A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care OCTOBER 2019 AUTHORS Len Finocchio, Matthew Newman, and Shawn Blosser Blue Sky Consulting Group Contents About the Authors 3 Executive Summary Len Finocchio leads the Blue Sky Consulting 5Introduction Group’s health care practice. He earned his Doctorate in Public Health from the University 8 Research Approach and Methods of Michigan. Matthew Newman is a founding partner with the Blue Sky Consulting Group. 1 0 Findings: Access, Quality, and He earned his Master of Public Policy degree Consumer Experience from Harvard University’s John F. Kennedy 2 3 Findings: Stakeholder Interviews School of Government. Shawn Blosser leads the Blue Sky Consulting Group’s data analy- 26 Discussion sis team. He earned his bachelor’s degree at 2 7 Considerations for Improvement Stanford University and did graduate work in economics at the University of Chicago. 29 Appendices A. Structured Interview Participants Acknowledgments B. Regression Analysis Methodology and Results The authors would like to acknowledge the expert contributions to this paper of Athena C. CAHPS Measures Comparison: GMC and Urban Counties Chapman, MPP; Caroline Davis, MPP; and Jill 33 Endnotes Yegian, PhD. About the Foundation The California Health Care Foundation is dedicated to advancing meaningful, measur- able improvements in the way the health care delivery system provides care to the people of California, particularly those with low incomes and those whose needs are not well served by the status quo. We work to ensure that people have access to the care they need, when they need it, at a price they can afford. CHCF informs policymakers and industry leaders, invests in ideas and innovations, and connects with changemakers to create a more responsive, patient-centered health care system. For more information, visit www.chcf.org. California Health Care Foundation www.chcf.org 2 Key findings include the following: Executive Summary $$ Access to care for Medi-Cal enrollees in GMC coun- ties appears to be no better than for enrollees in the Background comparison urban COHS and Two-Plan counties. In most counties in California, the Department of Health Care Services (DHCS) contracts with one or $$ The quality of care delivered by MCPs in GMC coun- two managed care plans (MCPs) to deliver services to ties was lower, on average, for 22 of 30 measures Medi-Cal enrollees. In just two counties, Sacramento compared with MCPs in the comparison counties. and San Diego, DHCS contracts directly with five or Quality scores were the same on six measures and more MCPs. This approach, the Geographic Managed better on two. The biggest difference was found Care (GMC) model, provides enrollees in those coun- in immunization rates for children and adolescents, ties with more options. It is unclear whether greater where GMC plans had lower rates by 8% and 7%, competition among MCPs has led to higher quality of respectively, than comparison plans. care, better access to services, or better experiences $$ Enrollees in the two GMC counties have many MCP for Medi-Cal enrollees and their providers, or whether options, and they exercise their ability to move market fragmentation leads to navigational challenges between MCPs at higher rates than enrollees in and poor coordination of care. similar counties. This multiplicity of MCP options, however, does not clearly manifest in better patient In 2020, DHCS is scheduled to release its Request satisfaction: Average scores for MCPs in GMC coun- for Proposals (RFPs) to begin the procurement pro- ties were higher on some satisfaction measures and cess to select commercial MCPs for the Medi-Cal lower on others. program starting in January 2023.1 This makes it the ideal time to assess the GMC model of managed care $$ Many providers in GMC counties contract with mul- and whether, and under what circumstances, DHCS tiple MCPs, resulting in overlap across MCPs. This is should continue to support it. Using a mixed-methods more common among primary care providers than approach that combines quantitative and qualitative specialists. Some providers, particularly specialists, analysis, this report examines how quality of care, may be found on only one MCP’s network listings, access to care, and patient satisfaction in GMC coun- resulting in some differentiation across networks. ties compare with similar urban counties that use a Consequently, enrollees moving between plans County Organized Health System (or COHS, a single and networks might be able to keep their primary public MCP) or the Two-Plan model (where a public care provider but are more likely to change their MCP and a commercial MCP compete). specialists. $$ Interviews with consumer groups, health care pro- viders, and other stakeholders in Sacramento and Results San Diego Counties noted that the multiplicity of This assessment reveals a mixed picture of GMC per- MCPs creates a confusing patchwork of networks, formance and its ability to accomplish purported goals providers, benefits, and services. As a result, they for Medi-Cal enrollees. This analysis of data on access report, some enrollees struggle to find and use to care, MCPs’ quality performance, and patient sat- services, particularly specialists. For providers, this isfaction suggests that the promise of GMC has not multiplicity of MCPs presents many redundancies in been fully realized. contracting, administrative requirements, and clini- cal approaches to care and quality improvement, which bring significant transaction costs. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 3 Considerations for Improvement By raising the bar on performance and other expecta- There are significant barriers to changing from the tions, these actions might naturally reduce the number GMC model to the COHS or Two-Plan model in either of MCPs interested in pursuing a contract with DHCS county.2 However, numerous approaches should be to serve Medi-Cal enrollees in Sacramento and San considered to improve access, quality, and consumer Diego Counties. If not, DHCS, working collaboratively and provider experience in GMC counties, including with San Diego and Sacramento, should consider the following: reducing the number of MCPs with which it contracts in these GMC counties. Although there was no con- $$ Demand more from participating MCPs. DHCS sensus among interviewees on what the ideal number could use the forthcoming procurement to raise should be, the authors also did not find evidence that expectations and demand more from MCPs bid- more MCP competition leads to improvements in any ding on a contract. For example, DHCS could: of the outcomes studied. By limiting the number of $$ Strengthen oversight of network adequacy by MCPs with which it contracts in a given county, DHCS calculating accurate physician-to-patient ratios could use its leverage to create greater competition for each MCP. Require MCPs to report, for each for those contracts. MCPs that receive a contract might in-network physician, the percentage of their be more willing to make the investments DHCS and practice spent seeing that MCP’s Medi-Cal San Diego and Sacramento County officials want than enrollees. they would if they had to divide up the market among four to six competitors. Moreover, enrollees’ health $$ Set measurable quality improvement targets care providers might benefit from greater efficiency and require MCPs to describe their approaches and improved navigational ease associated with work- for making significant and sustainable improve- ing with fewer MCPs. Were DHCS and the counties to ments in quality consistent with these targets. take this approach, they should also develop plans to $$ Require MCPs to make investments to expand minimize potential disruptions to enrollees and pro- access to care, strengthen the local delivery sys- viders during the transition period. tems, and address enrollees’ social determinants of health. The upcoming DHCS procurement of commercial MCP contracts in GMC and other model counties is $$ Adopt positive financial incentives tied to MCP an infrequent and important opportunity to catalyze performance. Several other states have adopted significant improvements in quality, satisfaction, and such programs to foster performance improve- health outcomes for Medi-Cal enrollees. California’s ments, whereas DHCS relies primarily on penalties experience with the GMC model provides use- for poor performance.3 ful lessons that should be applied to the upcoming procurement process and ongoing performance $$ Foster greater collaboration among county and expectations for Medi-Cal MCPs. state stakeholders. DHCS could commit resources to working collaboratively with county officials, consumer advocates, and MCP and provider rep- resentatives in San Diego and Sacramento to establish and advance improvement priorities and goals. Moreover, DHCS or the legislature could give Healthy San Diego and the Sacramento Medi- Cal Managed Care Advisory Committee, both of which are legislatively mandated, a more direct role in establishing procurement priorities, reviewing MCP bids, and overseeing MCP performance. California Health Care Foundation www.chcf.org 4 Introduction California pioneered the use of managed care for counties, DHCS is planning the release in 2020 with Medicaid in the 1970s. Over the next four decades, implementation scheduled for January 2023 (these the state progressively expanded managed care to dates are subject to change). This reprocurement pro- include most Medi-Cal enrollees across all 58 coun- vides an opportunity to review and evaluate the ways ties. As of March 2019, 10.6 million low-income in which managed care is implemented in California Californians were enrolled in Medi-Cal managed care. and to foster significant performance improvements. This represents 82% of all Medi-Cal enrollees.4 Figure 1. Managed Care Enrollment, by MCP Type, July 2019 The Medi-Cal managed care program is organized using three distinct approaches: (1) counties with a single public managed care plan (MCP), called a Imperial (1%) • • Cal MediConnect (1%) County Organized Health System (COHS); (2) counties Other* (<1%) • • Regional (3%) with competition between a public MCP and a com- mercial MCP (Two-Plan model); and (3) counties with GMC competition among two or more commercial MCPs 11% (Geographic Managed Care, Regional, and Imperial models). Managed care enrollment is mandatory for most Medi-Cal enrollees in 57 of 58 counties. In San Benito County, only one commercial MCP participates COHS 20% and enrollment is voluntary. Two-Plan 64% This study examines California’s experience with the Geographic Managed Care (GMC) model, which operates in two counties, Sacramento and San Diego, and accounts for 11% of Medi-Cal managed care enrollment (Figure 1). With its evolution over time and distinct complexity, California’s experience with the *Includes enrollment in the following programs: CCS Demonstration GMC model is not well understood. The availability of (0.00%), Primary Care Case Management (0.01%), Special Project (0.04%), PACE (0.05%), San Benito Model (0.07%), and SCAN (0.13%). multiple MCPs, and the competition among them, can Notes: CCS is California Children’s Services; CHHS is California Health offer unique advantages but may also come with chal- and Human Services; PACE is Program of All-Inclusive Care for the Elderly; SCAN is Senior Care Action Network. lenges and costs. This study assesses the GMC model’s Source: Department of Health Care Services, “CHHS Open Data,” overall performance relative to other managed care Medi-Cal Managed Care Enrollment Report, accessed August 9, 2019, models in urban counties, comparing quality of care, data.chhs.ca.gov. access to care, and the overall patient experience. The study was conducted to inform California’s repro- curement of commercial managed care contracts. Earlier this year, the Department of Health Care Services (DHCS) released a Request for Proposal (RFP) schedule to procure MCPs for each of the dif- ferent models in their respective counties.5 For GMC A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 5 Geographic Managed Care Potential Benefits and Challenges of The GMC model was created in 1992 by Assembly the GMC Model Bill 336 and Senate Bill 485. Its regulatory authority There are several potential benefits of the GMC model exists in Welfare and Institutions Code, Section 14089, in which multiple MCPs contract with DHCS to provide and California Code of Regulations, Title 22, Sections services to enrollees. They include the following: 53900-53928. The GMC model was