Original Articles
POPULATION HEALTH MANAGEMENT Volume 18, Number 4, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/pop.2014.0117
Lessons from Washington State’s Medical Home Payment Pilot: What It Will Take to Change American Health Care Reena A. Koshy, MD, MPH,1 Douglas A. Conrad, PhD,2 and David Grembowski, PhD 2
Abstract
The Washington State Multi-Payer Medical Home Reimbursement Pilot (Pilot) tested a payment method for the patient-centered medical home (PCMH) model intended to reduce avoidable emergency department (ED) and hospitalization rates. Very little is known about the primary care clinic (clinic) experience with various payment methods designed for the medical home model. The objective was to elicit and describe the primary care clinic experience among various medical groups in Washington State’s payment Pilot. This was a qualitative analysis of semi-structured interviews conducted in January 2014 to identify enabling features (or ‘‘facilitators’’) as well as barriers to successful implementation of PCMH in this multi-payer pilot. Participants were clinical and administrative staff of Pilot clinics representing various types of health systems under 8 parent organizations across Washington State. Pilot clinics across Washington State chose evidence-based population health strategies to achieve Pilot targets. Pilot clinics encountered more barriers than facilitators when implementing strategies. A key facilitator was having timely access to ED and hospital clinical data. A common barrier was the cost of infrastructure development to implement strategies. Other barriers included lack of data to guide interventions and insufficient payment for care management and quality improvement work. It will take more than just primary care transformation to improve health outcomes—a significant transformation in data collection, reporting and payment needs to match the change occurring in clinics. (Population Health Management 2015;18:237–245)
Introduction
T
he Patient-Centered Medical Home (PCMH) model takes a planned and evidence-based approach rather than a reactive approach to care.1 Care is tailored to the patient with the goal of improving outcomes for both the individual and population groups. The current fee-for-service (FFS) payment method, designed to pay for individual care after a visit occurs, creates financial incentives to increase volume of services over quality and outcomes.2 With an unsustainable trend of rising health care costs, many health reform strategies are seeking alternatives that pay for improvements and appropriate care.2 Payment-related research on PCMH focuses on cost of care and potential areas for savings rather than the impact on clinical practice.3 Because of the paucity of research in this area, the investigators set out to describe the primary care clinic experience with a new payment method.4 This paper
1 2
is significant because it describes the primary care clinic (clinic) experience with a medical home payment model from the perspective of a variety of organizations in Washington State. The objective of this study is to elicit and describe the clinic experience in the Pilot and understand what strategies they used and the facilitators and barriers encountered. The Pilot tested a payment method among 12 clinics and 7 public and private health plans, not including Medicare. Health insurers added an up-front per member per month (PMPM) payment to support clinics’ medical home approach to reduce avoidable emergency department (ED) and hospital utilization during a 32-month period. Clinics received quarterly PMPM payments of $2.50 during the first 8 months, and $2.00 PMPM for the remaining 24 months in addition to FFS payments for office-based care. Health plans had approximately 25,000 members distributed unevenly among clinics. Clinics chose reductions in avoidable event
Independent contractor for Washington Health Alliance, Seattle, Washington. Department of Health Services at University of Washington, Seattle, Washington.
237
238
KOSHY ET AL.
Table 1. Organizational Characteristics
Organization Type Health system
Number of providers Number of providers Number of parent Region in parent organization in Pilot (includes organizations location(s)* (count) DO, MD, NP, PA) represented by this type
Central East Multispecialty provider group Central Independent physician association Central Primary care practice West
1800–1200
4–18
5
500 200 4
8 8 4
1 1 1
*Regions in the Pilot are West, Central, and East DO, doctor of osteopathy; MD, doctor of medicine; NP, nurse practitioner; PA, physician assistant
rates with the goal of saving money equal to or exceeding the up-front payment received (see Supplementary Appendices A and B for Pilot design, available in the online article at www.liebertpub.com/pop). The PCMH model was chosen because its focus on increased clinical access and proactive care management was expected to reduce high-cost ED and hospital care. The Pilot used the New York University Emergency Department (NYUED) algorithm and Agency for Healthcare Research and Quality Prevention Quality Indicators to calculate avoidable event rates.5,6 Outcome reports were due every 6 months. Statistical attribution linked patients to providers quarterly. Real-time utilization or attribution reporting was not available during the Pilot. Quality measures were reported as a composite score of 7 measures at the end of each pilot year. Clinics had to maintain quality scores to receive shared savings. Design
This study used qualitative methods and semi-structured interviews with clinicians and administrators to examine and compare their experiences in implementing the Pilot.7,8,9 All interviews used a common set of questions successfully field-tested in a previous comparative case study of 7 payment reform projects across the United States.10,11 This qualitative study describes the clinics’ selected strategies for reducing avoidable ED and hospital utilization, enabling features (or facilitators) and barriers to implementing those strategies, and lessons learned. Participants
Investigators (RAK, DAC, DG) asked clinics to select representatives from their clinical and administrative teams to participate in 2 interviews. Clinical interviews typically included medical assistants, registered nurses, health coaches, or the medical director involved in the daily operations of the Pilot. Financial interviews most often included an administrator in the parent organization when more than 1 clinic participated in the Pilot. Pilot strategies were implemented at the clinic level and differed in some cases between clinics within the same parent organization. Parent organizations represented 4 distinct types: (1) a health system that includes a network of hospitals, EDs, urgent care centers, primary care and specialty outpatient clinics; (2) a multispecialty medical group that includes a network of primary and specialty clinics; (3) an independent physician association that is a physician-led legal entity for private practice physicians to negotiate contracts with health
insurance companies on their behalf; and (4) a primary care practice that operates independently of hospitals and clinics. Table 1 lists characteristics of each organization type. Pilot clinics operated in 3 different regions: (1) west of Puget Sound (West), (2) north and south Puget Sound (Central), and (3) east of the Cascade Mountains (East). Most were located in the Central region. The number of hospitals located near the Pilot clinics varied by region. The West region had 1 hospital and 1 Pilot clinic. The Central region had 23 hospitals located near 8 Pilot clinics. The East region had 8 hospitals near 2 Pilot clinics. Methods
Telephone interviews were conducted in January 2014 after the Pilot completed its final prospective payments and before final outcome and reconciliation payments were calculated. The lead interviewer asked participants 4 openended questions (see Table 2). Responses were digitally recorded, validated, coded, and compared by organizational type and region to find shared or diverging patterns and to derive themes (see Supplementary Appendix C for method details, available in the online article at www .liebertpub.com/pop).
Table 2. Interview Questions Interview Questions 1. What was your specific strategy for practice transformation (ie, implementing the patient-centered medical home) and, in particular, reducing avoidable ED visits and hospitalizations? 2. What were the facilitators (internal to your organization and in the external environment) for implementing that strategy and achieving your objectives of reduced avoidable ED visits or hospitalizations? 3. What were the barriers (internal to your organization and in the external environment) to implementing that strategy and achieving your objectives of reduced avoidable ED visits or hospitalizations? 4. What are the lessons you’ve learned in the course of implementing your strategies for practice transformation and reducing avoidable ED visits or hospitalizations? What would you do differently? What aspects of your strategy would you maintain and potentially even reinforce? ED, emergency department
LESSONS FROM A PCMH PAYMENT PILOT
239
Key results
hospitals to develop ED notifications during the Pilot. Some clinics used risk-stratification methods to allocate resources and tried to predict ED use among all their patients.
Clinic responses regarding strategies, facilitators, barriers, and lessons learned centered on 3 themes: (1) population health management, (2) PCMH development, and (3) financial impact. The most frequent strategies, facilitators, and barriers for each theme are listed in the following sections (see Tables 3–5 for a full list of coded responses).
