The Virtual Library: Out with print, in with the PC

When GlaxoWellcome merged with SmithKline Beecham in late December 2000, it created a global pharmaceutical giant with seven per cent of the world's drug market and a massive headache for the information management department. GlaxoSmithKline (GSK) spanned 24 research and development sites in seven countries, and its libraries met the information needs of 100,000 or so employees. Of these, the principal users were the company's 16,000 R&D staff who relied on 14 main libraries, which boasted a host of site-specific contracts with scientific journal publishers.

GSK's vice-president for information management Dr Melanie O'Neill, and her colleagues, had the task of making sense of all this. Added to that was the challenge of providing consistent access around the world to an increasingly mobile workforce. A visiting researcher to the UK would sometimes find that an electronic journal, available on their desktop in the US, could not be accessed any longer. All this needed to change, O'Neill told managers attending the World Pharma IT Congress in London in November this year. Earlier in the year, the decision was made to do away with the physical libraries altogether and to develop a 'virtual library' to deliver electronic access to nearly all books and journals. The project will also encompass in-house analysis of information and the capture and management of internal research data. The goal is to close all 14 libraries by the end of this year. The closure of a small library in 2002 had shown it was possible, and highlighted potential problems, said O'Neill. Since the project began in June, two main UK libraries - at Stevenage and Ware, both in Hertfordshire - have been closed and the project is on course to meet its target.

• When GlaxoSmithKline opened its state-of-the-art high-throughput chemistry facility at its Harlow, UK research centre (above) in October, it included the largest chemistry lab in the UK, but no library. The library at Stevenage (below) has followed it into oblivion. All GSK's libraries will be 'virtual' by the end of the year.

This was not a cover for some savage cost-cutting. The information management's funds aren't unlimited, but 'it's not a money-saving scheme,' O'Neill told Research Information. A tiny number of staff will be lost. It will, however, liberate valuable space. The R&D laboratories are on prime sites and the libraries often occupy some of the best locations in these buildings. The vacated space will now be freed for other uses. Shutting them down will be a wrench for some staff, conceded O'Neill. 'There is a lot of emotion attached to libraries,' she said. But people were simply using them less and less often. 'They had started to desert them.'

This trend not to visit the physical libraries was most notable among newly recruited staff, but included older researchers too. Other firms in the pharmaceutical sector have been making a gradual shift to all-electronic libraries, but the time seemed right for a bold stroke. 'A lot of electronic literature is now available,' said O'Neill. Some 3,000 electronic journals were already being delivered to GSK and building a 'Virtual Library' would be a step towards ensuring that all of the content was searchable from a single interface. Some scientific areas have particular problems; chemistry being one. Synthetic chemists sometimes need to refer to older literature for information about the physical properties of little-used compounds - perhaps even to scientific papers published in the early 1900s and earlier. So, recently completed projects by the American Chemical Society and the Royal Society of Chemistry, to digitise their earliest journals, played an important part in the decision to abandon print, said O'Neill.

One particular feature of the project that was guaranteed to cause some problems was the need to re-use published data and combine it with internal information in GSK's own databases. This meant that worldwide company licenses, with appropriate re-use clauses, needed to be negotiated with scientific publishers. This required a measure of trust and understanding between the two that had previously not existed. 'Publishers didn't understand how information moved around in the R&D department,' according to O'Neill. So visits to GSK sites were organised for staff from some of the major publishers. They were able to shadow scientists in their daily activities, and see how they used journals. The publishers should get something out of this too, said O'Neill, because it will help them develop new products that meet the needs of the GSK and other pharmaceutical companies adopting similar strategies. The company also wanted to make sure that back catalogues were being digitised quickly and, if not, to try and persuade publishers to accelerate the process.

The plan worked, and GSK has successfully negotiated re-use contracts with a small number of companies. The negotiations were protracted, conceded O'Neill. But they were 'helped by our global buying power,' as well as the fact that re-use was negotiated as part of a single overall contract 'and not as an add-on to an existing licence'.

'We have established a precedent,' she said. Information brokers are also doing similar deals with publishers, though these are mostly '[drug] target related' covering single therapeutic areas, and GSK is interested in these products too.