created to $$ More options for enrollees. To the extent that “improve the Medi-Cal program by increasing access, improving quality of care, reducing episodic care, and MCPs differentiate and enrollees are aware of and achieving an overall cost savings for the program.”6 understand these differences, the GMC model provides more options for enrollees to select an Currently, DHCS has contracts with five commercial MCP based on their individual preferences, such MCPs in Sacramento County and seven in San Diego as quality of care, provider network and avail- (Table 1). The most recent MCP entrants to GMC are ability, care models, community-based services, United Healthcare and Aetna, which began enrolling culturally appropriate care, and member services. Medi-Cal enrollees in both counties in 2017 following Furthermore, an enrollee can leave an MCP and a competitive procurement process in which DHCS select another each month based on their prefer- sought to expand the number of commercial MCPs ences. In most other counties, enrollees have only serving enrollees in these counties. United subse- one other MCP or no other MCP option. quently exited the Medi-Cal market in Sacramento in $$ Competition among MCPs. Beyond meeting mini- November 2018. mum contractual requirements to serve enrollees, MCPs may compete to increase their share of Medi- Cal enrollees, seeking to differentiate themselves by demonstrating better quality, larger networks for access to care, or better customer service than competitors. This competition might lead to better Table 1. GMC Enrollment, by MCP and County, July 2019 SACRAMENTO SAN DIEGO TOTAL GMC NUMBER % OF COUNTY NUMBER % OF COUNTY NUMBER % OF TOTAL Aetna 7,347 1.7% 9,737 1.4% 17,084 1.5% Anthem Blue Cross 176,889 41.4% 176,889 15.9% Blue Shield Promise (Care1st) 80,600 11.7% 80,600 7.2% Community Health Group 254,797 37.0% 254,797 22.8% Health Net 105,593 24.7% 67,130 9.8% 172,723 15.5% Kaiser Foundation 87,289 20.4% 49,289 7.2% 136,578 12.2% Molina Healthcare 50,339 11.8% 215,578 31.3% 265,917 23.8% United Healthcare 11,125 1.6% 11,125 1.0% Total 427,457 100.0% 688,256 100.0% 1,115,713 100.0% Source: Department of Health Care Services, “CHHS Open Data,” Medi-Cal Managed Care Enrollment Report, accessed August 9, 2019, data.chhs.ca.gov. California Health Care Foundation www.chcf.org 6 performance for individual MCPs or collectively $$ Administrative costs for providers. Providers must relative to other models. Competition might also meet myriad contractual and administrative require- drive down costs. ments of each MCP and the costs associated with them. While standard DHCS contract requirements $$ Greater leverage for DHCS over MCP perfor- and the role of independent physician associations mance. With enrollment spread across five to (IPAs) may mitigate the duplicative nature of many seven MCPs, DHCS may have greater leverage requirements, providers must establish, staff, and over poorly performing MCPs in that terminat- navigate care coordination protocols, information ing an MCP contract would be disruptive to fewer technology, formularies, pre-authorizations and enrollees than in a model with fewer MCPs. referrals, and claims processes for each MCP. In $$ More options for providers. MCPs may compete addition, MCPs may also have varying quality pri- for providers to have a broad network through bet- orities and payment incentive schemes. ter rates, more efficient administrative processes, $$ Administrative costs to DHCS. Likewise, there and better technological and other support. In may be additional costs to DHCS related to the addition, MCPs can attract providers by offering administration of contracts, financial compliance, access to Medicare and commercial lines of busi- quality reporting, and other requirements of mul- ness. With more MCPs offering potential contracts, tiple MCPs. providers would also have more negotiating lever- age on rates and other factors. $$ Provider leverage over MCPs and network frag- mentation. With multiple MCPs building their There are also potential challenges and costs to networks, providers may be able to avoid contract- a model with multiple MCPs. These include the ing with one or more MCPs if the terms are not to following: their liking, resulting in more provider leverage and potentially higher costs for MCPs (and ultimately $$ MCP selection challenges and delay of care. It the state). For patients, such provider leverage and may be difficult for enrollees to choose from among selective contracting may result in fragmented net- multiple MCP options. Research, discussed below, works and difficulties with access to care. has shown that enrollees with multiple options are more likely to delay enrollment, resulting in higher health care costs due to delaying needed care. Health Plan Choice in the Literature $$ Navigation and challenges with gaps in care when In private health care markets, increased competi- changing MCPs. The availability of multiple MCPs, tion and a higher number of competing MCPs are fragmented provider networks, and enrollees’ abil- associated with decreased costs.7 However, the links ity to change monthly bring logistical challenges between cost savings, quality of care, and patient sat- and transactional costs for everyone involved: isfaction are not always clear. Rivers and Glover’s (2008) Enrollees must undertake the MCP change require- review of studies examining competition could not ments (ideally, after comparing MCPs based on conclude whether competition-related cost savings their preferences and priorities) and, after switching were due to improved efficiency or decreased quality MCPs, learn the unique aspects of the new MCP’s of care.8 Enthoven and Baker (2018) found that pro- policies and procedures; MCPs must exchange and viding patients more options is generally associated process data for the disenrollment and enrollment; with higher patient satisfaction.9 Conversely, a study of and health care providers must process a new national MCPs found that quality of care is not neces- patient intake and associated health assessments. sarily positively impacted by competition.10 Critics of The challenges and costs could include gaps in private market competition among MCPs argue that a care, confusion from changing formularies, and larger number of MCPs is more complicated, involves duplication of services. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 7 more administrative costs, and increases consumer Research Approach confusion.11 and Methods Millet, Chattopadhyay, and Bindman (2010) found that enrollees with MCP options from which to choose Study Questions (consisting of GMC and Two-Plan models) were more Using a combination of quantitative and qualitative likely to take longer to select an MCP and more likely analysis, the authors assessed the effectiveness of to have shorter MCP enrollment than in counties with the GMC model and compared the results to those only one MCP (County Organized Health System).12 in comparable counties with different managed care They concluded that enrollees delaying MCP selec- delivery models. The authors were able to use data tion, when faced with multiple MCPs, were more likely analysis and interviews to assess a subset of the theo- to have higher adjusted hospital admission rates for retical rationales listed above. ambulatory care–sensitive conditions. The authors set out to answer the following questions Additionally, Bindman (2018) raised concerns about about the potential benefits of multiple MCPs: competition among Medicaid MCPs in decreasing $$ Do multiple MCPs, with potentially larger net- the network size of available providers, thus reduc- works, improve access to care for enrollees? ing enrollee access to care.13 A negative association between number of MCPs and quality of care was $$ Does competition between MCPs lead to higher also found among managed care competition in quality performance? New York’s State Children’s Health Insurance Program $$ Do enrollees take advantage of having more (SCHIP). Counties with a greater number of MCPs saw options by changing MCPs more frequently? lower Healthcare Effectiveness Data and Information Set (HEDIS) and Consumer Assessment of Healthcare $$ Do multiple MCPs and the ability of enrollees Providers and Systems (CAHPS) scores.14 to move between them lead to higher patient satisfaction? $$ Are MCPs’ provider networks differentiated, and do networks expand when additional MCPs enter the market? The authors also set out to answer these questions about the potential challenges of the GMC model: $$ Does the multiplicity of MCP options contrib- ute to higher default rates due to challenges for enrollees making a choice among them? $$ Are there navigational challenges for enrollees and providers with multiple MCPs, care models, and networks? $$ What are the types of transaction costs incurred because of the model’s complexity? California Health Care Foundation www.chcf.org 8 Methodology from stakeholders and leaders. These interviews were The authors collected and analyzed available data on aimed at assessing the benefits and challenges of the patient satisfaction and experience, access to care, GMC model and at surfacing recommendations for measures of MCP quality performance, and provider improvements that could be made. See Appendix A networks. Specifically, the authors collected and ana- for the list of interviewees. lyzed data from the following: $$ HEDIS Table 2. GMC Comparison Group Counties $$ CAHPS MCP(S) Alameda Anthem Blue Cross $$ California Health Interview Survey (CHIS) Alameda Alliance for Health $$ SelectedDHCS Medi-Cal Managed Care Contra Costa Anthem Blue Cross Performance Dashboard measures Contra Costa Health Plan $$ DHCS data on enrollee MCP selections and Los Angeles Health Net changes LA Care $$ DHCS MCP provider directory files Orange CalOptima Riverside Molina Health Care The authors compared the data analysis results from Inland Empire Health Plan GMC counties with those in a comparison group of counties. Counties were classified based on the per- San Bernardino Molina Health Care centage of the population that lived in urbanized Inland Empire Health Plan areas according to the 2010 census. San Diego and San Francisco Anthem Blue Cross Sacramento are both urban, so comparison counties San Francisco Health Plan were selected that were similarly urban. Table 2 lists San Mateo Health Plan of San Mateo the comparison counties and MCPs. Santa Barbara CenCal Health In order to assess the performance of the GMC model, Santa Clara Anthem Blue Cross the authors identified specific comparison metrics and Santa Clara Family Health Plan outcome measures and compared the results from Ventura Gold Coast Health Plan GMC counties with comparison urban group coun- ties. In addition, the authors conducted a regression Source: Blue Sky Consulting Group analysis, 2019. analysis using data from CHIS in which the authors assessed key access measures from the survey while controlling for factors such as patient demographic characteristics.15 The authors also conducted more than 30 structured interviews with stakeholders and experts from clinics, IPAs, hospital associations, counties, advisory groups (e.g., Healthy San Diego), MCPs, and consumer advo- cacy organizations. The interviews provide narrative insights and detail into the performance of the model A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 9 Findings: Access, The regression analyses, presented in Appendix B, do not show any statistically significant differences of the Quality, and Consumer managed care model on these selected CHIS access- to-care measures. That is, access to care in GMC Experience counties appears to be no better than in counties with only one or two MCPs. Access to Care in GMC Counties Is Not Better than in Comparison Table 3. CHIS Access-to-Care Measures, GMC and Comparison Counties, 2017 Counties The authors analyzed survey results from 2017 CHIS GMC URBAN to compare selected self-reported access-to-care Did not have usual source of care 18% 17% measures between GMC and other urban model Usual source of care: ER, some other 21% 21% counties.16 First, the authors compared how respon- place, no usual place dents answered selected access-to-care questions. Had difficulty finding primary care 11% 8% The authors report these comparisons using the survey question categories and language employed by CHIS. Had difficulty finding specialty care 27% 17% Second, to test initial findings, the authors developed Insurance not accepted by medical 22% 16% several regression models, as explained