CMS Analyzes Spending and Savings Across Alternative Payment Models
Leaders at the federal Centers for Medicare and Medicaid Services (CMS) have compiled an analysis of 21 different alternative payment models (APMs) that have been moving forward under the Medicare program, either as comprehensive models or as specialized models in terms of spending and savings encountered, including, among others, the Next Generation ACO Model (now discontinued), the Comprehensive Primary Care (CPC+) model, the Bundled Payments for Care Improvement (BPCI) Initiative, and others. The Medicare Shared Savings Program was not among the models analyzed.
As shared on the CMBS website, the document begins thus: “The goal of this study was to synthesize evaluation results across the portfolio of CMS Innovation Center models to inform future model development. We examined 21 Medicare models and demonstrations with at least two years of impact estimates that cover interventions operating between 2012 and 2020.” And, the unnamed analysts write that, “In this synthesis, we looked at measures that were available across multiple model evaluations. These included gross and net Medicare spending and measures of utilization (e.g., inpatient admissions, emergency department visits, post-acute care, inpatient readmissions) and quality of care (e.g., self-reported satisfaction with care, mortality). Results were summarized across models to gain a broader understanding of themes by interventions, care settings, provider types, and beneficiary target populations. Many of the results presented are either final estimates or represent at least four years of performance, which is generally sufficient time to start drawing conclusions. However, results for a few models represent only two to three years of performance and should be considered preliminary; conclusions may change over time.”
What have been the results? The Medicare analysts write that, “Across the 21 Medicare models, more than half (fourteen) demonstrated gross savings to Medicare. Changes in spending were driven by improvements in inpatient admissions (ten models) and/or post-acute care (fourteen models). For models that paid financial incentives to participants, six had net savings, six incurred net losses, and six models had no notable impacts on net spending. Beneficiary or caregiver self- reported experience of care remained relatively unchanged by the majority (nine) of models, improved in two models, and had unfavorable results (which were small in magnitude) for one model. Mortality remained unchanged for most (eight) models, suggesting that these interventions did not cause any harm, and there were notable improvements in mortality in four models. Models that focus on reducing acute or specialty care or that targeted specific populations1 (e.g., terminal illness, lower extremity joint replacements) were more likely to show gross savings and generally had larger, more favorable impacts on utilization relative to models focused on primary care and population management2 which generally serve broader, healthier populations.”
Given all that, they write, “It is possible that the higher baseline spending of sicker beneficiaries, the inclusion of institutional and specialty care providers, and the more narrowly focused target populations in the Acute or Specialty Care and Targeted Population models provided more room to reduce spending. Primary Care and Population Management models served large panels of relatively healthy, mostly low-cost Medicare beneficiaries and focused on preventing disease and improving care coordination. Longer time windows for investments in care coordination, staffing, clinical workflow redesign, health information technology, and data analytics, as well as greater engagement of primary and specialty care providers, may be needed to reduce spending in Primary Care and Population Management models. Even with successful evaluation results and transformation efforts, models may face other barriers to national expansion. Generous financial incentive payments, which helped ensure robust participation in models, made it difficult for many models to demonstrate net savings. Voluntary models also were encumbered by participants exiting the model prior to Medicare being able to realize returns or savings.”