Faced with new payment models from commercial payers and the Medicare Access and Children’s Health Insurance Program Reauthorization Act of 2015, U.S. health care systems are looking for ways to meet increasing market demand for value, defined as reduced total costs of care with improved clinical outcomes and patient experience.1 Strong incentives include the growth of risk-based financial contracts in which health care systems face potential financial losses if value improvement goals are not met across a population. With so much at stake for patients and health care systems, redesigning care to create value has become an urgent priority. Health care systems are struggling most with how to reduce total costs of care, as there are few existing models to adapt locally.
Effective approaches to creating value will incorporate an understanding of patients’ needs and meet them through interventions at all risk tiers to achieve short- and long-term returns.2 An approach that addresses only the most expensive 1% of patients (accounting for 22% of expenditures) may not reduce spending as these patients’ complex medical and social needs require significant resources.2 Such an approach also focuses efforts on individual patients rather than building the infrastructure for populations. Approaches for reducing long-term costs and unnecessary care must include efforts across all risk tiers, including middle-risk patients (the 19% of patients accounting for 58% or more of health care expenditures)3,4 and bottom-risk patients (the 80% of patients accounting for 20% of expenditures), who require health care screening and condition-specific preventive therapies. Many health care systems seek an understanding of how to effectively target their resources to subpopulations of patients who can most benefit from interventions. Progress toward identifying these subpopulations and implementing interventions, however, can be impeded by limited resources when health care systems’ departments function independently within silos.
Here, we report how we reconceptualized care delivery for patients with three expensive chronic conditions across our large health care system as part of a multipronged strategy (including bundles, pharmaceuticals, etc.) to reduce expenses. We believe that population health approaches such as ours can reduce total costs of care and improve value. At UCLA (University of California, Los Angeles) Health, population health means being accountable to our full population so that patients receive high-quality care at lower cost regardless of care site and whether they are seen primarily by primary care or specialty teams. We describe our system-wide Population Health Value (PHV) model designed to reduce expenses and create value through identifying causes of spending across populations; identifying key opportunities with the input of clinicians, staff, and patients; leveraging available data; and creating multispecialty, multidisciplinary care pathways based on risk tiers.
In August 2016, we began to develop a strategy to identify patient populations across UCLA Health with high expenses across all risk tiers and to promote proactive value-based care. UCLA Health is associated with the David Geffen School of Medicine at UCLA and encompasses 4 hospitals, 150 ambulatory clinics, and over 2,000 clinical faculty.
We focused on patients in the UCLA primary care network, which includes 40 UCLA-owned clinical practices caring for more than 350,000 patients. UCLA Health is accountable for approximately half of these patients through risk-based and accountable care organization contracts, spanning commercial health maintenance organizations and Medicare Advantage (57,500 patients and 9,300 patients, respectively), commercial preferred provider organizations (66,000 patients), and the Medicare Shared Savings Program (35,000 patients). We targeted high-cost subpopulations of patients with chronic conditions because these patients touch our system multiple times—each of which is an opportunity to intervene. For these subpopulations, a new leadership team focused on understanding and delivering holistic care to address patients’ complex medical, behavioral, and social needs.
We developed a four-step PHV framework to guide value creation within these subpopulations by reducing expenses while also improving clinical outcomes and the patient experience.
Step 1: Identify patient subpopulations with high expenses and reasons for spending
We identified three subpopulations with high expenditures—patients with dementia, chronic kidney disease (CKD), and cancer—by looking at total spending, spending per patient per month (PMPM), and patient-specific utilization data from internal and claims sources. Causes of spending were determined in commercial and governmental contracts, including acute and chronic care delivered in the inpatient, ambulatory, and postacute settings (Table 1). We reviewed these data in August 2016 (CKD) and in April 2017 (dementia and cancer), before launching interventions.
Step 2: Create design teams to understand the patient story
Process mapping demonstrated that patients in these subpopulations could interact with as many as 20 care team members within a few months. The care teams included office staff, clinicians, trainees, and patients and their caregivers. Existing primary care coordinators gained patient input on value improvement program development5 because drawing on the wisdom of these patients was important to understand key opportunities and engage patients.
We convened multispecialty, multidisciplinary PHV design teams beginning in August 2016 (CKD) and April 2017 (cancer and dementia). Each design team incorporated patient input in its decisions and included key care team members with whom patients regularly interact. For example, the dementia design team included program staff, primary care clinicians, and geriatricians—all of whom are vital to implementing initiatives that target large groups of lower-risk patients—as well as advance care planning and palliative care specialists, psychiatrists, and neurologists, who engage with late-stage patients (Figure 1).
In addition to patient care experts, the design teams included value improvement, analytics, operations, and health information technology (IT) experts. This merging of expertise enabled teams to make rapid decisions and, later, implement changes. Design teams were cochaired by the clinical leadership of each discipline and the health care system medical director of quality improvement (R.G.).
