队列
医学
统计
可比性
糖尿病
金标准(测试)
数学
内科学
内分泌学
组合数学
作者
Nicolene Steyn,H. Muller Rossouw,Tahir S. Pillay,Janine Martins
标识
DOI:10.1016/j.cca.2022.10.003
摘要
The gold standard for measuring LDL-C is impractical for routine use while direct measurements have numerous shortcomings. This study investigated the correlation of predictive equations with currently used enzymatic assays in populations where these have historically proven unreliable to determine whether equations may be used interchangeably. No reference measure was available for comparison in this study.We examined two analytical datasets from different platforms to evaluate the correlation of predictive equations for LDL-C (Friedewald, Sampson-NIH2, Martin-Hopkins, Extended Martin-Hopkins, Hattori and Anandaraja) with direct LDL-C assays in a large paediatric (n = 7598) and an adult cohort with uncontrolled diabetes (n = 57165). Non-parametric statistics were used for comparison.In the paediatric cohort, the Sampson-NIH2 equation correlated best with the direct LDL-C assays with the most values falling within desirable bias (35.9-44%) and TEa (68.6-72.9%) and the lowest RMSE (0.5904-0.6138) across platforms, but tended to underestimate LDL-C levels. The Martin-Hopkins equation is less likely to underestimate these values. In diabetes, the Martin-Hopkins equation correlated the best with values falling within acceptable bias (40.2-50.5%) and TEa (75-80.6%). In hypertriglyceridaemia the Extended Martin-Hopkins equation correlates best with the direct LDL-C assays.Different measurement platforms influence the results of predictive equations and directly measured LDL-C. We propose utilising the Martin-Hopkins equation as an alternative to dLDL-C assays in adults with diabetes and for screening purposes in paediatric populations to avoid underestimating cardiovascular risk.
科研通智能强力驱动
Strongly Powered by AbleSci AI