Summary
The fundamental variation to be taken into consideration in comparing clotting time
averages is that of the clotting time assay itself. A variation is ετ whose true variance is στ
2 and the experimental estimate of this variance sτ
2. Described is a clotting time machine automated with respect to temperature, thrombin
volume, thrombin injection, stirring and time determination so that the only subjective
observation is that of the clotting time endpoint. Statistical analysis of sτ
2 has been made using protocols in which measurements either were carried out in a
particular time sequence (regression type protocol) or were randomized (randomization
type). An experimental unit is a dilute thrombin solution, a dilute fibrinogen solution
and an operator-machine. The dilute reagents come from stock reagents. Characteristics
of the system are as follows. 1. For a wide variety of experimental units the ετ are time independent and normally distributed. The sτ
2 from single lots of stock reagents appear to be estimates of the same στ
2. If different stock reagents are used the measured sτ
2 show that στ
2 may change significantly. 2. The variance increases with the clotting time. An approximate
square relationship has been used for normalization. 3. The operator-machine and dilute
solutions of stock thrombin appear to be remarkably uniform. 4. A change in στ
2 between experimental units is due to fibrinogen and appears particularly when different
lots of fibrinogen, Armour’s Fraction I, are used to prepare the purified stock reagent.
5. When time dependent changes occur in experimental units, they are also most likely
due to fibrinogen instability. When the storage temperature of stock-F is decreased
from —20° to —95°, fibrinogen solubility is stabilized and regression decreased. 6.
When instabilities can be made small, a randomization type protocol is preferred since
fewer clotting times are required to characterize an experimental unit. When instability
is important, a number of assays must be invested in its detection. This number increases
approximately linearly with sτ
2 and decreases linearly as the time period for assays increases. The most important
limitation to the examination of regressions is the number of clotting times an observer
can perform on a single day.
