Interactive educational resources fall into three categories. Some technologies are designed to improve the effectiveness of care during face-to-face clinical encounters. For example, computer kiosks have been developed for clinic waiting rooms to evaluate patients' self-care so that diabetes educators can address patients' needs more effectively. Other technologies are designed to improve patients' ability to manage their diabetes independent of interactions with their health care team. Hand-held computers for assisting patients in adjusting their insulin fall into this category, as do diabetes-relevant video games and PC-based home learning tools. The third type of computer application includes technologies that are designed to improve patient-provider communication between face-to-face clinical encounters under the assumption that this may be one of the more effective mechanisms for improving self-care and health outcomes.

I focus here on this last category of interactive technologies, in particular, on systems that use automated telephone calls to enhance the link between diabetes educators and patients. Using these systems, patients can receive recorded educational messages, ask questions of health care providers via voicemail, and report health and self-care problems using their touch-tone telephone keypad. Because some of the routine assessment and information delivery is done via computerized calls, there are fewer demands on educators' time, leaving them free to focus their attention on patients who need them the most. Systems such as these, which augment interactions between patients and educators with communication between patients and automated messaging systems, are often referred to as Automated Telephone Disease Management (ATDM) systems.

Researchers have found that patients are able and willing to participate in ATDM communication and that the information patients provide is at least as reliable as information obtained during face-to-face clinical encounters. In a study of 200 cancer patients, researchers concluded that "[ATDM follow-up] has the potential to be a time- and labor-efficient strategy for ongoing assessment of the changing needs of large numbers of chronically ill outpatients, insuring timely professional intervention as needs arise."² Studies among patients with substance abuse disorders have found that, despite these patients' psychosocial problems, their ATDM-reported information is as reliable as data obtained through other methods and predicts recurrent acute episodes.³

ATDM has been shown to influence patients' health behavior. ATDM reminders can increase rates of vaccination,⁴ clinic attendance,⁵ and medication adherence.⁶ In a 6-month pre/post study of ATDM-supported diabetes care, patients experienced a threefold decrease in diabetes crises (home glucose readings >400 mg/dl or <50 mg/dl) and a 0.8% average decrease in HbA_1c levels (both P < 0.05).⁷

The few randomized trials of ATDM-supported chronic disease management have been encouraging. In one study, researchers randomized pregnant cocaine-using women to usual care or usual care supported by an ATDM intervention that included health assessments, mutual support opportunities, and health education.⁸ They found that patients used the ATDM system frequently and that it increased their use of drug treatment services. In another study,⁹ researchers randomized hypertensive patients to usual care or weekly ATDM monitoring with feedback of their assessment data to physicians. After 6 months, antihypertensive medication adherence and blood pressure control both improved.

Our own research indicates that ATDM assessment is feasible in populations of patients with diabetes and can improve outcomes. We conducted one randomized trial to determine the impact of ATDM-supported diabetes care among English- and Spanish-speaking patients treated in a county health care system. Intervention group patients received 12 months of biweekly ATDM calls with follow-up by a diabetes nurse educator. We conducted a second trial of a similar intervention tailored to care in Veterans Affairs (VA) health clinics.

During ATDM calls, patients in these two studies reported information about their glucose self-monitoring, symptoms, and self-care behavior. We found that they completed ATDM assessments consistently over the 12 months of their participation and were satisfied with their experience.¹⁰ Only 4% of attempted assessments were incomplete because the patient hung-up the telephone prematurely. In their 12-month follow-up survey, 87% of patients reported that they were "moderately" or "very" satisfied with the ATDM calls, 86% said that receiving such calls would make them more satisfied with their health care, and 79% said that they personally would choose to receive ATDM calls in the future.

The information that patients reported during their ATDM assessments was clinically meaningful and therefore could be used to focus patient education and other services. For example, we found that patients who reported that their glycemic control was "fair" or "poor" during their first ATDM assessment were much more likely to have had an HbA_1c >8% at study entry, a serum glucose of >240 mg/dl, and complications.

We also learned that patients were interested in receiving self-care education via ATDM calls.¹¹ On average, Spanish-speaking patients selected optional self-care education messages during 64% of their ATDM calls, and English-speakers selected these messages during 36% of their calls. After 12 months, most Spanish speakers and roughly one-fourth of English speakers continued to select educational messages, and both types of patients received a substantial amount of education as a result.

Results of both randomized trials suggest that ATDM can increase the effectiveness of diabetes care. In our county clinic study, intervention patients reported more frequent home glucose monitoring, foot inspection, and weight monitoring and fewer medication adherence problems during their 12-month follow-up interview than patients receiving usual care.¹² Glycosylated hemoglobin values were lower among intervention than control patients, and more than twice as many intervention patients were within the normal range for HbA_1c. On average, intervention patients reported fewer diabetic symptoms than usual care patients, including fewer symptoms of hyperglycemia and hypoglycemia.

