TABLE 4.

Replication experiment to evaluate precisiona

Concentration (copies/ml)Log10 concentration% CV (total)SD95% CI (log viral load)Log change 95% CIFold change 95% CI
250,0005.391.930.1045.11-5.670.563.6
5,0003.691.750.1303.43-3.950.533.3
3002.486.910.1642.15-2.810.664.6
  • a Three concentrations of analyte spanning the reportable range were used to evaluate the precision of a quantitative assay for human cytomegalovirus. Samples at each concentration were tested in duplicate in one run per day over 20 days using two operators. Between-day, operator-to-operator, and total imprecision were evaluated using analysis of variance of log10-transformed data. The coefficient of variation (CV) and standard deviation (SD) were calculated for each concentration. Total imprecision (% CV) values for the three concentrations ranged from 1.75% to 6.91%, indicating that the assay is less precise at lower concentrations. Relatively little variance was attributable to between-day or operator-to-operator components except for the low-concentration sample, for which the day-to-day imprecision was much greater than with the other concentrations (data not shown). Calculation of 95% confidence intervals (CI) revealed that a 3- to 4-fold change at mid- to high viral loads and about a 5-fold change at viral loads near the limit of detection may represent imprecision of the assay rather than a true biological difference.