A Dementia Care Management Intervention: Which Components Improve Quality?
Published Online: February 27, 2012
Joshua Chodosh, MD, MSHS; Marjorie L. Pearson, PhD, MSHS; Karen I. Connor, PhD, RN, MBA; Stefanie D. Vassar, MS; Marwa Kaisey, BS; Martin L. Lee, PhD; and Barbara G. Vickrey, MD, MPH
An estimated 5.4 million people living in the United States have Alzheimer disease (AD),1 a number expected to increase 50% over the next 2 decades. The devastating effects of AD and other dementias impact an additional 15 million people who engage as informal family caregivers.1 Because of accumulating functional disabilities and higher mortality, the need for healthcare delivery system supportive structures to assist patients and families is enormous. Several randomized controlled trials using coordinated care dementia management have provided strong evidence of meaningful health benefi ts for patients with dementia and for informal caregivers compared with usual care.2-6
Effective chronic disease care management interventions typically include multiple components with diverse types of healthcare and service providers, which are tested in combination versus usual care.2-4,6 These providers include clinicians, community agency staff, and specific care managers interacting with the patient-caregiver dyad, necessitating coordination among these entities. Some programs include nurse practitioners, for example, as care managers within specifi c clinics without built-in structural collaborations with outside agencies.3,5 Whereas the focus of care management is driven by the clinical condition and is thus similar across programs, the means of delivery by provider type and the intensity of delivery are most likely to vary between programs. Given the complexity of comprehensive care management for dementia, broad translation of new knowledge about effi cacious interventions into diverse practice settings is more likely to occur when complexity is minimized, thereby maximizing sustainability and potentially reducing program costs.
We analyzed how exposure to different care management program provider types, singly or in combination, and exposure intensity were associated with positive effects on care quality with a dementia coordinated care management intervention.2 This structured program emphasized collaborative care planning and enhanced communication across healthcare plan clinicians, social workers who served as care managers within healthcare organizations and community agencies, and the patient-caregiver dyad. We assessed the extent to which exposure to and interactions among these entities were associated with variation in achievement of higher dementia care quality.
The Alzheimer’s Disease Coordinated Care for San Diego Seniors (ACCESS) trial enrolled 408 patients with dementia paired with their informal caregivers (referred to as dyads). The intervention yielded higher receipt of recommended dementia care for 21 of 23 guideline-derived quality of care indicators at 18 months of follow-up, with the mean percentage per dyad of met indicators equal to 63.9% in the intervention arm versus 32.9% in the usual care arm. This cluster-randomized trial was a collaboration of 3 healthcare organizations and 3 community agencies in greater San Diego, California. A task force of clinical champions and community agency leaders met regularly to plan the care program including staffi ng structure and responsibilities, tools, and algorithms. Specially trained dementia care managers (primarily social workers) conducted structured assessments, identifi ed patient and caregiver problems, and further assessed problems to generate a care plan. Dyads were encouraged to participate in care planning. Dementia care managers coordinated and implemented chosen treatment actions and formally linked dyads to medical and community supports and services. The intervention also incorporated strategies for collaborative care planning and enhanced communication across healthcare plans and community agencies, with care coordination, ongoing followup, and decision support (Figure 1).2,7 Primary care provider education on specifi c problem areas was provided to intervention clinics.
The ACCESS intervention’s Chronic Care Model–based components7 also included a leased Web-based care management software (CaseTrakker; IMA Technologies, Sacramento, California), which was then tailored to the study’s dementia care management protocols. Care managers embedded within the healthcare system as well as those working at study-engaged community agencies (San Diego Alzheimer’s Association, San Diego Caregiver Resource Center, Meals on Wheels San Diego) were trained in and used this software to track and manage case loads, administer assessments, organize care management activities into care plans, generate letters with appropriately merged assessment and care planning data to share with providers, and automate work flow to increase effi ciencies.
Subjects and Setting
Eligible subjects were Medicare recipients who received care from 1 of 3 San Diego area healthcare organizations. Most care recipients (80%) were enrolled in Medicare managed care. Eligibility characteristics and recruitment methods are detailed elsewhere.2 The present analysis includes all 238 caregiver/care recipient dyads in the ACCESS intervention arm, including those who were randomized to care management but were unable to or elected not to receive it.
Sources of Data and Data Abstraction Methods
Exposures. All care management communications and encounters were recorded in CaseTrakker. Healthcare organization care managers used this Web-based system to record and guide initial assessments, reassessments, and all follow-up interactions. Community agency care managers (CACMs) accessed this system and recorded their interactions with dyads as well. All communications between care managers were also recorded in CaseTrakker, and every care manager relied upon this system as their electronic record of all individual and shared care management communications and encounters with dyads over the study intervention period.
Dementia Care Quality. Data sources we used for assessing dementia care quality were mailed caregiver surveys completed at baseline, 12 months, and 18 months, and abstracted medical records of enrolled patients with dementia.2
The CaseTrakker care management database was manually abstracted by 2 professional abstractors using an abstraction tool developed by study investigators.2 All recorded notes from healthcare organization and community care managers were abstracted using on-screen or printed versions of these records. To assess interrater reliability, 10% of records were randomly selected and reabstracted. Agreement for the 3 primary predictor variables ranged from 94.2% to 100%, and kappa statistics for these variables ranged from 0.82 to 1.00.
Independent Variable Creation: Exposures to Types of Providers
The 3 primary independent variables were (1) healthcare organization care manager (HOCM), (2) CACM encounters with dyads, and (3) healthcare organization primary care provider (HOPCP) care management program–related activities. Data elements used to construct all 3 variables were abstracted from the care management software database. (The contents of all documentation in the care management software database were printed or reviewed on-screen during the abstraction process, and a notation that a phone call was made but no one was reached, for example, was not coded as an encounter.)
The HOCM activities coded as encounters for all interactions with dyads included scheduled initial assessments and 6-month reassessments (usually occurring as in-person meetings), as well as follow-up telephone calls, e-mails, and mailed written communications.
The CACM encounters included similar interactions with dyads. Community agency care manager encounters occurred only if there was a referral from the HOCM (which was not counted as a CACM-dyad encounter, since it might not have resulted in a CACM interaction with a dyad). Subsequent encounters between CACMs and dyads were based on CACM assessments and individualized need, and were coded as CACM encounters.
The HOPCP encounters were coded from documented communications between HOPCPs and HOCMs or CACMs, or as HOPCP interactions with dyads that were motivated by HOCM/CACM notifi cation. Healthcare organization primary care provider encounters typically occurred after an HOCM assessment, although HOPCPs could contact HOCMs for assistance and if recorded in the database, this is also included as an encounter.
Because of the highly skewed nature of encounter data, we coded these variables as 1/0 for any encounter by either specific care manager or primary care provider type (referred to as exposure) during the care management program duration for that dyad. Numbers of encounters are also coded as continuous integers (0-X) as an intensity measure, quantified as the average number of encounters per month of care management program duration for that dyad.
Dependent Variables: Dementia Care Quality
PDF is available on the last page.