[In-patient treatment decisions]

The aim of this study was to identify the extent to which treatment decisions for medical in-patients in three hospitals in Catalunya are supported by published evidence.

In a retrospective, cross-sectional study, main diagnosis-treatment pairs (DTPs) were identified for 980 hospital discharges. The Cochrane library and MEDLINE from 1966 to 1998 were searched for systematic reviews or, in their absence, randomized controlled clinical trials (RCT) that supported the treatment given for each diagnosis. The level of evidence found for each DTP was classified. Descriptive analyses for patients and different DTP were performed.

A systematic review or RCT (level I) was found for 65.4% of the 980 patients. In 32.6% of cases, there was only non-experimental or consensus evidence (level II) and in 2% there was no good evidence for the treatment given (level III). An analysis of the 598 unique pairs reduced the percentage of cases supported by level I evidence (55.7) with a proportionate increase in level II (41%). There were some differences in the proportion of cases in each evidence level by diagnostic group.

For about two thirds of patients and a half therapeutic decisions in these three centres, there was a published RCT or a systematic review, with variations according to diagnostic groups. More robust and objective instruments are needed to assess the degree to which scientific evidence is applied in clinical practice.