by Georgina Ferry

(Posted August 15, 1997 · Issue 14; archived September 5, 1997)

"It is surely a great criticism of our profession that we have not organized a critical summary, by specialty or subspecialty, adapted periodically, of all relevant randomized controlled trials."
- Archie Cochrane, 1979

Twenty-five years ago Archie Cochrane, an epidemiologist from Scotland, spelled out for the first time the consequences of our collective ignorance about the effectiveness, or lack thereof, of the things doctors do to patients [1]. His book Effectiveness and Efficiency: Random Reflections on Health Services set out some simple principles: that resources for health care would always be limited, and that the most efficient way to use those resources would be to provide only those forms of care that had been shown to be effective in properly designed studies. Cochrane stressed that the best evidence for or against effectiveness came from randomized controlled trials.

Today his words seem no more than common sense. Yet between principle and practice there has been, and continues to be, a yawning gulf. To bridge that gulf is the goal of the Cochrane Collaboration, which grew out of a suggestion made by Cochrane himself in 1979. He saw that efficient delivery of health care depended upon information about effectiveness being readily available and up-to-date. Many people were influenced by his ideas, and during the 1980s a number of collaborative projects were set up. They used a statistical approach called meta- analysis to combine results from different randomized controlled trials, in areas such as breast cancer, cardiovascular disease, and maternal and perinatal care. This approach increased the power of the findings of studies that by themselves were too small to produce statistically reliable results. We now know, for example, that giving aspirin to heart attack patients within 24 hours saves lives. These pioneering examples led at the end of 1992 to the establishment of the U.K. Cochrane Centre in Oxford, and almost immediately afterward to the founding of the Cochrane Collaboration.

For anyone who might one day be a patient, the idea that the medical community is only now getting down to the job of finding out what works and what doesn't is somewhat alarming. But medicine has traditionally been an inexact science, with individual practitioners favoring some interventions over others for a wide variety of reasons, and failing to update their knowledge. The new mood, driven to a certain extent by the mismatch between increasing demand for health care on the one hand, and static resources on the other, is for an "evidence-based" approach. As a result, hard questions are being asked not only about new treatments, but about time-honored procedures such as routine removal of tonsils and adenoids in children, D&C for heavy periods, or routine episiotomy in childbirth.

The Cochrane Collaboration is "an international network of individuals and institutions committed to preparing, maintaining, and disseminating systematic reviews of the effects of health care." In less than four years it has established 13 centers in 10 countries, and its highly distributed organization ensures that its organic development continues. Funding comes from a huge range of sources. The U.K. Cochrane Centre is supported by R&D programs in the National Health Service, for example, while individual reviews receive grants from medical research charities, universities, and others. Income from subscriptions is also plowed back into supporting areas of the Collaboration that might not otherwise attract funds.

"It's an idea whose time has come," says Muir Gray, Secretary of the Cochrane Collaboration and director of R&D for the Anglia and Oxford Region of the National Health Service. What has enabled it to take off is the universal availability of computerized information management, particularly hypertext, electronic databases, and the World Wide Web. So how does the Collaboration work? Its product is a set of electronic databases called the Cochrane Library. The core element within this is the Cochrane Database of Systematic Reviews. The reviews are regularly updated reports on all known trials of, for example, antiplatelet medication following stroke, or physical therapy after head injury. Subscribers can receive regular updates of the Library on diskette or compact disc.

But the tiny, shiny form of the compact disc conceals a vast enterprise on the part of hundreds if not thousands of dedicated individuals all over the world. Each review is produced by Collaborative Review Group, which is responsible for updating it regularly. Anyone can start a review group, with advice from their local Cochrane Centre; if their plan is approved they are then registered as part of the Collaboration. Groups are required to have an international membership. Their task is not only to search the literature for published trials but to uncover, as far as possible, details of trials that for one reason or another were never published. The suppression of inconclusive or negative results, either by editors of journals or the trial researchers themselves, is one of the major obstacles to assessing accurately a medical intervention's effectiveness.

Other groups of individuals, known as "fields," have formed around shared interests in, for example, primary health care, the health of elderly people, or rehabilitation and therapy. These groups also organize literature searches in their field of interest, and make information about randomized controlled trials available to those conducting reviews. Searchers make use of databases such as MEDLINE, but there is also extensive traditional library searching to track down trials. Although most of the journals searched are in English, the Collaboration is gradually acquiring reviewers who can manually search journals in languages such as Chinese or Hungarian.

Each Cochrane Review is a structured report prepared according to an agreed protocol, evaluating and synthesizing the results of all known trials that meet the group's criteria. The review groups make use of specialized software developed by the Cochrane Collaboration to ensure a consistent format for inclusion in the main database. All in all, it's a long job. "Our guess is that there'll be several hundred thousand trials, and we're up to about 130,000, so there's another four or five years of slog," says Muir Gray. New trials that are published in the meantime are less of a problem. "If you've done a review of 18 trials, then adding the nineteenth is much easier than starting with a blank sheet of paper."

One of the best examples of this approach, Gray says, remains a review of seven trials of corticosteroids given to women about to give birth prematurely. Published in 1989, the review showed that the treatment reduced by 30-50% the chances that their babies would die. "There are still many places where women who go into preterm labor are not given steroids," Gray says. "But we know that thousands of deaths and handicapped babies could have been prevented by closing the gap between what we know and what we do."

Obviously it will take more than simply providing the information to close that gap. Good decisions about health care will depend on using information in the context of the resources available and the particular personal, regional, or other circumstances that apply. So who are the front-line consumers who will use Cochrane Reviews as ammunition? "It's a manager's job - the manager has a duty to find the best information," says Gray. "So the people who will use the Library most in the first decade will be the people who make knowledge into guidelines and training programs. We're also trying to aim it at health policy makers," says Gray. But they are also aiming the Cochrane Library at physicians and patients.

"I think at the moment even Cochrane information is too much for some people in the time they have available," he concedes. "In the future we should look to the sort of model that the busy family physician or internist or nurse working as a physician's assistant would access. What they want is a one-line answer; they should be able to get that within 15 seconds, and it shouldn't take more than 15 seconds to read. Then with hypertext you click and get a summary of a Cochrane Review, and you might do that over coffee; and then if you click again you see the whole review, and you might do that as part of an educational meeting; and then you could click again and see the primary research through MEDLINE."

But why stop at the health professionals? With access to this kind of information, patients themselves could help to turn the tide toward a more evidence-based approach. " This idea that clinicians necessarily know more than patients is crazy, particularly with the Web," Gray says. "What clinicians will have to face up to is that the amply informed patient, not necessarily the well-informed patient, will be the major driving force for change. We see that we are contributing to that, and we want to help the clinicians get ready. They can't stay ahead of literature, but they can say, 'Well, that's interesting evidence you've got there, but here's the Cochrane, that's the best evidence there is.'"

Georgina Ferry is a scientific journalist based in Oxford, England.

Send us your comments and ideas for future articles.

The Cochrane Collaboration - history, aims, and methods; links to newsletters of Cochrane Review Groups and to newsgroups tailored to the interests of members of the Collaboration.

The Cochrane Library - information about the databases produced by the Cochrane Collaboration and how to subscribe to them.

The Centre for Evidence-Based Medicine - history, aims, and objectives of this Oxford-based research center, with an introduction to evidence-based medicine and links to a journal on the subject.

Meta- Analysis: Methods of Accumulating Results Across Research Domains - Introduction - an introduction to meta-analysis. Part of a paper by Larry C. Lyons. Includes links to other meta- analysis sites.