A-232 Impact of Estimated Versus Direct Measured LDL-Cholesterol on Risk Group Categorization for Patients with Elevated Triglycerides

Background Use of the Friedwald equation for estimated LDL-Cholesterol (LDL-C) has been widespread for more than 4 decades. Since the original publication, it has been known that the Friedwald equation may not be accurate for patients with very high triglycerides (Tg) (>400 mg/dL) or very low calculated LDL-C (<70 mg/dL). Equations which are more accurate for estimation of LDL-C in patients with very high triglycerides (“Martin-Hopkins” and “Sampson-NIH” equations) have recently been published. To determine the potential impact of utilizing an updated equation as opposed to direct LDL-C measurement, we compared the direct-LDL-C measurements performed under our reflex algorithm to the estimated values by Martin or Sampson equations, and quantified the number of patients classified into low, medium, or high-risk groups based on LDL-C. Methods Results of lipid panels and direct LDL-C (Roche Cobas 8000 series) which resulted in a direct LDL-C measurement based on the reflex algorithm (Tg >400mg/dL, or Tg between 150 and 399mg/dL with LDL-C <70mg/dL, or Tg between 200 and 399 mg/dL and LDL-C between 70 and 99mg/dL reflex to direct LDL-C) were obtained from the medical record for a one month period (n=2513). Martin and Sampson equations were used to estimate LDL-C. Number of patients categorized into low (<100 mg/dL), medium (100-159 mg/dL) or high (=160 mg/dL) risk categories based on LDL-C by direct measurement or Martin or Sampson equation were compared by chi-square test (p<.05 considered significant). Post-hoc analyses comparing number classified as low vs. medium or medium vs. high by each equation compared to direct measurement were performed by chi-square test with Bonferroni correction for multiple comparisons. Results Number of patients classified into low, medium, and high-risk categories based on direct measurement, Martin, or Sampson equations varied significantly (p<.001). Martin equation demonstrated a tendency to categorize more patients at medium risk (597 patients, 24% of all results) versus low risk (1840, 73%) compared with direct measurement (medium risk: 450, 18%, low risk: 2012, 80%; p<.001). Number of patients categorized as high-risk (76, 3%) by Martin were not significantly different from the high-risk group identified by direct measurement (51, 2%; p=1). Sampson equation categorized fewer patients into moderate risk (369, 15%) than direct measurement (p=.008); number categorized high risk by Sampson (43, 2%) was again not significantly different from categorization by direct measurement (p=1). Conclusion In a patient population with triglyceride and/or LDL-C values associated with unacceptable biases in the Friedwald equation, there were significant differences in classification into low versus medium-risk groups based on LDL-C measured by direct method compared to either Martin or Sampson equations. Martin-Hopkins demonstrated a tendency to classify patients into higher risk groups than direct measurement, returning values of =100mg/dL for 53 patients who were reported as <100mg/dL by direct measurement. Sampson-NIH demonstrated the opposite trend, with 42 patients reported as =100mg/dL by direct LDL-C but <100mg/dL by Sampson-NIH calculation. Further study is required to determine if classification by either calculation or direct measurement is more effective at predicting long-term cardiovascular risk.