Validation of the Friedewald formula for the estimation of low density lipoprotein cholesterol in a sub-Saharan African population
Tasha Manases1,2,&, Andre-Pascal Kengne3, Jean Pierre Ndamefo4,5, Yannick Mboue-Djieka2, Christian Dimala2, Simeon-Pierre Choukem1,2,
1Departement of Internal Medicine and Pediatrics, Faculty of Health Sciences, University of Buea, Buea, Cameroon, 2Health and Human Development (2HD) Research Network, Douala, Cameroon, 3South African Medicine Research Council and University of Cape Town, Cape Town, South Africa, 4Biochemistry Unit, Douala General Hospital Laboratory, Douala, Cameroon, 5Department of Biomedical Sciences, Faculty of medicine and pharmaceutical Sciences, University of Douala, Douala, Cameroon
Tasha Manases, Department of Internal Medicine and Pediatrics, Faculty of Health Sciences, University of Buea, Buea, Health and Human Development (2HD) Research Network, Douala, Cameroon
low density lipoprotein cholesterol (LDL-C) levels are used to estimate cardiovascular disease (CVD) risk and to guide prescriptions. To circumvent the challenges of direct LDL-C measurement, guidelines recommend the use of Friedewald formula derived LDL-C levels. Despite reported limitations of this formula, its validity in sub-Saharan Africans has not been adequately investigated. The objective was to assess the validity of the Friedewald formula derived against directly measured LDL-C in adult Cameroonians.
we reviewed the fasting lipid profiles of 2500 patients, performed between March 2012 and January 2016 using enzymatic colorimetric method (reference), at the Douala General Hospital laboratory. The Friedewald formula was used to calculate LDL-C from total cholesterol, high density lipoprotein cholesterol and triglyceride levels. Calculated LDL-C values were compared to the reference values and clinical significance of differences between the two methods was assessed using total error allowable (TEa).
the difference between means of Friedewald-calculated and the reference LDL-C
values was neither statistically nor clinically significant (3.33±1.51 vs. 3.33 ± 1.25
mmol/l; p = 0.704). The calculated LDL-C correlated positively with the measured
LDL-C value (r = 0.749) and both methods showed a good agreement on Bland-Altman
plot. Conversely, there was only moderate agreement (kappa = 0.478, 95% CI:
0.455-0.502) between the two values in the stratification of cardiovascular
to the National Cholesterol Education Programme/Adult Treatment Panel III.
Consequently, 40.6% of the participants were misclassified.
Friedewald formula is technically accurate but has a modest clinical accuracy which can translate into a substantial misclassification of patients’ cardiovascular risk and subsequent inappropriate therapeutic decisions.