What if we had an Attribute Library for Preference Research?

One of the most interesting suggestions that came out of the recent commentary to the FDA PPI guidance was an idea of developing an Attribute Library. The concept of generating a way to gain consistency with attributes was brought up by several commenters in one way or another, including your author here. It was also suggested that FDA could help define commonly used attributes, this would create consistency across studies and contribute to reduced study burden for sponsors and participants. 

While this idea is a good one and could have significant benefit to those conducting patient preference studies, the concept raises a lot of questions. For example, who gets to decide on what attributes go into the Library? How are the attributes defined and their levels determined? It seems unlikely that industry would want FDA to dictate these attributes. But, perhaps a public-private partnership overseen by an organization like MDIC or the ISPOR Preference Research Working Group would be a good choice. Once the decision-makers are chosen, then we have to decide which attributes are appropriate for inclusion in the Library. 

Figure out which attributes to standardize is a harder question to answer. For example, let’s look at the attribute of “invasiveness”, which is used to describe how invasive a device or procedure is. In a preference study examining heart surgery, invasiveness might be used to distinguish open heart procedures from transcatheter procedures. But, in a different context, invasiveness might distinguish between a large incision and minor tissue damage from a cryoablation device. How do we ensure that the definition of invasiveness works for both types of studies? I think it would be difficult to land on a definition of invasiveness that would work for both of these, much less also define the different levels of invasiveness that could be used across studies in various therapeutic areas. What if, instead of one definition of invasiveness, we had multiple versions of the same attribute defined for various preference elicitation scenarios? 

Perhaps, instead of a Library, we have something more akin to a Knowledge Base that stores attributes as they’ve been used across various studies. This aggregation of attribute knowledge could be publicly available and attributes could be identified by the therapeutic area in which they were used and where one can find their publications. This Knowledge Base could help sponsors develop appropriate attributes for their studies, reduce the burden of creating attributes from scratch, and lead to greater consistency among preference research studies within therapeutic areas allowing for more consistent cross-study comparisons. 

Attribute development is one of the most critical aspects of designing a patient preference study. Ensuring alignment between preference study attributes and clinical outcome measures, meeting regulatory expectations, and generating meaningful results to support decision making is a tall order. I do believe there is merit in creating some measure of consistency across attributes to the extent that it will facilitate comparisons across studies and generate a reliable preponderance of evidence in a given therapeutic area. Having a knowledge base for existing attributes, their relationship with clinical and patient reported outcome measures, and their value to regulators may encourage more preference research in preference sensitive areas by making this part of the study development phase less burdensome. Certainly, this type of repository would help smaller sponsors develop more robust preference research to support their novel technologies, ensuring the patient voice continues to be included in the product lifecycle.