[Method and application limitations of decision support systems in acute pancreatitis]

Several decision support systems (DSSs) for acute pancreatitis (AP) were analyzed with reference to development methods, procedure limits and operational performances. Almost all the DSSs have been addressed to the early definition of severity, which appears the only decisional point in the approach to the management of AP. None of the three groups of methods, multifactor, scoring and bayesian systems, provides an explicit evidence of effectiveness. The multifactor systems (Ranson and successive) show inadequacy of design and operational limits which involve poor reliability and conflicting indications from the different centers. The scoring systems (APACHE, SAPS) have been projected and developed for clinical situations quite different from the AP at the onset, and seem more properly to be applicable to the monitoring of its complications. The bayesian systems, although the models used until now present important methodological shortcomings, are those which furnished the best results but are lacking in clinical validation and present a form hardly accepted by the clinician. Despite the disappointing operative results and their limited use in the AP, the DSSs probably constitute one of the most effective tools to improve the management of the severe forms, on condition that the methodology of design enad trial is correctly adjusted.