About "Mister VCM"
"Mister VCM" combines a graphical overview, showing explicitly the presence or the absence of information related to a given anatomical or functional system or etiology, with an interactive interface, allowing rapid access to the complete textual information. The overview of "Mister VCM" shows a graphical representation of a virtual patient with all the clinical problems described in the drug monograph. This overview is not only anatomical and functional, but also summarizes etiological and physiological elements. Drug interactions are included in the absolute contraindications "Mister VCM" in order to facilitate their reading, however it is not destined to replace the drug-drug interaction alert raising systems that are commonly proposed in medical software.
The anatomical part of "Mister VCM" helps organize information concerning numerous diseases and drugs, as the anatomical levels of medical classifications were considered during its design. The anatomical diagram in "Mister VCM" is very stylized: only one arm is detailed and the other limbs are not represented, and organs present in several copies are represented only once. However, with regard to drug knowledge, it is of no value to distinguish e.g. the bones in one arm from those in the other arm or the leg, or the left from the right kidney. This may need to be reconsidered for representing patient data in electronic records, or to refine the representation of particular diseases or symptoms, such as atherosclerosis of arteries of the lower extremities.
Several authors have used anatomical diagrams for presenting or entering data in electronic patient records, on an anatomical basis. J. Kirby et al.  proposed Pen&Pad, commercialized under the name "Clinergy". L. Stoicu-Tivadar et al.  included an anatomical diagram in their electronic record and reported that general practitioners were interested by it. P.J. McCullagh et al.  designed an application for multiple sclerosis. Sundvall et al.  have proposed to use Google Earth for positioning medical documents over a human body. The anatomical schemas found in the literature were designed for entering data in patient records, and therefore they tend to be very realistic, contrasting with "Mister VCM", which was designed for presenting medical knowledge and consequently involves a higher level of abstraction. Some previously published anatomical schemas use several views, e.g. for skeleton, muscles, vascular system, and other organs. However, we did not do this for "Mister VCM" so as to avoid the need for multiple images.
As anatomy is insufficient for identifying all diseases and drugs, we added an etiology axis, as in medical classifications which individualize, for example, infectious diseases. In addition, an area is devoted to the representation of physiological data and life habits. No other authors, as far as we are aware, have developed a graphical interface featuring a similar combined use of anatomy, etiology, physiological data and life habits.
"Mister VCM" uses the VCM graphical language, which the physician has to learn; however we have shown that this is possible in six hours . None of the existing anatomical diagrams described in the literature rely on a graphical language. The disease-based approach of VCM is particularly valuable, as it allows drugs to be presented according to the disease they treat; this in turn makes possible to organize drugs on the anatomical schema, and to represent contraindications and drug interactions on the same "Mister VCM". As the consequence, the three "Misters VCM" present on the graphical interface correspond to three levels of importance: absolute contraindications, relative contraindications and cautions for use, and adverse effects.
The detail in the graphical overview of "Mister VCM" can vary according to the complexity of the drug monograph. On "Mister VCM", the graphical representation of a given contraindication is context-dependent since it depends of the presence or the absence of other contraindications related to the same anatomical or functional system. For instance the representation of a contraindication with angina pectoris would be the VCM icon for angina pectoris if there is no other cardiac contraindication, and the more general VCM icon for cardiac diseases if there are other cardiac contraindications. The importance of context-dependent representation of medical information has been stressed by S.V. Pantazi . However, this implies a loss of detail. It may lead the physician to click on the icon and to read the associated textual excerpt, although the excerpt may not necessarily be relevant to the patient. The general nature of some of the VCM icons can lead to the same problem. Consequently, the specificity of the overview provided by "Mister VCM" can be low. However, when a pictogram of "Mister VCM" is gray, the physician can be certain that the monograph contains no information about the corresponding anatomical or functional location, without having to read the text.
In the field of information visualization, several methods have been proposed to ease the reading of a text, such as using a document lens  and greeking . These methods are independent of the knowledge expressed by the text, and they only highlight the structure or the aspect of the text, but not its content. These approaches, thus, are unlike "Mister VCM" which relies on the overview+detail technique, which consists of showing an overview of the whole monograph, and then displaying details on demand. Another technique is the deforming Fisheye, which deforms the document in order to show more detail from some areas. However, it has been demonstrated that the overview + detail method obtains better results when applied to texts . The overview + detail model used in "Mister VCM" is particularly efficient because the overview has fixed dimension and systematically presents the same kind of information at the same location, which allows the physician to find the right information quickly by memorizing these locations.
DOPAMINE  is another tool, developed by Wroe et al., for visualizing drug knowledge. It uses tables to represent drugs (columns) and their properties (lines), with various levels of granularity for both drugs and properties; thus, for example drugs can be collapsed into therapeutic classes. However, DOPAMINE was designed for verifying and authoring the Galen Drug Ontology, but not for clinical application. DOPAMINE relies on information visualisation techniques like table lens, which are not drug-specific; on the contrary, "Mister VCM" is based on anatomical and etiological considerations specific to the medical domain and therefore, in our opinion, this approach is more interesting for interface destined to physicians. However, the DOPAMINE table lens approach could be interesting for building a multiple-drug interface with "Mister VCM", in order to help the physician to compare the properties of several similar drugs, e.g. the various antihypertensives.
About the evaluation
We demonstrate that physicians obtained the information they need from the drug monographs more rapidly and accurately with "Mister VCM" than with a textual interface. Using "Mister VCM", the response time depended only weakly on the monograph length; this was presumably due to the fixed dimensions of "Mister VCM". A longer text implies more icons on "Mister VCM", but the size of "Mister VCM" itself is unchanged. In addition, the response time with "Mister VCM" is independent of the question type. "Mister VCM" represents implicit knowledge explicitly, so it provides explicit replies to all questions. By contrast, when using texts, questions with implicit replies require significantly more time. The severity of the errors recorded with the textual or the graphical interface was not significantly different.
We applied an evaluation methodology designed to avoid some of the bias which could have compromised the evaluation process. As the monographs were fictitious, the physicians could not use their personal knowledge to answer the questions. In addition, the monographs were generated at random, and therefore the designers of "Mister VCM" did not influence the monograph choice. Finally, the randomized order of questions prevents from a potential order effect on the results.
Several criteria can be used for evaluating the efficiency of drug information database interfaces, e.g. the number of clicks required to find the response. However, response time seems to be the most widely used and recommended criterion in the field of human-computer interactions .
For the evaluation, we compared the "Mister VCM" graphical interface to a textual interface which organizes the text into four sections: contraindications, drug interactions, cautions for use, adverse effects. This textual presentation is justified by the fact that it is the most basic and common way used in drug dictionaries for presenting drug monographs to physicians. However, as "Mister VCM" acts like an index, it would also have been possible to use a more sophisticated textual interface incorporating an index. However, it is difficult to determine precisely what is the reference indexed textual interface.