below. specialist in past year Sometimes/never able to get doctor’s 38% 35% The initial comparison of respondents’ self-reported appointment within two days access to care (Table 3) found differences for several Source: California Health Interview Survey (2017) and UCLA Center for measures, suggesting poorer access to care in GMC Health Policy Research, AskCHIS [online health query system], accessed model counties: August 9, 2019, ask.chis.ucla.edu. $$ Did not have usual source of care Quality of Care Is Generally Lower $$ Usualsource of care at emergency room (ER) or some other place in GMC Counties HEDIS is the most widely used measure set for evaluat- $$ Had difficulty finding primary and specialty care ing and comparing quality among MCPs. According to $$ Insurance not accepted by medical specialist in the Centers for Medicare & Medicaid Services (CMS), past year HEDIS measures can be used by MCPs “to identify opportunities for improvement, monitor the success $$ Had difficulty getting doctor’s appointment in of quality improvement initiatives, track improve- two days ment, and provide a set of measurement standards that allow comparison with other plans.”17 The state To further test these findings, the authors developed of California uses HEDIS to measure the effectiveness several regression models. Using these same 2017 of Medi-Cal MCPs, and publishes the results annually CHIS data, the authors specified a regression model in Medi-Cal Managed Care External Quality Review to test the effect of the managed care model on spe- Technical Report.18 cific dependent variables measuring access to care: no usual source of care, ER visits, delayed care, and DHCS collects and reports more than two dozen any access. The authors controlled for numerous HEDIS measures from Medi-Cal MCPs. Examples respondent demographics, self-reported health sta- include measures relating to immunization status, tus and health conditions, and self-reported mental cancer screening, heart disease and diabetes manage- health status. ment, emergency department utilization, and hospital readmissions. California Health Care Foundation www.chcf.org 10 In order to facilitate analysis of available data for this $$ All-cause readmissions. This measure is reported study, Medi-Cal MCP HEDIS measures for 2015–2018 in its original form. were summarized into four categories: To assess HEDIS quality measures in GMC and urban $$ All-measures average. This measure includes the model counties, the authors compared all measures simple average for all available measures.19 and aggregated the results here. Measures for Aetna $$ Child and adolescent access to primary care. This and United Healthcare, the newest MCPs to participate summary measure includes the average of the follow- in GMC counties, were not included in the 2017–2018 ing individual measures: Childhood Immunization measurement year.20 The authors also excluded Kaiser Status — Combination 3, Children and Adolescents’ from the analysis in GMC counties.21 In a subsequent Access to Primary Care Practitioners  12–24—  section below, the authors present quality scores in Months, Children and Adolescents’ Access to San Diego and Sacramento Counties with and without Primary Care Practitioners  25 Months–6 Years, —  Kaiser in the analysis. Children and Adolescents’ Access to Primary Care Practitioners  7–11 Years, Children and —  As shown in Table 4, quality of care among MCPs par- Adolescents’ Access to Primary Care Practitioners —  ticipating in GMC was worse, on average, compared 12–19 Years, Immunizations for Adolescents  —  with quality of care among MCPs in comparison urban Combination 2, and Well-Child Visits in the Third, counties on the all-measures average (69% GMC ver- Fourth, Fifth, and Sixth Years of Life. sus 72% urban), on access to primary care for children and adolescents (76% GMC versus 80% urban), and $$ Chronic disease management. This sum- on chronic disease management measures (68% GMC mary measure includes the average of the versus 70% urban). There were no differences in rates following individual measures: Annual Monitoring of All-Cause Readmissions between model types. for Patients on Persistent Medications  ACE—  Inhibitors or ARBs, Annual Monitoring for Patients Individual HEDIS measures show that GMC perfor- on Persistent Medications  Diuretics, Asthma —  mance, as reflected in these aggregate scores, was Medication Ratio — Total, Comprehensive Diabetes generally lower than the comparison urban counties Care  Blood Pressure Control (<140/90 mm Hg), —  (Table 5, page 12). Of the 30 measures, GMC counties Comprehensive Diabetes Care Eye Exam —  performed worse on 22 measures, the same on six, (Retinal) Performed, Comprehensive Diabetes and better on two. The biggest difference was found Care  HbA1c Control (<8.0%), Comprehensive —  in immunization rates for children and adolescents, Diabetes Care  HbA1c Poor Control (>9.0%), —  where GMC plans had lower rates (by 8% and 7%, Comprehensive Diabetes Care — Hemoglobin respectively) than comparison plans. The two mea- A1c (HbA1c) Testing, Comprehensive Diabetes sures where GMC plans outperformed comparison Care  Medical Attention for Nephropathy, and —  county plans were nutrition counseling for children Controlling High Blood Pressure. and adolescents and outpatient visits per 1,000 mem- ber months. Table 4. Aggregated HEDIS Measures for All Years, GMC and Comparison Counties, 2015–2018 ALL-MEASURES CHILD AND ADOLESCENT CHRONIC DISEASE ALL-CAUSE AVERAGE ACCESS TO PRIMARY CARE MANAGEMENT READMISSIONS GMC 69% 76% 68% 16% Urban 72% 80% 70% 16% Source: Blue Sky Consulting Group analysis of Department of Health Care Services, Medical Managed Care Quality Improvement Reports: External Quality Review Technical Reports with Plan-Specific Evaluation Reports (July 1, 2015–June 30, 2016; July 1, 2016–June 30, 2017; and July 1, 2017–June 30, 2018), www.dhcs.ca.gov. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 11 Table 5. Average HEDIS Score, by Quality Measure, 2015–2018 GMC URBAN Prenatal and Postpartum Care—Postpartum Care 58% 62% Prenatal and Postpartum Care—Timeliness of Prenatal Care 78% 82% Childhood Immunization Status—Combination 3 66% 74% Immunizations for Adolescents—Combination 2 28% 35% Children and Adolescents’ Access to Primary Care Practitioners—12–24 Months 91% 93% Children and Adolescents’ Access to Primary Care Practitioners—25 Months–6 Years 82% 85% Children and Adolescents’ Access to Primary Care Practitioners—7–11 Years 84% 87% Children and Adolescents’ Access to Primary Care Practitioners—12–19 Years 81% 84% Weight Assessment and Counseling for Nutrition and Physical Activity for Children and Adolescents— 73% 72% Nutrition Counseling—Total Weight Assessment and Counseling for Nutrition and Physical Activity for Children and Adolescents— 64% 64% Physical Activity Counseling—Total Well-Child Visits in the Third, Fourth, Fifth, and Sixth Years of Life 70% 74% Breast Cancer Screening 56% 59% Cervical Cancer Screening 52% 58% Annual Monitoring for Patients on Persistent Medications—ACE Inhibitors or ARBs 86% 87% Annual Monitoring for Patients on Persistent Medications—Diuretics 86% 86% Asthma Medication Ratio—Total 58% 58% Comprehensive Diabetes Care—Blood Pressure Control (<140/90 mm Hg) 58% 61% Comprehensive Diabetes Care—Eye Exam (Retinal) Performed 50% 56% Comprehensive Diabetes Care—HbA1c Control (<8.0%) 49% 52% Comprehensive Diabetes Care—HbA1c Poor Control (>9.0%) 40% 38% Comprehensive Diabetes Care—Hemoglobin A1c (HbA1c) Testing 83% 86% Comprehensive Diabetes Care—Medical Attention for Nephropathy 87% 87% Controlling High Blood Pressure 55% 58% All-Cause Readmissions 16% 16% Ambulatory Care—Emergency Department (ED) Visits per 1,000 Member Months 46.08 43.74 Ambulatory Care—Outpatient Visits per 1,000 Member Months 268.12 264.31 Avoidance of Antibiotic Treatment in Adults with Acute Bronchitis 35% 36% Screening for Clinical Depression and Follow-Up Plan—Performance Rate 30% 30% Screening for Clinical Depression and Follow-Up Plan—Reporting Rate 4% 3% Use of Imaging Studies for Low Back Pain 73% 79% Source: Blue Sky Consulting Group analysis of Department of Health Care Services, Medical Managed Care Quality Improvement Reports: External Quality Review Technical Reports with Plan-Specific Evaluation Reports (July 1, 2015–June 30, 2016; July 1, 2016–June 30, 2017; and July 1, 2017–June 30, 2018), www.dhcs.ca.gov. California Health Care Foundation www.chcf.org 12 When examining differences in all aggregated HEDIS Quality of Care Within San Diego measures between GMC and comparison group and Sacramento counties from 2015 through 2018, the GMC model MCP performance on aggregated HEDIS scores consistently had lower performance, on average, in across several years varied within each county each year (Table  6). There are many factors that can (Table  7, page  14). Kaiser consistently has higher affect MCP performance, and while the authors have aggregate scores than the other MCPs. This should attempted to compare GMC only to like counties, fac- be interpreted with caution because, as explained tors other than MCP performance or managed care earlier, Kaiser’s Medi-Cal member mix is different model likely explain a significant portion (or all) of from that of other MCPs as they have several con- observed differences. Moreover, the authors did not trols over enrollee entrance into their MCP. San Diego test the statistical significance of these differences. county MCPs appear to have performed somewhat Nevertheless, these findings are consistent with better than Sacramento MCPs except on All-Cause those from a contemporary study showing that qual- Readmissions.23 ity scores are generally superior for MCPs in COHS and Two-Plan counties, on average, than for MCPs in Individual MCP performance remained relatively con- competing commercial models, which include GMC, sistent year-to-year in both counties without large Regional, and Imperial models.22 The same study swings in performance (Table 8, page 15). In San found that many of these differences are statistically Diego, Community Health Group Partners and Molina significant. Healthcare showed modest improvements between 2015 and 2018. Molina’s performance also improved some in Sacramento. Table 6. All HEDIS Measures Average, GMC and Comparison Counties, 2015–2018 2015 2016 2017 2018 GMC 69% 69% 68% 70% Urban 72% 72% 71% 72% Source: Blue Sky Consulting Group analysis of Department of Health Care Services, Medical Managed Care Quality Improvement Reports: External Quality Review Technical Reports with Plan-Specific Evaluation Reports (July 1, 2015–June 30, 2016; July 1, 2016–June 30, 2017; and July 1, 2017– June 30, 2018), www.dhcs.ca.gov. