Irwig L, Glasziou P, Wilson A, Macaskill P
Department of Public Health, University of Sydney, New South Wales, Australia.
JAMA. 1991 Sep 25;266(12):1678-85.
An individual's blood cholesterol measurement may differ from the true level because of short-term biological and technical measurement variability. Using data on the within-individual and population variance of serum cholesterol, we addressed the following clinical concerns: Given a cholesterol measurement, what is the individual's likely true level? The confidence interval for the true level is wide and asymmetrical around extreme measurements because of regression to the mean. Of particular concern is the misclassification of people with a screening measurement below 5.2 mmol/L who may be advised that their cholesterol level is "desirable" when their true level warrants further action To what extent does blood cholesterol change in response to an intervention? In general, confidence intervals are too wide to allow decision making and patient feedback about an individual's cholesterol response to a dietary intervention, even with multiple measurements. If no change is observed in an individual's cholesterol value based on three measurements before and three after dietary intervention, the 80% confidence interval ranges from a true increase of 4% to a true decrease of 9%.
由于短期生物学和技术测量变异性，个体的血液胆固醇测量值可能与真实水平不同。利用血清胆固醇个体内和群体方差的数据，我们解决了以下临床问题：给定一个胆固醇测量值，个体的真实水平可能是多少？由于均值回归，真实水平的置信区间在极端测量值周围很宽且不对称。特别令人担忧的是，筛查测量值低于5.2 mmol/L的人群可能会被告知其胆固醇水平“理想”，而实际上他们的真实水平需要进一步干预。血液胆固醇对干预的反应会有多大变化？一般来说，即使进行多次测量，置信区间也太宽，无法就个体胆固醇对饮食干预的反应进行决策和向患者反馈。如果在饮食干预前后分别进行三次测量，未观察到个体胆固醇值有变化，那么80%的置信区间范围是真实增加4%到真实降低9%。
