Computer system compares well with clinicians in assessing, but not treating children with asthma
Research Activities, September 2011, No. 373
The recommendations of a computer decision-support system (CDSS) for the assessment and management of pediatric asthma, implemented as an add-on to a medical center's electronic medical record system, agreed with those of clinicians for most assessments of asthma control in returning patients. However, CDSS recommendations agreed with less than half of severity assessments and recommendations of treatment steps for new patients, according to a recent study of a pediatric pulmonology clinic.
Use of a CDSS has been proposed to improve adherence by clinicians to clinical practice guidelines, in part because treatment guidelines have not reduced the rate of emergency department visits and hospitalizations for children with asthma. However, there have been few studies to evaluate the accuracy and validity of advice given by such systems. For that reason, the researchers developed a CDSS for pediatric asthma, based on the most recent (2007) guidelines from the National Asthma Education and Prevention Program, and applied it to all clinic visits.
The system generated asthma control assessments for 74.3 percent of 1,032 returning patients and assessments of asthma severity for 59.9 percent and treatment step recommendations for 39.5 percent of 167 new patients. Clinician agreement with the CDSS was 70.8 percent for control assessments, but only 37 percent of severity assessments and 29 percent of treatment step recommendations. When outside experts analyzed the cases of disagreement between the clinicians and the CDSS, the source was judged to be CDSS error in 72 percent of control disagreements, 56 percent of disagreements on disease severity, and 76 percent of differences about control steps. However, pediatric pulmonologists were found to not follow the guidelines in 8 percent of return visits and 18 percent of new visits. Many of the CDSS errors, such as attributing all coughs to asthma, were easy to correct, the researchers noted. The study was funded in part by the Agency for Healthcare Research and Quality (Contract No. 290-08-10011).
More details are in "Accuracy of a computerized clinical decision-support system for asthma assessment and management," by Laura J. Hoeksema, M.D., Alia Bazzy-Asad, M.D., Edwin A. Lomotan, M.D., and others in the May 2011 Journal of the American Medical Informatics Association 18(3), pp. 243-250. Reprints (AHRQ Publication No. 11-R048) are available from the AHRQ Publications Clearinghouse.