The Real-World Consequences of Classification

Simply enter the question "are eggs dairy?" in your favorite search engine to experience how classification can shape our view of the world. The search results will include all kinds of articles, discussions, and even videos from people asking and trying to answer this question—generally because eggs are placed in the dairy section of most US grocery stores.

This implicit classification leads many people in the US to believe that eggs are dairy. This may be a surprising assumption to those in other countries, such as the UK, where eggs are kept in the dry goods section. Often, we are not aware of the way classification shapes our perceptions and decisions. Classification of eggs in the grocery store is, for most of us, a low-stakes problem. Classification of pharmaceutical products, on the other hand, can have massive clinical and economic effects, but for most individuals, the classification is just as invisible.

Classification of Data

When building the Context Matters data model, our team put a lot of thought into classification. We needed to strike a difficult balance: data needs to be both conveniently accessible and easy-to-find for many users, coming from different geographies and companies, among whom there may exist different ideas of how drugs should be classified. The data model also needed to be robust enough to be useful and applicable even when future changes in practice challenge the established classification paradigms.

For example, a recent Nature editorial made the bold claim that "breast cancers are rare diseases" because there are numerous rare subtypes of breast cancer "defined at a molecular level but combined under the site of origin in the breast, some if not all of which may represent rare diseases." Increasingly, regulatory and health technology assessment (HTA) agencies are considering cancer therapies based on their effectiveness in patients with or without certain biomarkers. 

Context Matters still groups these therapies by the point of origin, as that remains the easiest way to locate specific treatments in many cases. However, the evaluations of these therapies are divided according to the subgroups the HTA and regulatory agencies consider. Decisions, evidence, outcomes, and other relevant data are presented at the subgroup level, often determined by the presence of molecular markers. So, as treatments become more and more targeted—and as evaluations focus less on point of origin and more on biomarkers—the relevant data will remain easy to locate, access, and compare. 

Real-World Consequences of Classification

From the Context Matters perspective, classification is important because it determines how we store and locate information, and it impacts how we think about the items being classified. Do the consequences extend further? Are there any immediate, real-world consequences of classification?

"Precision medicine" or "personalized medicine," aims to identify clinically meaningful subgroups based on demographic traits or biomarkers. This can present difficulties in how we classify products in the data model, but it can also have serious consequences in trial design, and regulatory and HTA decision-making. 

Relatively common disease conditions may end up with numerous rare subgroups. It can be difficult to recruit trial populations that are sufficiently large for some of these small, rare subgroups. While it's important to measure the effectiveness of a drug in subgroups with specific markers, it remains necessary to construct subgroups such that the results can be generalized to a broader patient population. However, even this idea is not unassailable. There is growing interest in "N-of-1" or one-person" trials. The industry's focus on real-world evidence (RWE) and the upswing in risk-sharing agreements that have potential to drive innovation for rare disease therapies indicate the real potential for individualized medicine. Regardless, molecular profiling and careful screening of patients will be a key component of clinical trials in the future. 

The creation of these rare subgroups through classification can also present regulatory and reimbursement challenges: do these subgroups constitute rare diseases in their own right? Should any regulatory or funding benefits applied to drugs intended to treat rare diseases also apply to precision therapies that target rare subtypes of common diseases? 

Classification is not straightforward, but it is important. Classification of pharmaceutical products has never been easy, but precision medicine and a variety of boundary-crossing products are presenting fresh and difficult challenges today. At Context Matters, we always strive to be mindful of changing classification paradigms to ensure that our data model remains up-to-date and relevant.