Bundled Payments and Reference Pricing

There is a lot of attention around increasing healthcare coverage, but many people working on healthcare value argue for greater focus on controlling healthcare costs instead. I’ve highlighted research that ties much of the increase in healthcare costs over the last decade to rising provider and hospital prices rather than increasing utilization rates. One would assume that health insurance companies are in the best position to help moderate rises in provider and hospital costs. Working for a large health insurance company, I’ve come to realize that commercial healthcare insurance is an elegant interaction between what healthcare offerings an employer would like to offer its employees and what costs that employer is willing to bear to provide those offerings. Excluding a popular, but low-value service or a medical center with high costs to match their perceived value in the employees’ mind (e.g., academic centers, health systems with large marketing budgets) can be challenging.

Dividing Healthcare into Episodes Along a Care Journey

The individual’s symptom-based clinical journey might include: 1) Symptom(s) to Diagnosis, 2) Diagnosis to Treatment (resolution or control of the symptom[s]), and 3) Treatment to Surveillance. This paradigm ignores medical decision-making that might incorporate patient preferences without any meaningful impact on symptom control (e.g., behaviors that might reduce the onset of symptoms from an infectious [sexually transmitted disease] or non-infectious [weight gain, substance abuse] cause, actions that might prolong an individual’s life without a direct impact on that individual’s symptom burden).

The opportunity space for healthcare cost control from a clinical perspective might divide symptoms into:

  • Self-limiting – amenable to virtual care but could increase healthcare utilization without a concurrent increase in healthcare value (upper respiratory infections),
  • Untreatable – could be managed by virtual care, but would require enough trust from the patient to accept that no treatment exists to manage the condition (advanced glioblastoma), or
  • Linked to a diagnosis that is associated with acceptable treatments that can increase the individual’s quality of life – should be identified early and often (hypertension, advanced hip osteoarthritis).

The term “untreatable” and phrases “acceptable treatments” and “quality of life” vary from individual to individual and even vary within individuals over the course of their lives or specific circumstances. In the absence of a healthcare insurance model that can accomodate this level of variability, health insurance companies develop arbitrary coverage thresholds that can be supported by specific policies and premiums. But even with this caveat, health insurance companies and other entities paying for healthcare might be able to better manage healthcare spending by spending additional resources earlier in the care journey to to determine what type of symptom is driving the member’s utilization.

Controlling Costs When Managing Individuals From Diagnosis to Treatment

Reference pricing, when a payer establishes a standard price for a drug, procedure, service or bundle of services that require employees/members to pay any allowable charges above the set price (as an alternative to tracking out-of-pocket costs), is a variation on conventional network-based plan design (see White and Eguchi’s policy paper back in October 2014 for a deeper exploration). Reference pricing has been most commonly proposed to address shoppable commodity services (services that can be 1) scheduled in advance, 2) interchanged for one another in terms of quality, 3) delivered by more than one provider in a market and 4) compared by price across providers), typically laboratory and radiology testing, rather than a bundles of goods and services for a specific intervention (e.g., knee replacement). White and Eguchi estimated that 73 of the 100 highest spending DRGs (including eight of the 10 highest spending DRGs) and 90% of the 300 highest spending ambulatory procedures might be shoppable. In a review of autoworker claims data, shoppable services accounted for about 1/3 of total spending, with 3/4 of that 1/3 being spent in ambulatory care (physician visits, ambulatory procedures, imaging and lab tests). The researchers go on to point out that entities employing reference pricing would be most likely to see a savings of about two percent due to the large variation but small total spend on imaging and laboratory tests (where reference pricing might be effective) and the small variation but larger total spend on office-based services (where reference pricing is unlikely to be effective). In April 2017, California’s Public Employees Retirement System (CalPERS), proposed to expand their 2012 three procedure (colonoscopy, cataract surgery and arthroscopy) reference pricing pilot to include an additional 12 procedures (upper gastrointestinal endoscopy with biopsy, laparoscopic gallbladder removal, upper gastrointestingal endoscopy, esophagoscopy, sigmoidoscopy, hyesteroscopy uterine tissue sample, nasal/sinus submucous resection inferior turbinate, tonsillectomy and/or adenoidectomy [under age 12], nasal/sinus corrective surgery – septoplasty, lithotripsy – fragmenting kidney stones, hernia inguinal repair [age 5+, non-laparoscopic], repair of laparoscopic inguinal hernia). In November 2018, the American Academy of Actuaries published a position paper suggesting shoppable services might account for 43% of all healthcare spending among individuals under 65 covered by employer-sponsored health insurance. They suggested that reference pricing could reduce spending on shoppable services by 0-28% (accounting for 0-12% of spending on all services). Employers and payers could expect to see higher savings if high-priced providers dropped their prices in response to a published reference price and consumers shifted to lower-priced providers.

