INTERPRETATION OF BIOASSAY RESULTS
Bioassay measurements detecting plutonium or americium in workers can be initially
interpreted as indicating that occupational intakes may have occurred. Standard bioassay
procedures are not sufficiently sensitive to detect the worldwide environmental background
levels in an in vivo or in excreta. Since most plutonium and americium bioassay
measurement procedures include counting for radioactivity as the final step in the
measurement process, they are subject to the statistics associated with the counting process.
Two key questions associated with bioassay data are (1) When does a sample result
indicate the presence of something (i.e., when is the analyte detected)? and (2) What is the
overall capability of the bioassay method for continual assurance of detection of the
The decision level, Lc (also called the critical level for detection), is the level for a given
measurement that indicates the likely presence of the analyte. The Lc is dependent on the
probability of obtaining false positive results (type I, or alpha, error) that is acceptable to
the program. A 5% probability of false-positive results is a common design parameter of
measurement programs, implying that for a large number of measurements, 5% of the time
results will be indicated as positive when in fact there is no activity present. The Lc is
calculated from results of analyses of blank samples. Once a measurement is performed, it
is appropriate to compare it with the Lc to determine whether or not the result is "positive"
(i.e., the analyte is detected).
The MDA is the level at which continued assurance of detection can be provided. The
MDA is a function of the probabilities of both false positive and false negative (type II, or
beta) errors and is typically based on a 5% probability for each kind of error. The MDA is
also determined from analysis of blank samples, but is substantially higher than the Lc. The
MDA is appropriate for use in designing bioassay programs and as the basis for estimating
minimum detectable intakes and doses as indicators of program sensitivity. The MDA
should not be used as a comparison with actual measurements to determine whether or not
activity is present (i.e., <MDA is not an appropriate use of the concept).
Methods for calculating both Lc and MDA are given in HPS N13.30. (HPS, 1996).
As an alternative to the Lc and MDA of classical statistics, Miller et al. (1993) propose the
use of Bayesian statistical methods for evaluating bioassay data.
General follow-up actions to abnormal bioassay measurements should include data checks,
timely verification measurements, work history reviews, and performance of special in vivo
measurements or excreta sample analyses for intake and dose assessments.