Evaluating analgesia: the challenges.

Decisions must be made when assigning analgesics, and such decisions involve comparing efficacy and adverse effects. The rules for assessing efficacy by systematic review and by large clinical trials are becoming clearer all the time. It is known, for instance, that trials that are not randomized, are not double-blind, or are too small will exaggerate efficacy. These rules are relatively easy to apply to drug interventions, although there are glaring exceptions, such as in early intra-articular morphine studies. The rules may also be less easy to apply for analgesics available over-the-counter (OTC) if the drugs have not been studied in trials of high quality or if the OTC-recommended doses are below the doses studied in the trials. When there are at least 500 patients studied in trials of high quality, credible efficacy estimates for effective drugs can be derived; more patients are needed for less effective drugs. The number needed to treat and the number needed to harm can be used to compare the efficacy and safety of a treatment with placebo or with other treatments. A recent meta-analysis comparing the number needed to treat for paracetamol 1000 mg and paracetamol 600-650 mg in moderate to severe postoperative pain indicated that paracetamol 1000 mg is likely to be more effective than lower doses. Large data sets allow investigation into dose response, gender differences, and whether particular drugs work better for particular painful conditions. Purists would argue that databases of patient experience should not be used for these efficacy analyses. The rules on how to collect and use evidence on adverse effects are less clearly formulated than they are for efficacy. For safety at or above therapeutic dose, the rules of evidence-gathering are necessarily different, and randomized trials are rarely an adequate or sufficient source. Databases of patient experience, with all the caveats, may be the most reliable data from which to work.