Communicating and promoting comparative-effectiveness research findings.

T comparative-effectiveness research (CER) movement has sparked an important debate about who may communicate research findings, for what purposes, and using what methodologic standards.1-3 CER is intended to inform discussions about what works in health care. Much of the information comes from research using retrospective databases and quasi-experimental designs rather than randomized clinical trials. The Food and Drug Administration (FDA) prohibits drug companies from using such information to promote pharmaceuticals, requiring that promotions be supported by “substantial evidence” of purported effects (which generally means evidence from two well-controlled trials, though one randomized, controlled trial is permitted in certain circumstances).1,2 Pharmaceutical companies have complained about “asymmetry” between the strict rules for their industry and the absence of restrictions for other organizations — including public and private payers and agencies such as the new Patient-Centered Outcomes Research Institute (PCORI) — which are increasingly conducting CER and communicating its results.3 The counterargument is that permitting drug companies to freely promote CER findings, including those that don’t meet the substantial-evidence standard, opens the door for industry to mislead physicians and patients, potentially harming public health and safety.2 It would also remove incentives for companies to conduct confirmatory trials, effectively allowing them to circumvent the FDA requirements for drug approval.2 Moreover, there are existing channels for manufacturers to communicate CER findings, even if the data do not meet the substantial-evidence standard. For example, manufacturers can write letters to the editor in defense of public challenges, distribute peer-reviewed articles discussing unapproved uses (with certain restrictions), and respond to unsolicited requests for information.2 Industry representatives, however, respond that the rules for communication outside of the substantial-evidence standard are vague and that the lack of formal FDA guidance has restricted their actions.3 In part, the issue can be addressed with better standards for the conduct and translation of CER. FDA officials recently noted that such standards are a necessary prerequisite to ensuring that comparative-effectiveness information from observational studies will provide credible evidence.1 Several groups are developing standards for using observational data in CER and, more generally, for including nonrandomized studies in systematic reviews. Eventually, the FDA might be able to determine when such studies meet substantial-evidence requirements.2 But standards alone are unlikely to suffice. Though the field is improving, judging whether a study based on observational data is of high quality will always be challenging, given unmeasured confounders and investigator choices in design and analysis that can affect results.2 The advent of CER organizations such as the PCORI, which has a specific mandate to disseminate CER findings, calls for a more immediate response. A possible step forward would be for Congress to broaden the scope of a legislative provision — Section 114 of the Food and Drug Administration Modernization Act of 1997 — that enables drug companies to promote information related to health care economics that conforms to a broader “competent and reliable scientific evidence” standard rather than the substantial-evidence criterion, as long as the targets of that promotion are restricted to formulary committees or similar entities and the information is directly related to approved indications (see table).3,4 Extending Section 114 to include CER findings would permit pharmaceutical companies to promote the information using the competent-andreliable standard, though only to organizations such as health plans. The FDA could use the Federal Trade Commission’s definition of competent and reliable scientific evidence, which encompasses evidence based on the expertise of relevant professionals using generally accepted procedures, rather than requiring two well-controlled trials.3