Skip to main content
Have a personal or library account? Click to login
Accuracy of 23 Equations for Estimating LDL Cholesterol in a Clinical Laboratory Database of 5,051,467 Patients Cover

Accuracy of 23 Equations for Estimating LDL Cholesterol in a Clinical Laboratory Database of 5,051,467 Patients

Open Access
|Jun 2023

Figures & Tables

Figure 1

Diagram of Primary and Secondary Testing to Select the Highest Performing Equation. The figure shows the sequence of analysis, first comparing all 23 equations against ultracentrifugation for concordance in guideline-based LDL-C classification and overall magnitude of error (mg/dL units). In secondary testing, we compared Friedewald and the five equations that performed better than Friedewald in primary testing. These secondary tests evaluated performance at different levels of elevated triglycerides, by age (<18, 18–59, >60 years), sex, and fasting status strata, and across clinical subgroups (ASCVD, DM, HTN, CKD, abnormal TSH, high hsCRP). Finally, we assessed the top performing equation (Martin/Hopkins) for its impact in classification across an important clinical cutpoint in high risk patients (LDL-C 70 mg/dL) if a laboratory were to switch from the Friedewald equation. Patients highlighted in orange are ones with Friedewald LDL-C<70 mg/dL who have correct (confirmed by ultracentrifugation) upward reclassification to LDL-C >70 mg/dL by Martin-Hopkins. Patients highlighted in blue remain classified as LDL-C <70 mg/dL by both equations. Abbreviations: TGs = triglycerides; ASCVD = atherosclerotic cardiovascular disease; DM = diabetes mellitus; HTN = hypertension; CKD = chronic kidney disease; TSH = thyroid stimulating hormone; hsCRP = high-sensitivity C-reactive protein; LDL-C = low-density lipoprotein cholesterol.

Table 1

Demographic characteristics.

CHARACTERISTICSSTUDY POPULATION (N = 4,939,528)
Age, mean (SD), y56 (16)
Age category, no. (%), y
    <1859,838 (1.2)
    18–592,879,747 (58.3)
    ≥601,966,468 (39.8)
Not reported33,475 (0.7)
Sex, no. (%)
    Women2,635,486 (53.7)
    Men2,269,406 (46.3)
Fasting status, no. (%)
    Non-fasting586,256 (11.9)
    Fasting958,989 (19.4)
    Not reported3,394,283 (68.7)
Lipid values, median (IQR)
    TC, mg/dL193 (164–225)
    TG, mg/dL114 (81–164)
    HDL-C, mg/dL51 (42–63)
    LDL-C, mg/dL (ultracentrifugation)114 (90–141)
    Non-HDL-C, mg/dL138 (112–168)
    VLDL-C, mg/dL22 (17–29)
    Lp(a)-C, mg/dL6 (4–10)
    TC:VLDL-C ratio8.4 (6.5–10.9)
    TG:VLDL-C ratio5.0 (4.4–5.9)
    TG:TC ratio0.6 (0.4–0.9)

[i] IQR = interquartile range; TC = total cholesterol; HDL-C = high-density lipoprotein cholesterol; LDL-C = low-density lipoprotein cholesterol; VLDL-C = very low-density lipoprotein cholesterol; Lp(a)-C = lipoprotein(a) cholesterol.

Table 2

Percentage of patients correctly classified to LDL-C category.

EQUATIONLDL-C CATEGORY, MG/DL
<4040–5455–6970–99100–129130–159160–189≥190OVERALL
Martin/Hopkins78.681.284.291.790.889.286.691.189.6
Sampson66.071.877.890.589.085.480.584.486.3
Chen85.981.180.286.084.383.281.795.784.4
Puavilai62.569.775.089.387.283.177.079.284.1
DeLong66.872.376.289.686.481.474.476.883.3
Friedewald42.152.562.683.887.087.786.190.083.2
Molavi44.654.263.082.885.687.287.695.282.9
Saiedullah72.265.866.081.183.685.886.695.482.1
Vujovic75.776.377.089.283.676.668.071.280.3
Teerakanchana83.570.867.083.878.476.675.783.878.5
Orejon71.567.868.085.380.476.171.176.478.5
Dansethakul52.158.162.182.777.972.967.173.375.2
Bauer87.277.673.686.577.568.357.863.273.8
Rao42.354.261.882.776.467.356.462.771.0
Ghasemi21.628.938.868.075.580.583.397.869.1
Ephraim93.074.367.083.071.761.149.857.367.5
Hattori33.938.143.567.770.973.173.999.167.0
Lee and Hu81.463.457.874.066.063.461.581.966.9
Rasouli74.965.462.368.159.848.629.099.560.3
Cordova86.864.554.662.656.650.141.196.757.5
Anandaraja29.936.043.367.659.452.145.356.555.9
Choi88.356.144.473.057.946.736.249.253.4
Ahmadi27.526.031.858.749.134.818.816.535.1
Table 3

