Websites should specify level of evidence for information they publish, report says

In May 2000, the French Ministry of Health and the national medical association decided to assess the quality of health information available on the internet and to define a French code of ethics for health sites.

One of the objectives was to define criteria to assess the quality of information termed “sensitive” because it could influence a decision about the user's health. The main criterion chosen was the level of evidence (“niveau de preuve”) for the statement or claim being made.

Several thousand health related sites are listed in the Catalogue and Index of French language medical sites (CISMeF). Health information available on the net ranges from public health advice by leading medical experts and institutions, information from patients' associations and support groups, advertising, and gossip concerning medical rumours or the importance of astrology.

The computer and networks department of the Rouen University Hospital conducted a survey using this catalogue, and by June last year, more than 10 000 sites had been studied, of which nearly 3000 were considered to be sensitive.

The survey showed that a vast majority of French language internet resources do not provide a score for the level of evidence available for the information that they are disseminating. Only 4.7% of these “sensitive” sites do.

The studies most likely to provide the level of evidence available are medical guidelines and consensus conferences (18.1% and 10.7% respectively), but this drops to 0.5% for technical reports, and zero for teaching material and other resource types.

Dr Stéfan Darmoni, head of the computer and networks department of the Rouen University Hospital, said that the results of the survey showed that there was a need to ask publishers of sensitive information to give an indication of the level of evidence for the information they published. Some have started doing so—for example, the National Federation of Anti-Cancer Centres (FNCLCC) now does so for nearly 90% of the guidelines it has published.

A positive trend has been observed over recent years, but a more recent survey by Dr Darmoni and colleagues has shown much remains to be done. The survey showed that the level of evidence is not shown in any French language medical teaching material. Analysis of several of the teaching sites by experts, on the basis of criteria generated from existing codes of ethics in Europe and the United States, has shown widely diverging judgments, underlining the difficulty in evaluating health sites.