LAUSR

As alternative payment models take shape, Kim et al are to be commended for exploring the end-of-life (EoL) care in the patient population served by accountable care organizations (ACOs). The authors’ efforts enter an area where there is limited literature and understanding because commonly accepted quality metrics of EoL care such as those laid out by the National Quality Forum and American Society of Clinical Oncology are not among the 33 ACO quality measures set by the Centers for Medicare & Medicaid Services. The authors conduct this observational study across 21 hospital referral regions in the eastern United States by matching ACO-attributed beneficiaries with those non-ACO associated through propensity score matching. The authors report findings that were not as compelling as we would have expected (or hoped). Rather than seeing a robust difference in EoL care intensity that is not only statistically significant but also clinically relevant, the authors report that in fact there were no noteworthy differences between the non-ACO and ACO cohorts in three key quality metrics: chemotherapy in the last 14 days of life, hospice enrollment within the last 3 days of death or no hospice care, or two or more emergency department (ED) visits in the last month of life. The two measures of quality that did show a statistically significant difference were (1) intensive care unit (ICU) admissions, where the rates were higher in the ACO cohort vs the non-ACO group, and (2) two or more hospitalizations in the last month of life with the ACO cohort having a lower rate of multiple hospitalizations than the non-ACO cohort. These findings begged the question (although they are statistically significant), are these differences clinically and/or fiscally meaningful? The authors propose that the lower readmission rates may be attributable to better coordination of care within the ACO delivery model. However, one can argue that the two metrics of statistical significance would be of greater practical consequence if we knew the length of stay or number of days hospitalized in the last month of life. One can postulate with equal vigor that the ACO cohort has a lower rate of repeated hospitalizations because they are already admitted (and in the ICU), thereby not having the opportunity to be discharged and readmitted. The establishment of ACOs may reflect a local culture change in the way healthcare is delivered; consequently, the level of adoption of ACOs in a region may make a difference. If a large number (or even a majority) of providers and healthcare facilities coordinate care voluntarily through ACOs, then that market’s population may demonstrate different trends in practice and care delivery than a more traditional non-ACO fee for service model. Does the extent of adoption of ACOs within a geographic region make the difference in outcomes of the hospital referral region? The authors emphasize the value of studying the intensity of EoL care in the ACO beneficiaries when they report lower adjusted rates of both repeated ICU admissions (one or more ICU admits) and chemotherapy within the last 2 weeks of life in the markets with high ACO penetration, defined as the number of patients in the ACO group being more than 40% of the number in non-ACO group within that same regional market. Indeed, the findings presented from this crosssectional study further underscore the need for robust prospective studies to study whether this principle of patientcentered coordination of care through ACOs does in fact translate into more optimal EoL experience with decreased healthcare utilization, especially within the last month of life. Further stratification of outcomes prospectively by integrated health system–based ACOs vs nonhospital-based ACOs may provide further insight into the authors’ notably discordant findings of increased aggressive EoL care in the ICU within ACO patients as well as those where a change was hypothesized but not seen, such as ED visits, delayed or no hospice referrals, and anticancer therapy at the EoL. To conclude, the authors provide valuable early insight into the EoL experience of ACO beneficiaries with cancer and, in the process, highlight the critical need for the integration of EoL qualitymeasures into themeasures of performance of ACOs.