Data Analysis
Patterns of health care utilization may fluctuate with time, and enrollment may identify potential problems that otherwise would not have been found. The authors looked at health care use over 6-month and 1-year intervals before and after enrollment, examining Occupational Physical Assessment Test (OPAT) data and fee-based outpatient data for both inpatient events (including nursing home utilization and hospitalization events) and outpatient visits (including primary, specialty and mental health care services; home care visits, and emergency department [ED]). The total cost incurred by outpatient visits and inpatient care was then adjusted to 2011 dollars. The relationship between program enrollment and health care use and costs was examined with a multivariate regression analyses, controlling for age and health care use in the prior year among veterans who were in the “consent” group or “nonconsent” group.
Outcome variables included the number of primary care, ED, specialty care, home health care, mental health care clinic, and inpatient visits; total inpatient bed days; and the total costs of all the events. The authors fit different multivariate models for these events according to their distributions. Specifically, the Poisson model was used if the distribution of the outcome variable was not overdispersed (eg, ED, primary care).
A negative binomial was used if the distribution of these events was overdispersed (eg, specialty care visits). Further, because the occurrence of inpatient events is relatively rare, a logit model was used to examine the relationship between enrollment status and probability of any inpatient events, regardless of the number of events. Generalized estimating equation (GEE) model with gamma distribution and log link function was used to examine the relationship between cost and program enrollment. The authors also examined the program’s effect on medication use, focusing on high-risk medications in older adults, comparing both the number of unique medications, as well as frequency such medications prescribed 1 year before and after the enrollment.1
Results
Two hundred ninety-eight (298) veterans were referred to DEMO from a 150-mile radius of Baltimore. Of these veterans, 132 consented to participate in this study. The study participants largely were representative of the total group as well as both the overall veteran population and the more general population of community-dwelling individuals with dementia (Table).
Veterans in the DEMO program largely had mild-to-moderate cognitive impairment with mean Mini-Mental State Examination (MMSE) score of 22 and significant functional limitations (Table). Only 3% displayed a pattern of “pure” dementia typical of Alzheimer disease, and 11% of those referred did not have significant abnormalities on neurocognitive testing.
The team averaged 10.3 recommendations (range 3-22), which focused on a diverse set of issues related to additional diagnostic and therapeutic concerns. Although screening data, including basic laboratory results and imaging, was requested in the referral form, in 71% of cases further diagnostic investigations were suggested. With regard to therapeutic suggestions, not surprisingly, medicines were cited as targets in a majority of cases (eg, discontinuing high-risk medications, initiating/titrating medications to minimize cardiovascular risk). While remaining mindful of the time to benefit and competing morbidities, measures to modify cardiovascular risk factors were suggested in more than half and treatment of depression in 15% of cases. Similarly, addressing poor sensory input was suggested in 38% of cases, with other common recommendations focusing on multiple environmental and social interventions (> 50%) as well as supports/outlets/respite for the caregivers.
The full multidisciplinary DEMO group met only weekly to review cases, and due to travel and scheduling difficulties, feedback to the patients and their families often was delayed for weeks. Although initial plans included regularly scheduled follow-up phone calls, demand quickly outstripped program resources. Nonetheless, chart reviews and abstracted adherence and utilization data revealed that PCPs successfully implemented 52% of recommendations within 2 weeks, rising to > 60% by 3 months. When patients were reevaluated at 1 year, they were remarkably stable: Mini–Mental State Examination (baseline 22.2 ± 5.0 → 22.3 ± 5.7 at follow-up) and Instrumental Activities of Daily Living scores (15.5 ± 10.6 → 17.7 ± 11.4).
Feedback
This program was enthusiastically received by both patients and their caregivers—100% and 98%, respectively—reporting overall satisfaction with the services received and 93% of caregivers indicating satisfaction with how the program met their needs. Caregivers were happy with the amount of time the provider took to answer questions (100% satisfied with the amount of time the DEMO provider spent and that they explained “what they wanted to know,” with 98% responding “good” or “great” for both), as well as with services and amount of help received (83% and 77% “very satisfied,” respectively).