Facilitator—timely reporting of clinical data. Daily ED utilization reports helped identify target populations and patterns of ED use, particularly of high utilizers. The clinical information (eg, diagnosis, medications, imaging, lab results) included in ED notification helped allocate resources more efficiently because clinics could determine whether follow-up was needed and with whom. Timely reporting was not available for Pilot clinics from all EDs as some were not willing or able to establish an automated notification process. If clinics received ED notification within 24–48 hours, they could potentially prevent another admission by contacting the patient with follow-up phone calls. However, without ED notification, Pilot clinics could
Population health management Strategies. Clinics chose evidence-based population health management strategies to reduce avoidable ED and hospital events.1,2,12 The term ‘‘population’’ was used differently by respondents but generally referred to groups of people for whom interventions were designed. Clinic strategies identified people using the ED and targeted care to those at high risk of repeat use. All clinics worked with local
Table 3. Population Health Theme Organization type and region (W = west, C = Central, E = east) Population Health Theme—coded responses Strategies Developed ED notification with hospital Added follow-up phone calls to patients after ED visit Developed real-time ED notification Analyzed data and allocated resources by risk stratification Identified high utilizers of the ED Established EDIE connection with PCP Educated patients about when to use ED Tried to contact high utilizers frequently to reduce ED and UC visits Focused on upstream causes of ED and UC utilization Facilitators Timely (eg, daily) ED and inpatient utilization reports Integrated data systems EDIE connection with PCP State initiative to reduce Medicaid ED utilization rates Barriers Lack of timely ED notification Lack of actionable data (late pilot data, not detailed, not complete) Patients’ co-payments discourage primary care visit, favors ED care Disagreement with NYUED method Lack of access to dental care Lack of access to mental health care Lack of data integration across medical provider sites Discrepancies between pilot and practice quality metric results Patients prefer convenience of ED visit over primary care office visit Lessons Learned Clinic gained insights to population health interventions by pilot experience Clinic needed different strategies for various populations Clinic would maintain pilot interventions
PCP-W
IPA-C
Multispecialty-C
Health system-C
Health system-E
X X X X
X X X X
X X X X
X X X X
X X
X X X
X
X
X X X X
X
X X X X X
X X
X X
X X X
X
X X X X
X
X X
X X
X -
X
X
X
X X X X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association; ED, emergency department, EDIE, emergency deparment information exchange; UC, urgent care; NYUED, New York University Emergency Department
240
KOSHY ET AL.
Table 4. Patient-Centered Medical Home (PCMH) Development Theme Organization type and region (W = west, C = Central, E = east) PCMH Development Theme—coded responses
PCP-W
IPA-C
Multispecialty-C
Health system-C
Health system-E
X X
X X
X X
X X
X X
X
X
X
X
X
X X
X X
X X
X X
X X
X X X
X X
X X X
X X X
X X X
X
X
X
X
X
X
X
X
X X
X X X
X X
X
X
X
X
X X
X
Strategies Used additional funds for new staff salaries Created new care teams (health coach, social worker, pharmacist, MA, RN) Developed personal relationships with patients to improve communication Increased access to clinic visits Included non-pilot populations and targets in interventions Increased ratio of support staff to MDs/PA/ARNP Changed staff workflow and roles Increased monitoring of patients with chronic conditions. Instituted quality improvement process for preventive and quality measures. Hired RN to do new work Facilitators Pilot funds to hire staff Physicians support for implementation strategy Enabling patient communication with the clinic Barriers PMPM insufficient for infrastructure need (staff, training, data, analysis) Limitations of attribution method Difficulty of changing organizational culture Difficult for patients to contact providers with concerns Lessons Learned Clinic learned what capabilities were needed for PCMH Clinic learned about communication pathways between staff, patients, providers
X
X X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association; RN, registered nurse; MA, medical assistant; MD, doctor of medicine; ARNP, nurse practitioner; PA, physician assistant; PMPM, per member per month
not intervene in a timely manner or establish a complete picture of ED utilization patterns in their patient population. Facilitator—payment incentives for EDs. Starting in June 2012, Washington State launched a campaign, ‘‘ER is for Emergencies,’’ to reduce ED utilization among Medicaid members. The campaign encouraged hospitals to work with primary care providers across the state. EDs put in place 7 best practices to guide appropriate ED use, adopted a new ED information exchange (EDIE) system, and directed patients to a primary care physician (PCP) for follow-up care.13 The campaign was designed independently of the Pilot and did not connect primary care practices to the EDIE system initially, but the EDIE system enabled PCP communication. The state campaign added financial incentives for hospitals and EDs to show reductions in Medicaid ED utilization rates by June 2013 or face a no-payment policy for a specific list of conditions considered ‘‘not medically necessary.’’14 The state and Pilot created financial incentives for both the ED and Pilot clinics to reduce ED admissions during 2012–2013.
Facilitator—incentives to collaborate. In the West region, the state campaign brought together the single Pilot clinic, local hospital, and a regional health collaborative to identify high utilizers of the ED, develop care plans, and track utilization patterns. In the West and Central regions, 4 Pilot clinics established an EDIE connection to receive electronic ED notifications. In the East region, the state effort facilitated conversations between separate health systems, identifying communication gaps and opportunities to reduce readmission rates. Barrier—lack of actionable population health data. The biggest barrier for all Pilot clinics was the lack of actionable population health data for health plan members at the start of the Pilot. The Pilot used claims data to report ED utilization rates but did not provide specific information such as patient names, diagnosis, or ED locations. Reports were delayed as much as 18 months. Clinics designed population health interventions without adequate data describing ED and hospital utilization patterns. The frequency of ED utilization reporting increased from semiannually to quarterly
LESSONS FROM A PCMH PAYMENT PILOT
241
Table 5. Financial Impact Theme Organization type and region (W = west, C = Central, E = east) Financial Impact Theme—coded responses
PCP-W
Strategies Intended strategy to work for other payment models Held part of new money in reserve in case of calculated pilot shortfall Facilitators Increased financial reimbursement for clinical care Outcomes measures used in other payment strategies Barriers Financial barriers to do more patient evaluation in clinic Staff turnover Payment insufficient for quality improvement work Lessons learned Clinic learned how to think about new payment designs after this experience Clinic learned what data it needs
IPA-C
Multispecialty-C
Health system-C
Health system-E
X
X
X X
X
X
X X
X X
X
X X X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association
in January 2012, but the data were still more than 6 months old and did not reflect the clinics’ current patient population. By the time data were available to practices, the opportunity to make changes had passed. Practices needed accurate population health data prior to the launch of the pilot and at the beginning of each Pilot year to plan effective interventions. Practices found real-time information exchange (EDIE) beneficial as long as the connection was established early to enable action and the volume of data was manageable. Barrier—lack of reliable method to link patients and PCP. The Pilot claims-based attribution method lacked the
functionality of identifying a patient’s primary care provider to patients, EDs, or hospitals at the point of care and was another barrier to timely notification of an ED visit. Barrier—costs to patient. Interviewees reported hearing from patients that a co-pay was a financial barrier to seeking care in the office following an ED visit or for proactive care to prevent complications. For patients with little or no outof-pocket costs for ED care, respondents believed patients preferred to go to the ED instead of selecting a PCP. Clinics thought their patients chose to go to the ED after hours because they were not able or willing to miss work for a primary care visit during business hours. By one clinic report, 60% of ED visits for its population occurred when the primary care clinic was closed. Barrier—disagreement with determination of avoidable conditions. Clinics were harshly critical of the NYUED
algorithm classification method of an avoidable ED visit. The NYUED algorithm uses final diagnosis codes to estimate the probability that a diagnosed condition is avoidable or treatable in the primary care setting. Clinics stressed that a final diagnosis code is not reported initially; instead, patients presented with a chief complaint that required evaluation to determine the final diagnosis.15 Patient evaluations
take time, equipment, resources, and skills that may not be available in the clinic. Primary care providers had to determine the best place to evaluate patients with a chief complaint and reach the final diagnosis. Therefore, the clinics were frustrated with the use of the NYUED classifications for avoidable visits and the inability to challenge and refine outcome results. Lessons learned—clinics needed population health data to design interventions. Clinics learned how to design popu-
lation health interventions and reported wanting population health data to understand clinic trends and opportunities for intervention. Clinics learned to plan interventions with both clinical and financial input. All groups planned to maintain the changes in care team design and all felt they adopted the correct strategy for improving patient care. Clinics tried to take a whole practice approach but they could not apply one strategy for all populations. Clinics needed a different strategy for adults compared to children, and for Medicare and Medicaid populations compared to commercially insured populations. Clinics needed more than one approach to achieve target outcomes and they realized that the different strategies had different costs. PCMH development Strategies. All participating clinics used Pilot funding to develop PCMH core capabilities and hire new staff.16 Clinics created new care teams and changed roles and workflow to provide proactive care. All clinics increased same-day access to care by extending hours or increasing patient access to nurse consultation. Clinics employed staff to monitor patient registries and identify gaps in preventive care or chronic condition management that otherwise could lead to ED visits. 1,17–19 A common strategy was to increase a patient’s personal connection with the care team to encourage the patient to call the primary care clinic before
242
KOSHY ET AL.
going to the ED for care. Care teams varied at each clinic and included multiple disciplines such as medical assistants, registered nurses (RNs), health coaches, health facilitators, social workers, and pharmacists.
needed project management capabilities to oversee change and implement strategies.
