Long-term intraindividual variability of serum lipids in patients with type I and type II diabetes

The objective of this study was to estimate the long-term intraindividual variability of lipid levels in adult type I and type II diabetic patients. Total cholesterol, triglycerides, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and hemoglobin A1c were measured every 3-6 months in 135 patients attending the Austin Hospital diabetes clinic. Analysis was performed on 60 diabetic patients (33 type I and 27 type II) who had not been treated with lipid lowering drugs and who met the inclusion criteria of at least five measurements [mean +/- standard error of the mean (SEM), 9.5 +/- 0.4; range, 5-17] collected over a minimum of 4 years (5.1 +/- 0.1; 4-6.5 years). Total variability, expressed as coefficient of variation, was 8.8 +/- 0.4% for total cholesterol, 23.9 +/- 1.5% for triglycerides, 10.2 +/- 0.5% for HDL cholesterol, and 12.0 +/- 0.5% for LDL cholesterol. Biological variability, derived from total and analytical variability, was higher than previous estimates in nondiabetic subjects for total cholesterol and HDL cholesterol but similar for triglycerides and LDL cholesterol. No relationship was observed between total lipid variability and diabetes type, age, baseline or mean lipid levels, duration of follow-up, or the number of samples per patient. Men demonstrated greater variability than women for total cholesterol (men 9.5 +/- 0.5%, n = 34, women 7.9 +/- 0.5%, n = 26, p < 0.01) and triglycerides (men 26.5 +/- 2.2%, women 20.4 +/- 1.4%, p = 0.03). Total lipid variability was also unrelated to baseline or mean hemoglobin A1c or to the change in hemoglobin A1c during the study as a whole. However, the change in hemoglobin A1c was associated with the change in total cholesterol (r = 0.30, p < 0.03) and the change in LDL cholesterol (r = 0.27, p < 0.05). In conclusion, long-term intraindividual lipid variability in adult diabetic subjects is higher for total and HDL cholesterol than previously published values in nondiabetic subjects. Variability of triglycerides is at least double that of total cholesterol, HDL cholesterol and LDL cholesterol. Biological variability, not measurement error, accounts for the greatest proportion of total variability for all lipid parameters. Confidence levels calculated from these data have implications for the initiation of lipid lowering therapy and in monitoring the effects of intervention.