We thank Dr. Iraqi for his letter and welcome the opportunity to provide clarification. We conceptualized dementia care management as a reengineering of traditional, physician visit–centered care (or "usual care"). To create a new care model, we assembled a stakeholder group of physician champions from each participating health care organization and community agency leaders. The group established common dementia care goals and designed protocols for achieving them. The group judged incorporation of dementia care managers as key in implementation of care protocols, viewing many desired care management activities as better suited to providers with social work or nursing backgrounds. Care manager–designated tasks according to these protocols included not only actions to directly improve quality (for example, home safety evaluation) but also tasks to increase the likelihood that other providers would meet certain care quality goals, such as referring a patient to his or her physician for formal decision-making capacity assessment and informing the physician of the patient's unmet care needs. Our model's ultimate aim was to increase the likelihood that dementia care goals were achieved, regardless of who actually executed the activity. We anticipated that some goals would require interactions across physicians, care managers, and patient–caregiver dyads, whereas an individual (that is, physician or care manager) could complete other goals.

Our quality improvement intervention involved several components, including physician education on selected aspects of dementia care, deployment of trained care managers, and implementation of information systems to facilitate communication and referral. A near-universal question when any multifaceted intervention is found effective is, "Why?" Disentangling the effects of different intervention components must be explored with caution (unless components were separately tested in different randomization groups). Nine months after the intervention began, we found that intervention and usual care physicians differed on 2 knowledge or attitude variables, as Dr. Iraqi notes. However, the differences were modest, and the groups did not differ on 8 other knowledge or attitude measures (1). Thus, while we agree that some physicians may have modified their behavior in response to the intervention, the existing literature (2), the nature of many care management activities, and our finding of few differences in knowledge or attitudes across intervention and usual care providers suggest that the broad, large effects on dementia care quality that we observed are unlikely to be attributable primarily to changes in physician practices.

We previously reported and here reaffirm that the intervention and usual care groups did not differ at baseline on patients' disease severity (P = 0.10) and symptom duration (P = 0.25) or caregiver social support (P = 0.58).