As healthcare spending continues to rise in the United States, there is a strong push to mitigate healthcare costs and incentivize quality and efficiency by moving away from the classic fee-for-service model. The CMS BPCI-A alternative payment model represents an innovative approach to payment restructuring that may be necessary to meet the rising cost of healthcare. Under the BPCI model, all episodes of care have a fixed cost, with both savings and penalties placed on the hospital.3 However, surgery is nuanced, and patient needs and healthcare utilization can vary widely from one procedural area to another. Our analysis demonstrates that the major driver of 90-day healthcare utilization and costs in major bowel bundle patients is readmission at an acute care hospital during the bundle window. In contrast, the published experiences with comprehensive joint replacement found that much of the utilization was related to use of inpatient rehabilitation after discharge from an acute hospital.13–16
Based on our analysis of the major bowel bundled payment cohort, the greatest opportunity for improvement would be to reduce readmissions by accurately identifying patients who were high-risk for readmission and thus more likely to have high-cost care episodes. The ACS NSQIP SCR is a well-validated and widely available tool that has been studied in a number of broad surgical scenarios to predict readmission.17, 18 However, the calculator was unable to predict readmission risk or total care episode cost in our bundled payment patient cohort. There are several factors that may account for this failure. While the risk calculator was calibrated and validated for patients encompassing a wide range of ages, co-morbidities, and surgical conditions, our population was more complex and heterogeneous than the population utilized for the calculator.19, 20 In particular, our major bowel bundle surgical cohort was older, had higher ASA scores, higher rates of emergency surgery, and were more likely to have disseminated cancer compared to the ACS NSQIP population of colorectal surgery patients.21 The SRC might be useful for counseling a patient about their risk of various outcomes based on their co-morbidities when compared to the average NSQIP patient at the average NSQIP-affiliated hospital, but the SRC lacks the specificity and modifiability needed to risk stratify patients within specific contexts—such as single-institution bundled payment programs.22
An important caveat to the use of risk calculators concerns how these scores are used. While a score that quantifies the possibility of increased risk and cost would ideally be used to direct appropriate care and services to these patients, it also could be used to limit access to surgical care over the concern for penalties in a bundled payment model, as suggested by Tsai et. al.16 It is essential that appropriate risk adjustment techniques be employed to adjust for utilization of both in-hospital and post-discharge resources by high-risk patients. This recommendation is in line with the “Whole Person Care” approach that has been adopted by Medi-Cal, which might elucidate additional strategies for the management of patients who are more likely to incur high healthcare costs.23, 24 Assessing the impact of better integration and coordination of medical, behavioral, and social services to address the needs of high-cost patients warrants further study.
There are a number of limitations in the current study that are the result of retrospective data analysis. Miscoding or under-coding of medical conditions and complications may have occurred during the process of administrative data collection. Inaccurate or absent documentation within the electronic medical record also remains a concern. In the end, our analysis was performed on a unique but relatively small cohort, which increases the possibility for sampling bias and limits statistical power and generalizability. However, given the dearth of actual hospital experience with the Major Bowel Bundled Payment program and the accelerated integration of alternative payment models in the surgical landscape, the findings highlight the need for better study of the CMS bundle eligible general surgery patients with an emphasis on development of predictive risk models. Such models will be essential for hospitals to succeed in developing targeted interventions for those patients most at need of additional support for their transition from acute hospitalization back to the community.