Facilitator—financial support for PCMH approach. The new Pilot funds were used by all clinics to pay for essential staff to develop PCMH capabilities and shift the focus from FFS activities to population health management. Hiring RNs carried a significant cost for most practices and having access to new funds enabled organizations to pool funding sources and hire nurses.
strategies with patients, staff, providers, and with outside organizations during the Pilot. Timely response to patient queries was important to reduce ED visits and establish a positive connection with the clinic. Not all patients needed or responded well to follow-up ED calls. Information technology support was needed to automate notification processes. Communication worked well if integrated into electronic health records, but broke down if dependent on individuals sending faxes or making phone calls.
Barrier—inadequate funding for infrastructure development. The new PMPM payments did not pay for adequate
staff time to manage the whole clinic population and this was a significant barrier for all clinics. PMPM payments were made for a minority of the clinic population because of limited health plan participation and low attribution counts. The attribution method linked patients to providers using a statistical method rather than patient preference or sense of engagement with a clinic. Clinics felt this statistical method underestimated pilot population counts, resulting in insufficient funding for infrastructure to carry out populationbased strategies. Barrier—lack of defined patient population. Health plan attribution lists changed quarterly, making it difficult for clinics to know who was in the Pilot. The lists identified patients for quarterly payments but they were not necessarily the same patients included in the outcome measure calculations. This was a barrier for clinics because, with limited funds, they had to choose patients to include in population strategies but did not know whether the same patients or others, not in interventions, would be counted in outcome measures. The risk for clinics would be that outcome measures did not reflect their efforts. Lessons learned—clinics needed on-site assessment capabilities. The Pilot demonstrated the usefulness of on-
site clinical assessment capabilities to manage low-acuity conditions and prevent ED visits. RNs were better equipped to answer questions, address broader medical conditions, and provide trusted advice than medical assistants or receptionists without clinical training. Clinics with separate consulting nurse services found patients were referred to the ED more often than the clinic staff judged necessary, and referral information often was not shared with the clinic. Lessons learned—clinics needed data integration and analysis. The Pilot highlighted how much data integration
from different sources was required to create a composite picture of the clinic population. Data analysis capabilities were needed to identify patients, patterns of ED use, and target populations for interventions. The ability to classify risk helped clinics allocate nursing time, arrange clinic visits, or assign case management skills effectively. Lessons learned—clinics needed project management capability. Clinics often reported that organizational change
was difficult and required daily monitoring, assessment, and reminders to make progress and maintain results. Clinics
Lessons learned—clinics needed effective communication strategies. Clinics learned about effective communication
Financial impact Strategies. Clinics chose strategies that would work for all patients and be supported by other payment methods, thereby permitting sustainable changes. Clinics used Pilot funds to invest in care teams and add PCMH capabilities that would generate returns from a variety of quality improvement incentives. Facilitators—higher FFS. FFS incentives facilitated PCMH transformation. Independent of the Pilot, FFS reimbursement rates increased with the use of new billing codes for Medicare and with adoption of meaningful use criteria for electronic health records. The higher reimbursements influenced parent organization leaders to adopt new care coordination strategies and invest in data systems during the Pilot. Patients who went to the primary care clinic instead of the ED brought additional FFS payments to the clinic and reinforced outreach activities. One clinic found it easier to adopt the Pilot strategy because some Pilot measures were already being monitored for a Medicare FFS payment incentive program under way in the clinic. Barrier—inadequate payment to support quality improvement work. Pilot clinics reported that much of the work
done by medical assistants and RNs for quality improvement was not directly reimbursed; instead clinics typically must wait 2 to 3 years to receive payment for quality improvement. This was a financial barrier for the clinic because it was less likely to achieve quality targets in the long term if it could not pay for the daily work of quality improvement in the short term. In addition, clinics were paid less for the work of nurse practitioners and physician assistants but faced the same infrastructure costs to support mid-level providers in the clinic with nursing staff. Barrier—preauthorization requirements. One clinic reported that incentives were not aligned to enable acute evaluation and treatment in the outpatient setting. The clinic could not obtain preauthorization for imaging studies in the outpatient setting, but preauthorization was granted when the same patient was seen in the ED. This administrative barrier prevented evaluation of complex conditions in the primary care clinic. The cost of providing therapies such as intravenous hydration or medication administration in the clinic could have exceeded the reimbursement for delivering such care and was a barrier for outpatient treatment. For the patient, it was more efficient to be
LESSONS FROM A PCMH PAYMENT PILOT
evaluated and treated in the ED rather than wait for preauthorization of diagnostic studies. Lessons learned—clinics needed data to monitor progress. Clinics learned they needed more detailed data from
health plans to track progress on Pilot measures given financial consequences tied to outcomes and discrepancies between health plan data and parent organization quality reports. Clinics wanted to have more population data to understand trends in their own practice. To lower ED utilization rates, clinics needed real-time ED notification and clinical information to identify utilization patterns among various populations. Clinics needed timely feedback about clinic interventions. Lessons learned—cost of health interventions differed by populations. Clinics learned to quantify the cost of trans-
formation and the financial investments undertaken to achieve outcomes. Clinics learned what payment amounts are needed to manage a population with the PCMH model and specifically how payments differ by populations based on risk, patterns of use, and complexity of illness. Discussion
The myriad challenges faced by Pilot clinics also were valuable learning experiences in health care transformation. Investigators learned how payment shapes the health care system today and what obstacles need to be overcome to arrive at better outcomes for patients and providers. The Pilot chose a population health objective, to reduce costly ED and inpatient utilization rates, and practices responded with population health interventions. Practices reported more barriers than facilitators because the data and payment provided were not specifically designed for population health interventions. Clinical care transformation was ahead of data and payment transformation. The primary care Pilot experience teaches us what needs to change in data design and payment methods to achieve a whole health system transformation. Transforming data to support population health management
The biggest challenge for primary care practices was not having timely, detailed, or actionable data from claims billing sources. A transformation in data collection and reporting is needed that can identify patients, their primary care providers, specialists, and locations of care and describe utilization patterns that affect population outcomes. Bringing together information for patients and their providers at the point of care is needed for effective decision making. For example, to reduce ED utilization, end users needed a reliable and accessible source of health information describing ED use for all individuals seeking care in the state, available in real time, with information on trends of standardized measurement indicators. This level of data integration and detail is not currently available to all medical providers in the state. Hospital administrators, ED managers, and clinicians were not explicitly included in the design of the Pilot or involved in the informational meetings that accompanied implementation of the Pilot. With the benefit of hindsight,
243
the technical barriers (eg, lack of interoperability between hospital and clinic information systems to effectively exchange clinical information) might have been mitigated significantly by incorporating the hospital and ED perspective systematically during Pilot design and implementation. Future efforts to transform care and improve outcomes cannot be limited to 1 specialty. Population health data must reflect the care paths taken by the population, which includes care in the community, primary care, specialty care, and hospital-based services. Increased investments in data development can drive up health care costs if duplicate information systems are created. It is in the public interest to build a robust population health database and share the common resource. In addition, population health data must be all inclusive. Data that exclude groups (eg, the uninsured) limit the ability to understand patterns in health outcomes and improve the health system for all populations. What is needed to transform data to enable innovative approaches to health care delivery is an organization capable of creating data tools with this allinclusive population perspective. Transforming payment to support population health management
The incremental changes to FFS payments for a limited number of Pilot patients were inadequate to change outcomes for a larger population. Paying for population health interventions requires building the infrastructure to deliver proactive and coordinated care for populations and is different from paying for individual acute care services. This study identified the infrastructure needed to carry out population health strategies that included on-site clinical assessment, data and project management, communication capacity, and data integration across health care settings, specifically connecting primary care with EDs, hospitals, and specialty care. It is essential to change payments across the health care environment to gain meaningful transformation. Investigators learned that payment changes in both EDs and Pilot primary care clinics were effective in changing provider behavior and creating collaborative partnerships. Preliminary results show positive outcomes. Pilot clinics reduced the overall avoidable ED rate by 10.7% according to a report prepared by the University of Washington for the Washington Health Care Authority (Conrad D, Chan G. June 30, 2014, Report to the Health Care Authority). The state campaign reduced total Medicaid ED utilization by 9.9%.20 Health care costs for individuals likely affected personal use of health services. If mental health, dental care, or primary care visits are not included in a person’s insurance coverage, or if there is too high a co-pay or shared payment, this could lead individuals to the ED where they are guaranteed an evaluation. Transforming health care payments should take into account personal behavior with regard to cost and align incentives for patients as well as providers to achieve shared population health outcomes. Limitations
Responses describing the primary care experience with the Pilot payment method are limited to the views of the
244
Pilot clinics and may not identify all potential themes related to the study questions. Some interviewees joined their clinic partway through the Pilot and had limited knowledge of the strategies, facilitators, barriers, and lessons learned. The majority of Pilot participants represented large health systems because of barriers to entry for small independent clinics; responses may not capture the full experience of small independent practices.
