Patients' global evaluation of analgesia and safety of injected parecoxib for postoperative pain: a quantitative systematic review.

Parecoxib is the only parenterally administered cyclooxygenase-2-selective inhibitor available. We performed a systematic review, including full reports of randomized comparisons of parecoxib compared with any other analgesic intervention for prophylaxis or treatment of postoperative pain. Dichotomous data on patients' global evaluation of their analgesic regimen were extracted by means of the fraction of patients who rated their medication as "good" or "excellent." For safety analysis, data on any reported adverse effects were extracted. Relative risk (RR), number needed to treat (NNT), or number-needed-to-harm were calculated with 95% confidence intervals (CI). Data from 9 trials of 50 initially screened were finally analyzed. One thousand thirteen patients were randomized to receive parecoxib, 218 patients were allocated to an active control, and 507 patients received a placebo. With prophylactic administration, the pooled NNT to obtain the desired outcome ("good"/"excellent" rating) with parecoxib 20 and 40 mg compared with placebo was 4.5 (RR, 1.42; 95% CI, 0.91-2.24) and 4.0 (RR, 1.40; 95% CI, 1.10-1.79), respectively. In the treatment trials, the NNT to obtain the outcome of interest with parecoxib 20 mg was 2.1 (RR, 3.44; 95% CI, 1.49-7.96), 5.3 (RR, 1.43; 95% CI, 1.01-2.02), and -8.3 (RR, 0.85; 95% CI, 0.75-0.97) for the comparisons with placebo, morphine, and ketorolac, respectively. The corresponding NNT for treatment with parecoxib 40 mg was 1.7 (RR, 4.65; 95% CI, 2.04-10.61), 3.7 (RR, 1.62; 95% CI, 1.21-2.16), and 50 (RR, 1.03; 95% CI, 0.89-1.18) for the comparisons with placebo, morphine, and ketorolac, respectively. Overall adverse effects for parecoxib 20 and 40 mg were not different from those with placebo, morphine, or ketorolac. These results suggest a favorable profile for parecoxib compared with inactive or active controls. The optimal dose, timing, and frequency of administration need to be determined.