Can a Home-Based Collaborative Care Model Reduce Health Services Utilization for Older Medicaid Beneficiaries Living with Depression and Co-occurring Chronic Conditions? A Quasi-experimental Study.

Depression remains a major public health issue for older adults, increasing risk of costly health services utilization. While home-based collaborative care models (CCM) like PEARLS have been shown to effectively treat depression in low-income older adults living with multiple chronic conditions, their economic impact is unclear. We conducted a quasi-experimental study to estimate PEARLS effect on health service utilization among low-income older adults. Our secondary data analysis merged de-identified PEARLS program data (N = 1106), home and community-based services (HCBS) administrative data (N = 16,096), and Medicaid claims and encounters data (N = 164) from 2011 to 2016 in Washington State. We used nearest neighbor propensity matching to create a comparison group of social service recipients similar to PEARLS participants on key determinants of utilization guided by Andersen's Model. Primary outcomes were inpatient hospitalizations, emergency room (ER) visits, and nursing home days; secondary outcomes were long-term supports and services (LTSS), mortality, depression and health. We used an event study difference-in-difference (DID) approach to compare outcomes. Our final dataset included 164 older adults (74% female, 39% people of color, mean PHQ-9 12.2). One-year post-enrollment, PEARLS participants had statistically significant improvements in inpatient hospitalizations (69 fewer hospitalizations per 1000 member months, p = 0.02) and 37 fewer nursing home days (p < 0.01) than comparison group participants; there were no significant improvements in ER visits. PEARLS participants also experienced lower mortality. This study shows the potential value of home-based CCM for participants, organizations and policymakers. Future research is needed to examine potential cost savings.

Using State Administrative Data to Identify Social Complexity Risk Factors for Children.

PURPOSE: Screening for social determinants of health is challenging but critically important for optimizing child health outcomes. We aimed to test the feasibility of using an integrated state agency administrative database to identify social complexity risk factors and examined their relationship to emergency department (ED) use. METHODS: We conducted a retrospective cohort study among children younger than 18 years with Washington State Medicaid insurance coverage (N = 505,367). We linked child and parent administrative data for this cohort to identify a set of social complexity risk factors, such as poverty and parent mental illness, that have either a known or hypothesized association with suboptimal health care use. Using multivariate analyses, we examined associations of each risk factor and of number of risk factors with the rate of ED use. RESULTS: Nine of 11 identifiable social complexity risk factors were associated with a higher rate of ED use. Additionally, the rate increased as the number of risk factors increased from 0 to 5 or more, reaching approximately twice the rate when 5 or more risk factors were present in children aged younger than 5 years (incidence rate ratio = 1.92; 95% CI, 1.85-2.00) and in children aged 5 to 17 years (incidence rate ratio = 2.06; 95% CI, 1.99-2.14). CONCLUSIONS: State administrative data can be used to identify social complexity risk factors associated with higher rates of ED use among Medicaid-insured children. State agencies could give primary care medical homes a social risk flag or score to facilitate targeted screening and identification of needed resources, potentially preventing future unnecessary ED use in this vulnerable population of children.

Care coordination program for Washington State Medicaid enrollees reduced inpatient hospital costs.

Managing clinically complex populations poses a major challenge for state agencies trying to control health care costs and improve quality of care for Medicaid beneficiaries. In Washington State a care coordination intervention, the Chronic Care Management program, was implemented for clinically complex Medicaid beneficiaries who met risk criteria defined by a predictive modeling algorithm. We used propensity score matching to evaluate the program's impact on health care spending and utilization and mortality. We found large and significant reductions in inpatient hospital costs ($318 per member per month) among patients who used the program. The estimated reduction in overall medical costs of $248 per member per month exceeded the cost of the intervention but did not reach statistical significance. These results suggest that well-designed targeted care coordination services could reduce health care spending for Medicaid beneficiaries with complex health care needs.

Hospitalizations of nursing home residents in the last year of life: nursing home characteristics and variation in potentially avoidable hospitalizations.

