Fighting Health Care Fraud Like a Disease

An Epidemiological Approach to Combatting Fraudulent Practices

The emergence, spread, and evolution of many health care fraud schemes bear striking similarities to how public health threats emerge, proliferate, and mutate. Like disease outbreaks, health care fraud schemes often erupt in localized geographic areas, exploit weaknesses in defense mechanisms, spread through social contact, adapt to countermeasures by changing their nature, and can lie dormant for years before reemerging in force. As described below, approaching fraud through this epidemiological lens can help oversight agencies institute early warning measures, apply preventive strategies, contain the spread of fraud schemes, and ultimately deploy effective countermeasures to remove bad actors from the programs.

1. Predict the Outbreak

By understanding the precursors of known fraud schemes, oversight agencies can potentially develop surveillance techniques to detect when and where conditions are ripe for a new fraud scheme to erupt. Advanced data science approaches can help identify the features connected to emerging schemes; effective approaches may include:

• Fraudulent schemes may target novel technologies that lack robust tracking mechanisms or clear guidelines, exploiting unfamiliarity or gaps in regulation.
• Abrupt increases in provider enrollments may signal an emerging scheme, especially when linked to new services.
• Expansions of Local Coverage Determinations (LCDs), new billing codes, or relaxed oversight could create opportunities for exploitation.

2. Counter the Surge

Fraud often emerges in a relatively concentrated geography, driving up billing and payment in the region in a way that is out of step with the rest of the country. Left unchecked, the schemes may become so prevalent that they crowd out “normal practice,” and make it difficult to isolate and contain fraudulent providers. The Medicare Fraud Strike Force model was developed to combat this kind of regional fraud and provides inspiration for further evolution of analytic anti-fraud approaches. By continuously scanning administrative claims data for rapid, but geographically isolated, growth through techniques like those described below, oversight agencies can identify these schemes early before they become entrenched.

• Calculate disproportionate payments – how much more is paid in an area than expected based solely on the population, for every service category and core-based statistical area in the country, and monitor for growth.
• Develop peer group profiles for various types of health care providers and create algorithms that determine when several providers in a particular area modify their billing practices in unusual ways compared to their peers.
• Monitor the mix of health care conditions of the patient population in different geographic areas and develop proactive alerts when the condition profile changes, possibly indicating intentional manipulation of diagnosis codes to obtain coverage.
• Use social media, the Internet, and dark web monitoring to expose forums or groups where advice and/or patient lists used for fraudulent billing are shared by bad actors.

3. Stop the Spread

Like a disease, once a fraud scheme has taken root locally, it often expands through social and professional contacts. Sometimes the spread occurs through organized criminal groups that see health care fraud as safer and more profitable than other illicit activities, such as drug trafficking. Cultivating data sources that uncover connections between potential fraudsters and leveraging graph theory to identify fraudulent networks can help efficiently expand investigations and curb the propagation of criminal behavior. Potential approaches include:

• Examining shared ownership or management structures among providers to identify potential collusion or organized schemes, including identifying the use of common law firms or registered agents to enroll in Federal programs
• Investigating focal points such as hospitals, nursing homes, or clinics where fraudulent practices may originate or cluster.
• Identifying unusual network shapes, such as highly centralized or tightly clustered networks that may indicate suspicious activity or collusion among a small group of entities.
• Detecting anomalous connections between providers, such as frequent interactions with known fraudulent entities or unexplained links to geographically distant networks.
• Monitoring the evolution of networks over time to identify sudden changes, such as the rapid growth of a new cluster or the emergence of previously hidden connections.

4. Develop the Cure

Identifying fraudulent providers and criminal activity is only the starting point for successfully combating fraud. Investigators, prosecutors, and program staff must have the resources and tools to quickly cut off fraud at the source and remove bad actors from the programs through administrative, civil, and criminal action. Matthew Galeotti, the head of the DOJ Criminal Division, recently issued a memo identifying “waste, fraud, and abuse, including health care fraud” as a top priority for investigating and prosecuting white collar crime. Hopefully, this renewed focus will strengthen and augment the fraud-fighter’s toolkit.

Conclusion

By viewing health care fraud through the lens of disease and incorporating tools like graph theory, we can develop robust systems to detect, predict, and prevent its proliferation. Combining insights into its origins and mechanisms of transmission with advanced analytics and monitoring tools, we can protect vulnerable populations and preserve the integrity of health care systems. This innovative approach demands collaboration among regulators, health care organizations, and data scientists to stay ahead of ever-evolving fraud tactics.