Caution Needed When Assessing Impact of Dosing Errors in Clinical Trials
NEW ORLEANS, LA -- When addressing the issues of cause and
effect in clinical research, particularly when dosing errors
are involved, care must be taken to ensure that potential
confounding issues also are considered, according to Duke
University Medical Center researchers.
At issue is a group of drugs known as "clot-busters," which
are used by physicians to dissolve life-threatening blockages
in arteries around the heart. Many patients brought to
emergency rooms with symptoms of heart attacks are given these
drugs in an attempt to open blocked arteries and keep the heart
from being starved of oxygen-rich blood.
Earlier this year, a well-publicized study found that dosing
errors are common and that these errors are associated with
twice the mortality. The conclusion was that errors in doses
caused the higher mortality. These results were used to support
a shift to the more simple bolus (single injection) drugs and
to drugs that do not require weight-adjusted dosing.
In their new analysis, the Duke researchers determined that
the worst outcome among patients who had dosing errors appears
not to be caused by the dosing errors themselves. The sicker
patients tended to be treated more often with the wrong dose,
but their higher mortality can largely be explained by the fact
that they were sicker to begin with, according to Dr.
Christopher Granger, cardiologist at Duke's Clinical Research
Institute (DCRI), who prepared the results of his study for
presentation Tuesday at the 73rd annual scientific sessions of
the American Heart Association.
Therefore, although minimizing dosing errors remains an
important goal, this example underscores the importance of
careful analysis before proclaiming a health concern and a
possible solution, he said.
"The general issue of medical errors is important and is one
that is receiving a lot of attention," Granger said. "When we
are associating adverse outcomes with a particular therapy, we
need to be careful in ascribing cause without first assuring
ourselves that there aren't major unmeasured confounding issues
In his analysis, Granger looked at the database of ASSENT-2,
a multi-center international trial conducted from 1997 to 1998
that involved 16,949 patients brought to emergency rooms within
six hours of acute heart attack symptoms. Patients were
randomized to one of two clot-busters: t-PA, or TNK, a newer
and genetically modified version of t-PA.
The original ASSENT-2 trial was funded by Genentech, South
San Francisco, Calif., and Boehringer Ingelheim, Ingelheim,
Germany. Granger's analysis of the data was supported by the
ASSENT-2 was designed as a double-blind, double-dummy trial,
in which patients either received t-PA and placebo TNK, or TNK
and placebo t-PA. This was done because the two drugs are
administered differently, and trial designers wanted to ensure
that researchers wouldn't know which active drug each patient
While ASSENT-2 demonstrated that TNK was just as effective
in the acute treatment of heart attack as t-PA, Granger wanted
to find out the effects of dosing of the two drugs, which are
markedly different. For t-PA, physicians gave one injection of
the drug, followed by a 90-minute intravenous drip, the dose of
each being determined by a formula in which weight is an
important variable. Patients receiving TNK received one of five
pre-determined five- to ten-second doses, based on where they
fall in a range of weights.
Overall, 95.1 percent of the ASSENT-2 patients received the
correct dose of t-PA, while 96.4 percent received the correct
dose of TNK.
"If incorrect dosing was responsible for the adverse
outcomes, the relationship would be stronger for patients who
received the wrong dose of the active agent when compared to
those receiving the wrong dose of the placebo," Granger said.
"If there were major confounding issues, one might expect
similar outcomes for misdosing active drug and placebo."
As it turned out, the researchers found that 30-day
mortality was in fact higher when patients received either too
little (19.5 percent mortality) or too much (9.8 percent
mortality) t-PA, compared to the correct amount (5.4 percent
mortality). Surprisingly, however, the same relationship
existed for placebo - patients who were given too little or too
much had higher mortality than the correct amount (23.5 percent
or 10.0 percent compared to 5.4 percent).
"We saw the same relationship of misdosing and bad outcomes
with the active drug and the placebo, suggesting that most of
the association was due to confounding issues rather than the
result of wrong doses," Granger said.
Granger said he believes that part of the explanation is
that patients who got the wrong dose had higher-risk features
such as older age, lighter body weight and were more likely
female. The other part is due to the fact that during the
90-minute infusion of t-PA, if the patient became unstable with
bleeding or other instability, the drug may have been stopped,
but the bad outcome was due to the problem that led to the drug
being stopped, not to the discontinuation itself.
In general, fewer patients treated with TNK received the
wrong dose, which was likely related to its simple single 5-10
second dosing, researchers said.