Applying Human Factors Research on Warnings to a Decision Support for Primary Care Physicians

Abstract

Human Factors research on warning systems may have relevance beyond the specific domain of operators receiving warnings about potential problems in technological systems. We present an analysis of physicians' responses to letters generated by an automated decision support system in terms of findings in the human factors literature on warnings. The automated system mails reminders to primary medical teams about patients requiring screening or lipid-lowering drugs. The effectiveness of the system depends, among other factors, on physicians' responses to the reminders. Some properties of the reminder letter have parallels in warning systems. They should affect physicians' responses as they affect the responses of operators of technological systems. We report the results of preliminary analyses of data after a 16 months follow-up of 6,571 patients. The compliance rate to recommendations in the letters in the intervention group, in which physicians receive reminder letters, was significantly higher than in the control group. Physicians in the control group were found to be less compliant as the number of their patients requiring the medical care grows. In contrast, physicians in the intervention group, who received the reminder letters, maintained similar levels of compliance irrespective of the number of patients who required their care. Thus the letters appear to provide valuable information for physicians who have to cope with more difficult patient populations.