In an interview with Outsourcing-Pharma.com, Morgan described the "old" method of site selection being used as like "throwing mud at a wall to see if it sticks."

This is where CROs approach sites that they have used in the past that are stored in an internal database, in combination with sites who have responded to advertising, and any "preferred sites" provided to them by the sponsor, and as many of these sites that show a willingness to participate in the trial on offer, and are found eligible, are recruited. Those who don't eventually perform are shut down at a later date.

The problem with this method is that it is largely wasteful, because there is a general consensus within the industry that 50 per cent of sites that are recruited are suboptimal and, as a whole, 20 to 25 per cent of sites that are recruited recruit none or only very few patients, said Morgan

"The waste associated with recruiting suboptimal sites is one of the biggest budget items that CROs charge sponsors," he said.

This is because a lot of work goes into getting a site up and running - the site identification, inspection and qualification process all costs money, as does gaining ethics committee and regulatory approval for the trial.

But it is once a site is approved for a trial and initiated that "the money really starts to tick away," said Morgan.

"Once you initiate s site, it is expected to recruit, and so the logistical set up, such as organising drug supply and storage facilities as well as performing site monitoring etc. is costly."

If a site is not performing as expected it is normally pegged for shut down after a time period that has been pre-agreed with the sponsor, however, the site has a duty to continue running the trial until completion for any remaining patients and cannot close until then, and so the money clock keeps ticking.

This "old" method hasn't gone away, said Morgan, but the industry has been experiencing changes over the past couple of years in regard to site recruitment and about seven of the ten top CROs are now using a "new" method, selecting sites based around science, systems and analysis.

"Simply having a database is no longer enough," he said, describing it as a "blunt instrument that needs fine tuning."

Morgan said that Icon has been populating its database with historical data based on the past performance of sites and investigators, using pre-study and post-study reports.

This allows data searches to be done that rank the past performances of investigators in clinical trials in a particular country, therapeutic area, protocol design etc., relative to the performance of comparable peers, so as to minimise "noise factor."

On a micro level such a database is useful for making the best recommendations to sponsors on which investigators to use, based on past performance, but also on a macro level it is also important in helping to choose which countries to run a trial in, said Morgan.

"The use of data in this way needs to be tempered in judgement, as things can change with time, there may be new staff members, there may have been a logistical problem in getting drugs supplied to a site or regulatory approval in a country which may have skewed performance in the past," he said.

"But overall this process gives us stronger, more refined data and more confidence from the start in selecting sites and investigators."

Morgan said that the population of the database is an ongoing exercise, "as we continue to add information gained from studies… but in a years time we should be at a point where we have historical data input for over 90 per cent of our investigators."

Icon first started using this historical database, called ICOsite, three years ago and an enhanced functionality version was released in September 2007. The firm sets up, on average, approximately 15 trials a month of varying size.

"We use it in the pre-win phase, where we make our own assessment of appropriate sites for a study, and then again in the post-win phase before the trial starts, when we meet with the client and present them with a prioritised investigator list."

Looking to the future, Morgan said that everyone in the industry will eventually have to take a more fine-tuned approach to their databases. A more targeted approach to site selection from the start has the advantage of translating into clear cost and time benefits.

This puts small and mid-size CROs at somewhat of a disadvantage to the big players, as large investments are required to collect and maintain historical data: "Five year old data is worthless," he said, adding that for CROs who begin to find it difficult to compete with the large CROs, "consolidation or specialisation in a niche area would be the answer."