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 13 Table 7. Aggregated HEDIS Measures for All Years, San Diego and Sacramento MCPs, 2015–2018 CHILD AND ALL-MEASURES ADOLESCENT ACCESS CHRONIC DISEASE ALL-CAUSE AVERAGE TO PRIMARY CARE MANAGEMENT READMISSIONS San Diego County Care1st 68% 70% 69% 18% Community Health Group Partners 73% 79% 73% 16% Health Net, San Diego 68% 77% 69% 22% Kaiser SoCal 86% 86% 87% 15% Molina Healthcare, San Diego 72% 79% 72% 16% All plans 73% 78% 74% 17% All plans, excluding Kaiser 70% 76% 71% 18% Sacramento County Anthem Blue Cross, Sacramento 67% 74% 65% 16% Health Net, Sacramento 67% 72% 66% 16% Kaiser NorCal 82% 85% 82% 15% Molina Healthcare, Sacramento 68% 72% 68% 16% All plans 71% 76% 70% 16% All plans, excluding Kaiser 67% 73% 66% 16% GMC, Urban, and Statewide GMC 69% 75% 69% 17% Urban 72% 79% 71% 17% All plans statewide 71% 78% 70% 15% All plans, excluding Kaiser 70% 77% 69% 15% Source: Blue Sky Consulting Group analysis of Department of Health Care Services, Medical Managed Care Quality Improvement Reports: External Quality Review Technical Reports with Plan-Specific Evaluation Reports (July 1, 2015–June 30, 2016; July 1, 2016–June 30, 2017; and July 1, 2017–June 30, 2018), www.dhcs.ca.gov. California Health Care Foundation www.chcf.org 14 Table 8. All HEDIS Measures Average, San Diego and Providers Often Participate in Sacramento MCPs, 2015–2018 Multiple MCP Networks, Particularly 2015 2016 2017 2018 Primary Care Providers San Diego County One of the most important potential benefits of the Care1st 69% 68% 67% 68% GMC model relates to the size of available provider networks. To the extent that individual MCPs have Community Health 72% 72% 73% 75% unique provider networks, adding additional MCPs Group Partners would increase the available provider networks. Using Health Net, San Diego 68% 69% 67% 69% provider network files from DHCS for January 2017 Kaiser SoCal 89% 86% 84% 86% through November 2018, the authors assessed GMC MCPs’ networks and the extent of differentiation Molina Healthcare, 71% 71% 71% 73% San Diego therein by examining the percentage of providers that participate in multiple MCP networks. The authors also All MCPs 74% 73% 72% 74% assessed the number of providers participating in only All MCPs, 70% 70% 69% 71% one MCP. Finally, the authors examined whether avail- excluding Kaiser able provider networks expanded when new MCPs Sacramento County  joined GMC in 2017. Anthem Blue Cross, 67% 68% 67% 67% Sacramento Although the provider network files represent the best available data on the size of each MCP’s network, Health Net, 68% 67% 65% 67% Sacramento there are important caveats and limitations with these files. First, the provider files list all providers in the Kaiser NorCal 84% 83% 81% 82% MCP network, but they do not indicate whether these Molina Healthcare, 67% 67% 67% 70% providers actually provide services to or are accept- Sacramento ing new Medi-Cal patients. Second, these data are All MCPs 71% 71% 70% 72% not systematically audited or verified by DHCS and All MCPs, 67% 67% 66% 68% are known to contain inaccuracies. Nevertheless, they excluding Kaiser were the best data available for this analysis. GMC, Urban, and Statewide In order to prepare the files, data was limited to all pro- GMC 69% 69% 68% 70% viders with a Sacramento or San Diego address (some Urban 72% 72% 71% 72% MCPs listed providers from outside of the region). The authors examined only license types for physi- All MCPs statewide 71% 71% 70% 71% cians, physician assistants, and nurse practitioners All MCPs, 70% 70% 69% 71% and excluded behavioral health and other nonmedi- excluding Kaiser cal provider types. The files also designate physicians Source: Blue Sky Consulting Group analysis of Department of Health Care as primary care and/or specialists. Providers were Services, Medical Managed Care Quality Improvement Reports: External Quality Review Technical Reports with Plan-Specific Evaluation Reports matched across MCPs using the National Provider (July 1, 2015–June 30, 2016; July 1, 2016–June 30, 2017; and July 1, 2017– Identifier (NPI).24 For each MCP, the authors calculated June 30, 2018), www.dhcs.ca.gov. how many providers were unique to that MCP and how many providers were in that MCP and multiple other MCPs for the November 2018 period. For this analysis, the authors excluded Kaiser as their providers do not contract with other MCPs. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 15 While nearly 60% of providers listed in Sacramento In San Diego, there is much less provider exclusivity MCPs’ networks were exclusive to one MCP, there on networks, with only 30% of all providers listed on are notable differences between primary care provid- only one network (Table 10). Half of all primary care ers and specialists (Table 9). Primary care providers providers were listed on four MCPs’ networks, and were much more likely to be listed on multiple MCP more than one-fourth on five MCPs’ lists. Specialists networks; over half of primary care providers were showed a similar likelihood to be listed on multiple included in three MCPs’ network files. Furthermore, MCPs’ network filings with DHCS. Interestingly, one in one-quarter of primary care providers were included six specialists was included on all six MCPs’ provider on all four MCPs’ provider files. Specialists, on the listings, while more than one in four specialists was other hand, were less likely to be included in multiple exclusive to a single MCP. MCPs’ networks, with more than 60% being exclusive to one MCP’s listing. However, over one-third of spe- As shown in Tables 9 and 10, many providers par- cialists were included in two MCPs’ networks. ticipate in at least two MCP networks, although a significant fraction of providers (60% in Sacramento and 30% in San Diego) are exclusive to a single MCP. Table 9. Sacramento GMC MCPs’ Provider Network These data suggest, therefore, that while there is over- Listings (excluding Kaiser), November 2018 lap of providers across networks, there nevertheless NUMBER/PERCENTAGE OF MCPs can be some differentiation between MCPs. Again, UNIQUE PROVIDERS 1 2+ 3+ 4 it is important to reiterate the important caveat that while these files list the providers in an MCP’s network, All providers 3,233 1,945 1,288 692 300 they do not indicate the extent of a provider’s prac-     60.2% 39.8% 21.4% 9.3% tice time devoted to Medi-Cal patients and the MCP’s Primary care 458 134 324 242 110 adequacy to address access-to-care needs, either pri-   29.3% 70.7% 52.8% 24.0% mary care or specialty. Specialists 2,550 1,629 921 452 246     63.9% 36.1% 17.7% 9.6% Source: Blue Sky Consulting Group analysis of DHCS Provider Network Files provided April 11, 2019. Table 10. San Diego GMC MCPs’ Provider Network Listings (excluding Kaiser), November 2018 NUMBER/PERCENTAGE OF MCPs UNIQUE PROVIDERS 1 2+ 3+ 4+ 5+ 6 All providers 6,399 1,920 4,479 3,401 2,576 1,738 759     30.0% 70.0% 53.1% 40.3% 27.2% 11.9% Primary care 1,325 273 1,052 884 673 397 73     20.6% 79.4% 66.7% 50.8% 30.0% 5.5% Specialists 4,492 1,215 3,277 2,554 2,034 1,469 722     27.0% 73.0% 56.9% 45.3% 32.7% 16.1% Source: Blue Sky Consulting Group analysis of DHCS Provider Network Files provided April 11, 2019. California Health Care Foundation www.chcf.org 16 Provider Networks by MCP In San Diego, individual MCPs’ submitted lists also In Sacramento, network file submissions to DHCS revealed significant overlap of providers across MCP suggest that primary care and specialist provid- networks. Unlike Sacramento, there isn’t an MCP ers have contracts across multiple MCPs (Table 11). wherein a preponderance of providers are listed There is a notable exception, however: Anthem Blue exclusively with that MCP’s network. This supports Cross appears to offer a much larger network of spe- interviewees’ observations in San Diego that MCPs’ cialists who contract exclusively with their MCP. In networks were not very differentiated. Sacramento, network differentiation occurs largely with this one MCP, Anthem. Table 11. GMC MCP Provider Network Listings (excluding Kaiser), by MCP and County, November 2018   ALL PROVIDERS PRIMARY CARE SPECIALISTS Total Exclusive % Total Exclusive % Total Exclusive % Sacramento Aetna Better Health 950 88 9% 271 15 6% 623 54 9% Anthem Blue Cross 2,799 1,608 57% 384 71 18% 2,296 1,430 62% Health Net 1,178 174 15% 306 29 9% 765 88 12% Molina Healthcare 586 75 13% 173 19 11% 485 57 12% San Diego Aetna Better Health 3,312 349 11% 610 13 2% 2,624 297 11% Care1st 2,454 175 7% 490 1 0% 2,015 159 8% Community Health Group 4,226 542 13% 994 53 5% 2,800 199 7% Health Net 2,891 191 7% 577 2 0% 2,116 113 5% Molina Healthcare 3,008 274 9% 866 114 13% 2,375 161 7% United Healthcare 3,461 389 11% 867 90 10% 2,618 286 11% Source: Blue Sky Consulting Group analysis of DHCS Provider Network Files provided April 11, 2019. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 17 New MCPs Can Bring Network compared with specialists. The data indicate that Expansion these migrating providers largely ended up contract- The departure of United Healthcare from Sacramento ing with one new GMC MCP. GMC allowed for an investigation of how many pro- viders did not migrate to other MCPs when United left An analysis of the providers who were exclusive to the market (i.e., the extent to which United’s presence one of the two new GMC model entrants (United and in the market expanded the available provider net- Aetna) in San Diego indicates that the overwhelm- work). The addition of United and Aetna in San Diego ing majority of these providers were new to GMC provided an opportunity to investigate how many and were not drawn from other MCPs. In the case of providers were drawn from other MCPs as opposed United (Table 13), of its 363 exclusive providers, only to how many were added as a result of these MCPs 45 (or about 12%) had contracts with other Medi-Cal entering the market. MCPs prior to United’s entrance into the market. In the case of Aetna (Table 14, page 20), only about 18% (or As shown in Table 12, only about one in five providers 59) of its exclusive providers were previously part of who had been exclusive to United’s network migrated another MCP’s network. These findings suggest that to other GMC MCPs after United left the market in new MCP entrants can bring networks that increase November 2018 (i.e., nearly 80% of United’s provid- the number of new providers available to Medi-Cal ers did not remain available to Medi-Cal patients). enrollees. Relatively more primary care providers had migrated Table 12. Migration of United-Exclusive Providers to Other GMC MCPs, Sacramento, 2017–18 NUMBER/PERCENTAGE OF OTHER SACRAMENTO GMC MCPs NUMBER OF PROVIDERS IN WHICH PROVIDER WAS LISTED, NOVEMBER 2018 EXCLUSIVE TO UNITED, TOTAL NOVEMBER 2017 1 2 3 4 5 MIGRATION All providers 412 70 11 9 4 0 94     17.0% 2.7% 2.2% 1.0% 0.0% 22.8% Primary care 83 27 2 5 1 0 35     32.5% 2.4% 6.0% 1.2% 0.0% 42.2% Specialists 297 41 9 3 3 0 56     13.8% 3.0% 1.0% 1.0% 0.0% 18.9% Table 13. San Diego GMC Providers Exclusive to United Healthcare in November 2018 NUMBER/PERCENTAGE OF OTHER SAN DIEGO GMC MCPs NUMBER OF PROVIDERS IN WHICH PROVIDER WAS LISTED AS OF JANUARY 2017 EXCLUSIVE TO UNITED, TOTAL NOVEMBER 2018 1 2 3 4 5 MIGRATION All providers 363 29 11 4 1 0 45     8.0% 3.0% 1.1% 0.3% 0.0% 12.4% Primary care 88 8 6 3 0 0 17     9.1% 6.8% 3.4% 0.0% 0.0% 19.3% Specialists 261 20 5 1 1 0 27     7.7% 1.9% 0.4% 0.4% 0.0% 10.3% Source (Tables 12 and 13): Blue Sky Consulting Group analysis of DHCS Provider Network Files provided April 11, 2019. California Health Care Foundation www.chcf.org 18 Table 14. San Diego GMC Providers Exclusive to Aetna in November 2018 NUMBER/PERCENTAGE OF OTHER SAN DIEGO GMC MCPs NUMBER OF PROVIDERS IN WHICH PROVIDER WAS LISTED AS OF JANUARY 2017 EXCLUSIVE TO AETNA, TOTAL NOVEMBER 2018 1 2 3 4 5 MIGRATION All providers 322 37 10 9 3 0 59     11.5% 3.1% 2.8% 0.9% 0.0% 18.3% Primary care 8 1 0 2 0 0 3     12.5% 0.0% 25.0% 0.0% 0.0% 37.5% Specialists 271 28 7 7 3 0 45     10.3% 2.6% 2.6% 1.1% 0.0% 16.6% Source: Blue Sky Consulting Group analysis of DHCS Provider Network Files provided April 11, 2019. MCP Selection by Enrollees Is comparison Two-Plan counties. In addition, more than 11% were enrolled in a prior MCP. The Sacramento Similar to Two-Plan Counties default rate was therefore correspondingly lower, at New or returning Medi-Cal enrollees in counties with 21%, compared with 33% in comparison counties. multiple MCP options have three enrollment pathways into an MCP25: Figure 2. Enrollment in GMC and Comparison Counties, September 2016–September 2018 $$ Active selection. All enrollees may select an MCP by submitting an enrollment form (DHCS refers to this as “regular” enrollment). ■ Auto-Assigned ■ Prior Plan ■ Active Selection $$ Passive/prior MCP. Enrollees who do not select an MCP may be assigned to one based on prior 68% enrollment or because other family members are 61% 59% enrolled in that MCP. $$ Auto-assigned. Those not enrolled in an MCP using the other two pathways are assigned to one using an algorithm that employs eight 35% HEDIS and safety-net measures to enroll a higher 33% percentage of this group into MCPs with higher performance scores.26 21% The authors analyzed MCP selection data (Figure 2) 11% to compare enrollee pathways in GMC and Two-Plan 6% 6% counties.27 San Diego enrollment over two years was Sacramento San Diego Comparison Group quite similar to Two-Plan counties, with roughly 59% Source: Blue Sky Consulting Group analysis of DHCS auto-assignment data of enrollees making an active selection among their provided June 6, 2019. options, just under 6% enrolling into a prior MCP, and roughly 35% defaulted by the auto-assignment algo- rithm. In Sacramento, just under 68% made active MCP selections, which was higher than the rate in A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 19 These data suggest that the selection pathways for Specifically, nearly 2,800 enrollees (or 7.6% per enrollees in GMC counties are not much different year) changed MCPs each month, on average, in from those in comparison Two-Plan counties, although Sacramento over the 24-month period. In San Diego the rate of active selection is higher in Sacramento. over this same period, nearly 4,800 enrollees (or 7.9% The authors did not have data on the enrollees’ MCP per year) changed MCPs each month. In total for the selection experience, such as time taken or diffi- two