There is ample evidence that, in the days leading up to December 7, 1941, sufficient credible information was available to intelligence analysts to support a firm conclusion that Pearl Harbor was about to be attacked. Yet, despite this, total surprise was achieved. This happened, not from any act of treachery nor yet from identifiable incompetence but rather as a result of an overload of irrelevant and competing information, along with interdepartmental rivalries, which contributed significantly to the failure to integrate the information into a coherent picture. In many respects the clinical intensivist is in a position analogous to that of the military planner in 1941.

The term monitoring is derived from the Latin monere, a transitive verb that means to warn. Thus, for the clinician, a physiologic monitor is a device or system that receives data concerning the status of an organ or organ system and by some form of transduction, makes information derived from these data available to the clinician. Advances both in physiologic understanding and in technology have conspired to produce a situation in which the volume of information to be presented to the clinician is so great that there is real danger of information overload.<sup>29 Price D.J. The use of computers in neurosurgical intensive care Bailliere's Clinical Anesthesiology London: Bailliere and Tindall (1987).  533

Cliccare qui per andare alla sezione Riferimenti</sup>

It is self-evident that technology has provided the intensivist with an ever-increasing array of physiologic parameters that can be made available either continuously or at least with great frequency (Table 1). There is, alas, much less specific evidence to show that this wealth of information has resulted in better patient care or improved patient outcome.<sup>19 Lee J., Turner J.S., Morgan C.J. , e al. Adult respiratory distress syndrome: Has there been a change in outcome predictive measures? Thorax 1994 ;  49 : 596

Cliccare qui per andare alla sezione Riferimenti</sup>

This paradox arises from the wide disparity between the sophistication of current biological signal transduction technology and the relatively poor understanding (at least at a level at which effective intervention is possible) of the pathophysiologic mechanisms that threaten our patients. An agrarian society, lacking the technologic means of defense, has little need of a missile early warning system. On the other hand, access to an accurate long-range weather forecast would be invaluable.

In his gloomy but insightful article in the Journal of Intensive Care Medicine, Buchman<sup>3 Buchman T.G. Computers in the Intensive Care Unit; promises yet to be fulfilled Intensive Care Med 1995 ;  10 : 234

Cliccare qui per andare alla sezione Riferimenti</sup> discusses the reasons for the hitherto unfulfilled promises of the computer age from the point of view of the intensivist. Fundamentally, it is the disparity between our ability to observe (which is reasonably good), our abilities to understand (which are variable), and our abilities to intervene, which in many cases are minimal.

These difficulties are further compounded by the lack of an efficient, context-based taxonomy to allow accumulated experience to be more effectively integrated into the knowledge base. Thus, the significance of physiologic observations is enormously context dependent. A value as straightforward as blood pressure or heart rate will have significance modulated widely by fixed factors such as age, size, and preexisting medical conditions and by acute factors such as temperature, vascular tone, concomitant therapy, state of alertness, and so on. The experienced clinician, often unconsciously, is able to integrate this information almost effortlessly in a manner that it is still hard for the best software to mimic.

In a group of leading European critical care units, a recent attempt was made to carry out a preliminary, multicenter survey of the prevalence of life-threatening events in four arbitrary critical illness domains.<sup>23 Morgan C.J., Takala J., DeBacker D. , e al. Definition and detection of alarms in critical care Comput Methods Programs Biomed 1996 ;  51 : 5  [cross-ref]

Cliccare qui per andare alla sezione Riferimenti</sup> This group started with the premise that there are physiologic patterns that may be detectable in the hours before an overt clinical crisis and that early detection of such patterns would prima facie offer opportunities to improve patient outcome. Despite the use of liberal definitions of an event and the availability of an enormous amount of automatically acquired data, there were great difficulties. During the building of a data library of events, it rapidly became clear that it was much harder to achieve coherence in the data than the authors had anticipated. Analysis was further confounded by the very large number of missing cells. In addition, the deficiencies of taxonomy pointed out by Buchman<sup>3 Buchman T.G. Computers in the Intensive Care Unit; promises yet to be fulfilled Intensive Care Med 1995 ;  10 : 234

Cliccare qui per andare alla sezione Riferimenti</sup> became apparent. This was particularly problematic in dealing with ongoing processes that, by general agreement, were considered to be of pathologic significance but that were then difficult to classify as events. After this sobering experience, the group retrenched and concentrated on data gathering in a single ICU where a working clinical information system appeared to offer the best hope of complete data collection. Although this will undoubtedly result in a more coherent data set, it immediately raises the question of generalizability of the findings—a problem that, from the outset, has plagued the serious study of computer applications to critical care.

This group has continued its work however and has now published a description of the data library.<sup>26 Nieminen K., Langford R.M., Morgan C.J. , e al. A clinical description of the IMPROVE Data Library IEEE Eng Med Biol Mag 1997 ;  16 : 21

Cliccare qui per andare alla sezione Riferimenti</sup> If this dataset derived essentially from experience in a single unit can be shown to be applicable to other units in a similar clinical setting, this will represent a significant step forward in the taxonomy of events and alarms for the critical care community.

Despite this critical introduction, there is no doubt that the computer in the critical care unit is an increasingly valuable tool in two broad types of application:

This article is intended as a review of the developments in both of these broad classes of function. I will also attempt to provide a framework by which the clinician may judge the potential utility of newer monitoring modalities as they become available and integrated into the broader clinical picture.