Alternate Approaches to CER
The addition of erlotinib to gemcitabine chemotherapy for patients with advanced pancreatic cancer has been shown to provide a statistically significant survival benefit over treatment with gemcitabine alone (0.33 months; p = 0.023), but the incremental cost of this therapy is calculated at $430,000–$510,000 per quality-adjusted life year.
Clinicians and patients face a practical question: "What is the balance between the survival benefit and the cost of adding erlotinib to gemcitabine, and how should this affect our medical decision making?" This type of inquiry could logically be expected to be addressed through CER, and much of the initial enthusiasm over this research paradigm derived from its promise to provide more complete answers to such dilemmas. However, careful consideration of this seemingly straightforward question in the context of recent federal legislation highlights the limitations of the current approach to CER and the need for an alternative.
There are two fundamentally different approaches through which CER could be used to address real-world questions, such as the one posed by the erlotinib example. The first approach is a safe form of CER, focused strictly on comparing outcomes both in the setting of clinical trials and in post hoc analyses of real-world therapy. We use the term "safe" because these analyses avoid the technically and ethically difficult issue of cost effectiveness. Safe CER analyses can have 1 of only 2 potential outcomes: identification of either a superior strategy or a condition of equipoise. Either is palatable and leaves all involved parties feeling content with their efforts and conclusions.
Recent legislation essentially guarantees that safe CER is the type that will be pursued in the US. Congress enacted legislation to create the Patient-Centered Outcomes Research Institute (PCORI) as the nation's de facto CER entity, and then specifically prohibited PCORI from using …dollars-per-quality adjusted life year (or similar measure…) as a threshold to establish what type of health care is cost effective or recommended." Any latitude that might have been found in this definition has been eliminated by PCORI Executive Director Joe Selby, who subsequently stated that "You can take it to the bank that PCORI will never do a cost-effectiveness analysis."
How much does this safe form of CER, which we believe is synonymous with PCOR, actually help clinicians and patients? The benefit is likely limited. Despite assertions to the contrary, it appears that PCOR is fundamentally a methodology of comparative efficacy. Although consideration is given to results from clinical trials and from real-world settings, the basic issue under study remains the difference in efficacy between alternate strategies. In this scenario, the illustrious efficacy versus effectiveness distinction becomes semantic and the underlying investigative approach is not novel; similar research has been conducted for years as "Phase IV trials," or "comparator trials." Rather than being the much-anticipated, transformative research paradigm in American health care, PCOR may be arguably little more than a multibillion dollar exercise in rebranding extant models of post–Phase III research.
Many who refute this position highlight two characteristics that they contend to be demonstrative of PCOR's novelty. First, they argue that PCOR places new emphasis on QOL assessment. While QOL considerations should certainly be included in any multidimensional CER analysis, there is little evidence to suggest that the current CER enterprise is achieving this objective. A Medline search on "comparative effectiveness" and "quality of life" from the inception of the US CER enterprise (2009) through February 2012 identifies only 54 nonreview, English-language publications. These publications represent less than one-third of articles on comparative effectiveness, a statistic that makes it difficult to argue that QOL assessment is currently "a major part of CER" in the US. Second, PCOR proponents claim that these investigations focus on real-world settings. This is a challenging position to defend when cost is excluded from the definition of "value," unless this purported real world is one in which both patients and clinicians have unrestricted access to limitless resources.
While PCOR advocates may argue that the quantity of QOL data will increase in the coming years, they cannot refute the fact that PCOR is, by design, incapable of comprehensively informing value-based discussions between clinicians and patients that include the trade-offs between cost and efficacy, or QOL. Such discussions are becoming more common and more important, because in the actual real-world setting of finite assets, both parties are progressively influenced by the cost of therapy. Comparative, value-based decision making has become a part of life for the informed American consumer, and it appears imprudent to suggest that patients should be denied the information required to apply similar thought processes to the management of their own health.
This limitation is not inherent to CER but rather to PCOR, its safe incarnation. An alternate approach is one where the "effectiveness" dimension of comparative effectiveness is made capable of critically examining efficacy, QOL cost, and their subvariables in the context of the basic value equation (value = benefits/costs). We term this critical form of CER "comparative, value-based effectiveness research" (CVER) and believe that, because it facilitates semiquantitative comparisons across the domains of cost, benefit, and value, it represents a more robust opportunity for physicians and patients to use objective data to inform the complex process of real-world medical decision making. In contrast, because PCOR excludes cost considerations, it cannot be used in conjunction with this type of mathematical model. The semiquantitative analytical capabilities of PCOR are limited to those performed within a single dimension, while its cross-dimensional comparisons must remain qualitative.