Computer Based Decision Support System (MEDAN)
for Abdominal Septic Shock

A Randomized Controlled Multicenter Trial


Principal Investigators:

  • Ernst W. Hanisch, MD
    Department of Surgery
    Asklepios Kliniken Langen (Germany)

  • Rüdiger Brause, PhD
    Department of Biology and Computer Science
    J.W.Goethe-University Frankfurt/Main (Germany)


Contact Address:

  • MEDAN Study Secretary
    University Hospital
    Department of General and Vascular Surgery
    Theodor Stern Kai 7
    60590 Frankfurt/Main (Germany)

    Tel:  +49 69 - 798 - 2 39 77
    61 03 - 912 - 13 09
    FAX: +49 69 - 798 - 2 12 28


Participating Centres:


Data Monitoring Committee:


Statistical Centre:

  • Hartmut Hecker, PhD
    Institut für Biometrie
    Medizinische Hochschule Hannover (Germany)



Septic shock still has a very high mortality. Numerous studies have failed in the last decade and no magic bullet is in sight, although PROWESS results are promising. Nevertheless, their generalization to abdominal septic shock patients awaits further clarification, given the fact that only a small proportion of surgical patients were included.

Computer based decision support systems potentially may improve clinical outcomes. Thus, MEDAN (MEdical Data Analysis by Neural networks) aims to establish a system based on neural network technology analysing a large database containing exclusively abdominal septic shock patients.



Information provided in time by a neural network (in the sense of a computer based decision support system) about abdominal septic shock patients can improve outcome.



Patients presented on the first day of abdominal septic shock (definition according to the consensus conference ACCP/SCCM) are randomly assigned (web-based, on-line; for technical details see here) to either a "information" or a "no information" group. Information is given with respect to prognosis (i.e. no warning, critical, very critical state), diagnostic and therapeutic options. Physicians are not obliged to follow the information provided by our information system. The function of the neural network information system is that of a "watch dog" and reminder. Once randomized to the "information" arm of the study, information can be provided at any time during the complete ICU-stay of the patient. The major end point of the study is mortality.

At least one interim analysis is planned by the statistical center and the data-monitoring committee.

Ethical approval was given by the ethics committee of the JW-Goethe university. According to the ethics committee no informed consent of patients or their relatives is necessary.

The neural network's analysis (see here) is based upon the MEDAN DATA BASE including 583 exclusively abdominal septic shock patients (definition according to the consensus conference).

Sample size is calculated using a sequential approach and requires 600 patients detecting a 15% absolute risk reduction of dying from septic shock with 90% power and with an alpha error of 5%.



German Research Foundation (HA 1456/7-2)


Start Date: 01.04.2002

Expected End Date: 30.08.2005