Introduction: The Centers for Medicaid and Medicare Services (CMS) Bundled Payments for Care Improvement (BPCI) initiative was implemented in 2015. At our private practice, we implemented a retrospective payment model 2 for a 90‐day episode of care for cervical spine and other orthopedic procedures. Under these retrospective payment models, Medicare continues to make fee‐for‐service (FFS) payments but reconciles the total expenditures for the episode with a bundled payment amount as determined by CMS. A payment or recoupment amount is then made by Medicare reflecting the aggregate expenditures compared to the target price. The purpose of the study is to assess the value of the cervical spine CMS bundle at our private practice.
Methods: We utilized the data provided by CMS to compare the total expenditures of cervical spine diagnosis related groups (DRGs) of 471, 472, and 473. Medicare patients who underwent cervical spine surgery between January 2009‐December 2012 were defined as non‐BPCI (n = 88) and were compared to Medicare BPCI patients (n = 40) who had surgery between January 2015‐December 2015. Post‐acute events within the 90 day episode including admission to an IRF or SNF as well as home health (HH) and readmissions were analyzed. Expenditures were converted to 2016 dollars using Consumer Price Index (CPI). Normality of expenditures was assessed using the Manning & Mullahy method and expenditures were subsequently log transformed. Wilcoxon tests and a multivariate generalized estimated equation were used to determine differences between BPCI and non‐BPCI patients as well as assess the independent effects of post‐acute events.
Results: The median expenditure for non‐BPCI patients was $16,566 (IQR $14,604 ‐ $19,951) compared to $18,510 (IQR $15,936‐$23,371) for BPCI patients (p = .02). Compared to non‐BPCI patients BPCI patients had a higher rate of SNF admissions (non‐BPCI 6% vs 7.5% BPCI; p = .23), IRF admissions (non‐BPCI 1% vs. 5% BPCI; p = .68), HH (non‐BPCI 14% vs. 15% BPCI; p = .79) and readmissions (non‐BPCI 9% vs. 12.5% BPCI; p = .54). At the multivariate level, the significantly higher expenditure for BPIC patients persisted and all post acute events were significant, independent drivers of increased cost. After controlling for post acute events, BPCI patients had a 10% increase in expenditures (p = .02). Admissions to an IRF or SNF increased cost 93% (p < .001) and 56% (p < .001), respectively. HH utilization increased expenditures 26% (p < .0001) and 90‐day readmissions increased costs by 45% (p < .0001).
Conclusion: The objective of the BPCI initiative was to improve the value of health care, e.g. decreasing cost while improving outcomes. Our institution was only managing the post acute care expenditures and not the acute hospital expenditures. In spite of our best efforts to contain costs with clinical practice guidelines, patient navigators and a BPCI management team, the expenditures were significantly higher for BPCI patients. Furthermore, the outcomes defined as post acute events were not improved. The variability of surgical procedure complexity included in the 471, 472 and 473 DRGs, cervical spine bundles may not be appropriate. We have discontinued BPCI for cervical spine DRGs and are focusing our efforts on defining bundles by specific Current Procedural Terminology® codes.