The Centers for Medicare and Medicaid Services (CMS) recently tested four types of bundled payment (Bundled Payment for Care Improvement [BPCI]) that could support reference pricing in the future:

  • Model 1: Inpatient stay in an acute care hospital (discounted amount compared to Medicare’s Inpatient Propsective Payment System with separate payments for physician services),
  • Model 2: Inpatient stay plus post-acute care and all related services up to 90 days after hospital discharge (retrospective),
  • Model 3: Initiation of post-acute care services with a skilled nursing facility, inpatient rehabilitation facility, long-term care hospital or home health agency (retrospective), and
  • Model 4: Prospectively determined bundled payment for all services furnished by the hospital, physicians and other practitioners during the entire inpatient stay.

In October 2018, the Lewin Group published their analysis of episodes enrolled between the fourth quarter of 2013 and the four quarter of 2016 in Models 2-4. The overwhelming number of cases enrolled across the three models were for major joint replacement of the lower extremity (Model 2: 225,378/694,382; Model 3: 20,602/88,680; Model 4: 5849/13,551). The Lewin Group categorized clinical episodes into: High proportion of total baseline episode payments driven by post-acute care, high proportion of total baseline episode payments driven by the anchor inpatient stay, chronic episodes, and planned episodes. Here’s a sampling of the observed spending for each of the four categories:

Episode [Model] Managing Group Risk-adjusted difference in differences for total allowed payment amount included in the bundle definition
Major Joint replacement of lower extremity (driven by post-acute care) [2] Hospitals -$1230**
Major joint replacement of lower extremity (driven by post-acute care) [2] Physician Group Practice -$1958**
Major joint replacement of lower extremity (driven by post-acute care) [3] Skilled Nursing Facility -$1849**
Major joint replacement of lower extremity (driven by post-acute care) [3] Home Health Agency -$512
Urinary tract infection (driven by post-acute care) [2] Hospitals -$937**
Urinary tract infection (driven by post-acute care) [2] Physician Group Practice -$683
Urinary tract infection (driven by post-acute care) [3] Skilled Nursing Faclity -$2312**
Spinal fusion (non-cervical, driven by anchor stay) [2] Hospitals -$1497*
Spinal fusion (non-cervical,driven by anchor stay) [2] Physician Group Practice -$2507**
Percutaneous coronary intervention (driven by anchor stay) [2] Hospitals $710
Percutaneous coronary intervention (driven by anchor stay) [2] Physician Group Practice -$438
Congestive heart failure (chronic episode) [2] Hospitals -$231
Congestive heart failure (chronic episode) [2] Physician Group Practice -$200
Congestive heart failure (chronic episode) [3] Skilled Nursing Facility -$1436*
Congestive heart failure (chronic episode) [3] Home Health Agency -$791*
Chronic Obstructive Pulmonary Disease (COPD), bronchitis & asthma (chronic episode) [2] Hospitals -$338
COPD, bronchitis & asthma (chronic episode) [2] Physician Group Practice -$73
COPD, bronchitis & asthma (chronic episode) [3] Skilled Nursing Facility $1217
Cardiac valve (planned episode) [2] Hospitals -$374
Major bowel procedure (planned episode) [2] Hospitals -$1491*

*p < 0.1, **p < 0.05.

The bundled payment intervention’s association with lower costs seems to be strongest for episodes driven by post-acute care. The relationship is weaker for episodes driven by the anchor admission. There does not appear any relationship between the bundled payment intervention and costs for chronic conditions or planned episodes. The Lewin Group made the following conclusions:

  • Medicare payments declined for three quarters of the clinical epsiode combinations evaluated with little change in quality of care.
  • Self-reported changes in functional status did not differ between beneficiaries with BPCI episodes and comparison beneficiaries; slightly less favorable views of care experiences and satisfaction under BPCI.
  • BPCI-participating providers differed from the average provider in ways that may have contributed to their ability or willingness to engage in the initiative.
  • Evidence suggests that for some clinical episode strata, patient resource needs may have declined for skilled nursing facility and home health agency episode initiators relative to the change for comparison providers.
  • BPCI had no impact on total market volume of non-fracture major replacement lower extremity procedures.
  • Reconciliation payments offset the reductions in payments due to BPCI, resulting in a net loss to Mediare of $202.1 million ($268/episode) under Model 2 and $85.2 million ($921/episode) under Model 3.

