Assessment of decision support for blood test ordering in primary care - Arandomized trial

Background: Different methods for changing blood test-ordering behavior in primary care have been proven effective. However, randomized trials compari ng these methods are lacking. Objective: To compare the effect of two versions of BloodLink, a computer-b ased clinical decision support system, on blood test ordering among general practitioners. Design: Randomized trial. Setting: 44 practices of general practitioners in the region of Delft, the Netherlands. Participants: 60 general practitioners in 44 practices who used computer-ba sed patient records in their practices. Intervention: After stratification by solo practices and group practices, p ractices were randomly assigned to use BloodLink-Restricted, which initiall y displays a reduced list of tests, or BloodLink-Guideline, which is based on the guidelines of the Dutch College of General practitioners. Measurements: Average number of blood tests ordered per order form per prac tice. Results: General practitioners who used BloodLink-Guideline requested 20% f ewer tests on average than did practitioners who used BloodLink-Restricted (mean [+/-SD], 5.5+/-0.9 tests vs. 6.9 +/- 1.6 tests, respectively; P = 0.0 03, Mann-Whitney test). Conclusions: Decision support based on guidelines is more effective in chan ging blood test-ordering behavior than is decision support based on initial ly displaying a limited number of tests. Guideline-driven decision support systems can be effective in reducing the number of laboratory tests ordered by primary care practitioners.