Error between estimated LDL-C and VAP ultracentrifugation LDL-C.

EQUATIONERROR, MEDIAN (IQR), MG/DLRELATIVE ERROR, MEDIAN (IQR), %
Friedewald–0.2 (–4.4 to 2.6)–0.2 (–4.0 to 2.2)
Martin/Hopkins0.3 (–1.6 to 2.4)0.2 (–1.4 to 2.1)
Sampson1.7 (–1.3 to 4.1)1.6 (–1.2 to 3.5)
Puavilai3.3 (0.5 to 5.5)2.9 (0.4 to 4.9)
Vujovic5.5 (3.2 to 7.7)4.8 (2.8 to 6.9)
Hattori–7.6 (–11.8 to –4.6)–6.4 (–10.0 to –4.1)
Anandaraja5.9 (–4.3 to 17.5)5.1 (–3.6 to 16.1)
Chen–2.9 (–5.6 to –0.2)–2.6 (–4.4 to –0.3)
Cordova–10.8 (–17.2 to –4.5)–9.8 (–13.4 to –4.7)
Teerakanchana5.2 (1.7 to 8.9)4.6 (1.3 to 8.7)
Ahmadi18.7 (–2.6 to 50.6)16.4 (–2.3 to 44.6)
DeLong4.1 (1.4 to 6.2)3.5 (1.2 to 5.5)
Rao7.7 (2.3 to 11.6)6.8 (2.1 to 9.9)
Ephraim9.4 (7.3 to 12.2)8.4 (6.4 to 11.1)
Saiedullah–3.9 (–6.0 to –1.5)–3.4 (–5.3 to –1.3)
Dansethakul6.9 (2.7 to 9.7)5.8 (2.2 to 8.8)
Rasouli–10.1 (–16.6 to –4.0)–8.8 (–12.3 to –4.3)
Ghasemi–5.8 (–12.0 to –1.8)–5.1 (–11.0 to –1.6)
Lee and Hu3.8 (–3.8 to 12.0)3.4 (–3.0 to 12.0)
Orejon6.2 (3.4 to 8.3)5.3 (2.7 to 8.0)
Bauer7.6 (5.6 to 10.1)6.8 (4.9 to 9.1)
Molavi–1.1 (–5.2 to 1.7)–0.9 (–4.5 to 1.5)
Choi14.5 (11.8 to 17.2)12.6 (9.8 to 16.0)
Figure 2

Upward Reclassification of Patients with Friedewald LDL-C < 70 mg/dL When Using an Alternative LDL-C Equation. The figure displays the percentage of patients with Friedewald LDL-C < 70 mg/dL who are reclassified to LDL-C > 70 mg/dL by top performing equations and confirmed to have a correct reclassification by ultracentrifugation. Reclassification for patients with TG levels of <400 mg/dL (n = 539,575) is highlighted in blue and for patients with TG levels of 150–399 mg/dL (n = 183,455) is highlighted in dark blue. Abbreviations: LDL-C = low-density lipoprotein cholesterol; TG = triglyceride.

DOI: https://doi.org/10.5334/gh.1214 | Journal eISSN: 2211-8179
Language: English
Submitted on: Dec 15, 2022
Accepted on: May 26, 2023
Published on: Jun 19, 2023
Published by: Ubiquity Press
In partnership with: Paradigm Publishing Services
Publication frequency: 1 issue per year

© 2023 Christeen Samuel, Jihwan Park, Aparna Sajja, Erin D. Michos, Roger S. Blumenthal, Steven R. Jones, Seth S. Martin, published by Ubiquity Press
This work is licensed under the Creative Commons Attribution 4.0 License.