KOSHY ET AL.
3.
4.
Conclusions
The Pilot tested an alternative payment method to improve population outcomes. Clinics chose evidence-based population health interventions to achieve Pilot targets. The alternative data reporting and payment methods were insufficient to realize the potential of the PCMH population health approach. This study identifies where further changes are needed. Data for population health interventions must: (1) define the population, (2) be accessible to health care providers, and (3) have an all-inclusive perspective to describe the health needs of an entire population. Payment methods need to create incentives that work across the health system and support infrastructure development for population health interventions, including integrated information exchange between primary care, specialty care, EDs, and hospitals. Changes in data integration and payment in both the ED and primary care resulted in better outcomes from preliminary results. It will take more than just primary care transformation to improve health outcomes—a significant transformation in data collection, reporting, and payment must match the change occurring in clinics.
5. 6.
7.
8. 9. 10. 11.
Author Disclosure Statement
Drs. Koshy, Conrad, and Grembowski declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Koshy worked as an independent contractor for the Washington Health Alliance and retained full independence to conduct the research. She was project manager for the Pilot from 2010 to 2013. Drs. Conrad and Grembowski report no potential conflicts of interest. Informed consent was obtained from all participants to proceed with interviews, recording, and use of quotes. No patient identifiable data were collected and interview responses were considered secondary data from Pilot operations and exempt from human subjects or institutional review board approval. This research was supported and funded by the Washington Health Alliance (WHA, formerly the Puget Sound Health Alliance). The WHA is a regional health collaborative in Seattle, co-convener of the Washington State MultiPayer Medical Home Payment Pilot, and provided project management for the Pilot. References
1. Wagner EH, Coleman K, Reid RJ, Phillips K, Sugarman JR. Guiding Transformation: How Medical Practices Can Become Patient-Centered Medical Homes. New York: The Commonwealth Fund; February 2012. 2. Bailit M, Phillips K, Long A. Safety Net Medical Home Initiative. Paying for the Medical Home: Payment Models to Support Patient-Centered Medical Home Transformation
12.
13.
14.
15.
16. 17. 18.
19.
in the Safety Net. Seattle, WA: Bailit Health Purchasing and Qualis Health; October 2010. Grumbach K, Grundy P. The Outcomes of Implementing Patient-Centered Medical Home Interventions: A Review of the Evidence from Prospective Evaluation Studies in the United States. Updated November 16, 2010. http://www .cms.org/uploads/GrumbachGrundy2010OutcomesPCPCC .pdf. Accessed July 30, 2014. Baillit Health Purchasing, LLC. Facilitators and Barriers to Payment Reform: Market based, Governmental, Organizational and Design Considerations. http://www.bailit-health .com/articles/092713_bhp_rwjf_facilitators.pdf. Accessed July 30, 2014. NYU Wagner. NYU ED algorithm. http://wagner.nyu.edu/ faculty/billings/nyued-background. Accessed July 30, 2014. Agency for Health Care Research and Quality. Prevention Quality Indicators overview. http://www.qualityindicators .ahrq.gov/modules/pqi_resources.aspx. Accessed July 30, 2014. Firestone WA, Herriott RE. Multisite qualitative policy research: some designs and implementation issues. In: Fetterman DM, ed. Ethnography in Educational Evaluation. Beverly Hills: Sage Publications; 1984:63–88. Pettigrew AM. Longitudinal field research on change: theory and practice. Organization Science. 1990;1:267–292. Stake RE. Multiple Case Study Analysis. New York: The Guilford Press; 2006. Conrad D, Grembowski D, Gibbons C, et al. A report on eight early-stage state and regional projects testing valuebased payment. Health Aff (Millwood). 2013;32:998–1006. Conrad DA, Grembowski D, Hernandez SE, Lau B, MarcusSmith M. Emerging lessons from regional and state innovation in value-based payment reform: balancing collaboration and disruptive innovation. Millbank Q. 2014;92: 568–623. Care Continuum Alliance. Implementation and Evaluation: A Population Health Guide for Primary Care Models. 2012. http://selfmanagementalliance.org/wp-content/uploads/2013/ 11/A-Population-Health-Guide-for-Primary-Care-ModelsCareContinAlliance.pdf. Accessed July 30, 2014. Washington State Hospital Association. ER is for Emergencies results in better care coordination and cost savings. http://www.wsha.org/ERemergencies.cfm. Accessed July 30, 2014. Washington State Hospital Association. Background information: Medicaid emergency department use. http://www .wsha.org/files/ER%20Background%20Information%20 April%2016.pdf. Accessed July 30, 2014. Raven MC, Lowe RA, Macelli J, Hsia RY. Comparison of presenting complaint versus discharge summary for identifying ‘‘nonemergency’’ emergency department visits. JAMA. 2013;309:1145–1153. Safety Net Medical Home Initiative. Change concepts overview. http://www.safetynetmedicalhome.org/changeconcepts. Accessed July 30, 2014. Morgan SR, Chan AM, Alquatari M, Pines JM. Non-emergency department interventions to reduce ED utilization: a systematic review. Acad Emerg Med. 2013;20:969–985. Safety Net Medical Home Initiative. Practice Results. Number of days to third next available appointment. http:// www.safetynetmedicalhome.org/sites/default/files/PracticeResults-Enhanced-Access.pdf. Accessed July 30, 2014. Safety Net Medical Home Initiative. Practice Results. Contacting patients within two days of their hospital discharge.
LESSONS FROM A PCMH PAYMENT PILOT
http://www.safetynetmedicalhome.org/sites/default/files/ Practice-Results-Care-Coordination.pdf. Accessed July 30, 2014. 20. Washington State Health Care Authority. Report to the Legislature. Emergency Department Utilization: Update on Assumed Savings from Best Practices Implementation. March 20, 2014. http://www.hca.wa.gov/Documents/Emergency DeptUtilization.pdf. Accessed July 30, 2014.