OBJECTIVES: To examine the incidence of, variations in, and costs of potentially avoidable hospitalizations (PAHs) of nursing home (NH) residents at the end of life and to identify the association between NH characteristics and a facility-level quality measure (QM) for PAH. DESIGN: Retrospective. SETTING: Hospitalizations originating from NHs. PARTICIPANTS: Long-term care NH residents who died in 2007. MEASUREMENTS: A risk-adjusted QM was constructed for PAH. A Poisson regression model was used to predict the count of PAH given residents' risk factors. For each facility, the QM was defined as the difference between the observed facility-specific rate (per 1,000 person-years) of PAH (O) and the expected risk-adjusted rate (E). A logistic regression model with state fixed-effects was then fit to examine the association between facility characteristics and the likelihood of having higher-than-expected rates of PAH (O-E > 0). QM values greater than 0 indicate worse-than-average quality. RESULTS: Almost 50% of hospital admissions for NH residents in their last year of life were for potentially avoidable conditions, costing Medicare $1 billion. Five conditions were responsible for more than 80% of PAHs. PAH QM across facilities showed significant variation (mean 12.0 ± 142.3 per 1,000 person-years, range -399.48 to 398.09 per 1,000 person-years). Chain and hospital-based facilities were more likely to exhibit better performance (O-E < 0). Facilities with higher nursing staffing were more likely to have better performance, as were facilities with higher skilled staff ratio, those with nurse practitioners or physician assistants, and those with on-site X-ray services. CONCLUSION: Variations in facility-level PAHs suggest that a potential for reducing hospital admissions for these conditions may exist. Presence of modifiable facility characteristics associated with PAH performance could help us formulate interventions and policies for reducing PAHs at the end of life.

Site of death among nursing home residents in the United States: changing patterns, 2003-2007.

CONTEXT: The proportion of US deaths occurring in nursing homes (NHs) has been increasing in the past 2 decades and is expected to reach 40% by 2020. Despite being recognized as an important setting in the provision of end-of-life (EOL) care, little is known about the quality of care provided to dying NH residents. There has been some, but largely anecdotal evidence suggesting that many US NHs transfer dying residents to hospitals, in part to avoid incurring the cost of providing intensive on-site care, and in part because they lack resources to appropriately serve the dying residents. We assessed longitudinal trends and geographic variations in place of death among NH residents, and examined the association between residents' characteristics, treatment preferences, and the probability of dying in hospitals. METHODS: We used the Minimum Data Set (NH assessment records), Medicare denominator (eligibility) file, and Medicare inpatient and hospice claims to identify decedent NH residents. In CY2003-2007, there were 2,992,261 Medicare-eligible NH decedents from 16,872 US Medicare- and/or Medicaid-certified NHs. Our outcome of interest was death in NH or in a hospital. The analytical strategy included descriptive analyses and multiple logistic regression models, with facility fixed effects, to examine risk-adjusted temporal trends in place of death. FINDINGS: Slightly more than 20% of decedent NH residents died in hospitals each year. Controlling for individual-level risk factors and for facility fixed effects, the likelihood of residents dying in hospitals has increased significantly each year between 2003 through 2007. CONCLUSIONS: This study fills a significant gap in the current literature on EOL care in US nursing homes by identifying frequent facility-to-hospital transfers and an increasing trend of in-hospital deaths. These findings suggest a need to rethink how best to provide care to EOL nursing home residents.

Cost benefits of a peer-led asthma self-management program for adolescents.

OBJECTIVE: Frequent use of health-care services associated with pediatric asthma places substantial economic burden on families and society. The purpose of this study is to examine the cost-saving effects of a peer-led program through reduction in health-care utilization in comparison to an adult-led program. METHODS: Randomly assigned adolescents (13-17 years) participated in either peer-led (n = 59) or adult-led (n = 53) asthma self-management program. Health-care utilization data were collected at baseline and at 3-, 6-, and 9-months post-intervention. Negative binomial regression models were conducted to examine the effects of the peer-led program on health-care utilization. Net cost savings were estimated based on the differences in program costs and health-care utilization costs between groups. RESULTS: Significant group differences were found in acute office visits and school clinic visits after controlling for race and socioeconomic status. The incidence rate of acute office visits was 80-82% less for the peer-led group during follow-ups. The peer-led group was four to five times more likely to use school clinics due to asthma than the adult-led group during follow-ups. The non-research cost of peer-led program per participant was lower than the adult-led program, $64 versus $99, respectively. The net cost saving from the reduction in acute office visits and the lower program costs of the peer-led program was estimated $51.8 per person for a 3-month period. CONCLUSIONS: An asthma self-management program using peer leaders can potentially yield health-care cost savings through the reduction in acute office visits in comparison to a traditional program led by health-care professionals.