Big Data in Cancer Care: Limitations of Drug Claims

Big Data in Cancer Care: Limitations of Drug Claims
VBCC – March 2016, Vol 7, No 2Employers’ Perspective

Catherine E. Cooke, PharmD, BCPS
F. Randy Vogenberg, PhD, RPh
Partner, National Institute of Collaborative Healthcare, and Access Market Intelligence; Principal, Institute for Integrated Healthcare
Greenville, SC

As one part of an ongoing macrotheme regarding big data, understanding the benefits and limitations of using drug claims in employer-based population health is an important topic for employers in their health plan coverage strategies. Pharmacy claims are a common source of data used to describe the uses of cancer medications by employees and by their covered family members. Despite the availability of these data and a plethora of analyses, however, several limitations must be considered when interpreting drug use data.

Decoding Claims Data for a Drug’s Use and Cost

Simply put, a claim for a prescription drug is a proxy. Employers can be confident of the drug’s name and strength, but determining its use requires some assumptions before it can be used for population health strategies and perspectives. Deciding whether medications used in cancer care are appropriate is even more challenging, because drug claims data lack important clinical information to assess patient outcomes. Other limitations to big data include number errors and repackaging of drugs with different National Drug Code numbers.1

As one example, when examining drug use and cost, all medications categorized as “chemotherapy” agents may be included to determine cancer drug utilization and costs. However, several medications used in the treatment of different cancers are also used in the treatment of noncancerous conditions.

For example, methotrexate (Trexall), a common drug used in the treatment of patients with leukemia, breast cancer, lung cancer, and many other cancers, is also indicated for the treatment of patients with rheumatoid arthritis or psoriasis.2 Cyclophosphamide is a cancer therapy, but is also used for severe lupus or autoimmune kidney diseases. Another example involves more expensive cancer therapies such as the monoclonal antibody rituximab (Rituxan), which is used in B-cell non-Hodgkin lymphoma as well as in noncancerous autoimmune diseases. Investigating a prescriber’s specialty (which is not consistently available on all pharmacy claims), or calculating the expected dose of a medication, may provide clues to the drug’s uses, but it does not alleviate the problem of interpreting the data and the cost.

If medical claims data can be linked to pharmacy data, the diagnoses can lend support to determine the actual intended use of the drug. Alternatively, audits of coding may provide insight into coding or other errors, which will affect the interpretation of a drug’s use.3

Assessing Health Plan Performance Key to Employers’ Coverage Strategies

Another challenge is the way in which drug use data are presented. Many third-party or administrative service companies provide spreadsheets when discussing drug utilization or costs to employers, which makes it difficult to visualize any relationships and conduct comparative assessments.

The promise and goal of examining drug use are to obtain insight into the overall health plan’s performance while using the detailed pharmacy benefit data. In reality, determining an accurate assessment of a plan performance as it relates to cancer care or to employer-based population health requires a more comprehensive approach to the data.

Understanding what drug claims provide and do not provide lends a framework for determining the plan’s performance. Analyses of even large data sets of drug claims cannot adequately assess the appropriate use of cancer medications and the outcomes of their appropriate use.

It is incumbent on employer plan sponsors to properly plan and implement the analyses of the health plan performance, just as they attempt to do in designing employees’ benefits coverage. Successful performance assessment can lead to improved coverage strategies and benefit management tactics that are valuable to employers and to their employees.


  1. Suko J. Demystifying source NDC codes and repackaged products. October 31, 2012. KNOWHOW: The FDB Blog. Accessed January 14, 2016.
  2. Hospira. Methotrexate package insert. October 2011. Accessed February 25, 2016.
  3. Anderson BN. Are your pharmacy benefits being adjudicated properly? A PBM audit can tell you. Benefits Mag. 2014;51:22-27.