Strategies for making analgesia safer: The role of comparative effectiveness research

The development of the selective cyclooxygenase 2 (COX-2) inhibitors and their infamous fall from grace accelerated rheumatologists’ need for a safe analgesic for their patients. With ~20% of adults using an analgesic on a weekly basis (1), the quest for safer analgesia continues. Many patients presenting to a rheumatologist report having pain, requiring that safer analgesia be part of our focus. There are several feasible strategies for finding a safer analgesic. First, there are a variety of nonpharmacologic analgesic modalities that have exhibited some efficacy in trials. Physical therapy has shown positive effects for the pain of osteoarthritis and rheumatoid arthritis (2,3). Tai chi and acupuncture reduce pain in osteoarthritis (4,5). Chiropractic manipulation and cognitive behavior therapy have been found to be effective in selected clinical trials of low back pain (6,7). While not all patients will initiate or stick with these strategies, more effort could be directed toward improving patient adherence with nonpharmacologic therapies. Second, the side effects of current pharmacologic therapies may be avoidable if they are administered in lower dosages in combination with other complementary analgesic strategies. For example, combining nonsteroidal antiinflammatory drugs (NSAIDs) with agents directed at neuropathic pain has been shown to be effective in some models of pain (8). Further trials with combinations of treatments may provide new safer avenues of analgesia. Third, new drug development will fill some of the need. Novel therapeutic mechanisms, such as nerve growth factor antagonist (tanezumab), or refinements of existing strategies, such as nitric oxide–donating NSAIDs (naproxcinoid), may play important roles in the future. In addition to these agents, which are in various stages of phase III development, other agents may soon become available (e.g., tapentadol, ralfinamide). Fourth, current analgesics would likely be made safer if patient selection criteria could be refined. Factors potentially useful for risk stratification include genetic alleles (pharmacogenetics) or clinical variables, such as age, concomitant medications, or comorbidities. Investigators have demonstrated the potential value of such a risk stratification strategy with celecoxib. A pooled analysis of placebo-controlled trials of celecoxib showed that cardiovascular risk with celecoxib occurred only in subjects with at least 1 cardiovascular risk factor (9). If replicated, this finding may provide an important strategy for improving the safety of celecoxib. Finally, it is possible to identify individual agents and dosing regimens within analgesic categories that may be safer or riskier. Not all NSAIDs or opioids are the same. One study of the fracture risk associated with opioids suggested possible differences between agents (10). A study assessing the blood pressure effects of NSAIDs showed that indomethacin users experienced a larger increase in blood pressure than users of ibuprofen or sulindac (11). In addition, other studies of cardiovascular risk demonstrated that not all selective or nonselective NSAIDs conferred the same risk (12). The report by Masso Gonzalez et al in this issue of Arthritis & Rheumatism (13) helps to further clarify which NSAIDs may be safer in terms of gastrointestinal (GI) effects and sheds light on why there may be a difference in safety across agents. The investigators brought together broad expertise in prostaglandin pharmacology as well as NSAID epidemiology, reflected in their far-reaching study. They combined primary meta-analyses with pharmacologic assay data to attempt to explain why certain NSAIDS, nonselective and selective, are associated with differential risk of upper GI bleeding. The meta-analysis focused on epidemiologic studies and demonstrated that agents such as ibuprofen, rofecoxib, and celecoxib were associated with a lower risk than ketoralac or piroxicam. The authors explored the hypotheses raised from their meta-analysis in pharmacologic studies assessing the degree of inhibition of whole blood COX-1 and COX-2 in vitro. They found that agents with coincident inhibition of both COX-1 and COX-2 were the ones associated with the highest relative risk of upper GI bleeding. Masso Gonzalez and colleagues’ study sheds light on which NSAIDs may be safer with respect to GI bleeding and suggests that COX selectivity may explain part of the GI safety profile. This research can be considered part of the larger comparative effectiveness research agenda. According to the US government, “The purpose of comparative effectiveness research is to provide information that helps clinicians and patients choose which option best fits an individual patient’s needs and preferences. Clinicians and patients need to know not only that a treatment works on average but also which interventions work best for specific types of patients (e.g., the elderly, racial and ethnic minorities). This information is essential to translating new discoveries into better health outcomes for Americans, accelerating the application of beneficial innovations, and delivering the right treatment to the right patient at the right time” (14). While part of comparative effectiveness research promises to find simpler, less expensive, and more generalizable methods for conducting randomized controlled trials (sounds too good to be true!), a large part of the comparative effectiveness research agenda utilizes rigorous epidemiologic methods. The key limitation of observational methods for studying comparative effectiveness lies in the potential for confounding, a bias introduced by variables that direct certain patients to a given treatment but also predispose to a given outcome. For example, patients with a history of a duodenal ulcer are more likely to be prescribed a selective COX-2 inhibitor but this factor also predisposes them to future ulcers, resulting in selective COX-2 inhibitors appearing riskier to the GI system in some epidemiologic studies. Although some warn that this confounding is unavoidable in observational comparative effectiveness studies, methods for limiting this sort of bias have been developed. When the methods are carefully applied, results of such studies complement other comparative effectiveness designs and must be incorporated into treatment recommendations. Perhaps a broader issue that we should consider is what safety and benefit information would be most useful to patients and clinicians in deciding on the best treatment option, and then how to communicate this information. Individual risks should not be considered in isolation. We need to know about the broad range of rare but clinically disastrous events, and the more common adverse effects that many doctors consider nuisances but patients complain about bitterly. At the same time that we look for a composite safety measure, drug benefits need to be considered in the decision-making equation. How much risk is a given patient willing to tolerate compared with the potential benefits promised by a given agent? This is a very personal decision for each patient to weigh. We as clinicians often do not possess the information to allow patients to make a high-quality treatment decision, and most of us were never educated on how to facilitate such decisions in the context of a typical clinical visit. Comparative effectiveness research will clearly not resolve all of these challenges, but it may improve the evidence basis for many treatment decisions and refocus the research agenda on improving clinical care.