245
Address correspondence to: Reena Koshy MD MPH 4464 Fremont Ave N Suite #103 Seattle, WA 98103 E-mail: [email protected]
POPULATION HEALTH MANAGEMENT Volume 18, Number 4, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/pop.2014.0117
Lessons from Washington State’s Medical Home Payment Pilot: What It Will Take to Change American Health Care Reena A. Koshy, MD, MPH,1 Douglas A. Conrad, PhD,2 and David Grembowski, PhD 2
Abstract
The Washington State Multi-Payer Medical Home Reimbursement Pilot (Pilot) tested a payment method for the patient-centered medical home (PCMH) model intended to reduce avoidable emergency department (ED) and hospitalization rates. Very little is known about the primary care clinic (clinic) experience with various payment methods designed for the medical home model. The objective was to elicit and describe the primary care clinic experience among various medical groups in Washington State’s payment Pilot. This was a qualitative analysis of semi-structured interviews conducted in January 2014 to identify enabling features (or ‘‘facilitators’’) as well as barriers to successful implementation of PCMH in this multi-payer pilot. Participants were clinical and administrative staff of Pilot clinics representing various types of health systems under 8 parent organizations across Washington State. Pilot clinics across Washington State chose evidence-based population health strategies to achieve Pilot targets. Pilot clinics encountered more barriers than facilitators when implementing strategies. A key facilitator was having timely access to ED and hospital clinical data. A common barrier was the cost of infrastructure development to implement strategies. Other barriers included lack of data to guide interventions and insufficient payment for care management and quality improvement work. It will take more than just primary care transformation to improve health outcomes—a significant transformation in data collection, reporting and payment needs to match the change occurring in clinics. (Population Health Management 2015;18:237–245)
Introduction
T
he Patient-Centered Medical Home (PCMH) model takes a planned and evidence-based approach rather than a reactive approach to care.1 Care is tailored to the patient with the goal of improving outcomes for both the individual and population groups. The current fee-for-service (FFS) payment method, designed to pay for individual care after a visit occurs, creates financial incentives to increase volume of services over quality and outcomes.2 With an unsustainable trend of rising health care costs, many health reform strategies are seeking alternatives that pay for improvements and appropriate care.2 Payment-related research on PCMH focuses on cost of care and potential areas for savings rather than the impact on clinical practice.3 Because of the paucity of research in this area, the investigators set out to describe the primary care clinic experience with a new payment method.4 This paper
1 2
is significant because it describes the primary care clinic (clinic) experience with a medical home payment model from the perspective of a variety of organizations in Washington State. The objective of this study is to elicit and describe the clinic experience in the Pilot and understand what strategies they used and the facilitators and barriers encountered. The Pilot tested a payment method among 12 clinics and 7 public and private health plans, not including Medicare. Health insurers added an up-front per member per month (PMPM) payment to support clinics’ medical home approach to reduce avoidable emergency department (ED) and hospital utilization during a 32-month period. Clinics received quarterly PMPM payments of $2.50 during the first 8 months, and $2.00 PMPM for the remaining 24 months in addition to FFS payments for office-based care. Health plans had approximately 25,000 members distributed unevenly among clinics. Clinics chose reductions in avoidable event
Independent contractor for Washington Health Alliance, Seattle, Washington. Department of Health Services at University of Washington, Seattle, Washington.
237
238
KOSHY ET AL.
Table 1. Organizational Characteristics
Organization Type Health system
Number of providers Number of providers Number of parent Region in parent organization in Pilot (includes organizations location(s)* (count) DO, MD, NP, PA) represented by this type
Central East Multispecialty provider group Central Independent physician association Central Primary care practice West
1800–1200
4–18
5
500 200 4
8 8 4
1 1 1
*Regions in the Pilot are West, Central, and East DO, doctor of osteopathy; MD, doctor of medicine; NP, nurse practitioner; PA, physician assistant
rates with the goal of saving money equal to or exceeding the up-front payment received (see Supplementary Appendices A and B for Pilot design, available in the online article at www.liebertpub.com/pop). The PCMH model was chosen because its focus on increased clinical access and proactive care management was expected to reduce high-cost ED and hospital care. The Pilot used the New York University Emergency Department (NYUED) algorithm and Agency for Healthcare Research and Quality Prevention Quality Indicators to calculate avoidable event rates.5,6 Outcome reports were due every 6 months. Statistical attribution linked patients to providers quarterly. Real-time utilization or attribution reporting was not available during the Pilot. Quality measures were reported as a composite score of 7 measures at the end of each pilot year. Clinics had to maintain quality scores to receive shared savings. Design
This study used qualitative methods and semi-structured interviews with clinicians and administrators to examine and compare their experiences in implementing the Pilot.7,8,9 All interviews used a common set of questions successfully field-tested in a previous comparative case study of 7 payment reform projects across the United States.10,11 This qualitative study describes the clinics’ selected strategies for reducing avoidable ED and hospital utilization, enabling features (or facilitators) and barriers to implementing those strategies, and lessons learned. Participants
Investigators (RAK, DAC, DG) asked clinics to select representatives from their clinical and administrative teams to participate in 2 interviews. Clinical interviews typically included medical assistants, registered nurses, health coaches, or the medical director involved in the daily operations of the Pilot. Financial interviews most often included an administrator in the parent organization when more than 1 clinic participated in the Pilot. Pilot strategies were implemented at the clinic level and differed in some cases between clinics within the same parent organization. Parent organizations represented 4 distinct types: (1) a health system that includes a network of hospitals, EDs, urgent care centers, primary care and specialty outpatient clinics; (2) a multispecialty medical group that includes a network of primary and specialty clinics; (3) an independent physician association that is a physician-led legal entity for private practice physicians to negotiate contracts with health
insurance companies on their behalf; and (4) a primary care practice that operates independently of hospitals and clinics. Table 1 lists characteristics of each organization type. Pilot clinics operated in 3 different regions: (1) west of Puget Sound (West), (2) north and south Puget Sound (Central), and (3) east of the Cascade Mountains (East). Most were located in the Central region. The number of hospitals located near the Pilot clinics varied by region. The West region had 1 hospital and 1 Pilot clinic. The Central region had 23 hospitals located near 8 Pilot clinics. The East region had 8 hospitals near 2 Pilot clinics. Methods
Telephone interviews were conducted in January 2014 after the Pilot completed its final prospective payments and before final outcome and reconciliation payments were calculated. The lead interviewer asked participants 4 openended questions (see Table 2). Responses were digitally recorded, validated, coded, and compared by organizational type and region to find shared or diverging patterns and to derive themes (see Supplementary Appendix C for method details, available in the online article at www .liebertpub.com/pop).
Table 2. Interview Questions Interview Questions 1. What was your specific strategy for practice transformation (ie, implementing the patient-centered medical home) and, in particular, reducing avoidable ED visits and hospitalizations? 2. What were the facilitators (internal to your organization and in the external environment) for implementing that strategy and achieving your objectives of reduced avoidable ED visits or hospitalizations? 3. What were the barriers (internal to your organization and in the external environment) to implementing that strategy and achieving your objectives of reduced avoidable ED visits or hospitalizations? 4. What are the lessons you’ve learned in the course of implementing your strategies for practice transformation and reducing avoidable ED visits or hospitalizations? What would you do differently? What aspects of your strategy would you maintain and potentially even reinforce? ED, emergency department
LESSONS FROM A PCMH PAYMENT PILOT
239
Key results
hospitals to develop ED notifications during the Pilot. Some clinics used risk-stratification methods to allocate resources and tried to predict ED use among all their patients.
Clinic responses regarding strategies, facilitators, barriers, and lessons learned centered on 3 themes: (1) population health management, (2) PCMH development, and (3) financial impact. The most frequent strategies, facilitators, and barriers for each theme are listed in the following sections (see Tables 3–5 for a full list of coded responses).
