Medicare Fraud – Inspecting Quality into a process rarely works well
I just watched a story on 60 Minutes on how simple it is to game the Medicare system out of millions of dollars. Essentially, with a few pieces of static data a criminal can bill the system for medical equipment without having even met the Medicare patient. The government will pay the charges in 30 days and if the criminal gets a hint that they are under investigation, they just close shop and open up another business reusing the same list of patients.
The criminals purchase lists of patients for $10 per patient and bill Medicare for any number of devices or drugs. In the report they interview a person whose credentials were used to bill the system from two prosthetic arms. The man had his real arms and only learned of the fraud from a quarterly statement from Medicare.
What I found most interesting is that the solution being employed to reduce the fraud is more law enforcement and more auditing. To me this is backwards. The government is trying to inspect quality into the billing and payment process. No matter how successful the government is, given the simplicity of the crime, I doubt the effectiveness of their efforts.
Instead of inspecting quality at the back end of the process, it is always better to prevent the defect. From what I know of the process based on the report, it seems that the main flaw is the fact that it is so simple to acquire the static patient credentials needed to bill the government.
While there are probably more ways to prevent fraud, the solution I thought of immediately is giving each patient a single use token device like Secure ID. If the patient is involved in a medical transaction, they would be required to provide the provider a single use number which the provider would use to authorize that transaction with the government. While this would not prevent fraudulent companies from billing for unnecessary equipment, it would immediately render the patient lists useless since companies could not use them without the single use token.
A second line of defense would be to employ data mining software like those used by credit card companies. I imagine it would be possible to develop algorithms that would catch potential fraud before invoices are paid. For example, prior to paying for prosthetics, the system would assure that amputations had been billed.
It is usually simpler to add more inspections to the end of a process, but it rarely has the impact that error prevention does.