|
PHARMACOVIGILANCE AND THE ROLE OF META-ANALYSIS: RISK OF MYOCARDIAL INFARCTION
ASSOCIATED WITH SELECTIVE COX-2 INHIBITORS
Chen L, Ashcroft DM
School of Pharmacy and Pharmaceutical Sciences, The University of Manchester,
Oxford Road, Manchester, M13 9PL, UK li-chia.chen@manchester.ac.uk
Introduction
Randomised clinical trials (RCTs) are accepted as the gold standard
for investigating drug efficacy, and methods for assessing overall effect
though meta-analysis are well developed. Far less is known about the utility
of applying meta-analysis to monitor drug safety.[1] This study aimed
to examine the risk of myocardial infarction (MI) associated with selective
COX-2 inhibitors (coxibs) using a meta-analysis of RCTs, and identify
the main challenges to this approach of pharmacovigilance.
Methods
RCTs were identified by computerised searches of MEDLINE (1966-November
2005), EMBASE (1980-2005) and supplemented by searching review articles,
conference reports and proceedings of the relevant FDA advisory panels.
Studies which used at least one coxib intervention for more than 4 weeks
were included and the proportions of patients experiencing MI during treatment
were extracted. The summary estimates of the MI risk (risk ratio) comparing
coxibs against placebo, naproxen and non-naproxen non-steroidal anti-inflammatory
drugs (NSAID), and their corresponding 95% confidence interval (95%CI)
from each trial were pooled using standard and cumulative random-effect
meta-analysis. All calculations were performed using STATA v8.0.
Results
In all, we identified 83 RCTs which included 104,249 patients with
chronic arthritis, lower back pain, colorectal adenomas and mild cognitive
impairment. The results are summarised in the table, which presents the
risk ratio (95% CI) of MI comparing coxibs against placebo, naproxen and
non-naproxen NSAIDs and the number of trials included in each comparison.
Comparator
Coxib
|
placebo
|
naproxen
|
non-naproxen NSAID
|
| celecoxib |
2.08 (0.71, 6.08), n=3 |
0.99 (0.04, 24.21), n=1 |
1.51 (0.73, 3.00), n=2 |
| rofecoxib |
1.34 (0.85, 2.11), n=12 |
1.52 (0.13, 18.05), n=2 |
0.96 (0.36, 2.54), n=7 |
| etoricoxib |
3.00 (0.12, 73.37), n=1 |
1.58 (0.06, 38.66), n=1 |
1.60 (0.76, 3.39), n=1 |
| valdecoxib |
Not available |
Not available |
0.21 (0.04, 1.16), n=3 |
| lumiracoxib |
0.87 (0.26, 2.88), n=7 |
1.73 (0.82, 3.64), n=5 |
1.41 (0.45, 4.43), n=1 |
The evidence from RCTs presented a trend of higher MI risk associated
with certain coxibs comparing to placebo or NSAIDs though there were no
significant difference between any of the comparisons. Comparing coxibs
with individual NSAIDs, only valdecoxib showed less risk of MI when compared
against diclofenac. However, the quality of reporting of serious adverse
events was often inadequate in many of the published reports.
Conclusions
Prospective meta-analysis can be a useful method to track the emerging
safety profile of medicines during their development, licensing and subsequent
launch onto the healthcare market. This study has shown a trend towards
an increased risk of MI associated with coxibs, but the difference in
rates was not significantly different when compared with more traditional
NSAIDs. Limitations to this approach relate to the poor quality of reporting
of adverse events in the published literature. Future work should examine
whether evidence synthesis of both experimental and observational study
designs could improve the evaluation of drug safety data.
Reference
1. Etminan M, Carleton B, Rochon PA. Quantifying adverse drug events
: are systematic reviews the answer? Drug Safety 2004; 27(11): 757-761.
Presented at the HSRPP Conference 2006, Bath
|
