Article type
Year
Abstract
Background: Despite decades of research on computerized clinical decision support systems (CCDSSs), results from rigorous evaluations remain mixed and little is known about what makes an effective system. Previous systematic reviews seeking characteristics important for success were limited by small sample sizes and poor quality of primary studies.
Objective: To determine if CCDSSs are effective at improving care processes or patient outcomes, and to identify characteristics associated with effectiveness.
Methods: A systematic review of randomized trials comparing use of CCDSSs to usual practice or non-CCDSS controls in clinical care settings. We conducted literature searches to January 2010 in bibliographic databases and scanned reference lists. Guided by partnerships with clinicians and senior hospital administrators, we considered over 50 trial and system characteristics. Authors of all included primary studies were contacted to provide additional information and to help select features potentially associated with effectiveness. Two outcome categories were analyzed independently: 'process of care outcomes' (such as appropriate monitoring of disease) and 'patient outcomes' (such as values of test results).
Results: In the 166 included trials, success rates for process of care and patient outcomes averaged 60% and 24%, respectively, across 6 categories of application: toxic drug monitoring and dosing, acute care, primary preventive care, chronic disease management, diagnostic test ordering, and drug prescribing. We will use univariate and multivariate analyses to identify features significantly associated with effectiveness.
Conclusions: CCDSSs affect the process of care but have limited benefit for patients. Our analysis will provide empirical guidance on effectiveness, optimal design and implementation.
Objective: To determine if CCDSSs are effective at improving care processes or patient outcomes, and to identify characteristics associated with effectiveness.
Methods: A systematic review of randomized trials comparing use of CCDSSs to usual practice or non-CCDSS controls in clinical care settings. We conducted literature searches to January 2010 in bibliographic databases and scanned reference lists. Guided by partnerships with clinicians and senior hospital administrators, we considered over 50 trial and system characteristics. Authors of all included primary studies were contacted to provide additional information and to help select features potentially associated with effectiveness. Two outcome categories were analyzed independently: 'process of care outcomes' (such as appropriate monitoring of disease) and 'patient outcomes' (such as values of test results).
Results: In the 166 included trials, success rates for process of care and patient outcomes averaged 60% and 24%, respectively, across 6 categories of application: toxic drug monitoring and dosing, acute care, primary preventive care, chronic disease management, diagnostic test ordering, and drug prescribing. We will use univariate and multivariate analyses to identify features significantly associated with effectiveness.
Conclusions: CCDSSs affect the process of care but have limited benefit for patients. Our analysis will provide empirical guidance on effectiveness, optimal design and implementation.