Article type
Year
Abstract
Introduction/Objective: To systematically review the evidence from controlled clinical trials evaluating the effects of computer-based clinical decision support systems (CDSSs) on physician behaviour and patient outcomes.
Methods: Controlled, clinical trials were identified by searching the electronic bibliographic databases MEDLINE and EMBASE. An initial search encompassed the years 1974 - 1992, with subsequent updates occurring in 1995 and in January 1997. The Science Citation Index, conference proceedings, and reference lists were also reviewed each time, and authors of relevant studies were contacted. Prospective, controlled studies were included if they evaluated the use of a CDSS (a program that matched patient characteristics with a knowledge base to produce patient-specific assessments or recommendations to be used as an aid in decision making) by health-care practitioners in a clinical setting and assessed the effects on practitioner performance or patient outcomes. The quality of the studies was evaluated using a scoring system that assessed the unit and method of allocation to groups, the baseline similarity of the groups, the type of outcome measures, and follow-up. Scores between 0-10 were possible.
Results: A total of 53 studies were identified that evaluated the effects of CDSSs on drug dosing, diagnosis, preventive care, and active medical care. The methodological quality scores ranged from 2 - 10 with more recent studies rating more highly (correlation coefficient: 0.43). Physician performance was assessed in 49 studies, and 31 found a benefit (63%). These included 6/9 studies on dosing systems, 1/6 studies on diagnostic aids, 11/17 preventive care systems, and 13/17 studies evaluating CDSSs for active medical care. Sixteen studies assessed the effect of using a CDSS on patient outcomes. Four studies found a benefit. Of the remaining 12 studies, only 1 had a power of greater than 30% to detect a small yet clinically significant effect.
Discussion: Over 50 controlled trials have evaluated the effects of using CDSSs and demonstrated that physician performance at such tasks as drug dosing, health care maintenance, and ongoing medical care can be improved. They do not support a role, at present, for using COSSs in diagnosis. The effects of using CDSSs in terms of patient outcomes have been insufficiently studied
Methods: Controlled, clinical trials were identified by searching the electronic bibliographic databases MEDLINE and EMBASE. An initial search encompassed the years 1974 - 1992, with subsequent updates occurring in 1995 and in January 1997. The Science Citation Index, conference proceedings, and reference lists were also reviewed each time, and authors of relevant studies were contacted. Prospective, controlled studies were included if they evaluated the use of a CDSS (a program that matched patient characteristics with a knowledge base to produce patient-specific assessments or recommendations to be used as an aid in decision making) by health-care practitioners in a clinical setting and assessed the effects on practitioner performance or patient outcomes. The quality of the studies was evaluated using a scoring system that assessed the unit and method of allocation to groups, the baseline similarity of the groups, the type of outcome measures, and follow-up. Scores between 0-10 were possible.
Results: A total of 53 studies were identified that evaluated the effects of CDSSs on drug dosing, diagnosis, preventive care, and active medical care. The methodological quality scores ranged from 2 - 10 with more recent studies rating more highly (correlation coefficient: 0.43). Physician performance was assessed in 49 studies, and 31 found a benefit (63%). These included 6/9 studies on dosing systems, 1/6 studies on diagnostic aids, 11/17 preventive care systems, and 13/17 studies evaluating CDSSs for active medical care. Sixteen studies assessed the effect of using a CDSS on patient outcomes. Four studies found a benefit. Of the remaining 12 studies, only 1 had a power of greater than 30% to detect a small yet clinically significant effect.
Discussion: Over 50 controlled trials have evaluated the effects of using CDSSs and demonstrated that physician performance at such tasks as drug dosing, health care maintenance, and ongoing medical care can be improved. They do not support a role, at present, for using COSSs in diagnosis. The effects of using CDSSs in terms of patient outcomes have been insufficiently studied