Evaluating the Impact of Treatment Effectiveness and Side Effects in Prescribing Choices - Innovation and Marketing in the Pharmaceutical Industry - page 175

6.2.2 Analysis of Clinical Trials Data

In a typical clinical trial, researchers collect data on participants' health for a defi ned
time period. These data include multiple measurements such as vital signs, concen-
tration of the studied drug in blood, and other indicators of participants' health
improvement. Researchers then perform statistical tests to compare the mean or
median of these measurements across drugs, based on which researchers make
inferences about a drug's effectiveness and safety in comparison to a placebo or
other treatments. Although less common, researchers may sometimes also calculate
the variance of these measurements across participants. The VIGOR (Vioxx
Gastrointestinal Outcomes Research) trial is an example that helps illustrate the
standard approach of data analysis in the medical literature.
The VIGOR trial was designed to assess the effectiveness of Vioxx, a selective
inhibitor of cyclooxygenase-2, relative to that of naproxen, a nonselective NSAID
(nonsteroidal anti-infl ammatory drugs), in treating rheumatoid arthritis (Bombardier
et al. 2000 ). Treatment effectiveness was measured by the incidence of clinical
upper gastrointestinal events (i.e., gastroduodenal perforation or obstruction, upper
gastrointestinal bleeding, and symptomatic gastroduodenal ulcers). Researchers
employed a randomized controlled double-blind study. More than 8,000 patients
were randomly assigned to receive either 50 mg of Vioxx daily or 500 mg of
naproxen twice daily. The incidence of clinical upper gastrointestinal events was
recorded over a number of periods until the end of the study. For data analysis, the
researchers used Cox proportional-hazards model to estimate the occurrence of the
incidence. Results showed that the risk of confi rmed upper gastrointestinal events
for patients in the Vioxx group relative to those in the naproxen group was 0.5 (95 %
confi dence interval, 0.3-0.6; P < 0.0001), and the risk of complicated upper gastro-
intestinal events was 0.4 (95 % confi dence interval, 0.2-0.7, P = 0.005). The rate of
discontinuation of treatment owing to a lack of effi cacy was not signifi cantly differ-
ent across the two groups (6.5 % for the Vioxx group and 6.3 % for the naproxen
group). Based on these fi ndings, the researchers concluded that Vioxx is associated
with a signifi cantly lower incidence of clinical gastrointestinal events, therefore
more effective in treating rheumatoid arthritis than naproxen.
In this study, the researchers also assessed the safety of these two drugs by com-
paring the proportion of various adverse events in each treatment groups. They
found that the overall mortality rate (including death from gastrointestinal events
and from cardiovascular events) was similar in the two treatment groups. The rate
of ischemic cerebrovascular events was also similar in these two groups at 0.2 %,
whereas myocardial infarctions occurred in only 0.1 % of the participants in the
naproxen group but at a signifi cantly higher 0.4 % in the Vioxx group (95 % confi -
dence interval for the difference, 0.1-0.6). There are fi ve most common adverse
events that led to the discontinuation of treatment, including dyspepsia, abdominal
pain, epigastric discomfort, nausea, and heartburn. The researchers found there was
a signifi cantly lower rate of discontinuing treatment due to any one of the fi ve
adverse events in the naproxen group than in the Vioxx group (3.5 % vs. 4.9 %).
These fi ndings suggested that Vioxx is associated with a higher risk than naproxen.