distribution of log residue concentrations at each time. Estimates are made of
between-animal variance and of measurement error variability, which are subse-
quently used to calculate a statistical tolerance limit at each time point. The
estimated withdrawal period is the first time where the upper 95% confidence
limit of the 99% of residue concentrations is at or below the MRL (Fig. 4 ).
The US FDA approach requires that a minimum of 20 animals are studied and
milk samples are analysed in triplicate. For veterinary drug products used to treat
mastitis, the method assumes that no more than one third of the milk is derived from
treated animals, which is reflected in the calculation of the daily intake of milk by
consumers by incorporating a factor of 1/3. The approach uses only those points in
the final phase of the depletion curves when fitting the regression lines. Measure-
ments reported as “below the LOQ” (limit of quantification), and data from time
points with less than three remaining values are also excluded.
The time-to-safe-concentration (TTSC) method (EMEA-CVMP, 2000 ) is pre-
sented in the “Note for guidance for the determination of withdrawal periods for
milk” (EMEA/CVMP/473/98-Final). It is recommended as the harmonised method
for calculating milk withdrawal periods for new veterinary medicinal products in
the EU. The use of alternative statistical approaches may be justified when a dataset
is not suitable for analysis using the TTSC method. The latter assumes a log-normal
distribution of individual times to safe concentration, and calculates the upper 95%
confidence limit of the 95% of individual times to comply with MRL (i.e. the 95/95
statistical tolerance limit). The withdrawal period is calculated as the 95/95 toler-
ance limit (Fig. 4 ).
The US FDA approach and the TTSC approach are fundamentally different. In
particular, the distributional assumption of the US FDA method and the TTSC
Log marker residue concentrationMilk WPLog marker residue concentrationMRLTTSCTime after last treatment Time after last treatmentLOQ LOQStatistical tolerance limitLinear regression
Statistical tolerance limit
ToleranceFig. 4Schematic diagram contrasting the US FDA approach (left panel) and the EMEA approach
(right panel) for the calculation of a milk withdrawal period (WP). The US FDA approach
calculates the withdrawal period as the first time where the upper 95% confidence limit of the
99% of residue concentrations is at or below the tolerance. The EMEA approach calculates the
milk WP (also known as the time-to-safe-concentration [TTSC]) as the first time where the upper
95% confidence limit of the 95% of individual times complies with the MRL
282 P.T. Reeves