GMC counties, over 7,500 enrollees (or 7.8% per culty making an MCP selection, so the authors can’t year), on average, moved between MCPs each month. comment on enrollees’ ease or challenges selecting among five or seven MCPs. While transactional costs are challenging to quantify, each MCP change entails such costs. For enrollees, the cost of changing MCPs can include the disrup- Enrollees in GMC Counties Change tion of care and time invested in learning a new MCP’s MCPs More Frequently network, services, pharmacy formularies, and other As noted earlier, one rationale for multiple MCPs is to features. For MCPs, these transactional costs can offer choice to enrollees who can exercise this right include enrollment and patient care data exchange monthly. While offering enrollees the opportunity to between MCPs, new patient intake assessments, and use specific providers or seek better services, these data exchange with primary care and other providers. MCP changes do have transactional costs. Providers also incur similar transactional costs when enrollees change MCPs, most notably, the possible The authors examined DHCS MCP change data from interruption or complete loss of a clinical relationship. September 2016 through September 2018, specifi- cally examining MCP changes within the respective These costs are known to DHCS, providers, and con- GMC county. The authors compared these with MCP sumer advocates. Policy discussions surface with some change rates in all other counties where enrollees can regularity about moving away from the monthly right change MCPs with the county.28 Perhaps not surpris- to change MCPs to an annual open enrollment, which ingly, enrollees in GMC counties are more likely to then locks members into an MCP for a year. MCPs and change MCPs than those in counties with only two providers generally prefer the annual open enrollment choices (Table 15). Across all counties with choice, and lock-in. Advocates, however, have long worked about 5% of enrollees changed MCPs annually. In to maintain the monthly change right for enrollees, GMC counties, nearly 8% changed MCPs over the allowing them to seek specific providers or express course of a year. other preferences. Table 15. Enrollee MCP Changes Within County, September 2016–September 2018 CHANGES AS A PERCENTAGE OF ENROLLMENT AVERAGE NUMBER OF MCP CHANGES PER MONTH Minimum Maximum Monthly Average Annualized All counties (with GMC) 34,699 0.34% 0.56% 0.41% 4.88% All counties (without GMC) 27,113 0.30% 0.53% 0.37% 4.42% Sacramento 2,794 0.54% 0.86% 0.63% 7.58% San Diego 4,792 0.50% 0.83% 0.66% 7.91% All GMC counties 7,586 0.52% 0.82% 0.65% 7.79% Source: Blue Sky Consulting Group analysis of DHCS ad hoc plan changes data provided March 18, 2019. California Health Care Foundation www.chcf.org 20 For other measures, however, GMC percentile scores GMC Enrollees’ Satisfaction with were higher when compared with urban percentile Their Care Is Mixed Compared with scores. GMC MCPs earned higher patient satisfaction Enrollees in Other Counties scores for ratings of personal doctor, getting needed Consumer Assessment of Healthcare Providers and care, getting care quickly, and customer service. Systems (CAHPS) is a patient satisfaction survey con- ducted every three years. The most recent survey was fielded in 2016 and published in January 2018.29 The Other Measures Also Paint a Mixed CAHPS survey is administered to patients in all Medi- Picture of GMC Performance Cal MCPs and covers patient satisfaction with both In addition to HEDIS, CAHPS, and CHIS data, the their MCP and providers. Results are summarized by authors also examined several measures from the MCP, allowing for a comparison across MCPs or for DHCS Managed Care Performance Dashboard. These aggregation of data across managed care models.30 In measures included mild-to-moderate mental health order to calculate the results in Table 16, the authors visits, medical exemption requests, and grievances took the average score across all MCPs in the respec- filed. The GMC model delivered more mild-to-moder- tive model counties and then calculated the percentile ate mental health visits per 1,000 members (Table 17) rank represented by that score when compared with than comparison model counties. all MCPs statewide.31 Table 17. Selected Managed Care Performance Table 16. Percentile Rank of Aggregated CAHPS Dashboard Measures, 2017–18 Measures, 2016 MEASURE GMC URBAN GMC URBAN All MCPs Rating of all health care 37% 44% Mild-to-moderate mental health visits 17.6 13.4 per 1,000 members Rating of personal doctor 53% 46% Medical exemption requests 4.7 0.7 Rating of specialist seen most often 46% 48% per 10,000 members Getting needed care 59% 46% Grievances per 1,000 member months 116.9 65.6 Getting care quickly 64% 46% Excluding Aetna and United How well doctors communicate 48% 51% Mild-to-moderate mental health visits 17.5 13.4 Customer service 66% 49% per 1,000 members Source: Blue Sky Consulting Group analysis of DHCS data from Medical exemption requests 1.0 0.7 Department of Health Care Services, Managed Care Quality and per 10,000 members Monitoring Division, 2016 CAHPS Medicaid Managed Care Survey Summary Report, January 2018, www.dhcs.ca.gov. Grievances per 1,000 member months 89.3 65.6 Source: Blue Sky Consulting Group analysis of DHCS Managed Care Performance Dashboard data provided March 12, 2019. These results reveal a mixed picture of patient satis- faction. According to these percentiles, GMC MCPs earned some scores below urban comparison MCPs when compared with all MCPs statewide. For exam- ple, having a 37th percentile score means that 63% of scores for other statewide MCPs were higher. These lower GMC percentile scores, compared with urban percentiles, were for ratings of all health care, specialist seen most often, and how well doctors communicate. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 21 GMC model counties had significantly more medical across networks. In San Diego, MCP networks were exemption requests per 10,000 members and griev- less differentiated than in Sacramento. The high ances per 1,000 member months than in comparison percentage of providers who were exclusive to one counties. This suggests that enrollees experienced of the two new GMC model entrants (United and more challenges with MCP services, accessing spe- Aetna) in San Diego indicates that the overwhelm- cific providers, and other issues for which they sought ing majority of these providers were new to GMC formal redress than enrollees in other urban coun- and not drawn from other MCPs. How much care ties with different models. Without Aetna and United these new providers provide to Medi-Cal enrollees in these data, the rates are lower for both measures could not be determined from the data. but still higher than in comparison counties, suggest- $$ MCP selection. New or returning enrollees to Medi- ing that adding MCPs may have contributed to some Cal in GMC counties have the same or higher active enrollees’ struggles to understand and navigate MCP MCP selection rates than comparison counties with networks and services. two plans from which to choose. In Sacramento, notably, active MCP selection is modestly higher than in comparison counties. Summary of Quantitative Findings The performance of MCPs participating in GMC coun- $$ MCP switching. Enrollees change MCPs within the ties was, on average, generally similar to or worse than GMC counties at a higher rate than in non-GMC the performance of MCPs participating in similarly comparison counties. While important for choice, urban COHS or Two-Plan model counties. Specifically: the multiplicity of MCPs does bring difficult-to- measure transactional costs to MCPs, providers, $$ Access to care. Medi-Cal enrollees in GMC coun- and enrollees themselves. ties were more likely to report difficulty accessing specialty care than enrollees in other urban coun- $$ Member satisfaction. These data from 2016 show ties; however, following a regression analysis, the a mixed picture in which GMC MCPs score higher managed care model type does not appear to be a on some measures but lower on others. While significant factor in explaining differences in access these satisfaction data do not paint a compelling to care across counties. picture one way or the other of enrollees’ reported satisfaction, GMC model counties had significantly $$ Quality of care. MCPs in the two GMC counties higher rates of medical exemption requests and performed worse across the majority of HEDIS grievances than in comparison counties, suggest- measures than other MCPs; this lower performance ing that enrollees experienced more challenges for was consistent across years from 2015 through which they sought formal redress. 2018. Quality performance was somewhat better, on average, among MCPs in San Diego when com- pared with Sacramento MCPs. $$ Choice of providers. It is impossible to draw much meaning from data on the adequacy of MCP net- works without data on what share of physician time is spent caring for that MCP’s Medi-Cal members. Nevertheless, the data show that providers con- tract with multiple MCPs, resulting in some provider overlap across networks. This is more common among primary care providers. Providers, particu- larly specialists, may be found on only one MCP’s network files, resulting in some differentiation California Health Care Foundation www.chcf.org 22 Findings: Views Were Mixed on the Benefits of Adding Two MCPs to GMC Stakeholder Interviews Counties in 2017 After a competitive procurement process, in 2015 Interviewees Were Generally Aetna and United Healthcare, both large commercial Supportive of the GMC Model MCPs and new to Medi-Cal, were added to San Diego There was general agreement among interviewees and Sacramento GMC in early 2017. Interviewees that MCP choice, in principle, can benefit both pro- observed that the MCPs brought the benefits of viders and enrollees by fostering competition among national experience and scale, showed innovative MCPs. Many consumer advocates, in particular, prefer approaches to service, and made investments to enter to give Medi-Cal enrollees multiple MCP options, so the GMC market. Their entrance also appears to have that they can switch MCPs to access a specific hospital brought new providers into Medi-Cal, although the or specialist or to receive better benefits (e.g., care numbers added were modest, and it is not clear how coordination, transportation). Moreover, attempts by meaningful their participation is. MCPs to differentiate themselves to enrollees through their provider networks, care models, member ser- Some provider interviewees noted, however, that vices, language accessibility, and community-based these additional MCPs haven’t added much and that services are viewed positively. Interviewees also noted the overall quality of care and access to services hasn’t that MCPs can bring unique expertise and economies yet improved. A Sacramento interviewee observed of scale and scope to services, particularly if the MCPs that although there are four major health systems have numerous lines of business (e.g., commercial, in the county, their specialists don’t meaningfully Medicare) across the state or nation. participate in GMC MCPs’ networks. To build suf- ficient enrollment and financial viability, these MCPs To build a network, MCPs may try to distinguish them- relied on default enrollments but also used market- selves competitively with providers through an overall ing to differentiate their MCPs to new enrollees and “value proposition.” This can include higher payment to draw enrollment from existing MCPs. Due to low rates, access to other lines of business (e.g., commercial enrollments, interviewees observed, the MCPs don’t or Medicare Advantage), organizational efficiencies, yet have the lives, scale, and risk mix to innovate. and better quality-improvement resources and tech- Indeed, a poor risk mix and loss of the UC Davis hos- nical assistance. This competition may give providers pital contract contributed to United’s departure from increased leverage over MCPs, especially when com- Sacramento in November 2018. pared with COHS counties, where some interviewees noted that providers have little leverage with the MCP. Providers agreed that MCPs can be more responsive Interviewees Perceive “Diminishing to them when they have to compete for contracts. Returns” from Adding More MCPs While interviewees agreed, in principle, that provid- DHCS officials also share the view that the GMC model ing more MCP options for enrollees can foster greater with multiple, competing MCPs has the potential to competition, many also noted that there is a point of achieve better outcomes. In addition, with numerous diminishing returns with additional MCPs, and that MCPs, the state has more leverage over poorly per- this threshold is difficult to discern. Adding more forming MCPs as the industry responds to purchasing MCPs to GMC counties increased administrative and power and the threat of lost business. Interviewees navigational complexity and, according to several suggested that the relatively poor performance of interviewees, “diluted and diffused” the benefits of existing MCPs in Sacramento motivated the decision differentiation and competition. This administrative of DHCS to procure additional MCPs in 2015. