Q&A: CNA's Innovative SIU's Next Step Is to Adopt Social Networking Analysis

Insurance & Technology recently caught up with Tim Wolfe, Director of CNA's fraud-fighting Special Investigative Unit (SIU). Wolfe left a career in law enforcement in the U.K. and Bermuda about 22 years ago to join the insurance industry. He describes how CNA's SIU has gone from the typical industry "gumshoe" model, heavy in field investigators, to a small core group that has mastered vendor management and complex case investigation and seeks to leverage the latest technology developments to detect and prevent fraud.

Anthony O'Donnell, Executive Editor, I&T: Give us some idea of what insurance companies are up against in terms of fraud today.

Tim Wolf, Director, SIU, CNA: Insurance fraud is much more sophisticated today than in the past. We're seeing more grandiose, sophisticated schemes involving not just doctors and lawyers, but other types of providers associated with the claims process, such as people billing for interpreting services when only English is spoken, billing for transportation to medical appointments that wasn't needed, or billing for wheelchairs or other medical devices that are not appropriate to the case.

The fraudsters are very slick in their approaches and know how to exploit our weaknesses. You can't overestimate what lengths they may go to in order to bilk us. For example, in one recent, well-publicized case, fraudsters preyed upon medical professionals at industry conferences, set them up with prostitutes and blackmailed them into joining their fraud network.

Tim Wolfe

AO: What are some of the advantages technology is creating in anti-fraud activities? How has it helped to reshape the SIU at insurance companies?

TW: The traditional SIU model consisted -- and still does at many companies -- of a team of field investigators who investigate suspicious claims based on referrals from adjusters. The problem with that model is that you have to rely on the judgment of adjusters to identify red flags for fraud. Naturally, some adjusters are better than others at this, and fraud is not always a priority for them, so you don't know how much fraud you're missing. Technology, and especially predictive analytics, helps us to consistently identify more subtle indicators of fraud and get a better idea of how much fraud is out there.

AO: How has CNA's SIU evolved in recent years in response to changing business conditions and advances in technology?

TW: In the early 2000s, we saw the dot-com bust and the events of 9/11, which set the country back in a lot of ways, especially economically. At CNA we also had a new CEO (Steve Lilienthal) who divested the company of entities that weren't profitable in order to focus on our core property/casualty business. Between 2002 and 2005, our SIU went through three reductions in force. Because we were focusing on more profitable business with less risk, claims volume decreased and referrals decreased proportionately, so we could no longer maintain the same staff levels.

AO: How big was CNA's SIU prior to the staff reductions, and how did your focus change after them?

TW: We were 200-strong with a large staff of field investigators, a surveillance group and a team dedicated to data investigations. We reduced that to about 20. We also outsourced all but the most specialized investigations to four national vendors. That led to our focusing resources into four teams: one manages vendors, another focuses on large-scale, organized fraud, the third manages intake and triage, and the last team handles training and regulatory compliance, process improvement and metrics to ensure that we're focusing our efforts in the right areas to save money and add value to the claims process.

AO: What was the result of that reorganization?

TW: We increased our return on investment four-fold. As I suggested earlier, in the conventional SIU model, in-house field investigators predominate, but we've found that, with a well-structured quality assurance program and clearly defined service levels, you can manage investigative vendors to reduce expenses and get better results.

AO: How did the role of technology change with the changing shape of your SIU?

TW: For us the watershed moment in terms of technology came in 2007 in connection with major investigations focusing on fraud by medical and legal providers. We purchased i2's [acquired by IBM, Oct. 2011] Analyst Notebook link analysis tool and brought in a qualified data analyst experienced in working with that software. It gave us the ability to pull in data from various sources, cross-reference it, run it against public records and look for commonalities between claims across all lines of business.

Prior to that we had worked with NetMap for Claims from [Jersey City, N.J.-based] ISO, with some, limited success, but we didn't have staff with the required skill-sets who could fully take advantage of it.

AO: What was the essence of the improvement you enjoyed with the addition of the i2 product and the data analyst?

TW: We went from being reactive to adjuster referrals to being proactive. We also joined the NICB [National Insurance Crime Bureau; Des Plaines, Ill.], which generates "Forewarn" alerts based on cases they either generate themselves or pass on from other member carriers. We can also comb the data ourselves, for example, searching for the doctors who are billing us predominantly with the maximum CPT [current procedure terminology] codes that generate the highest billing amounts.

We are also engaged with NICB in a new program known as the Aggregated Medical Database, which provides for analysis of medical bill data from CNA and other participating carriers. By aggregating this data and establishing a baseline, we can isolate medical practices that diverge from the norm and look for patterns of questionable behavior, such as physicians billing for excessive face time with a patient or treatment rendered on weekends or holidays. We also monitor news releases from the Coalition Against Insurance Fraud for information on people who have recently been arrested or indicted for fraud to see if we have exposure.

AO: What are some of the improvements that you've put in place, or hope to implement in the near future?

TW: On the individual fraud side we've done several things. First we built our own internal predictive model, based on our historical data, for workers' compensation and implemented it in early 2008. This consisted of the predictive model itself and the application of business rules.

The success of the internal model was limited in as much as it generated many false positives. It was taking so much time sorting through them that we reached a point where returns were minimal, so we started looking at implementing a more robust long-term solution. In Dec. 2007 we brought in eight vendors and invited them to demonstrate their products. We then selected three to perform a proof-of-concept with our own data. The best results came from SAS [Cary, N.C.], whom we finally decided to engage.

This year we began implementing four models in our four main lines of business, workers' comp, general liability, property and auto. Three are now up-and-running; property is still in the testing phase.

AO: What kinds of results are you enjoying in the three lines of business that are live?

TW: We've already seen some good returns. Overall we're seeing a hit rate of about 26% from the alerts that SAS generates. Our team reviews those alerts to determine whether they are actionable cases. About one in four of flagged cases are viable investigation targets.

AO: What are your next steps in applying technology to CNA's fight against fraud?

TW: SAS is building a Social Networking Analysis (SNA) piece that is scheduled to be delivered in December. It works in concert with SAS's weekly run of our data and applies link analysis to score groups of associated claims. The social networking piece involves identifying relationships between parties to a claim that might not be otherwise apparent.

AO: What kind of results do you expect from the SNA piece?

TW: Our business case projects 12 major cases annually, which translates to a savings of approximately $500,000 per year.

Anthony O'Donnell has covered technology in the insurance industry since 2000, when he joined the editorial staff of Insurance & Technology. As an editor and reporter for I&T and the InformationWeek Financial Services of TechWeb he has written on all areas of information ... View Full Bio