T HE COMPLEXITY OF DIABETES CARE and the flood of newly diagnosed patients in the face of reimbursement cutbacks have created an almost unmanageable clinical load. Automation and self-management may be the most viable solution to this problem. For example, if the average endocrinologist cares for 400 patients with diabetes and if they attain the goal of checking their blood glucose (BG) four to six times per day and send them each to be reviewed by the health care team, that creates more than a half million data points to review each year. The current standard of care is to review BG data at clinic visits while relying on patients to call the health care team if there are "problems" or 911 if there are perceived emergencies. This system allows broad interpretation by the patient, potential long delays in contacting the health care team during illness, and more frequent emergency department and/or hospital usage rather than timely phone management. In addition this standard of care renders a huge portion of the data useless at analysis, since the review may be months after the collection.The logical solution then would be to encourage data collection and provide the knowledge and tools required to react independently to the data collected. When each data point is used to make a decision, this in itself motivates collection of the data and influences frequent dose adjustments and behavior changes and can lead to improved metabolic control. While the paper by Leu et al. 1 in this issue of DTT did not specifically discuss BG data management or dose adjustments, the program did generate alerts for out-of-range data sent by pager for review. The patient was not required to send data.The automated system in this paper is used as a reminder to perform scheduled tasks, such as exercise, to take medications, or to check BG levels, as preselected by the patients. The patient could then reply with a canned or typed response. The program sent approximately 500 automated messages to each patient, and generated (when the two high responders were excluded) a 14% response rate: 0.1% of all sent messages generated responses that were noncanned and therefore went directly to the investigator, and 14% of those required further action. For 21 patients in the pager group 114 calls requiring investigator attention over 5 months is fairly high. The ability to type messages increases the attention needed to maintain the system. This is especially evident by the two highresponding patients who each replied to over 85% of their messages. By restriction of the responses to more varied canned responses with acknowledged requests for investigator attention the time required by the health care team may have been reduced by this program. The majority of messages requiring attention were about technical problems.Though the authors conclude that the patients felt more supported in the pager group