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 23 multiplicity also brings increased transactional costs, The GMC Model Also Creates which are discussed below. Administrative Challenges and In San Diego, interviewees struggled to describe how Transaction Costs for Providers adding additional MCPs improved enrollees’ and pro- For providers, contracting with multiple MCPs can viders’ experiences. Several interviewees observed bring significant redundancy and transaction costs that having two additional MCPs hadn’t made much due to MCPs’ idiosyncratic and duplicative contracting difference in the delivery or quality of services. In and administrative requirements. There are multiple Sacramento, interviewees also agreed that the GMC business and administrative requirements that often model was conceptually beneficial but operationally differ by MCP for service pre-authorizations, billing problematic. and claims adjudication, information technology, and quality monitoring and improvement approaches. The GMC Model Introduces Both providers and enrollees also have to understand Navigational Complexities and navigate multiple networks, benefit plans, and for Enrollees care processes, which may differ by population (e.g., dual eligibles). This navigational maze includes differ- Numerous interviewees observed that the multiplicity ences in specialist networks, pharmacies and urgent of MCPs results in a fragmented system of networks, care options, prescription drug formularies, care coor- providers, and services. They added that this complex dination and management approaches, and member “patchwork” of choices contributes to enrollees’ con- services and portals. One interviewee noted that care fusion and difficulties selecting MCPs in the first place, coordinators in their clinic created “road map” algo- and then navigating an MCP’s providers and services rithms for navigating each MCP’s service approval thereafter. Interviewees noted that GMC is a challeng- requirements. ing system for any consumers to navigate, let alone for refugees, the homeless, those with limited English proficiency, and those with serious mental illness or Multiple MCPs Result in Diffuse substance use disorders. A Sacramento interviewee opined that the GMC model is “a ragtag system that Initiatives Meant to Incentivize we’ve patched together and made work for patients.” Providers and Benefit Patients This organizational patchwork also has the effect of These navigational complexities and resulting confu- diffusing the impact of specific initiatives by MCPs, sion can lead to enrollees changing MCPs frequently. particularly quality improvement and pay-for-perfor- One interviewee shared that one MCP’s network and mance incentives. For these to work, interviewees medical groups were particularly confusing and hard to observed, MCPs and providers must make significant understand, leading enrollees to erroneously assume investments over time with a stable population. The that providers were in other MCPs’ networks and then ability of enrollees to change MCPs monthly makes for to change MCPs. In another example, when a provider an often-evolving patient panel. leaves one MCP’s network and joins another, it may take months before that provider appears available The promise and impact of pay-for-performance and on the new network, leaving enrollees seeking that other quality improvement programs can be diluted provider in limbo. The frequency of MCP changes by with a multiplicity of priority HEDIS and other mea- enrollees, and a discussion of potential costs, was pre- sures, different incentive payments, and varying expert sented in a previous section. and information technology support for clinicians. An interviewee noted that MCPs’ selection of performance priorities feels arbitrary and not at all coordinated with other MCPs. In the end, the interviewee noted, these California Health Care Foundation www.chcf.org 24 initiatives all become “a jumble to understand and for enrollees.34 In addition to the committee itself, there effectuate,” and providers lose interest and commit- is an additional work group focused on care coordina- ment. Population health interventions also become tion. Compared with San Diego, interviewees painted diluted and poorly coordinated; another interviewee a different picture of the role played by this advisory opined that “one plan’s QI or population health initia- body. Interviewees observed that the committee tive won’t get my attention.” largely served as a communication venue for MCPs, providers, and other stakeholders and that some col- laboration occurred here. One interviewee noted that Collaboration Differs in San Diego the committee often served as a venue for providers and Sacramento Counties and advocates to express their frustration to MCPs. Both counties have statutorily created GMC advisory Working relationships between MCPs and providers groups convened and led by the county health depart- sometimes developed outside of the committee. The ment, though their evolution and oversight roles differ committee has one dedicated staff person and some considerably. Healthy San Diego (HSD) was created additional staff support from the Sacramento County in 1995 to inform the county’s implementation of, Department of Health Services. and contracting for, the GMC model.32 Interviewees labeled HSD as having a “quasi-governance” role Interviewees in both counties remarked that neither built on a history of MCP communication and collabo- committee had “real oversight or enforcement author- ration to standardize and streamline contracting and ity” around MCP performance. This oversight and other processes. compliance authority ultimately fell to DHCS, and the counties’ roles were largely to convene stakehold- Healthy San Diego is composed of the Joint Consumer ers and advise MCPs. While San Diego stakeholders and Professional Advisory Committee, two sub- lauded HSD collaboration, they also admitted that committees, and two advisory groups. Interviewees MCPs were frequently reluctant to share specific observed that MCPs “collaborate and compete,” details about operations or innovations. contributing to a Medi-Cal managed care market that generally serves enrollees well. Several interviewees did observe that tensions can arise when county and Clinic Consortia and IPAs Offer MCP priorities or implementation approaches differ. Some Efficiencies The county deputy chief medical officer and an admin- Numerous interviewees described how independent istrative secretary staff HSD. MCPs make an annual physician associations (IPAs) and clinic consortia pro- $1,500 contribution to HSD, and funds are used to vide some efficiencies in the GMC model through host trainings and other such events. economies of scale and scope and the standardization of administrative procedures for providers, and may When three new MCPs bid to win contracts in 2015 reduce some of the transaction costs. Their role, how- in San Diego, HSD members interviewed them and ever, differs in the two GMC counties. made a recommendation to the county to approve specific MCPs and then communicate this to DHCS for In San Diego, Health Center Partners of Southern their consideration. This process, whereby the county California is a nonprofit clinic consortium that rep- approves those MCPs with which DHCS can contract, resents and supports 10 community-based health is unique to San Diego and included in their original center organizations as a clinically integrated network statute.33 through its subsidiary Integrated Health Partners.35 IHP brings negotiating leverage for clinics and man- The Sacramento Medi-Cal Managed Care Advisory ages risk, contracting, billing, and other fundamental Committee, created in 2010 by SB208 (Steinberg), is issues, thereby insulating Federally Qualified Health charged with improving services and health outcomes Centers (FQHCs) from the administrative complexity A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 25 of dealing with multiple MCPs. The negotiating lever- age of IHP with MCPs is considerable as it serves as Discussion the largest clinic network in the area with both primary This assessment reveals a mixed picture of the perfor- care and specialists. IHP has also prepared these clin- mance of the GMC model and its ability to realize the ics to take on value-based payments and to participate potential benefits of greater competition and more in alternative payment methodologies. options for enrollees. The data analysis and inter- views conducted for this study show that, despite the Sacramento, however, does not have a clinic consor- theoretical advantages of competition among mul- tium organized around, and in support of, FQHCs and tiple MCPs, Medi-Cal enrollees do not receive better other clinics. Interviewees noted that Sacramento’s access or quality of care from the GMC model than FQHC community is less well developed and less if they were served by a model with only one or two collaborative than San Diego’s. FQHCs only prolifer- MCPs. To the contrary, across most measures, quality ated in the Sacramento market over the last decade. of care appears to be worse, on average, for MCPs Instead, non-Kaiser MCPs in Sacramento delegate operating in GMC counties than for MCPs operating 97% of lives to four commercial IPAs: River City, Hill in urban COHS and Two-Plan counties. Furthermore, Physicians, Imperial Health Holdings, and Nivano there was little to no improvement in quality scores for Physicians.36 River City IPA plays, according to some most MCPs in GMC counties from 2015 through 2018. interviewees, an outsized role in Sacramento, where it These findings are consistent with several other stud- serves 72% of all delegated lives. Interviewees noted ies, noted earlier, about multiple competing Medicaid that IPAs in Sacramento, particularly River City, played managed care plans. a role in standardizing some administrative and con- tracting processes rather than having to manage these The large numbers of Medi-Cal MCPs participating in with all of the MCPs individually. Several interview- Sacramento and San Diego Counties have also created ees noted that the large delegated role to IPAs has a confusing patchwork of networks, providers, and ser- made accountability for access to care, quality, and vices for enrollees who can, as a result, struggle to find patient satisfaction a challenge and has contributed to and use services, particularly specialists. For providers, enrollee confusion. this multiplicity of MCPs presents many redundancies in contracting administrative requirements and clinical approaches to care and quality improvement, which bring significant transaction costs. This assessment of the GMC model isn’t singular; there are differences in how the model operates in San Diego and Sacramento. Stakeholders in San Diego express more support for the model than those in Sacramento, and provider adaption to GMC in Sacramento is less mature than in San Diego. In both counties, however, stakeholders agreed that improve- ments to the model were warranted. In Sacramento particularly, the county health director, stakeholders, and State Senator Richard Pan have held convenings to address concerns about GMC and potential solu- tions. These concerns, laid out in a recent issues and options paper, include poor access to primary and specialty care, fragmentation and navigational chal- lenges, uneven quality, and the lack of local control.37 California Health Care Foundation www.chcf.org 26 Considerations for $$ Require MCPs to demonstrate capacity and experience with all Medi-Cal enrollees, includ- Improvement ing persons dually eligible for Medi-Cal and Medicare and persons with serious mental illness This assessment of California’s experience with man- or substance use disorders. aged care in GMC counties calls into question the rationale that