ATDM-supported diabetes care had positive impacts on patient-centered outcomes.¹³ Compared to control patients, intervention patients reported greater satisfaction with care, especially with the technical quality of the services they received, their choice of providers and continuity of care, their communication with providers, and the quality of their health outcomes. Compared to patients receiving usual care, those receiving the intervention also reported fewer symptoms of depression at follow-up and fewer days in bed due to illness.

Data from the VA randomized trial are currently being analyzed. Preliminary findings indicate that impacts on patients' self-care, glycemic control, and satisfaction with care are similar to those described above for the county clinic trial.

ATDM-supported patient education can have any one of several goals, including improved self-care behavior (e.g., foot self-exams, self-monitoring of blood glucose, and medication adherence), improved health outcomes (e.g., glycemic control and symptoms), and more appropriate use of health services (e.g., fewer missed appointments and decreased use of urgent care for nonurgent events). Selection of the outcome targets will drive the way a system is designed. For example, a system designed to increase the number of patients who have annual eye exams will require much less frequent messages than one designed to change dietary behavior and might work best if physicians, as well as diabetes educators, receive patient reports. Moreover, an ATDM-based program targeting ophthalmologic exams may not save an organization money in the short term, and decision-makers should be clear about desired evaluation criteria up front.

Regardless of the goal, ATDM systems have two potential roles in diabetes education: 1) gathering up-to-date information about patients so that the efforts of diabetes educators can be targeted more effectively, and 2) promoting change in patient behavior directly by delivering information and reminders. It may be difficult to implement an ATDM system that fills both of these roles simultaneously. Priorities should be set to select the optimal role of the system in the context of other available services.

There are companies that can be located over the Internet that provide the hardware and software required for implementing ATDM within an ambulatory care setting. Other companies are clinical service providers and offer ATDM-supported patient management on a per-patient, per-month basis. The choice of whether to provide ATDM in-house or through a contracted service agency is important, and both options have their advantages.

An advantage of in-house ATDM is that the health care organization can have more control over the delivery of the service. For example, there is more opportunity to add or remove patients from the system, modify the ATDM messages, or design and change clinical reports. Educators and other clinicians often want the flexibility to modify patients' ATDM messages based on their own style of patient interaction and on patients' desires. One issue to explore when considering contracted ATDM is that making such changes could be more difficult and costly due to distance barriers and conflicting financial incentives. Changes may need to be negotiated with the vendor and require additional payments. Time-to-implementation for such changes can vary from days to weeks.

A disadvantage of in-house ATDM is that the responsibility for managing the technical aspects of the system falls on the health care organization. Although one company has produced an ATDM system that can be mastered rapidly with only limited experience with PCs, other systems require the use of specialized programming languages. In-house ATDM also requires an organization to purchase the necessary equipment and ensure that the integrity of the messages is maintained. Although it is not clear whether in-house or contracted ATDM is less costly in the long-term, equipment purchase for in-house ATDM may represent a significant initial investment, while contracted costs can be more constant and predictable. In managing the message development process in-house, care must be given so that unqualified staff do not cause the system to crash or create messages that have undesirable consequences such as generating unnecessary diagnostic work-ups.

An important issue to address when using contracted ATDM services is that they may complicate care coordination, especially if the vendor includes its own set of educators and others who will provide some follow-up when a patient problem is identified. Because these individuals typically do not have access to patients' medical records, they could make inappropriate changes to a patient's treatment plan if clear lines of communication are not established. Without clear protocols for information exchange, even appropriate advice or medication adjustments by outside staff may not be communicated to patients' primary care team or documented in patients' chart. Because many patients already receive care from multiple providers, clear and consistent instructions to patients must be maintained so that they do not become confused as to where they should turn for help when they have a clinical problem or a problem using the ATDM system.

Regardless of whether ATDM services are provided within an organization or through an outside vendor, it is important to consider who will be the primary coordinator between the service and patients' overall care. The model we have used includes a single nurse educator who reviews all ATDM patient information, provides the first line of clinical follow-up, and refers patients back to their usual care providers based on predetermined protocols. Another model is to provide ATDM information directly to patients' primary care nurses and physicians and allow them to communicate with patients directly.

In evaluating whether to integrate ATDM systems into diabetes education, several factor should be considered. The most important issue is the selection of the outcome(s) that the system is designed to improve. This decision should provide a context for other key decisions such as whether the ATDM is to be used primarily for patient assessment or direct education, whether the service is to be provided in-house or through a contract with an outside vendor, and how ATDM care is to be integrated into patients' overall care plan.