Training needed a big shake up. Publishers had developed e-learning modules around an individual product - their own. What GSK wanted to do was to provide training around clusters of products through a mixture of class-based courses and e-learning modules. The training was to be developed to meet the needs of individual departments and, after discussion with R&D managers, new courses have been set up, sometimes on a worldwide basis.

Improvements are being measured, to make sure there is a real return on investment. 'Now every piece of training has a feedback form', said O'Neill. 'And there's a follow up questionnaire sent to course attendees after one to three months. It's only short but a key question is "have you saved time?".'

How Glaxo aggregated and distributed incoming information also underwent a major rethink. 'We needed to analyse it and put GSK's "spin" on published literature so that it is useful in formulating strategy,' said O'Neill.

Existing reports and bulletins were being produced for each disease and technology by an army of information specialists. But much of this was on a monthly basis and was out-of-date by the time it reached its destination. Some of this function will be replaced by alerting services available from publishers. However, there is a need for products that will merge alerts across publishers, said O'Neill.

In future, GSK plans to embed analysts within project teams, alongside medics, biologists and chemists. The secret of success is that they are able to 'speak the shorthand' of the scientists, said O'Neill. As a result, they are being recruited from R&D staff because it's easier to train them in information skills than to train information scientists in the necessary science.

Some way of measuring the effect of analytical reports was also called for, to make sure analysts were providing the information required by decision makers. GSK now compiles data on the percentage of analytical reports with 'significant business impact.' This has helped those producing them understand the impact their reports make, and there has been steady improvement in the figures over the past year - helped by some additional training. 'In the past, reports were often too verbose,' which meant that finding important information was made more difficult than it needed to be, says O'Neill. With this in mind, experienced journalists were brought in to teach analysts how to get their message across quickly and with maximum effect, by way of executive summaries and other methods.

The 'Virtual Library' project is part of a move to oversee information management on a global basis. The department acts across the whole company, which means that there is a clear focus on GSK's business priorities. There is a single person or place for publishers or other suppliers to negotiate 'enterprise-wide' licenses or discuss possible new products. 'This doesn't always work in practice,' said O'Neill, not least, because suppliers themselves don't organise their business in this way.

With no local librarian, simple queries will be handled at the computer desktop, which means that intuitive software products are needed. 'The expectation is that only difficult queries will require the intervention of one of the 300 information management staff available to help scientists.' To ensure that this is the case, GSK is exploring the available text-mining tools to see which ones might best meet the needs of researchers.

There are still some obstacles. 'Books won't have a place in the corporate environment,' said O'Neill, but not all of them are available electronically. GSK has undertaken a massive review - including representatives from across R&D - of which ones to keep, she said. Books from the libraries that will still be needed are going to be stored centrally in two or three depositories. They will be available to order, much as science books are borrowed and delivered from the British Library's document supply centre in Boston Spa.

It is possible that there will be an increased amount of printing. This may especially be the case with longer documents, which can be tiring to read on computer screens. So far there have been no signs of this, but it may happen. If it does, it should be a phase that will disappear eventually, she believes. Improved browsing software, designed for longer documents so that users 'get to the bit they need to look at', and adjustable monitors with better screens, should help with this.

The capture and management of intellectual property generated by GSK's own researchers, faces related problems. Moving to a fully electronic laboratory notebook will involve changing the behaviour of discovery scientists. They will need to think about putting their first thoughts about an experiment or drug target in electronic form, suggested O'Neill. Such information is important to patent claims.

There remains a problem with some of science's print legacy. The American Chemical Society's project digitised all its journals back to Volume 1 of each title. This extended the full-text electronic access back to the nineteenth century (in the case of the Journal of the American Chemical Society, 1879). But some important early literature has still not been digitised, and there are no immediate plans to do so. Chemists sometimes refer to older German journals, and this is a gap that needs to be filled, not just for GSK but all pharmaceutical companies. 'We may have to do it ourselves and sell it to our rivals,' said O'Neill jokingly. What is certain is that, like GSK, other pharmaceutical firms will be soon be mothballing their real libraries and replacing them with fully electronic ones.