Full or partial refund for non-responders In 2007, after NICE determined that Janssen-Cilag's myeloma drug Velcade (bortezomib) didn't meet cost-effectiveness criteria, the company proposed a risk-sharing arrangement. The deal permits certain patients with progressive multiple myeloma to use Velcade. After four cycles of therapy, response to treatment is measured using serum-M protein. If the protein is reduced by at least 50 percent, NHS continues to fund treatment. If not, therapy is stopped, and Jannsen-Cilag refunds the cost of the drug already administered.

Initial discounts with outcomes tracking In Italy, Bayer-Schering and Pfizer entered into an agreement that offered the country's health service a 50 percent discount for their oncology drugs Nexavar and Sutent for the first two and three months of use, respectively. At the end of this initial period, treatment continues only for patients who respond, and the discount ends. This initial discount program, with clinical outcomes tracking, ensures drug use in the appropriate patient population where an effective pre-treatment biomarker is not available. At the same time, programs such as this limit payers' financial burden during the treatment phase, when responders are still being identified.

A COMPLEX TOOL: Risk-sharing agreements are one of the most difficult forms of contract for both manufacturer and payer.

There are many reasons why outcomes-based risk sharing has been slower to develop in the United States—not least among them the fact that there is no single central health system to negotiate with. Until recently, most risk-sharing deals took the form of money back guarantees used to gain first mover advantage and build market share. Examples include Proscar (Merck-Frost) in 1994, Clozaril (Sandoz) in 1995, Zocor (Merck) in 1998, and Diovan (Novartis) and Cialis (Lilly/ICOS), both in 2004. Key differences between these arrangements and those in Europe include that these were used for highly competitive therapeutic classes, they were primarily targeted toward patients (e.g., copay refunds), and there was extensive experience with regard to clinical efficacy of the products.

But there are signs that the US market is moving toward more sophisticated risk-sharing agreements. The past few months have seen the announcement of several agreements that adapt the technique to the needs of a system dominated by private-sector payers.

In April, Procter & Gamble and Sanofi-Aventis announced the launch of a pilot program with Health Alliance, a small Midwestern insurer, covering the anti-osteoporosis drug Actonel. Under the arrangement, the manufacturers promise to reimburse Health Alliance's average treatment costs for certain bone fractures suffered by women in the plan who have been taking Actonel correctly for at least six months.

Just weeks later, Merck and the insurance company Cigna announced an innovative deal covering Merck's diabetes drugs Januvia and Janumet. Where most performance-based arrangements pay the drug company more for a more effective drug, this deal took the opposite approach, calculated to drive patient adherence and volume. Cigna will review A1C lab values for its patients who take any oral anti-diabetic drugs, not just Merck's. If the A1C values improve over the course of a year, Cigna gets a bigger discount on Januvia and Janumet. If claims data show that patients who have been prescribed Januvia or Janumet are taking the drugs as prescribed, the discount increases. The deal suggests the shape that performance-based risk sharing may take in this country—providing financial incentives to payers to treat patients better and focus on results.