Trends in sepsis and pneumonia during COVID-19: lessons from BPCIA.

OBJECTIVES: The COVID-19 pandemic affected care delivery nationwide for all patients, influencing cost and utilization for patients both with and without COVID-19. Our first analysis assessed changes in utilization for patients with sepsis without COVID-19 prior to vs during the pandemic. Our second analysis assessed cost and utilization changes during the pandemic for patients with sepsis or pneumonia both with and without COVID-19. STUDY DESIGN: A retrospective case-control study was utilized to determine differences in cost and utilization for patients with sepsis or pneumonia, relative to a COVID-19 diagnosis. METHODS: Claims data from 8 teaching hospitals participating in sepsis and pneumonia episodes in the Bundled Payments for Care Improvement Advanced (BPCIA) model were utilized. BPCIA is a Medicare value-based care bundled payment program that aims to decrease costs and increase quality of care through a 90-day total cost of care model. RESULTS: The first analysis (N = 1092) found that non-COVID-19 patients with sepsis had 26% higher hospice utilization (P < .05) and 38% higher mortality (P < .0001) during the pandemic vs the prepandemic period. The second analysis (N = 640) found that during the pandemic, patients with sepsis or pneumonia with COVID-19 had 70% more skilled nursing facility (SNF) use (P < .0001), 132% higher SNF costs (P < .0001), and 21% higher total episode costs (P < .0001) compared with patients without COVID-19. CONCLUSIONS: COVID-19 has affected care patterns for all patients. Patients without COVID-19 postponed care and used lower-acuity care settings, whereas patients with COVID-19 were more costly and utilized postacute care at a higher rate. These analyses inform future care coordination initiatives, given the ongoing pandemic.

Succeeding in Medicare's newest bundled payment program: Results from teaching hospitals.

BACKGROUND: In 2018, Medicare implemented a successor to its Bundled Payments for Care Improvement (BPCI) program, BPCI Advanced, with stricter participation rules and new financial incentives to reduce spending. METHODS: Using claims-based episode data from thirteen participants, we compared spending and utilization in the first fifteen months of the new program (October 2018 to December 2019) to hospital- and episode-specific target prices, with a deep dive into clinical correlates for the most commonly-selected clinical episodes, sepsis and congestive heart failure. RESULTS: Twelve out of thirteen participants in a collaborative of teaching hospitals achieved shared savings for both Medicare and their own institution. Aggregate hospital shared savings were 5.8% of benchmark prices across 6,131 patients in 16 clinical episodes (p<0.001), appreciably higher than the reference savings rates reported after the first period of Medicare's predecessor BPCI program. Differences in shared savings across hospitals for sepsis and congestive heart failure correlated with reductions in patients' use of post-acute care, including reductions in skilled nursing facility, readmission, and home health rates. Evidence is presented showing reductions in patient utilization for cost-intensive post-acute settings accompanied increases in the proportion of patients exclusively utilizing non-institutional care after discharge from an anchor stay or procedure. CONCLUSIONS: These findings provide an example of the fulfillment of a core promise of bundled payments to uncover new opportunities for reduced spending. LEVEL OF EVIDENCE: Non-random cohort of hospitals.

Evaluating the Addition of Clinical and Staging Data to Improve the Pricing Methodology of the Oncology Care Model.

PURPOSE: The Oncology Care Model (OCM) is the largest value-based care model focusing on oncology, but the current pricing methodology excludes relevant data on the cancer stage and current clinical status, limiting the precision of the risk adjustment. METHODS: This analysis evaluated 15,580 episodes of breast cancer, lung cancer, and multiple myeloma, starting between July 1, 2016, and January 1, 2020, with data from a cohort of OCM practices affiliated with academic medical centers. The authors merged clinical data with claims for OCM episodes defined by the Center for Medicare and Medicaid Innovation to identify potential quality improvement opportunities. The regression model evaluated the association of the cancer stage at initial diagnosis and current clinical status with variance to the OCM target price. RESULTS: Cancer stage at the time of initial diagnosis was significant for breast and lung cancers, with stage IV episodes having the highest losses of -$6,700 (USD) for breast cancer (P < .001) and -$18,470 (USD) for lung cancer (P < .001). Current clinical status had a significant impact for all three cancers in the analysis, with losses correlated with clinical complexity. Breast cancer and multiple myeloma episodes categorized as recurrent or progressive disease had significantly higher losses than stable episodes, at -$6,755 (USD) for breast (P < .001) and -$19,448 (USD) for multiple myeloma (P < .001). Lung cancer episodes categorized as initial diagnosis had significantly fewer losses than stable episodes, at -$3,751 (USD) (P = .001). CONCLUSION: As the Center for Medicare and Medicaid Innovation designs and launches new oncology-related models, the agency should adopt methodologies that more accurately set target prices, by incorporating relevant clinical data within cancer types to minimize penalizing practices that provide guideline-concordant cancer care.

Patient Outcomes in a Medicaid Managed Care Lock-In Program.

BACKGROUND: Prescription drug abuse is a growing epidemic in the United States, and opioids are among the most commonly abused and misused controlled substances. Managed care organizations can use pharmacy lock-in programs to limit patients' access to opioids by requiring that they receive all scripts from 1 prescriber, potentially reducing inappropriate use. OBJECTIVE: To evaluate opioid use patterns among patients in a Medicaid managed care lock-in program limiting opioid coverage to prescriptions written by assigned prescribers. METHODS: This retrospective cohort study included all patients enrolled in the lock-in program at Blue Care Network (BCN) of Michigan Medicaid managed care from March 2008 through May 2013, with outcomes assessed through August 2013. BCN medical and pharmacy claims, the Michigan Automated Prescription System, and Blue Cross Complete Controlled Substance Committee reports were used to assess outcomes at 6, 12, 24, and 36 months after enrollment. Patients were defined as "stable" if they exclusively filled opioid prescriptions from assigned prescribers or received treatment for opioid dependence and "unstable" if they purchased prescription opioids with cash or submitted opioid claims not prescribed by assigned providers. RESULTS: Of the 59 patients enrolled in the program, over half (55.9%) dropped BCN coverage, and 1 died while enrolled. The proportion of patients who dropped coverage fell as time in the program increased, from 29% in the first 6 months to 11% semiannually after 24 months. Among those who remained enrolled, the proportion of stable patients increased from 31% at 6 months to 78% at 36 months. The small sample size did not permit formal statistical analysis. CONCLUSIONS: The finding that most patients exited the program by dropping coverage was an unintended consequence meriting further investigation. Conversely, the finding that patients who remained enrolled largely achieved desired outcomes indicates that this program played an important role in addressing opioid abuse.