RESUMO
INTRODUCTION: Most estimates of biologic variation (sb ) are based on periodically acquiring and storing specimens from reference subjects, followed by analysis within a tightly controlled analytic run. We demonstrate that reliable estimates of sb can be derived for virtually all constituents of the CBC from previously obtained paired patient results and summary QC data. METHODS: A laboratory data repository provided all of the outpatient CBC results measured over 20.5 months at a large Canadian referral laboratory. These CBC measurements were taken on one of four Beckman Coulter LH analyzers. A total of 1852 different patients had CBCs repeated at least twice within 84 h. We tabulated the pairs of intrapatient constituents that were separated by 0-6, 6-12, 12-18, 72-78, and 78-84 h. The standard deviations of duplicates (SDD) of the paired data were then regressed against time. The y-intercept represents the sum of sb and short-term analytic variation (sa ): y0 =(s(2) a +s(2) b )(1/2) . The short-term imprecision was determined from normal range Coulter quality control specimens. RESULTS: Patient sb for hematocrit, MCH, absolute monocytes, and absolute neutrophils are extremely close to those determined by biologic variation experiments using healthy volunteers. Most of the other estimates of sb tended to be slightly lower than literature estimates. CONCLUSIONS: We describe a novel approach to deriving sb . The ratio of the sb to sa (a measure of sigma) indicates that the Beckman Coulter LH is extremely suitable for CBC monitoring of outpatients as well as for inpatients, whose sb is generally higher.
