How to use it 01 Enter fasting insulin and glucose. Glucose accepts mmol/L or mg/dL; the panel shows the converted pair so the unit can be checked before reading anything. 02 Read HOMA-IR first, then the band. The band is wide and population-dependent, and the panel's own note says the quoted 2.5 cut-off is indicative only. 03 Enter the lipid panel. Total cholesterol, HDL and triglycerides in the same unit system — the divisor changes with the unit, 5 for mg/dL and 2.2 for mmol/L. 04 Watch the triglyceride limit. At or above 400 mg/dL (4.52 mmol/L) the LDL card dims and an error-level warning says the number must not be used. 05 Read non-HDL alongside. It is total minus HDL, it does not depend on the triglyceride estimate, and the panel offers it as the robust alternative when triglycerides are high.
Worked readouts Fasting insulin 10 µU/mL with glucose 5.5 mmol/L (99.1 mg/dL); lipids 200 / 50 / 150 mg/dL:
Card
The panel reads
HOMA-IR
2.44 — upper-normal band (1.5–2.5)
HOMA-%B
100.0
QUICKI
0.3338
LDL-C (Friedewald)
120 mg/dL — valid
VLDL-C estimate
30 mg/dL (TG ÷ 5)
Non-HDL cholesterol
150 mg/dL
A more resistant profile — insulin 25 µU/mL, glucose 6.1 mmol/L — reads HOMA-IR 6.78 , the marked-resistance band, with HOMA-%B 192.3 and QUICKI 0.2908 . And the same lipid panel with triglycerides at 500 mg/dL makes the TG ÷ 5 term subtract 100 mg/dL of VLDL, so the formula prints an LDL of 50 mg/dL — a number the panel explicitly marks invalid, because at that triglyceride level the Friedewald estimate must not be used at all.
The formulas and what they assume HOMA-IR is insulin × glucose ÷ 22.5 when glucose is in mmol/L; it is a product, so both inputs must be fasting and the insulin assay must be comparable. HOMA-%B = 20 × insulin ÷ (glucose − 3.5) estimates beta-cell output on the same pair, and QUICKI = 1 ÷ (log₁₀ insulin + log₁₀ glucose in mg/dL) is a log transform of the same information. All three are surrogate indices, not measurements of what the pancreas or muscle actually does.
Friedewald's equation is simpler arithmetic with a stronger assumption: LDL = total − HDL − TG ÷ 5 (mg/dL), where TG ÷ 5 estimates VLDL cholesterol from triglycerides. The fixed ratio holds reasonably well at ordinary triglyceride levels but breaks as triglycerides rise and as LDL falls — a large direct-measurement study found the equation under-reading LDL by a median of 9 mg/dL when triglycerides were 150–199 and by 18.4 mg/dL at 200–399, in samples whose estimated LDL was below 70. The engine implements the two published limits: invalid at TG ≥ 400 mg/dL and a low-precision warning below LDL 70 mg/dL, pointing at the Martin-Hopkins equation.
Assumptions and limits
Insulin assays are not standardised. Absolute insulin values differ between laboratories, so there is no universal HOMA-IR cut-off; the panel keeps the frequently quoted 2.5 but labels it as indicative. Treat the index as a comparison within one laboratory and one protocol, not as a diagnosis.
Friedewald is an estimate with named failure modes. High triglycerides, non-fasting samples, type III hyperlipoproteinaemia and low LDL all break the fixed TG:VLDL ratio. The mmol/L form uses the same assumption with a divisor of 2.2. What it does not do: it does not diagnose insulin resistance or dyslipidaemia, does not decide treatment, and does not replace a directly measured LDL or a full risk assessment. For the cardiovascular risk scores use the Cardio panel ; for kidney and dosing arithmetic use the Renal panel .
One number, two units, three warnings The unit toggle is the first trap: the mmol/L card needs the converted inputs, and re-entering the usual rounded conversions — TC 5.2 , HDL 1.3 , TG 1.7 — reads LDL 3.13 , VLDL 0.77 and non-HDL 3.90 at divisor 2.2; the exact conversions of 200 / 50 / 150 mg/dL (5.172 / 1.293 / 1.694 ) read LDL ≈ 3.11 instead. Mixing a mg/dL triglyceride value into the mmol/L field understates LDL by a wide margin. The second trap is trusting a printed number past the validity limit: at TG 500 mg/dL the card still computes 50 mg/dL , but it is dimmed and carries an error warning because the fixed ratio no longer describes the sample. The third is the low-LDL zone: a panel of 120 / 50 / 100 returns LDL 50 mg/dL with a warning that Friedewald systematically under-reads below 70, which is precisely where high-risk decisions are made. The engine's habit is to keep showing the arithmetic and name its limits — read the limits.
Where it helps Comparing two fasting panels over time Run both dates through the same indices; HOMA-IR and QUICKI move in opposite directions for the same physiological change, which is a useful cross-check that the inputs were entered consistently.
Understanding a lab report The panel shows where VLDL, non-HDL and the TG/5 assumption come from, so a reported LDL can be traced back to the three measured values instead of arriving as an unexplained number.
Privacy Insulin, glucose and lipid values are computed in your browser; nothing is uploaded, and no result is stored beyond this device.
References
Matthews DR et al., Homeostasis model assessment , pubmed.ncbi.nlm.nih.gov (访问日期:2026-10-07)— the HOMA-IR and HOMA-%B derivation from fasting insulin and glucose.
Friedewald WT et al., Estimation of the concentration of low-density lipoprotein cholesterol in plasma , pubmed.ncbi.nlm.nih.gov (访问日期:2026-10-07)— the original TC − HDL − TG/5 equation.
Martin SS et al., Friedewald-estimated versus directly measured low-density lipoprotein cholesterol , pubmed.ncbi.nlm.nih.gov (访问日期:2026-10-07)— measured underestimation at low LDL and raised triglycerides.
Wikipedia, Homeostatic model assessment , en.wikipedia.org (访问日期:2026-10-07)— HOMA-IR and HOMA-%B as fasting surrogate indices.
Calculators in this hub
Hand-picked tools, one click away. The mini versions compute live and carry your values into the full calculator.
Sources & review
Reviewed by CalcX Editorial Team
Updated 2026-10-07