Step 3: Create custom analytics and spending-based risk stratifications
For each subpopulation, we created custom analytics that use administrative and clinical data (e.g., International Classification of Disease Revision–9 or –10 [ICD-9 and ICD-10] codes, problem lists and discharge diagnoses, natural language processing, medications, labs) to define targeted populations, better measure care, and understand causes of spending. This process occurred during August–September 2016 (CKD) and April–May 2017 (dementia and cancer).
Customized metric development at UCLA Health involved conceptualizing and validating the definitions for populations, goals, and process measures with clinicians.6 The design teams identified improvement opportunities based on their clinical experience and case-level analysis. Each measure incorporated all payer types because we aimed to provide equitable care regardless of insurance type. These data could be shared with primary and specialty clinicians since they used the same electronic health record (EHR).
We then developed risk stratification metrics specific to the subpopulation, which underwent iterative improvements given that risk modeling is a developing field.7 Patient- and caregiver-reported outcomes were incorporated when available. For example, in dementia, we included behavioral problems, severe functional impairment, and access to resources in addition to utilization and concurrent condition data. We defined five patient-expenditure risk strata that correlated with our natural distribution of spending for the three conditions: the top 1% (tier 1), 2%–5% (tier 2), 6%–20% (tier 3), 21%–60% (tier 4), and 61%–100% (tier 5) of spending.
Step 4: Develop targeted care pathways based on spending risk tiers
We then developed interventions to address patient needs identified at each risk tier. These were implemented during October–December 2016 (CKD) and June–September 2017 (dementia and cancer). We believed that the greatest opportunities to produce value would be to focus in the short run on middle-risk patients (tiers 2 and 3) and in the long run on low-risk patients (tiers 4 and 5).3 We identified that patient care in our health care system was often fragmented between primary and specialty clinicians, potentially leading to care duplication or missed opportunities during care transitions. Therefore, the multispecialty, multidisciplinary PHV design teams mapped comanagement strategies and care pathways by primary and specialty practitioners and their care teams for each risk level (Chart 1). Key elements included leveraging midlevel practitioners and care coordinators, health IT infrastructure, and other shared resources to reach the subpopulations of patients who may benefit most from specific interventions.8 In dementia care, the model focused on reorganizing the care team to optimize system efficiency and to leverage specialists’ expertise (e.g., freeing neurologists to see complex patients by moving memory testing to geriatricians) (Figure 1 and Chart 1).
In contrast to the cautiousness with which improvement projects are often received, the PHV design teams were enthusiastic from the start. We believe that this positive reaction was due to several factors:
- orienting value improvement efforts around the patient experience and clinical outcomes, which assured team members that efforts would serve patients meaningfully;
- breaking down silos and working across specialties, which created a team atmosphere;
- focusing on clinical issues in the design teams, while the analytics team supported the technical work; and
- advancing research, data-driven educational programs and growing national recognition of UCLA Health in value improvement.
Interventions for each subpopulation affected the entire primary care patient population and the UCLA Health providers caring for them. Each subpopulation had unique characteristics, causes of spending, and complexities in approaching system-wide value improvement, as detailed below.
Before implementation of the PHV interventions, patients with dementia (n = 4,348) incurred $1,768 PMPM spending (Table 1). Key reasons for spending included inpatient and intensive care unit (ICU) bed days. These patients often presented with acute infections and altered mental status. Although hospitalization was clinically indicated, our analysis showed that individuals with known care goals had less intensive therapy and reduced length of stay. In response, the dementia design team stratified patients with dementia into five risk tiers and developed three care pathways.
Patients in tiers 4 and 5 (low risk) receive dementia education that leverages patient portals and online materials to better align patient and caregiver expectations about disease progression and to monitor care. Once patients develop advanced disease or inpatient utilization, they enter tier 3, and the EHR will trigger a suggested referral to the Alzheimer’s and Dementia Care (ADC) program. The ADC program targets patients in tiers 2 and 3 (middle risk) and provides nurse practitioner co-management of dementia care.9 These patients also receive increased social services, goals of care discussions, and referrals to neurology and behavioral health when appropriate. Patients who enter tier 1 (top 1% of spending) often require intensive care management. The ADC team reviews these individual patients’ clinical and social needs and creates nuanced, creative solutions to coordinate care across primary and multiple specialty care teams, reduce high utilization, and initiate palliative care (Figure 1 and Chart 1).
The dementia intervention was implemented across the UCLA primary care network between June and September 2017. Early evaluation compared average monthly inpatient bed days per 1,000 patient-years between the 18 months prior (December 2015–May 2017) and 12 months after (October 2017–September 2018) intervention implementation. This preliminary analysis showed a 1% monthly reduction in inpatient bed days (incident rate ratio [IRR], 0.99; 95% confidence interval [CI], 0.98–1.00; P < .03) among the 4,348 patients.
Before implementation of the PHV interventions, patients with CKD (n = 17,172) accounted for $5,559 PMPM spending (Table 1). Key reasons for spending included hospitalizations and emergency department (ED) visits. Patients with stage 4 and 5 CKD were often hospitalized for emergent dialysis without proactive care coordination and protocols to expedite ambulatory evaluation or placement of catheters. The CKD design team included multidisciplinary staff with patient input, nephrologists, interventional radiology, and dialysis center staff. They defined five risk tiers with three associated care pathways.
