 499-502.
  3. Martin, S.S., et al. 2013. Friedewald-estimated versus directly measured low-density lipoprotein cholesterol and treatment implication. Journal of the American College of Cardiology 62:8 732-739.
  4. Martin, S.S., et al. 2013. Comparison of a Novel Method vs the Friedewald Equation for Estimating Low-Density Lipoprotein Cholesterol Levels From the Standard Lipid Profile. Journal of the American Medical Association. 310(19): 2061–2068.