Facilitator—timely reporting of clinical data. Daily ED utilization reports helped identify target populations and patterns of ED use, particularly of high utilizers. The clinical information (eg, diagnosis, medications, imaging, lab results) included in ED notification helped allocate resources more efficiently because clinics could determine whether follow-up was needed and with whom. Timely reporting was not available for Pilot clinics from all EDs as some were not willing or able to establish an automated notification process. If clinics received ED notification within 24–48 hours, they could potentially prevent another admission by contacting the patient with follow-up phone calls. However, without ED notification, Pilot clinics could
Population health management Strategies. Clinics chose evidence-based population health management strategies to reduce avoidable ED and hospital events.1,2,12 The term ‘‘population’’ was used differently by respondents but generally referred to groups of people for whom interventions were designed. Clinic strategies identified people using the ED and targeted care to those at high risk of repeat use. All clinics worked with local
Table 3. Population Health Theme Organization type and region (W = west, C = Central, E = east) Population Health Theme—coded responses Strategies Developed ED notification with hospital Added follow-up phone calls to patients after ED visit Developed real-time ED notification Analyzed data and allocated resources by risk stratification Identified high utilizers of the ED Established EDIE connection with PCP Educated patients about when to use ED Tried to contact high utilizers frequently to reduce ED and UC visits Focused on upstream causes of ED and UC utilization Facilitators Timely (eg, daily) ED and inpatient utilization reports Integrated data systems EDIE connection with PCP State initiative to reduce Medicaid ED utilization rates Barriers Lack of timely ED notification Lack of actionable data (late pilot data, not detailed, not complete) Patients’ co-payments discourage primary care visit, favors ED care Disagreement with NYUED method Lack of access to dental care Lack of access to mental health care Lack of data integration across medical provider sites Discrepancies between pilot and practice quality metric results Patients prefer convenience of ED visit over primary care office visit Lessons Learned Clinic gained insights to population health interventions by pilot experience Clinic needed different strategies for various populations Clinic would maintain pilot interventions
PCP-W
IPA-C
Multispecialty-C
Health system-C
Health system-E
X X X X
X X X X
X X X X
X X X X
X X
X X X
X
X
X X X X
X
X X X X X
X X
X X
X X X
X
X X X X
X
X X
X X
X -
X
X
X
X X X X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association; ED, emergency department, EDIE, emergency deparment information exchange; UC, urgent care; NYUED, New York University Emergency Department
240
KOSHY ET AL.
Table 4. Patient-Centered Medical Home (PCMH) Development Theme Organization type and region (W = west, C = Central, E = east) PCMH Development Theme—coded responses
PCP-W
IPA-C
Multispecialty-C
Health system-C
Health system-E
X X
X X
X X
X X
X X
X
X
X
X
X
X X
X X
X X
X X
X X
X X X
X X
X X X
X X X
X X X
X
X
X
X
X
X
X
X
X X
X X X
X X
X
X
X
X
X X
X
Strategies Used additional funds for new staff salaries Created new care teams (health coach, social worker, pharmacist, MA, RN) Developed personal relationships with patients to improve communication Increased access to clinic visits Included non-pilot populations and targets in interventions Increased ratio of support staff to MDs/PA/ARNP Changed staff workflow and roles Increased monitoring of patients with chronic conditions. Instituted quality improvement process for preventive and quality measures. Hired RN to do new work Facilitators Pilot funds to hire staff Physicians support for implementation strategy Enabling patient communication with the clinic Barriers PMPM insufficient for infrastructure need (staff, training, data, analysis) Limitations of attribution method Difficulty of changing organizational culture Difficult for patients to contact providers with concerns Lessons Learned Clinic learned what capabilities were needed for PCMH Clinic learned about communication pathways between staff, patients, providers
X
X X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association; RN, registered nurse; MA, medical assistant; MD, doctor of medicine; ARNP, nurse practitioner; PA, physician assistant; PMPM, per member per month
not intervene in a timely manner or establish a complete picture of ED utilization patterns in their patient population. Facilitator—payment incentives for EDs. Starting in June 2012, Washington State launched a campaign, ‘‘ER is for Emergencies,’’ to reduce ED utilization among Medicaid members. The campaign encouraged hospitals to work with primary care providers across the state. EDs put in place 7 best practices to guide appropriate ED use, adopted a new ED information exchange (EDIE) system, and directed patients to a primary care physician (PCP) for follow-up care.13 The campaign was designed independently of the Pilot and did not connect primary care practices to the EDIE system initially, but the EDIE system enabled PCP communication. The state campaign added financial incentives for hospitals and EDs to show reductions in Medicaid ED utilization rates by June 2013 or face a no-payment policy for a specific list of conditions considered ‘‘not medically necessary.’’14 The state and Pilot created financial incentives for both the ED and Pilot clinics to reduce ED admissions during 2012–2013.
Facilitator—incentives to collaborate. In the West region, the state campaign brought together the single Pilot clinic, local hospital, and a regional health collaborative to identify high utilizers of the ED, develop care plans, and track utilization patterns. In the West and Central regions, 4 Pilot clinics established an EDIE connection to receive electronic ED notifications. In the East region, the state effort facilitated conversations between separate health systems, identifying communication gaps and opportunities to reduce readmission rates. Barrier—lack of actionable population health data. The biggest barrier for all Pilot clinics was the lack of actionable population health data for health plan members at the start of the Pilot. The Pilot used claims data to report ED utilization rates but did not provide specific information such as patient names, diagnosis, or ED locations. Reports were delayed as much as 18 months. Clinics designed population health interventions without adequate data describing ED and hospital utilization patterns. The frequency of ED utilization reporting increased from semiannually to quarterly
LESSONS FROM A PCMH PAYMENT PILOT
241
Table 5. Financial Impact Theme Organization type and region (W = west, C = Central, E = east) Financial Impact Theme—coded responses
PCP-W
Strategies Intended strategy to work for other payment models Held part of new money in reserve in case of calculated pilot shortfall Facilitators Increased financial reimbursement for clinical care Outcomes measures used in other payment strategies Barriers Financial barriers to do more patient evaluation in clinic Staff turnover Payment insufficient for quality improvement work Lessons learned Clinic learned how to think about new payment designs after this experience Clinic learned what data it needs
IPA-C
Multispecialty-C
Health system-C
Health system-E
X
X
X X
X
X
X X
X X
X
X X X
X
X
X
X
X
X
X
X
X
X
X
PCP, independent primary care practice; IPA, independent physician association
in January 2012, but the data were still more than 6 months old and did not reflect the clinics’ current patient population. By the time data were available to practices, the opportunity to make changes had passed. Practices needed accurate population health data prior to the launch of the pilot and at the beginning of each Pilot year to plan effective interventions. Practices found real-time information exchange (EDIE) beneficial as long as the connection was established early to enable action and the volume of data was manageable. Barrier—lack of reliable method to link patients and PCP. The Pilot claims-based attribution method lacked the
functionality of identifying a patient’s primary care provider to patients, EDs, or hospitals at the point of care and was another barrier to timely notification of an ED visit. Barrier—costs to patient. Interviewees reported hearing from patients that a co-pay was a financial barrier to seeking care in the office following an ED visit or for proactive care to prevent complications. For patients with little or no outof-pocket costs for ED care, respondents believed patients preferred to go to the ED instead of selecting a PCP. Clinics thought their patients chose to go to the ED after hours because they were not able or willing to miss work for a primary care visit during business hours. By one clinic report, 60% of ED visits for its population occurred when the primary care clinic was closed. Barrier—disagreement with determination of avoidable conditions. Clinics were harshly critical of the NYUED
algorithm classification method of an avoidable ED visit. The NYUED algorithm uses final diagnosis codes to estimate the probability that a diagnosed condition is avoidable or treatable in the primary care setting. Clinics stressed that a final diagnosis code is not reported initially; instead, patients presented with a chief complaint that required evaluation to determine the final diagnosis.15 Patient evaluations
take time, equipment, resources, and skills that may not be available in the clinic. Primary care providers had to determine the best place to evaluate patients with a chief complaint and reach the final diagnosis. Therefore, the clinics were frustrated with the use of the NYUED classifications for avoidable visits and the inability to challenge and refine outcome results. Lessons learned—clinics needed population health data to design interventions. Clinics learned how to design popu-
lation health interventions and reported wanting population health data to understand clinic trends and opportunities for intervention. Clinics learned to plan interventions with both clinical and financial input. All groups planned to maintain the changes in care team design and all felt they adopted the correct strategy for improving patient care. Clinics tried to take a whole practice approach but they could not apply one strategy for all populations. Clinics needed a different strategy for adults compared to children, and for Medicare and Medicaid populations compared to commercially insured populations. Clinics needed more than one approach to achieve target outcomes and they realized that the different strategies had different costs. PCMH development Strategies. All participating clinics used Pilot funding to develop PCMH core capabilities and hire new staff.16 Clinics created new care teams and changed roles and workflow to provide proactive care. All clinics increased same-day access to care by extending hours or increasing patient access to nurse consultation. Clinics employed staff to monitor patient registries and identify gaps in preventive care or chronic condition management that otherwise could lead to ED visits. 1,17–19 A common strategy was to increase a patient’s personal connection with the care team to encourage the patient to call the primary care clinic before
242
KOSHY ET AL.
going to the ED for care. Care teams varied at each clinic and included multiple disciplines such as medical assistants, registered nurses (RNs), health coaches, health facilitators, social workers, and pharmacists.
needed project management capabilities to oversee change and implement strategies.
