Predicted effects of hemoglobin A1c assay precision on a patient population distribution of serial hemoglobin A1c difference values

Background: Interpretation of serial measurements of % hemoglobin A1c includes an assessment of differences from preceding values (DHbA1c). We examined predicted effects of different assay precisions on an observed population distribution for DHbA1c. Methods: Primary data were 5260 DHbA1c values from sequential HbA1c measurement pairs obtained within 1 calendar year. Each DHbA1c was replaced by a distribution obtained from sampling each component HbA1c value according to a normal distribution characterized by a fixed coefficient of variation (CV) of either 1%, 3% or 5% (forming data sets A, B and C, respectively). Data sets B and C, with inferior precision, were compared with the reference data set A (highest precision). Results: Using DHbA1c bin widths of 0.5% HbA1c, differences in assay precision caused significant redistribution of numbers within bins. For instance, for CV=5%, there was a 7.2% decrease in the number of results within the DHbA1c bin=(-0.5 to <= 0.0)% compared with the number for CV=1%, and a 6.4% increase in numbers of results for DHbA1c > 0.5. Conclusion: Different HbA1c assay CVs can significantly affect the fraction of patients within different clinical categorizations for DHbA1c and consequently may differently influence patient care recommendations. (c) 2006 Elsevier B.V. All rights reserved.