The health system hired a CKD care coordinator, who focuses on expediting ambulatory care and increasing access to ambulatory interventional radiology services to expedite evaluation of malfunctioning catheters. This individual coordinates with the extensivist primary care physicians who manage patients in tier 1 (top 1% of spending) (Chart 1).
The CKD intervention was implemented across the UCLA primary care network during October–December 2016. Early evaluation compared average monthly hospitalizations per 1,000 patient-years between the 18 months before (April 2015–September 2016) and 12 months after (January–December 2017) intervention implementation among 1,502 patients with stage 4 or 5 CKD. This preliminary analysis showed a nearly 2% reduction in monthly hospitalizations (IRR, 0.98; 95% CI, 0.98–0.99; P < .0001).10
Before implementation of the PHV interventions, there were 27,757 patients with cancer who accounted for $3,723 PMPM spending (Table 1). In addition to ED visits, hospitalizations, and bed days, imaging and pharmaceuticals accounted for large proportions of spending, especially in end-of-life care. Breast, lung, and colorectal cancer patients in their last six months of life drove spending. These patients were frequently offered chemotherapy (37%), were often hospitalized (47% with 22% ICU), and had poor documentation of care goals (40% documented). Across all risk tiers and cancer stages, they sometimes received unnecessary duplicative or surveillance imaging in short intervals because of lack of coordination and protocols. The cancer design team included key staff with patient input, oncologists, surgeons, radiologists, and radiology oncologists who focused on spending drivers and defined three care pathways for five risk tiers (Chart 1). An initial evaluation will be conducted when preliminary data are available.
We are using this PHV strategy across other high-cost subpopulations with chronic conditions, and we are iteratively improving the CKD, dementia, and cancer care pathways. We are developing models for future capture of missing data (e.g., patient-reported outcomes). We are also developing models to recalculate components of spending based on standardized prices within our subpopulations because some payers provide limited data transparency. We are focusing on opportunities that can maximize financial benefit overall, and we believe these efforts can create long-term value. However, some efforts may reduce revenues to certain system entities.
Our four-step PHV framework for creating value has helped us begin to organize care across specialties, build capacity, and grow a culture for value at UCLA Health. We believe this model is a novel approach that could be adapted by other health care systems to improve quality and cost-effectiveness of care for high-expense subpopulations.
The authors thank the other team members including Joyce Kimori, Darcie Miller, Varsenik Papazyan-Gutierrez, Eric Cheung, Verna Porter, John Glaspy, Chris Pietras, Neil Wenger, Anne Walling, and Sun Yoo. The authors also thank Rei Cates, Vilay Khandelwal, and Nick Mabe for diligent programming as part of the value analytics team.
1. Jarousse LA. Take a look at how market forces will impact health care. Hosp Health Netw. September 8, 2014. https://www.hhnmag.com/articles/4012-take-a-look-at-how-market-forces-will-impact-health-care
. Accessed March 18, 2019.
2. McWilliams JM, Schwartz AL. Focusing on high-cost patients—The key to addressing high costs? N Engl J Med. 2017;376:807–809.
3. Hayes SL, Salzberg CA, McCarthy D, et al. High-need, high-cost patients: Who are they and how do they use health care? A population-based comparison of demographics, health care use, and expenditures. Issue Brief (Commonw Fund). 2016;26:1–14.
4. Conwell LJ, Cohen JW. Characteristics of People With High Medical Expenditures in the U.S. Civilian Noninstitutionalized Population, 2002. March 2005. Washington, DC: Agency for Healthcare Research and Quality; Statistical brief 73. https://meps.ahrq.gov/data_files/publications/st73/stat73.shtml
. Accessed March 18, 2019.
5. Clarke R, Bharmal N, Di Capua P, et al. Innovative approach to patient-centered care coordination in primary care practices. Am J Manag Care. 2015;21:623–630.
6. Clarke R, Hackbarth AS, Saigal C, Skootsky SA. Building the infrastructure for value at UCLA: Engaging clinicians and developing patient-centric measurement. Acad Med. 2015;90:1368–1372.
7. Edwards ST, Bitton A, Hong J, Landon BE. Patient-centered medical home initiatives expanded in 2009–13: Providers, patients, and payment incentives increased. Health Aff (Millwood). 2014;33:1823–1831.
8. Garcia ME, Uratsu CS, Sandoval-Perry J, Grant RW. Which complex patients should be referred for intensive care management? A mixed-methods analysis. J Gen Intern Med. 2018;33:1454–1460.
9. Reuben DB, Evertson LC, Wenger NS, et al. The University of California at Los Angeles Alzheimer’s and Dementia Care program for comprehensive, coordinated, patient-centered care: Preliminary data. J Am Geriatr Soc. 2013;61:2214–2218.
10. Gupta R, Skootsky S, Abtin F, et al. A Population Health Value approach to high-cost populations: A system-wide intervention to reduce hospitalization among chronic kidney disease patients. 2019. [Unpublished manuscript.]