Facilitator—financial support for PCMH approach. The new Pilot funds were used by all clinics to pay for essential staff to develop PCMH capabilities and shift the focus from FFS activities to population health management. Hiring RNs carried a significant cost for most practices and having access to new funds enabled organizations to pool funding sources and hire nurses.
strategies with patients, staff, providers, and with outside organizations during the Pilot. Timely response to patient queries was important to reduce ED visits and establish a positive connection with the clinic. Not all patients needed or responded well to follow-up ED calls. Information technology support was needed to automate notification processes. Communication worked well if integrated into electronic health records, but broke down if dependent on individuals sending faxes or making phone calls.
Barrier—inadequate funding for infrastructure development. The new PMPM payments did not pay for adequate
staff time to manage the whole clinic population and this was a significant barrier for all clinics. PMPM payments were made for a minority of the clinic population because of limited health plan participation and low attribution counts. The attribution method linked patients to providers using a statistical method rather than patient preference or sense of engagement with a clinic. Clinics felt this statistical method underestimated pilot population counts, resulting in insufficient funding for infrastructure to carry out populationbased strategies. Barrier—lack of defined patient population. Health plan attribution lists changed quarterly, making it difficult for clinics to know who was in the Pilot. The lists identified patients for quarterly payments but they were not necessarily the same patients included in the outcome measure calculations. This was a barrier for clinics because, with limited funds, they had to choose patients to include in population strategies but did not know whether the same patients or others, not in interventions, would be counted in outcome measures. The risk for clinics would be that outcome measures did not reflect their efforts. Lessons learned—clinics needed on-site assessment capabilities. The Pilot demonstrated the usefulness of on-
site clinical assessment capabilities to manage low-acuity conditions and prevent ED visits. RNs were better equipped to answer questions, address broader medical conditions, and provide trusted advice than medical assistants or receptionists without clinical training. Clinics with separate consulting nurse services found patients were referred to the ED more often than the clinic staff judged necessary, and referral information often was not shared with the clinic. Lessons learned—clinics needed data integration and analysis. The Pilot highlighted how much data integration
from different sources was required to create a composite picture of the clinic population. Data analysis capabilities were needed to identify patients, patterns of ED use, and target populations for interventions. The ability to classify risk helped clinics allocate nursing time, arrange clinic visits, or assign case management skills effectively. Lessons learned—clinics needed project management capability. Clinics often reported that organizational change
was difficult and required daily monitoring, assessment, and reminders to make progress and maintain results. Clinics
Lessons learned—clinics needed effective communication strategies. Clinics learned about effective communication
Financial impact Strategies. Clinics chose strategies that would work for all patients and be supported by other payment methods, thereby permitting sustainable changes. Clinics used Pilot funds to invest in care teams and add PCMH capabilities that would generate returns from a variety of quality improvement incentives. Facilitators—higher FFS. FFS incentives facilitated PCMH transformation. Independent of the Pilot, FFS reimbursement rates increased with the use of new billing codes for Medicare and with adoption of meaningful use criteria for electronic health records. The higher reimbursements influenced parent organization leaders to adopt new care coordination strategies and invest in data systems during the Pilot. Patients who went to the primary care clinic instead of the ED brought additional FFS payments to the clinic and reinforced outreach activities. One clinic found it easier to adopt the Pilot strategy because some Pilot measures were already being monitored for a Medicare FFS payment incentive program under way in the clinic. Barrier—inadequate payment to support quality improvement work. Pilot clinics reported that much of the work
done by medical assistants and RNs for quality improvement was not directly reimbursed; instead clinics typically must wait 2 to 3 years to receive payment for quality improvement. This was a financial barrier for the clinic because it was less likely to achieve quality targets in the long term if it could not pay for the daily work of quality improvement in the short term. In addition, clinics were paid less for the work of nurse practitioners and physician assistants but faced the same infrastructure costs to support mid-level providers in the clinic with nursing staff. Barrier—preauthorization requirements. One clinic reported that incentives were not aligned to enable acute evaluation and treatment in the outpatient setting. The clinic could not obtain preauthorization for imaging studies in the outpatient setting, but preauthorization was granted when the same patient was seen in the ED. This administrative barrier prevented evaluation of complex conditions in the primary care clinic. The cost of providing therapies such as intravenous hydration or medication administration in the clinic could have exceeded the reimbursement for delivering such care and was a barrier for outpatient treatment. For the patient, it was more efficient to be
LESSONS FROM A PCMH PAYMENT PILOT
evaluated and treated in the ED rather than wait for preauthorization of diagnostic studies. Lessons learned—clinics needed data to monitor progress. Clinics learned they needed more detailed data from
health plans to track progress on Pilot measures given financial consequences tied to outcomes and discrepancies between health plan data and parent organization quality reports. Clinics wanted to have more population data to understand trends in their own practice. To lower ED utilization rates, clinics needed real-time ED notification and clinical information to identify utilization patterns among various populations. Clinics needed timely feedback about clinic interventions. Lessons learned—cost of health interventions differed by populations. Clinics learned to quantify the cost of trans-
formation and the financial investments undertaken to achieve outcomes. Clinics learned what payment amounts are needed to manage a population with the PCMH model and specifically how payments differ by populations based on risk, patterns of use, and complexity of illness. Discussion
The myriad challenges faced by Pilot clinics also were valuable learning experiences in health care transformation. Investigators learned how payment shapes the health care system today and what obstacles need to be overcome to arrive at better outcomes for patients and providers. The Pilot chose a population health objective, to reduce costly ED and inpatient utilization rates, and practices responded with population health interventions. Practices reported more barriers than facilitators because the data and payment provided were not specifically designed for population health interventions. Clinical care transformation was ahead of data and payment transformation. The primary care Pilot experience teaches us what needs to change in data design and payment methods to achieve a whole health system transformation. Transforming data to support population health management
The biggest challenge for primary care practices was not having timely, detailed, or actionable data from claims billing sources. A transformation in data collection and reporting is needed that can identify patients, their primary care providers, specialists, and locations of care and describe utilization patterns that affect population outcomes. Bringing together information for patients and their providers at the point of care is needed for effective decision making. For example, to reduce ED utilization, end users needed a reliable and accessible source of health information describing ED use for all individuals seeking care in the state, available in real time, with information on trends of standardized measurement indicators. This level of data integration and detail is not currently available to all medical providers in the state. Hospital administrators, ED managers, and clinicians were not explicitly included in the design of the Pilot or involved in the informational meetings that accompanied implementation of the Pilot. With the benefit of hindsight,
243
the technical barriers (eg, lack of interoperability between hospital and clinic information systems to effectively exchange clinical information) might have been mitigated significantly by incorporating the hospital and ED perspective systematically during Pilot design and implementation. Future efforts to transform care and improve outcomes cannot be limited to 1 specialty. Population health data must reflect the care paths taken by the population, which includes care in the community, primary care, specialty care, and hospital-based services. Increased investments in data development can drive up health care costs if duplicate information systems are created. It is in the public interest to build a robust population health database and share the common resource. In addition, population health data must be all inclusive. Data that exclude groups (eg, the uninsured) limit the ability to understand patterns in health outcomes and improve the health system for all populations. What is needed to transform data to enable innovative approaches to health care delivery is an organization capable of creating data tools with this allinclusive population perspective. Transforming payment to support population health management
The incremental changes to FFS payments for a limited number of Pilot patients were inadequate to change outcomes for a larger population. Paying for population health interventions requires building the infrastructure to deliver proactive and coordinated care for populations and is different from paying for individual acute care services. This study identified the infrastructure needed to carry out population health strategies that included on-site clinical assessment, data and project management, communication capacity, and data integration across health care settings, specifically connecting primary care with EDs, hospitals, and specialty care. It is essential to change payments across the health care environment to gain meaningful transformation. Investigators learned that payment changes in both EDs and Pilot primary care clinics were effective in changing provider behavior and creating collaborative partnerships. Preliminary results show positive outcomes. Pilot clinics reduced the overall avoidable ED rate by 10.7% according to a report prepared by the University of Washington for the Washington Health Care Authority (Conrad D, Chan G. June 30, 2014, Report to the Health Care Authority). The state campaign reduced total Medicaid ED utilization by 9.9%.20 Health care costs for individuals likely affected personal use of health services. If mental health, dental care, or primary care visits are not included in a person’s insurance coverage, or if there is too high a co-pay or shared payment, this could lead individuals to the ED where they are guaranteed an evaluation. Transforming health care payments should take into account personal behavior with regard to cost and align incentives for patients as well as providers to achieve shared population health outcomes. Limitations
Responses describing the primary care experience with the Pilot payment method are limited to the views of the
244
Pilot clinics and may not identify all potential themes related to the study questions. Some interviewees joined their clinic partway through the Pilot and had limited knowledge of the strategies, facilitators, barriers, and lessons learned. The majority of Pilot participants represented large health systems because of barriers to entry for small independent clinics; responses may not capture the full experience of small independent practices.
