How have market fluctuations impacted CNA's approach to investigations?

The shift here was more gradual. From 2000 to 2005, we utilized surveillance operations, a data investing group, and a large staff of field investigators. Then the underwriting appetite changed, especially after September 11, 2001 terrorist attacks. We found that claim volume reduced significantly, while SIU referrals waned as well.

There just wasn't enough work to keep the field staff busy. So, over the course of three years, we had downsized to the current model of 25 professionals. Since we retained a small core group, we solely rely on outsourcing for the majority of our investigation and surveillance activities. Surveillance is very common, especially when handling workers' compensation or liability claims. In 2002, we opted to completely outsource surveillance functions. Additionally, in 2005 we began restricting field investigations — which entails going to the loss site, taking statements, interviewing witnesses, and so forth. This is now outsourced almost exclusively; however, we have retained a small team for “major investigations.” This team deals with provider fraud (doctors, attorneys) cases that require special expertise. We also hired four nationwide vendors to handle all surveillance and approximately 95 percent of field investigations. While we have reduced the number of nationwide vendors, we continue to operate in this fashion.

What are the caveats of wholly outsourced investigations?

Many companies still embrace the model of fully outsourcing these functions. However, there are various problems associated with that approach. Without having a clear oversight of all vendors, companies may end up paying exorbitant prices for services while lacking a quality control structure. Sure, completely outsourcing investigations may be cheaper than hiring your own field staff. But you want to make sure that you are only paying what is truly owed, and that each vendor is performing investigations to your specific standards.

Are there major challenges associated with a hybrid model?

One persistent concern is the quality of the investigations being conducted by vendors. CNA has a very structured quality-control function in place. We assembled a team whose primary task is to audit the contracted vendors. This team views a sampling every month, scoring vendors, and relaying results. It is important to note that the experience level of investigators can vary because vendors are constantly hiring. This mix of rookie and veteran investigators can yield a mixed bag of results. We are selective in the investigators we use and assign cases accordingly.

Is it possible to maintain a high degree of quality control when farming out certain functions?

As I noted earlier, CNA started with four vendors in 2005 and then whittled that down to two in 2007. We are constantly evaluating results and have detailed service level agreements in place that stipulate vendor expectations. We even went so far as to provide reporting format samples for each type of case and fully expect vendor compliance.

It's crucial to take a proactive approach. For instance, we provide questionnaires that are broken down by line of business outline expectations clearly. We expect questions to be covered at an absolute minimum. In addition, every quarter we hold a stewardship meeting with each vendor. We discuss results, complaints, and kudos. This also provides a forum for us to offer feedback aimed toward making improvements on both our end and on the vendor side.

How does CNA alter the current model to combat fraud?

Medical provider fraud is noticeably rising. Last year, half a billion dollars in bogus diagnostic tests were billed. In response, we have increased staff on the major investigations team, introduced technologies, and hired a data analyst. This allows us to focus on people who are bilking not just on one claim at a time but on claims across multiple lines of business. Common scenarios are billing for services not rendered and medical identity theft. There's also been some new technology piloting recently that would identify any sudden trends.