MeSH Keywords:

The Internet offers a mixture of webpages, bulletin boards and databases covering an enormous scope of information which ranges from the highly significant through to the trivial and the obscene.

Because there are no quality controls and virtually anybody can publish anything, it is difficult for searchers to take information retrieved from the Internet at face value.

In most technical and scientific areas, such as for example quantum physics, this problem is at most a problem for a limited group of professionals, who usually have the skills to recognize the reliability of a given source and to distinguish between questionable and valuable information. But the provision of medical information is a particularly sensitive area, as also lay persons, who do not have these skills, increasingly turn to the Internet to find answers for their health problems [1], with the result that incorrect or misleading medical information may lead to potentially dangerous health behavior.

A further recent concern has been the unregulated advertising of medicines of unknown quality direct to the public via the Internet, the presentation of investigational treatments to the public (which may be intended for researchers or investors in the first place, but which may be accessed also by consumers) and lack of clear distinction between promotional and educational texts in the context of medical products; problems, which have led the FDA (US Food and Drug Administration) to look into possibilities to regulate advertising of medical products on the Internet [2]. In a resolution at the 50th World Health Assembly in May 1997, delegates expressed their concern at the advertising, promotion and uncontrolled sale of medical products by electronic communication [3]. At the time of writing, in September 1997, a first case of a patient "poisoned" by a product purchased through the Internet was reported in the New England Journal of Medicine [4].

Medical information on the Internet must not even be necessarily incorrect to harm: For example, patients reading material intended for medical professionals may misunderstand information, get wrong expectations regarding treatment options by reading about highly experimental treatments or draw their own conclusions regarding the prognosis of their disease from the statistical/epidemiological data provided (rather than taking into account their individual situation). Another problem is, that health care systems differ from country to country as does sometimes medical practice - a therapy or diagnostic procedure being a gold standard in country A may not be a standard in country B, for example for economic, cultural or for safety/regulatory reasons (medicines may have been approved by the national authorities in country A but not in country B); patients reading medical material on the Internet sometimes cannot clearly identify if the information is applicable to their local conditions.

Apart from the objective quality problems of medical information on the Internet, users may subjectively experience a quality problem, if they don’t find what they are looking for.

Taken together, the quality problem of is a result of

• lack of quality control at the stage of information production
• the information not generally being associated with a unified controlled vocabulary, which affects recall and precision of retrieval.

Both are consequences of the anarchic structure of the Internet. The medical community is increasingly aware of the "quality problem" and the "Internet’s challenge to health care provision" [5]. Several studies have evaluated the quality of medical information on various Internet venues such as websites [6] or in newsgroups [7], or medical advice given by webmasters and doctors via email [8] or by professional "ask-the-doctor" services [9]. The aim of quality control mechanisms on the Internet have been defined as "primum nil nocere" [10] - first, Internet information shall not harm.

Several organizations and individuals have created quality criteria which might help publishers to design their documents and also the public to assess and determine the quality of Internet-documents [11-18]. The question still remains, whether and by whom these quality criteria shall be enforced or controlled.

A centralistic quality control mechanism, such as "a single or centralized review process, institution, or agency (to ensure quality) is neither desirable or realistic, since the Internet is a decentralized, global medium." [19]. Thus, a "distributed" quality control model has been proposed by us: In contrast to a centralized, "upstream" filtering approach, users should be empowered to perform an individual quality control by "downstream" filtering [10].

A prerequisite for automatic downstream filtering is that the information is "labeled" with metainformation consisting of a machine-readable description associated with that information.

"Filtering" can mean to block access completely, but also for example to display warnings if the filtering criteria are not met.

Thus, the above outlined problems could be overcome by providing two kinds of additional information (metainformation) for each information (e.g. webpage) on the Internet:

In this paper we present a model, in which third parties, so called rating services, label information with value-neutral metainformation (labels). This would empower users to set their own quality requirements and to "filter" the information accordingly. One key difference to other metainformation initiatives which base on HTML-embedded keywords is that metainformation can be assigned and distributed by a third party, rather than only by the author. Our proposed model is based on a technical standard, which has been recently recommended by the W3-Consortium at MIT's Laboratory for Computer Science, the so-called "Platform for Internet Content Selection" (PICS; see http://w3.org/PICS/) [20-24]. PICS allows labels to be applied to Internet content. PICS was originally intended for protecting minors from inappropriate material on the Internet. However, the PICS standard is flexible enough to "label" any kind of information on the Internet and is already in use for other applications such as intellectual property rights management.

The PICS conventions are "analogous to specifying where on a package a label should appear, and in what font it should be printed, without specifying what it should say" [22]. To make use of PICS for medical applications we developed a PICS-conform rating vocabulary (rating system) for medical information, which we call med-PICS.

In our vision, medical professionals, associations, bodies and institutions could in the future act as decentralized "rating services" to label (i.e. to describe and/or to evaluate) other webpages with med-PICS labels or slightly modified own vocabularies. For example, a Public Health Institution in charge for preventive medicine could publish labels that identify harmful health information, scientific associations could publish labels as "seal of approval" for web sites offering credible information. Users would subscribe to a multitude of these rating services thereby getting automatically ratings from different perspectives whenever they retrieve medical Internet information.

In addition, webauthors may use med-PICS-labels to describe their websites with a standardized vocabulary, to facilitate searching the Web.

National drug-regulatory institutions (such as the FDA) could legally require producers of medical products which are regulated by respective laws (such as the Food and Drug Act) to include descriptive metainformation in their webpages, for example to clearly state the audience, to state, whether the product is investigational or approved etc.