KOSHY ET AL.
3.
4.
Conclusions
The Pilot tested an alternative payment method to improve population outcomes. Clinics chose evidence-based population health interventions to achieve Pilot targets. The alternative data reporting and payment methods were insufficient to realize the potential of the PCMH population health approach. This study identifies where further changes are needed. Data for population health interventions must: (1) define the population, (2) be accessible to health care providers, and (3) have an all-inclusive perspective to describe the health needs of an entire population. Payment methods need to create incentives that work across the health system and support infrastructure development for population health interventions, including integrated information exchange between primary care, specialty care, EDs, and hospitals. Changes in data integration and payment in both the ED and primary care resulted in better outcomes from preliminary results. It will take more than just primary care transformation to improve health outcomes—a significant transformation in data collection, reporting, and payment must match the change occurring in clinics.
5. 6.
7.
8. 9. 10. 11.
Author Disclosure Statement
Drs. Koshy, Conrad, and Grembowski declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Koshy worked as an independent contractor for the Washington Health Alliance and retained full independence to conduct the research. She was project manager for the Pilot from 2010 to 2013. Drs. Conrad and Grembowski report no potential conflicts of interest. Informed consent was obtained from all participants to proceed with interviews, recording, and use of quotes. No patient identifiable data were collected and interview responses were considered secondary data from Pilot operations and exempt from human subjects or institutional review board approval. This research was supported and funded by the Washington Health Alliance (WHA, formerly the Puget Sound Health Alliance). The WHA is a regional health collaborative in Seattle, co-convener of the Washington State MultiPayer Medical Home Payment Pilot, and provided project management for the Pilot. References
1. Wagner EH, Coleman K, Reid RJ, Phillips K, Sugarman JR. Guiding Transformation: How Medical Practices Can Become Patient-Centered Medical Homes. New York: The Commonwealth Fund; February 2012. 2. Bailit M, Phillips K, Long A. Safety Net Medical Home Initiative. Paying for the Medical Home: Payment Models to Support Patient-Centered Medical Home Transformation
12.
13.
14.
15.
16. 17. 18.
19.
in the Safety Net. Seattle, WA: Bailit Health Purchasing and Qualis Health; October 2010. Grumbach K, Grundy P. The Outcomes of Implementing Patient-Centered Medical Home Interventions: A Review of the Evidence from Prospective Evaluation Studies in the United States. Updated November 16, 2010. http://www .cms.org/uploads/GrumbachGrundy2010OutcomesPCPCC .pdf. Accessed July 30, 2014. Baillit Health Purchasing, LLC. Facilitators and Barriers to Payment Reform: Market based, Governmental, Organizational and Design Considerations. http://www.bailit-health .com/articles/092713_bhp_rwjf_facilitators.pdf. Accessed July 30, 2014. NYU Wagner. NYU ED algorithm. http://wagner.nyu.edu/ faculty/billings/nyued-background. Accessed July 30, 2014. Agency for Health Care Research and Quality. Prevention Quality Indicators overview. http://www.qualityindicators .ahrq.gov/modules/pqi_resources.aspx. Accessed July 30, 2014. Firestone WA, Herriott RE. Multisite qualitative policy research: some designs and implementation issues. In: Fetterman DM, ed. Ethnography in Educational Evaluation. Beverly Hills: Sage Publications; 1984:63–88. Pettigrew AM. Longitudinal field research on change: theory and practice. Organization Science. 1990;1:267–292. Stake RE. Multiple Case Study Analysis. New York: The Guilford Press; 2006. Conrad D, Grembowski D, Gibbons C, et al. A report on eight early-stage state and regional projects testing valuebased payment. Health Aff (Millwood). 2013;32:998–1006. Conrad DA, Grembowski D, Hernandez SE, Lau B, MarcusSmith M. Emerging lessons from regional and state innovation in value-based payment reform: balancing collaboration and disruptive innovation. Millbank Q. 2014;92: 568–623. Care Continuum Alliance. Implementation and Evaluation: A Population Health Guide for Primary Care Models. 2012. http://selfmanagementalliance.org/wp-content/uploads/2013/ 11/A-Population-Health-Guide-for-Primary-Care-ModelsCareContinAlliance.pdf. Accessed July 30, 2014. Washington State Hospital Association. ER is for Emergencies results in better care coordination and cost savings. http://www.wsha.org/ERemergencies.cfm. Accessed July 30, 2014. Washington State Hospital Association. Background information: Medicaid emergency department use. http://www .wsha.org/files/ER%20Background%20Information%20 April%2016.pdf. Accessed July 30, 2014. Raven MC, Lowe RA, Macelli J, Hsia RY. Comparison of presenting complaint versus discharge summary for identifying ‘‘nonemergency’’ emergency department visits. JAMA. 2013;309:1145–1153. Safety Net Medical Home Initiative. Change concepts overview. http://www.safetynetmedicalhome.org/changeconcepts. Accessed July 30, 2014. Morgan SR, Chan AM, Alquatari M, Pines JM. Non-emergency department interventions to reduce ED utilization: a systematic review. Acad Emerg Med. 2013;20:969–985. Safety Net Medical Home Initiative. Practice Results. Number of days to third next available appointment. http:// www.safetynetmedicalhome.org/sites/default/files/PracticeResults-Enhanced-Access.pdf. Accessed July 30, 2014. Safety Net Medical Home Initiative. Practice Results. Contacting patients within two days of their hospital discharge.
LESSONS FROM A PCMH PAYMENT PILOT
http://www.safetynetmedicalhome.org/sites/default/files/ Practice-Results-Care-Coordination.pdf. Accessed July 30, 2014. 20. Washington State Health Care Authority. Report to the Legislature. Emergency Department Utilization: Update on Assumed Savings from Best Practices Implementation. March 20, 2014. http://www.hca.wa.gov/Documents/Emergency DeptUtilization.pdf. Accessed July 30, 2014.
245
Address correspondence to: Reena Koshy MD MPH 4464 Fremont Ave N Suite #103 Seattle, WA 98103 E-mail: [email protected]
