Curing Big Pharma Compliance Woes with Data-Driven Medicine

June 1, 2016 conduentblogs

Recently, Rob Hellewell of Conduent and Aimee Hamilton of TMS Health, a Conduent Company, analyzed in Westlaw Journal How Data-Driven Medicine Can Cure Big Pharma’s Compliance Woes – that is, how pharmaceutical companies can prevent compliance issues (think off-label promotions, payments under the
Anti-Kickback Statute, physician referrals under the Stark Law, unapproved patient populations, unsupported client claims, improper sales practices, and more) by using innovative data-driven analytics programs with specific adaptations that highlight potential risk areas. In this post, we summarize the challenges and remedies being adopted by leading pharmaceutical companies and their outside counsel.

Compliance infractions are leading to more and more costly government investigations and corporate integrity agreements (CIAs), but data itself is becoming a powerful resource for Big Pharma in detecting risk early on—it doesn’t need be just a liability.

The Westlaw article explains in greater depth why these CIA measures are so challenging, but in essence measures covering company accountability, education and training, monitoring, and reporting, along with new areas including executive compensation and cross-border operations, coming into the spotlight, are necessitating new, data driven approaches that draw on companies’ data to identify emerging enterprise risks. The article explains that while traditional compliance aspects, including codes of conduct, policies, and external audits, fraudulent activity can still slip beneath the cracks, undetected. That’s where data-driven strategies come in.

Here are salient excerpts from the Westlaw article about strategies forward-think pharmaceutical companies
are adopting:

  • “Keyword Searches. Internal audit software can fall short when it comes to unstructured data, such as e-mail, chat and instant messaging, where risk indicators are most likely to lurk. While software can examine structured data for concerning patterns, it cannot do the same with unstructured data. However, keyword searches are a basic way to mine unstructured data for red flags… Even then, keywords alone often do not detect indicia of risk, as many people may use shorthand, jargon, or other coded language to hide their transgressions, which remain invisible in a standard keyword search.
  • Sampling. Many CIAs require organizations to sample their data, such as claims, for a limited period and then expand their review if errors are found. Forward-thinking organizations have taken a more proactive approach, identifying areas or facilities that pose higher risk and increasing not only the sample size but also the frequency of sampling. Transparency reports are a popular source for scrutiny, but increasingly organizations are fully assessing the range of data sets available to determine the range of possible insights; they also are considering whether pairing data sets would yield even more robust information. Possible data sources include sales records, clinical trial data, budgets, reimbursement records, and grant and donation databases…
  • Advanced Analytical Approaches. Some organizations have turned to data analytics tools capable of finding meaning beyond the surface and transform seemingly unrelated data into meaningful patterns. For example, sophisticated technology-assisted review (TAR) algorithms, typically used in an electronic discovery context, can prioritize a data set based on the likelihood that documents contain relevant information… Other advanced tools can help detect subtle patterns in data. For instance, anomaly detection tools can scan records for irregular payments or a high volume of payments made to a particular physician or hospital: this is particularly valuable now that prosecutors are studying data from the Centers for Medicare & Medicaid Services’ Open Payments Program for infractions. Data visualization tools allow organizations to pinpoint irregularities in relationships between employees and third parties and rank them. For example, they might find risky interactions between sale representatives and physicians based on data from travel and entertainment receipts, fee-for-service transactions that exceed established caps, sample distributions, medical information requests, or prescription volume data. Linguistic analysis techniques can identify surreptitious meaning in seemingly innocuous discussions. Concept clustering can identify hidden patterns within documents that appear to have no relationship on their surface.

    Additionally, organizations are adopting approaches that combine core eDiscovery capabilities with advanced analytics. By working with data science experts to develop and refining custom algorithms in an iterative process, legal and compliance teams are creating processes to identify key terms of interest and potential indicia of risk on both a look-back and proactive basis. As new features—that is, characteristics beyond text—are identified as important to a compliance issue, the algorithms can be refined. When run over document populations, a “heat map” of activity for specific pharmaceutical products or issues is created, and a subset of the potentially relevant data is routed to legal and compliance for review and possible remediation.”

By understanding trends in CIAs, companies are creating a stronger foundation for compliance, including the design and implementation of innovative data-driven programs with analytics at their core. Read the full story here.

Rachel Teisch is vice president, marketing at Conduent. She can be reached at

About the Author


Previous Article
How can healthcare achieve value-based care?

The keys to good health outcomes for value-based care are available in population health management.

Next Article
Have More TAR with Your Tea and Crumpets, Please

In February, we wrote about Pyrrho Investments Ltd. v. MWB Property Ltd., a case before the High Court of J...