Validation of the friedewald and martin-hopkins equations against direct LDL-C measurement in Brazilian adolescents

Background Dyslipidemia in adolescents poses significant cardiovascular risks. Direct measurement is optimal, yet cost-effective alternatives are necessary for broader application. Standard predictive formulas, such as the Friedewald formula, frequently underestimate LDL-C in patients with high triglycerides. The Martin-Hopkins equation provides greater accuracy across a range of triglyceride levels. Nonetheless, research on lipid outcomes in children and adolescents remains limited. Purpose We aimed to validate LDL-C estimation using the Friedewald and Martin–Hopkins formulas in a large cohort of Brazilian adolescents. Methods We used data from the São Luís birth cohort, initiated in 1997–1998, with a systematic hospital-based sample (1 in 7 births), totaling 2,493 live births across 10 maternity services. At ages 18–19 (2016–2017), 687 participants from the original cohort were evaluated, and a retrospective expansion added 1,828 adolescents born in 1997 (randomly sampled from the Brazilian National Birth Registry System and also recruited via schools/universities and social media), yielding 2,515 participants overall. Lipid profile parameters were assessed, and LDL-C was estimated using the Friedewald and Martin–Hopkins formulas. These indirect estimates were validated against direct LDL-C measurements, and incorrect and missed dyslipidemia diagnoses were quantified. Agreement between methods was evaluated using Bland–Altman analysis. Linear association was assessed with Pearson correlation. Mean differences between directly measured and estimated LDL-C were assessed using two-sided paired Student’s t-tests with a 95% confidence interval (95%CI). Diagnostic performance for high LDL-C (≥110 mg/dL) was evaluated by calculating sensitivity, specificity, and accuracy for LDL-C classification. Results After excluding 238 participants with missing lipid profile data, we included 2,277 adolescents (aged 18-19 years), of whom 51.93% were male. Mean directly measured LDL-C was 89.71 ± 26.27 mg/dL. The mean LDL-C was 89.96 ± 26.61 mg/dL for the Friedewald formula and 90.00 ± 26.84 mg/dL for the Martin–Hopkins formula. The mean difference between Friedewald-estimated and directly measured LDL-C was 0.22 mg/dL (95%CI: −0.07 to 0.51), and 0.26 mg/dL (95% CI: −0.07 to 0.59) for Martin–Hopkins, indicating minimal bias. The accuracy of the Friedewald formula was 98.77%, with a sensitivity of 96.92% and a specificity 99.23%. For the Martin-Hopkins formula, the accuracy was 96.75%, the sensitivity of 94.50%, and the specificity of 97.31%. No statistical difference was observed between the real LDL-C and Friedewald formula (p=0.140) and Martin-Hopkins formula (p=0.127). Conclusion Both formulas showed minimal bias and high diagnostic performance compared with direct LDL-C measurement. Calculated LDL-C represents a reliable and cost-effective strategy for cardiovascular risk screening in adolescents.