POSTOPERATIVE ARTIFICIAL NUTRITION Overuse or Misuse? Cyrille Colin Denise Lanoir University Claude Bernard Lyon I Cecile Chambrier Hotel-Dieu, Hospices Civils de Lyon Joan Wilkinson Klim McPherson London School of Hygiene and Tropical Medicine Paul Bouletreau Hotel-Dieu, Hospices Civils de Lyon The practice of postoperative artificial nutrition (PAN) in elective surgery was covered by a consensus conference in France (December 16,1994). Artificial nutri- tion was defined as the intake of at least two macronutrients (protein, lipid, and carbohydrate) through an artificial pathway (enteral/parenteral). The guidelines resulting from the conference (2) recommended prescribing artificial nutrition for only malnourished patients, patients with insufficient postoperative nutrient intake lasting 7 or more days , and patients with severe postoperative complications. These were similar to American guidelines produced in 1993 (1). We are currently performing a 4-year before/after observational study through 1998 to evaluate the impact of the conference on medical practice. The study included a control group in England. Two samples of patients, one national and the other regional, were collected in France, both before and after the conference. Data were extracted from medical records by the investigators in the national mail survey and by a trained physician in the regional survey (Rhone Alpes). In England, a retrospective cross-sectional sample was collected in four hospitals in London over the same period and with the same data sheets. All of the patients included underwent elective upper or lower abdominal surgery (esophagectomy, gastrec- tomy, duodenopancreatectomy, colectomy, or rectum resection) in public (61%) or private (39%) hospitals (Table 1). The most noticeable results observed among the four French data sets were the high rates of artificial nutrition in the postoperative period, especially for non- malnourished patients. In view of the recommendations produced by the consensus conference, these high rates (77-86%) remained unexplained after taking into account age, weight, nutritional status, type of operation, and postoperative fasting period in a logistic regression model. Even though the English data come from a slightly different patient sample, they show a great difference from French practice. Results show that the consensus conference procedure in France can lead to guidelines that are highly inconsistent with current medical practice. These discrep- ancies between practice and evidence-based recommendations could arise for two reasons. First, the methodology of the consensus conference remains controversial, and second, the recommendations are based on relatively few clinical trials (4;5;6). Despite these limitations, the recommendations were strongly supported by experts INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 13:3, 1997 471