multiple MCPs competing for Medi- $$ Require MCPs to work collaboratively with other Cal enrollees drives quality and patient satisfaction Medi-Cal MCPs to coordinate administrative higher among competitors.38 Short of changing the and other requirements to ease navigational and model type to a Two-Plan or COHS model, numer- transactional challenges for both providers and ous approaches should be considered to improve the enrollees. GMC model’s performance.39 $$ Adopt positive financial incentives tied to MCP DHCS should clarify its purchasing goals and objec- performance. Several other states have adopted tives and strengthen quality monitoring of MCP such programs to foster performance improve- performance in GMC counties. Other recent papers ments, whereas DHCS relies primarily on penalties have proposed approaches DHCS could embrace to for poor performance.41 strengthen its purchasing power and move toward value-based purchasing.40 These approaches, among $$ Foster greater collaboration among county and others, should be considered to improve access, qual- state stakeholders. DHCS could commit resources ity, and consumer and provider experience in GMC to working collaboratively with county officials, counties. Specifically, DHCS should undertake the consumer advocates, and MCP and provider rep- following: resentatives in San Diego and Sacramento to establish and advance improvement priorities and $$ Demand more from participating MCPs. DHCS goals. Moreover, DHCS or the legislature could could use the forthcoming procurement to raise give Healthy San Diego and the Sacramento Medi- expectations and demand more from MCPs bid- Cal Managed Care Advisory Committee, both of ding on a contract. For example, DHCS could: which are legislatively mandated, a more direct role $$ Strengthen oversight of network adequacy by in establishing procurement priorities, reviewing calculating accurate physician-to-patient ratios MCP bids, and overseeing MCP performance. for each MCP. Require MCPs to report, for each in-network physician, the percentage of their By raising the bar on performance and other expecta- practice spent seeing that MCP’s Medi-Cal tions, these actions might naturally reduce the number enrollees. of MCPs interested in pursuing a contract with DHCS to serve Medi-Cal enrollees in Sacramento and San $$ Set measurable quality improvement targets Diego Counties. If not, DHCS, working collaboratively and require MCPs to describe their approaches with San Diego and Sacramento, should consider for making significant and sustainable improve- reducing the number of MCPs with which it contracts ments in quality consistent with these targets. in these GMC counties. Although there was no con- $$ Require MCPs to make investments to expand sensus among interviewees on what the ideal number access to care, strengthen the local delivery sys- should be, the authors also did not find evidence that tems, and address enrollees’ social determinants more MCP competition leads to improvements in any of health. of the outcomes studied. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 27 By limiting the number of MCPs with which it contracts are administrative and other costs lower due to in a given county, DHCS could use its leverage to cre- economies of scale and scope and to competition? ate greater competition for those contracts. MCPs that Finally, what are the administrative costs and bur- receive a contract might be more willing to make the dens to DHCS of having multiple MCP contracts in investments DHCS and San Diego and Sacramento GMC counties? Can economies of scale and scope County officials want than they would if they had for DHCS be quantified? to divide up the market among four to six competi- $$ Network adequacy. The state’s analysis on the tors. Moreover, enrollees’ health care providers might adequacy of MCP networks does not reflect provid- benefit from greater efficiency and improved naviga- ers’ time serving Medi-Cal enrollees. For example, tional ease associated with working with fewer MCPs. one provider listed on an MCP’s network may see Were DHCS and the counties to take this approach, only a handful of patients annually, whereas other they should also develop plans to minimize potential providers on the network list may see a high vol- disruptions to enrollees and providers during the tran- ume of Medi-Cal patients every month. To better sition period. understand MCP network contributions to Medi- Cal, DHCS could require MCPs to track and report The upcoming DHCS procurement of commercial providers’ Medi-Cal encounters on a routine basis. MCP contracts in GMC and other model counties is an infrequent and important opportunity to catalyze $$ MCP switching. This study finds that enrollees significant improvements in quality, satisfaction, and move between and among MCPs in GMC coun- health outcomes for Medi-Cal enrollees. California’s ties; however, the reasons for this are less clear. experience with the GMC model provides use- Interviewees described many navigational chal- ful lessons that should be applied to the upcoming lenges for enrollees. It would be valuable to better procurement process and ongoing performance understand enrollees’ experiences with the GMC expectations for Medi-Cal MCPs. model and its benefits and drawbacks from their perspective. $$ Provider satisfaction. This study did not assess Further Research providers’ satisfaction with individual MCPs or the In addition to these actions, several avenues for future GMC model generally; however, providers inter- research were identified, mostly related to the costs viewed did express frustration with the multiplicity associated with the GMC model. They include the of MCPs and confusion about the roles of IPAs. following: MCPs and IPAs could survey providers though an $$ Leverage with providers. Interviewees noted that independent survey to reveal providers’ satisfaction providers have leverage when there are multiple and dissatisfaction. MCPs, creating the possible drawback of higher costs. Does this leverage result in higher provider prices and rates, on average, when compared with other model counties where MCPs have more leverage? If so, how does this impact the total costs of care? $$ Administrative costs. How large is the additional administrative burden on health care providers of contracting with multiple MCPs, each with its own policies and procedures related to utilization review, claims payment, and other processes, and how does this impact the total cost of care? Conversely, California Health Care Foundation www.chcf.org 28 Appendix A. Structured Interview Participants Dimitrios Alexiou, President/CEO, Hospital Jane Ogle, Consultant and Former Deputy Director Association of San Diego & Imperial Counties for Healthcare Delivery Systems at the Department of Health Care Services Sean Atha, Senior Vice President, Business & Network Development, River City Medical Group Jonathan Porteus, Chief Executive Officer, Wellspace Health Sacramento William Barcelona, Senior Vice President of Government Affairs, America’s Physician Groups Kiran Savage, Deputy Director, California Pan-Ethnic Health Network Athena Chapman, Chapman Consulting George Scolari, Chair, Healthy San Diego Behavioral Abbi Coursole, Senior Attorney, National Health Health Work Group, Community Health Group Law Program Margaret Tatar, Managing Principal, Health Jack Daily, HCA Coordinator, Consumer Center for Management Associates Health Education and Advocacy, Legal Aid Society of San Diego, Inc. Liza Thantranon, Regional Counsel and Managing Attorney, Legal Services of Northern California, Sandy Damiano, PhD, Deputy Director, Department Sacramento of Health Services, Sacramento County Abbie Totten, Medi-Cal Program Officer, Health Net Sarah De Guia, Executive Director, California Pan- Ethnic Health Network Jennifer Tuteur, MD, FAAFP, Deputy Chief Medical Officer, Health and Human Services Agency, Paul Durr, Senior Vice President and Chief Executive County of San Diego Officer, Sharp Community Medical Group Henry Tuttle, President and Chief Executive Officer, Jeff Dziedzic, Chief Operating Officer, Aetna Better Health Center Partners of Southern California Health of California Chet Uma, Chief Executive Officer, Aetna Better Jeff Gering, Vice President of Support Services & Health of California Planning, Family Health Centers of San Diego Christy Ward, Chief Executive Officer, One Britta Guerrero, Chief Executive Officer, Sacramento Community Health, Sacramento Native American Health Center Bobbie Wunsch, Founder and Partner, Pacific Health Brian Jensen, Regional Vice President, Hospital Consulting Group Council of Northern and Central California Judith Yates, Senior Vice President, Hospital Kevin Kandalaft, Chief Executive Officer, Association of San Diego & Imperial Counties UnitedHealthcare Community Plan of California Nicholas Yphantides, MD, MPH, Chief Medical Greg Knoll, Healthy San Diego Board Chair, CEO/ Officer, Health and Human Services Agency, Executive Director/Chief Counsel, Legal Aid County of San Diego Society of San Diego, Inc. Kim Lewis, Managing Attorney, National Health Law Program Meaghan McCamman, California Primary Care Association Branch McNeal, Senior Partner, Mercer A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 29 Appendix B. Regression Analysis Methodology and Results The authors conducted regression analyses using necessary to use a specialized form of regression called patient-level data from the California Health Interview a logistic (or logit) regression, where the dependent Survey (CHIS). The CHIS survey is a random-dial variable is categorical rather than continuous. Using telephone survey conducted by the UCLA Center these responses as dependent variables, the authors for Health Policy Research in collaboration with the constructed logistic models that included a dummy California Department of Public Health and the variable to indicate whether the member belonged Department of Health Care Services, and includes over to an MCP in a GMC county (based on respondent’s 20,000 Californians each year across all 58 counties. county of residence). The authors also included a The survey includes adults, teens, and children, and variety of other explanatory variables, including it collects detailed demographic information from the demographic variables such as the member’s age, respondents, such as age, gender, and level of educa- gender, race, income, and level of educational attain- tional attainment. The survey also asks questions on a ment, in addition to variables to capture whether the variety of health-related topics, such as health insur- member was married or had a partner, was a native ance coverage and access to health-related services. English speaker or had a high level of English profi- The data used in regressions included annual survey ciency, worked full-time, was clinically obese, or was responses for the years 2014 through 2017. a smoker. The authors also included health variables such as whether the member had diabetes, asthma, The authors tested numerous models to compare high blood pressure, heart disease, or psychological members of MCPs in GMC counties against mem- distress in the past year or needed help for emotional bers of MCPs in the comparison group counties. or mental issues or alcohol or drug problems. Finally, Specifically, the authors tested whether these MCP the authors also included dummy variables for the members differed with regard to their responses for year of the survey. the following survey questions: The authors tested numerous specifications using $$ Member had used the ER in the past 12 months various combinations of these explanatory variables, for any reason and in all these analyses the authors found no sta- $$ Member had a usual place to go to receive tistically significant difference in outcomes based on health care when feeling sick or needing health the respondent’s Medi-Cal managed care model (i.e., advice GMC Model versus other urban counties with COHS or Two-Plan managed care delivery models). An exam- $$ Member had a preventive care visit in the past ple of one model the authors tested is presented in 12 months Table B1 (see page 31). $$ Member had difficulty getting a doctor’s appoint- ment within two days (if needed) Table B1 presents numerous statistics from the logistic regression. The coefficient estimate is calculated using $$ Member had difficulty finding a primary care maximum likelihood estimation, or MLE. The odds provider ratio is the exponential of the coefficient estimate and $$ Member had difficulty finding