8127/frame/fm

 
i.
Standards and controls.
Any reviewer should check that all controls
have performed as expected, and that no spurious results were obtained. This
8127/frame/ch09 Page 236 Friday, July 21, 2000 11:44 AM

---

Good Laboratory Practice — Establishing Validity and Reliability
237
includes verifying the results from known samples and confirming that neg-
ative controls have failed to give reportable results.
 
ii.
Second blind read.
The reviewer should verify that no significant dis-
crepancies exist between the results recorded by the primary analyst and a
second reader. This assumes that the second reader has independently read
and recorded the results rather than simply signing-off on the primary analyst’s
read. Minor subjective differences that do not alter the interpretation or
conclusions would generally not be considered significant.
Although many DNA laboratories have now incorporated blind second
reads as part of their standard protocol, other disciplines have been slower
to jump on this particular bandwagon. The absence of a second intralabo-
ratory blind read exacerbates the need for the outside reviewer to have access
to the original data. If weak or ambiguous data are relied upon to form a
conclusion, lack of verification by a second qualified analyst would weaken
any conclusions drawn from those data.
 
e.
Calculations
Calculations are perhaps the most fertile ground in a report for errors.
Nobody likes to recheck numbers and, even if the rest of the report has been
critically reviewed, the last residuum for errors may be in the calculations.
The reviewer should, if possible, check the source of any numbers used in
the calculation and independently verify the calculations themselves.
One of our students brought into class a report from a private DNA labo-
ratory as part of a presentation (Figure 9.2). The work had been performed
for the prosecution. The student, one of the attorneys for the case, put a
copy of the page detailing the population statistics on an overhead projector.
It was immediately obvious to us that each alternating “total” row was a
cut-and-paste replica of the one just before it and had no relationship
whatsoever to the actual sum of the numbers in the column. The final
calculations at the bottom of each column were, in fact, correct, but could
not be derived from the intermediate sums. Not only did the laboratory
technical reviewer (if there was one) fail to catch the error, so did anyone
from the defense team. The trial had long since concluded with a conviction
of the defendant and apparently no one the wiser.