Bundled payment for gastrointestinal hemorrhage

The Bundled Payment for Care Improvement (BPCI) program built on these results in 2013 by expanding to include Part A and B services rendered up to 90 days after discharge, and as of January 2016, it includes 1,574 participants across the country. On a voluntary basis, hospitals, physician groups, and postacute providers and conveners were able to participate in 1 of 4 bundled payment models that were anchored on an inpatient for any of 48 clinical conditions that were based on MS-DRG (Table 1).

• Model 1 defined the episode as the inpatient hospital stay and bundled the facility and physician fees, similar to prior demonstration projects.

• Model 2 is a retrospective bundled payment for Part A and B services in the inpatient hospital stay and up to 90 days after discharge.

• Model 3 is a retrospective model that starts after hospital discharge and includes up to 90 days. (Models 1-3 maintain the current payment structure and retrospectively compare the actual reimbursement with target values that are based on historical data for that hospital with a 2%-3% payment reduction.)

• Model 4 makes a single, prospectively determined global payment to a hospital that encompasses all services during the hospital stay.

Opportunities in inpatient and postacute bundled payments

Participation in bundled payments requires a new set of analytic and organizational capabilities.

• The first step is to identify the patient population on the basis of inclusion and exclusion criteria and to measure the current cost of care through external claims data and internal hospital data. This includes payments for hospital inpatient services, physician fees, postacute care, readmissions, other Part B services, and home health services. The biggest opportunity for postacute bundles is shifting site of service from postacute care to lower-cost settings and reducing readmission rates.

• Subsequently, they need to identify areas of opportunity to reduce expenditure, while also demonstrating consistent or improved quality and outcomes.

• On the basis of this, the team can identify variation in care within the cohort and in comparison with benchmarks across the country.

• After identifying areas of opportunity, the team needs to develop strategies to improve value such as care pathways, information technology tools, care coordination, and remote services.

Of the 48 clinical conditions in BPCI, 4 could be described as related to GI: esophagitis, gastroenteritis, and other digestive disorders (Medicare Severity–Diagnosis Related Group [MS-DRG] 391, 392); gastrointestinal hemorrhage (MS-DRG 377, 378, 379); gastrointestinal obstruction (MS-DRG 388, 389, 390); and major bowel procedure (MS-DRG 329, 330, 331). After evaluating the GI bundles, it was apparent that these were created for billing purposes and were not clinically intuitive, which is why our institution immediately excluded the broad category of esophagitis, gastroenteritis, and other digestive disorders. GI obstruction and major bowel surgery relate to the care of gastroenterologists, but surgeons are typically primary drivers of care for these patients. Thus, we believed that GI hemorrhage was most appropriate because gastroenterologists drive care for this condition, and there is substantial evidence about established guidelines and pathways during this episode.

Bundled payment for gastrointestinal hemorrhage

We built a multidisciplinary team of physicians, data analysts, clinical documentation specialists, and care managers to start developing a plan for improving the value of care in this population. This included data about readmissions and site of postacute care for this population, which were supplemented by chart review of financial outliers and readmissions. We quickly learned about some of the challenges to medical bundles and the GI hemorrhage bundle in particular. It is difficult to identify these patients early in the hospital stay because inclusion is based on a billing code. Many of these patients also have cardiovascular disease, cancer, or cirrhosis, which makes it hard to identify which patients will end up with primary GI hemorrhage coding until after the patient is discharged. They are also on many different inpatient services; in our hospital, there were at least 12 different admitting services. In addition, almost one-third of the patients actually had an admission before this hospitalization, often for different clinical conditions.

Most importantly, it was very challenging to develop protocols to improve the value of care in this population. Most of the patients had many comorbid conditions, so a GI hemorrhage pathway alone would not be sufficient to alter care. The two main areas of opportunity for cost savings in postacute bundled payments are postacute site of service and readmissions, both of which are hard to change for medical GI patients. For medical patients, they have many comorbidities before admission, so postacute site of service is typically driven by which site they were admitted from. This is different from surgical patients who are in SNF or rehabilitation facilities for limited time frames, and there may be more discretion to shift to lower cost settings. In addition, readmissions have not been studied much in GI hemorrhage, so it is not clear how to improve them. On the basis of these factors and the limited sample size for this condition, our health system opted to stop taking financial risk for this population.