a specialty care can be used to compare the relative importance of provider (if needed) the explanatory variables. The percentage increase in odds is the transformation of the logit coefficient $$ Member had difficulty understanding his or her using the formula 100(eb – 1), where b is the logit doctor coefficient, and expresses the result as a percent- age. Therefore, if this value is X, one may say “each Note that these responses are all binary, or yes/no additional unit of the explanatory variable results in an answers to the survey question. Because of this, it was California Health Care Foundation www.chcf.org 30 increase of about X% in the odds of the dependent responses from Medi-Cal members in GMC coun- event occurring.” Finally, the “Wald Prob > Chi-Sq” ties or in similar urban counties, and 1,290 (16.6%) of value represents 1 minus the confidence level at which those respondents said they did not have such a place the hypothesis that the coefficient value equals zero to go. Of the explanatory variables tested, the only cannot be rejected. Thus, a value of 0.05 indicates that significant explanatory variables were age, gender, the coefficient estimate is statistically significant at the English proficiency, whether the respondent ever had 95% confidence level. high blood pressure, and whether the respondent had an emotional or drug problem. The variable denoting In this model, the dependent variable was assigned a whether the respondent was a member of an MCP in 1 if the member’s survey response indicated he or she a GMC county (“GMC Plan Member”), however, was had no place to go to receive health care when sick not statistically significant. or in need of health advice. The CHIS data had 7,785 Table B1. Regression Results $$ Dependent variable: 1 if member had no usual place to go to receive healthcare $$ Number of observations: 7,785 $$ Pseudo R-square: 0.06090 $$ Number of observations where dependent variable is 1: 1,290 $$ Max rescaled R-square: 0.09663 PERCENT CHANGE COEFFICENT ODDS RATIO IN ODDS WALD PROB>CHI SQ Intercept (0.3213) 0.2842 GMC Plan Member (0.0634) 0.9386 (6.1388) 0.7230 Year = 2015 (0.1067) 0.8988 (10.1187) 0.5737 Year = 2016 (0.0719) 0.9307 (6.9346) 0.7191 Year = 2017 0.0241 1.0244 2.4379 0.9211 Age (0.0133) 0.9868 (1.3204) 0.0167† Gender = Male 0.6799 1.9736 97.3619 0.0000 † Race = White (0.0828) 0.9205 (7.9459) 0.6231 Diabetes (0.3832) 0.6817 (31.8346) 0.0731* Asthma (0.0811) 0.9221 (7.7930) 0.6514 High Blood Pressure (0.6894) 0.5019 (49.8099) 0.0000 † Emotional or Drug Problem (0.1804) 0.8349 (16.5074) 0.2678 Married or Has Partner (0.2008) 0.8181 (18.1929) 0.1813 English Speaker (Well/Very Well) (0.8623) 0.4222 (57.7803) 0.0000 † Education of BA or higher 0.1225 1.1303 13.0307 0.4783 Works full time 0.1606 1.1742 17.4236 0.2825 *Indicates signficance at the 90% level. † Indicates statistical significance at the 95% level. Source: Blue Sky Consulting Group Analysis of California Health Interview Survey data, 2019. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 31 Appendix C. CAHPS Measures Comparison: GMC and Urban Counties For patient satisfaction survey responses, a three-point This three-point global rating mean was the sum of mean score is calculated for each CAHPS measure the response scores (1, 2, or 3) divided by the total (Table C1).42 For the global ratings, scores are deter- number of responses to the global rating question. mined in this manner: For composite measures, response values were scored $$ Response values of 9 and 10 were given a score as follows: of 3. $$ Responses of “Always” were given a score of 3. $$ Response values of 7 and 8 were given a score of 2. $$ Responses of “Usually” were given a score of 2. $$ Response values of 0 through 6 were given a $$ All other responses were given a score of 1. score of 1. The three-point composite mean was the average of the mean score of each question included in the composite. Table C1. Aggregated CAHPS Scores, GMC and Urban Counties, 2016 GMC URBAN Adults Children Adults Children Rating of all health care 2.3 2.5 2.3 2.5 Rating of personal doctor 2.4 2.6 2.4 2.6 Rating of specialist seen most often 2.5 2.6 2.5 2.6 Getting needed care 2.1 2.1 2.1 2.3 Getting care quickly 2.2 2.3 2.1 2.4 How well doctors communicate 2.6 2.5 2.5 2.6 Customer service 2.4 2.5 2.4 2.4 Source: Blue Sky Consulting Group analysis of DHCS data from Department of Health Care Services, Managed Care Quality and Monitoring Division, 2016 CAHPS Medicaid Managed Care Survey Summary Report, January 2018, www.dhcs.ca.gov. California Health Care Foundation www.chcf.org 32 Endnotes 1.Department of Health Care Services, Medi-Cal Managed 13.Andrew Bindman, “Redesigning Medicaid Managed Care Request for Proposal (RFP) Schedule by Model Type, Care,” JAMA 319, no. 15 (April 17, 2018): 1537–1538, updated March 11, 2019, www.dhcs.ca.gov (PDF). jamanetwork.com. 2.For a discussion of the policy and feasibility issues related 14.Hangsheng Liu and Charles E. Phelps, “Nonprice to changing the managed care model type to a Two-Plan Competition and Quality of Care in Managed Care: The or County Organized Health System, please see the New York SCHIP Market,” Health Services Research 43, no. 3 appendix in the authors’ companion paper, A Close Look (June 2008): 971–987, www.ncbi.nlm.nih.gov. at Medi-Cal Managed Care: Quality, Access, and the 15.Additional details of the data analysis approach and Provider’s Experience Under the Regional Model, available regression results are presented in Appendix B. at www.chcf.org. 16.Importantly, these are not the access standards to 3.Bailit Health Purchasing, Paying Medi-Cal Managed Care which plans are contractually held by DHCS. For the time Plans for Value: Quality Goals for a Financial Incentive and distance access standards by provider type, refer to Program, California Health Care Foundation, April 26, 2019, Department of Health Care Services, Medicaid Managed www.chcf.org. Care Final Rule: Network Adequacy Standards, July 19, 2017, 4.Department of Health Care Services, Medi-Cal Monthly www.dhcs.ca.gov (PDF). Enrollment Fast Facts, March 2019, www.dhcs.ca.gov (PDF). 17.Centers for Medicare & Medicaid Services, Healthcare 5.Department of Health Care Services, Medi-Cal Managed Effectiveness Data and Information Set (HEDIS), modified Care Request for Proposal (RFP) Schedule by Model Type, July 6, 2017, www.cms.gov. updated March 11, 2019, www.dhcs.ca.gov (PDF). 18.Department of Health Care Services, Managed Care 6.Department of Health Care Services, Medi-Cal Managed Quality and Monitoring Division, Medi-Cal Managed Care Care Geographic Managed Care Expansion: Request for External Quality Review Technical Report (July 1, 2017– Application: Sacramento/San Diego 2015, August 31, 2015. June 30, 2018), April 2019, www.dhcs.ca.gov (PDF). 7.For example, a retrospective study looking at prices paid 19.For measures where a lower score is better by private PPOs for office visits between 2003 and 2010 (e.g., hospital readmissions and HbA1c Poor Control (>9.0%), in 50 states showed lower costs where there was more the score was rescaled to make it comparable with the other competition: Lawrence C. Baker, M. Kate Bundorf, Anne B. measures by subtracting the reported value from 1. Royalty, et al., “Physician Practice Competition and Prices 20.Aetna became operational in January 2018 and United Paid by Private Insurers for Office Visits,” JAMA 312, no. 16 Healthcare in October 2017. The external quality review (October 22/29, 2014): 1653–1662, jamanetwork.com. organization (EQRO) review requires that members be 8.Patrick A. Rivers and Saundra H.Glover, “Health Care enrolled continuously for 11 of 12 months to collect HEDIS Competition, Strategic Mission, and Patient Satisfaction: measures during the Measurement Year 2017. Research Model and Propositions,” Journal of Health 21.The authors excluded Kaiser from the analysis as it limits Organization and Management 22, no. 6 (2008): 627–641, enrollment in both GMC counties and is not an option for www.ncbi.nlm.nih.gov. enrollees to select unless they meet certain conditions. As 9.Alain C. Enthoven and Lawrence C. Baker, “With Roots in a result, Kaiser’s risk mix is generally different from that of California, Managed Competition Still Aims to Reform other plans. Health Care,” Health Affairs 37, no. 9 (September 2018), 22.Andrew Bindman, Denis Hulett, and Taewoon Kang, www.healthaffairs.org. A Close Look at Medi-Cal Managed Care: Statewide 10.Dennis P. Scanlon, Shailender Swaminathan, Woolton Quality Trends from the Last Decade, California Health Care Lee, and Michael Chernew, “Does Competition Improve Foundation, September 25, 2019, www.chcf.org. Health Care Quality?” Health Services Research 43, no. 6 23.Care1st became Blue Shield Promise starting in 2019. (December 2008): 1931–1951, www.ncbi.nlm.nih.gov. Because the data used for these analyses did not include 11.Liran Einav and Jonathan Levin, “Managed Competition 2019, the authors use the plan name Care1st. in Health Insurance,” Journal of the European Economic 24.Note that for some periods more than 10% of provider Association 13, no. 6 (December 2015): 998–1021, records were missing the NPI. web.stanford.edu. 25.Department of Health Care Services, Managed Care 12.Christopher Millett, Arpita Chattopadhyay, and Andrew Performance Monitoring Dashboard Report: Glossary, B. Bindman, “Unhealthy Competition: Consequences of March 2019, www.dhcs.ca.gov (PDF). Health Plan Choice in California Medicaid,” American Journal of Public Health 100, no. 11 (November 2010): 2235–2240, 26.Department of Health Care Services, Auto Assignment www.ncbi.nlm.nih.gov. Incentive Program, accessed July 9, 2019, www.dhcs.ca.gov. A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under Geographic Managed Care 33 27.The authors excluded COHS counties in the comparison 40.Bailit Health Purchasing, How California Can Use as enrollees do not have a choice of plans. Purchasing Power and Oversight to Improve Quality in Medi-Cal Managed Care: Raising the Bar, California Health 28.These results are likely underestimated as the authors Care Foundation, April 26, 2019, www.chcf.org. assumed one person changing plans (it could have been as many as five) where data were suppressed due to small 41.Bailit Health Purchasing, Paying Medi-Cal Managed reporting cell sizes. Care Plans for Value: Quality Goals for a Financial Incentive Program, California Health Care Foundation, April 26, 2019, 29.Department of Health Care Services, Managed Care www.chcf.org. Quality and Monitoring Division, 2016 CAHPS Medicaid Managed Care Survey Summary Report, January 2018, 42.Department of Health Care Services, Managed Care www.dhcs.ca.gov (PDF). Quality and Monitoring Division, 2016 CAHPS Medicaid Managed Care Survey Summary Report, January 2018, 30.More CAHPS measure detail is available in Appendix C. www.dhcs.ca.gov (PDF). 31.GMC county CAHPS scores do not include Kaiser for reasons described earlier. Furthermore, these scores do not include Aetna and United as they were not under contract with DHCS in 2016. 32.County of San Diego Health and Human Services Agency, Healthy San Diego: A Brief History, June 28, 2018, www.sandiegocounty.gov (PDF). 33.California Legislative Information, Welfare and Institutions Code, Article 2.91. Geographic Managed Care Pilot Project, Section 14089.05(h)(3), effective January 1, 2013, leginfo.legislature.ca.gov. 34.Sacramento County Department of Health Services, Medi-Cal Managed Care Stakeholder Advisory Committee Charter, November 1, 2018, www.dhs.saccounty.net. 35.Laura Hogan and Bobbie Wunsch, Forces for Change: A Landscape of the Statewide and Regional Clinic Consortia in California, Pacific Health Consulting Group, October 2018, pachealth.org (PDF). 36.Sacramento County Department of Health Services, Medi-Cal Managed Care: IPA Enrollment by GMC Plan, March 25, 2019, www.dhs.saccounty.net (PDF). 37.Peter Beilenson, Improving the Medi-Cal Managed Care System of Sacramento County, Sacramento County Department of Health Services, June 2019, www.dhs.saccounty.net (PDF). 38.Sacramento County Board of Supervisors, Sacramento County GMC Update from the Department of Health Care Services, June 19, 2018. 39.For a discussion of the policy and feasibility issues related to changing the managed care model type to a Two-Plan or County Organized Health System, please see the appendix in the authors’ companion paper, A Close Look at Medi-Cal Managed Care: Quality, Access, and the Provider’s Experience Under the Regional Model, available at www.chcf.org. California Health Care Foundation www.chcf.org 34