How A Medicare Payment Model is Affecting Care for Older Adults with Alzheimer’s Disease and Related Dementias

Older adults with Alzheimer’s disease and related dementias (ADRD) who need joint replacement may not undergo surgery because their ADRD may be considered a contraindication.¹ Not only that, older adults with ADRD (compared to those without) are more likely to experience costly postsurgical complications, such as delirium, prolonged length of stay, and functional impairment.^2,3 Policies that may impact provisions for older adults with ADRD, like the Comprehensive Care for Joint Replacement (CJR) Model, should be evaluated to ensure appropriate care.

Starting in 2016, the CJR Model was developed to reduce spending while maintaining or improving the quality of postsurgical care for joint replacements. In the Model, postsurgical care provided by clinicians, such as in skilled nursing facilities (SNF), is paid for as part of a 90-day bundle. Outside of Model-participating hospitals, such care is paid for separately. Given that hip and knee replacements are costly and the most common surgeries for Medicare, the rationale for bundling payments within the CJR Model is to incentivize hospitals to decrease expenditures while being held financially accountable to maintain or improve quality of care.^4,5 There may be concern that bundled payments could lead to ‘cherry-picking’ of healthier patients less likely to incur high expenditures after surgery; however, this has not been demonstrated in a prior study evaluating the impact of the CJR Model on the receipt of joint replacement across all Medicare patients.⁶ Still, it is possible that the CJR Model could lead to ‘cherry-picking’ in patients with ADRD because of the burden of postsurgical complications.

This issue’s study by Kim and colleagues addresses this concern and sheds light on the impact of the CJR Model on older adults with ADRD.⁷ Using Medicare administrative claims data from 2013 to 2017 and applying a differences-in-differences analysis, the authors compared the number of joint replacements per 1,000 enrollees per quarter in older adults with ADRD residing in Model-participating areas versus non-participating areas, and found no association between the CJR model and receipt of joint replacement. This finding did not change when stratifying by type of joint replacement surgery—elective knee replacement, elective hip replacement, or post-fracture hip replacement. Even among more medically complex ADRD subpopulations, such as nursing home (NH) residents and NH residents with severe cognitive and/or functional impairments, the CJR model was not found to have an impact on the receipt of joint replacement.

This study also assessed the impact of the CJR Model on Medicare spending and different types of healthcare resource use (i.e., institutional post-acute care, home health, and hospice) among older adults with ADRD who received surgery. The CJR Model led to a modest reduction in total spending (−$1,029 per episode, 95% confidence interval [CI]: −$1,577 to −$481) in older adults with ADRD who received surgery. This reduction in spending was primarily driven by fewer discharges to and days spent in post-acute care settings (−1.62% discharges, 95% CI: −3.17% to −0.07% and −1.79 days among those discharged to institutional post-acute care settings, 95% CI: −2.89 to −0.70, respectively). When comparing spending on types of post-acute care settings, the reduction in spending was largely driven by decreased spending on SNF (−$754 per episode, 95% CI: −$1,164 to −$345).

When examining the subpopulations who received an elective joint replacement versus those with a post-fracture hip replacement, both subpopulations had reduced spending (−$1,375 per episode, 95% CI: −$2,125 to −$624 in those with elective joint replacement and −$1,063 per episode, 95% CI: −$1,825 to −$300 in those with post-fracture hip replacement) after the CJR Model was implemented. However, only patients who received elective joint replacement surgery were less likely to be discharged to post-acute care settings (−4.01% discharges, 95% CI: −7.1% to −0.92% among those with elective surgery versus −1.01% discharges, 95% CI: −2.7% to 0.68% among those post-fracture), suggesting that post-acute care was necessary after fracture.

When comparing ADRD patients seen at hospitals participating versus not participating in the CJR Model, there were no differences across postsurgical outcomes, including functional status, 90-day readmissions, 90-day emergency department use, and 90-day mortality. In subpopulations, 90-day readmissions improved for those who received an elective joint replacement (−2.22%, 95% CI: −3.92% to −0.52%) and did not change for those with a post-fracture hip replacement (−1.01%, 95% CI: −2.7% to 0.68%).

This study’s findings are reassuring, but also highlight areas of improvement. First, the CJR Model has not led to reduced joint replacement surgeries in ADRD patients, a finding consistent with another recent study.⁸ Second, this study demonstrates that the CJR Model has decreased spending among ADRD patients while maintaining quality of care, at least when evaluating 90-day hospital readmissions, 90-day emergency department use, 90-day mortality, and functional status improvement as measured by ADL score. Now, the CJR Model should focus on not only maintaining, but also improving quality of care. In addition to the bundled payments in the CJR Model, the Centers for Medicare and Medicaid Services (CMS) provide tools to Model-participating hospitals to help them improve coordination of care and facilitate sharing of best practices between participant hospitals, which could improve quality of care.⁴ Yet, the only instance identified in this study of improved quality in postsurgical outcomes was in reduced 90-day readmissions among the subpopulation with an elective joint replacement. This calls attention towards improving 90-day readmissions in patients with post-fracture hip replacement, and improving other quality of care measures, such as 90-day mortality, 90-day emergency department visits, and functional improvement for the entire ADRD population. Additionally, the present study did not report on spending or postsurgical outcomes in the more medically complex ADRD populations, such as NH residents with or without severe cognitive or functional impairments. Future examination of these subpopulations could reveal further room for improvement in quality of care delivered through the CJR Model.

Additionally, this study showed that reduced spending was primarily driven by reduced use of institutional post-acute care. However, reduced use of post-acute care may not be beneficial for all older adults with ADRD. If older adults with ADRD are increasingly discharged home (with or without home health), the amount of informal caregiving they need will likely increase, and so will the burden on caregivers. In the first year after diagnosis, community-dwelling ADRD populations already receive on average 25 caregiving days per month.⁹ Caregiver burden can lead to decreased care provision, as well as physical and psychological health deterioration for the caregiver, with in-home caregiving being associated with greater psychological burden and worse mental health than out-of-home caregiving.^10–13 Shifting post-acute care to informal caregivers could also exacerbate other problems, such as worsening the already enormous financial burden on families of ADRD patients, which is estimated at 70% of the total lifetime cost burden of ADRD.¹⁴

While the current CJR Model does not appear to promote ‘cherry-picking’ within the ADRD population, geriatricians should be cognizant of the potential impact of CJR in other patient subgroups. First, patients from minority populations may receive differential treatment. Two studies found that the CJR Model was associated with differences in receipt of joint replacement by race and ethnicity, among the broader Medicare population (i.e., not specific to those with ADRD).^15,16 Second, geriatricians should consider the impact of the CJR Model on patients who are seen at a hospital that is no longer participating in the Model. Over time, the number of hospitals participating in the CJR Model has varied, increasing from 689 to 718 between 2016 and 2018, and then eventually dropping to 324 in 2021.^4,5 Thus, patients going to the same hospital for their procedures may have different experiences and care trajectories depending on whether the hospital is currently participating in the CJR Model or not.

The present study found that from 2013 to 2017, the CJR Model reduced spending while maintaining care quality, with no negative impacts on the receipt of joint replacement among older adults with ADRD. Further studies in ADRD subpopulations, and a more holistic view encompassing caregiver burden, could reveal ways to improve the quality of care for older adults with ADRD and their families within the CJR Model.