Model 2

  • The relative resource intensity of BPCI patients did not change during the intervention, expect in two clinical episode strata (non-cervical spinal fusion [hospital-initiated], major joint replacement lower extremity [medical group- and hospital-initiated]). In both cases, less resource-intensive patients were recruited into BPCI.
  • Hospitals were able to reduce episode payments for vulnerable beneficiaries (e.g., eligible for both Medicare and Medicaid, have Alzheimer’s disease or related dementias, received institutional care within five days prior to the anchor hospitalization.
  • Lower use of institutional post-acute care contributed to reduced Medicare payments; there were also reductions in home health agency payments for physician group-initiatied episodes.
  • Reductions in institutional post-acute care use and unplanned readmissions, as well as reduced patient complexity, were associated with greater per-episode net payment reconciliation amount.
  • Claims-based measures generally indicated that quality of care did not change.

Model 3

  • For skilled nursing facility clinical episodes, declines in skilled nursing facility payments contributed to total payment reduction; payment results were mixed for home health agency clinical episodes.
  • The model had limited, mixed impact on quality of care.

Model 4: Because participation was low, it achieved little impact.

CMS deployed a follow-up model (BPCI-Advanced) that started in October 2018 and will run through December 2023. BPCI-Advanced is a single retrospective payment with a 90-day clinical episode duration that will include 31 inpatient episodes (e.g., major joint replacement of the lower extremity; renal failure; urinary tract infection; acute myocardial infarction; congestive heart failure; COPD, bronchitis, asthma; inflammatory bowel disease) and four outpatient episodes (back and neck pain except spinal fusion, cardiac defibrillator, major joint replacement of the lower extremity, and percutaneous coronary intervention). Acute care hospitals and physician group practices participate on a voluntary basis. Target prices are specificied a priori with payments linked to quality measure performance. It is not clear if this newest iteration of bundled payments will demonstrate savings for CMS.

Walmart’s Ventures into Managing the Clinical Journey

Walmart recently launched a more comprehensive telehealth program with Doctors on Demand that decreases copays from $49 to $4 for employees in Minnesota with an expectation that these visits can serve as “virtual primary care.” Walmart is also working with Grand Rounds, and HealthSCOPE Benefits (a company purchased by UnitedHealthGroup, my employer).

Walmart’s Centers of Excellence program (download the 2020 Associate Benefits Book and go to page 77) uses an aggressive network design that pays nothing for services delivered outside the Centers of Excellence for specific procedures in the program except for emergencies. The employer’s Center of Excellence program includes:

  • Certain heart conditions (ages 18 and up)
  • Certain spine conditions
  • Hip replacement surgery
  • Knee replacement surgery
  • Medical record review for
    • Specific cancers (breast, colorectal, lung, prostate, blood [including myeloma, lymphoma, leukemia])
    • Outpatient kidney dialysis or end-stage renal disease to determine if an on-site evaluation for kidney transplant evaluation is recommended
  • Liver, kidney, heart, lung, pancreas, multiple organ, and bone marrow/stem cell transplants (including CAR T-cell treatment)
  • Gastric bypass and gastric sleeve weight loss surgeries (ages 18 and up)

Walmart appears to be addressing the “Symptom to Diagnosis” with increasing access to telemedicine for employees and re-evaluating placing primary care clinics near their physical locations for customers. Their remote chart review and Centers of Excellence are additional methods to improve diagnostic accuracy without relying on providers or health systems with varying levels of quality or reliability. The company is also facilitating “Diagnosis to Treatment” by restricting their payments for employees to receive select procedures at Centers of Excellence.


Bundled payments and reference pricing may be restricted to a limited set of use cases within a clinical journey. Some employers are testing other approaches to consider other parts of the clinical journey with and without virtual medicine. From my perspective, virtual medicine might be more effective at addressing “Treatment to Surveillance” and facilitating more care providers to compete in the “Diagnosis to Treatment” segment of the clinical journey. Artificial intelligence (including machine learning) might augment the clinician-patient interaction for “Symptom to Diagnosis.”