Why is LDL estimated rather than measured?
LDL cholesterol is the main target in cardiovascular prevention, but measuring it directly (by β-quantification, the ultracentrifugation reference method) is slow and expensive. So in clinical practice LDL is almost always calculated from a standard lipid panel: total cholesterol (TC), HDL and triglycerides (TG). Every estimate starts from the same idea: the cholesterol that is not HDL is split between LDL and very-low-density lipoproteins (VLDL).
The methods differ in how they estimate VLDL: Friedewald assumes a fixed fraction of TG, whereas Sampson-NIH and Martin-Hopkins model it in a personalized way.
Friedewald (1972)
The classic method assumes a fixed TG:VLDL ratio of 5 (in mg/dL):
It is simple and works reasonably with normal TG, but that fixed ratio of 5 stops being realistic as TG rise. That is why it is invalid when TG > 400 mg/dL and tends to underestimate LDL in patients with low LDL or high TG.
Reference: Friedewald, Levy & Fredrickson (1972), Clinical Chemistry
Bidirectional calculation: estimating TG from direct LDL
The calculator also works in reverse. If instead of triglycerides you have a directly measured LDL (from a homogeneous lab assay), you can switch to "I have direct LDL" mode and solve for triglycerides from Friedewald's own equation:
This is the algebraic inverse of Friedewald: once estimated, that TG feeds the rest of the models and indices. It is still an estimate, valid within Friedewald's regime (TG < 400 mg/dL). Since LDL can never exceed Non-HDL cholesterol, if the direct LDL approaches or exceeds it, the estimated TG would tend to zero or a negative value (impossible): in that case the calculator warns you instead of showing a meaningless result.
Sampson-NIH (2020)
It replaces the fixed ratio with a quadratic VLDL model that depends on both TG and Non-HDL, fitted against β-quantification:
It keeps its accuracy up to TG ≈ 800 mg/dL, making it the preferred method in hypertriglyceridemia and when LDL is low. It is the value this calculator highlights by default.
Reference: Sampson et al. (2020), JAMA Cardiology
Martin-Hopkins (2013)
It keeps Friedewald's form but replaces the 5 with an adjustable factor read from a 180-cell table based on the patient's TG and Non-HDL strata (the factor ranges roughly from 3.1 to 11.9):
It markedly improves accuracy at low LDL and moderately elevated TG compared to Friedewald, without measuring LDL directly.
Reference: Martin et al. (2013), JAMA
Non-HDL and VLDL
Non-HDL cholesterol bundles every atherogenic particle (LDL, VLDL, IDL, lipoprotein(a)) into a single number. Because it does not depend on TG, many guidelines prefer it as a secondary target, especially when TG are high.
Here VLDL is derived from the Sampson LDL, the most accurate base available.
Atherogenic indices
The Castelli indices relate total or LDL cholesterol to protective HDL. The TG/HDL ratio is an indirect marker of insulin resistance: above ~3.0 (in mg/dL) it suggests a predominance of small dense LDL particles (phenotype B), which are more atherogenic. Note that this ratio is unit-dependent, so the 3.0 threshold corresponds to mg/dL values.
Indicative reference ranges
- •Optimal LDL < 100 mg/dL (lower in high risk).
- •Desirable Non-HDL < 130 mg/dL.
- •Castelli I < 4.5 · Castelli II < 3.0.
- •TG/HDL < 3.0 (mg/dL).
These are indicative thresholds: the real LDL goal depends on each person's overall cardiovascular risk and must be set by a professional. This tool is educational and does not replace medical evaluation.