Sir

The problems inherent in continuing the current system of local ethics reviews for multicentre trials are appropriately highlighted in your News Feature 'Trial and error' (Nature 448, 530–532; 2007). However, you state that the Central Institutional Review Board of the US National Cancer Institute (NCI) adds bureaucracy by its efforts to centralize review. We disagree.

A unique feature of our model is that both central and local boards can be involved; the extent of the local review remains the prerogative of the local board, and their task is facilitated by access to the central board's review. This model has had widespread acceptance: 66% of NCI-designated cancer centres have now joined. The institute set up a central board for adult and paediatric cancer treatment trials because of the redundancy that occurs when each site's local institutional review board performs a separate ethics review. For example, 50 to 200 sites typically participate in NCI-sponsored phase III trials.

We do not agree that this centralization delays the time it takes to activate a study. The institute's Cancer Therapy Evaluation Program reveals that, for more than two-thirds of the protocols reviewed by the central board, activation is not delayed, as numerous other activities continue in parallel, such as creation of case report forms, finalization of contracts with pharmaceutical partners and drug shipment plans.

We believe that centralized ethics review, with local context supplied by the local board, is the preferred way to conduct review of multisite protocols. The NCI is currently supporting a formal cost analysis by an independent third party. In view of the high local costs cited in your News story, such data should compel administrators of local boards to reconsider their position on centralized review.