Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness JULY 2021 AUTHORS Erika Siao and Julie Silas, JD Contents About the Authors 3Introduction Erika Siao, is a research associate, and Julie 3Methodology Silas, JD, is directing attorney at Homebase, a nonprofit organization of legal, policy, and 4 Why Homelessness and Health Care Providers subject matter experts who work at the com- Share Data munity, state, and national level to build capacity and develop and implement effec- 6 What Was Learned: Emerging Stories tive programs and systems to prevent and California Leads the Way end homelessness.* California Communities in Action Acknowledgments 11 Common Challenges to Cross-Sector Homebase and the California Health Care Data Sharing Foundation have partnered on this report Privacy to help homeless response and health care Relationships and Collaboration system providers undertaking cross-sector Interoperability data sharing to more effectively serve their Data Quality clients. The authors would like to thank staff from the many agencies and organizations 23 Policy Opportunities: New Efforts in Place or interviewed who shared their experiences on the Horizon (see Appendix C for the full list). The authors Opportunities for Policy would also like to thank Advisory Group mem- bers, who contributed guidance, insights, and 26 Conclusion feedback along the way (see Appendix D for the full list). 28 Appendices A. National Examples - Communities in Action About the Foundation B. Privacy Primer - Relevant California and Federal Laws The California Health Care Foundation is C. Table of Interviewees dedicated to advancing meaningful, measur- D. Advisory Group Members able improvements in the way the health care delivery system provides care to the people of 35 Endnotes 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. *The content in this report is provided for informational purposes only and does not constitute legal advice. Homebase does not enter into attorney-client relationships. California Health Care Foundation www.chcf.org 2 Introduction The homelessness and health care sectors realize the $ Relationships and collaboration interconnectedness of the housing and health care $ Interoperability needs of individuals and communities. Given that housing status is a key social determinant of health, $ Data quality both sectors recognize the role stable housing has in improving and maintaining health, as well as reduc- This report examines each challenge and a spectrum ing unnecessary emergency room use and hospital of potential opportunities to overcome them, with admissions. At the same time, research indicates that concrete examples from local communities that have addressing the health-related needs of people experi- had direct experience with cross-sector data sharing encing or at risk of homelessness is crucial to accessing (varying in size, geography, and type and stage of and sustaining housing. data sharing efforts). While there are no uniform ways to address the common challenges, communities Purposeful collaborations between the health care and have creatively employed strategies and taken advan- homeless systems of care address the important rela- tage of opportunities to continue pushing forward tionship between health care and housing. This report data sharing efforts. These opportunities prove to be focuses on the various ways in which the two sectors most effective when tailored to each community's own in California are sharing data with each other to bet- needs, structures, relationships, and motivations. ter coordinate and support mutual clients within their communities, most often at the county level. Lessons This report is intended to serve as a guide to those from throughout the state illustrate that data sharing at any stage of undertaking cross-sector data sharing has been pivotal in breaking down silos and coordi- efforts, including those ready to start such efforts for nating between the two systems to better address the first time. While the report is situated in the midst clients' needs. of the COVID-19 pandemic, the hope is that lessons and insights gained during this time can carry forward This report was written in the midst of the COVID-19 for years to come. From conversations with coun- pandemic, when communities were facing ties, Continuums of Care (CoCs), health systems, and unprecedented challenges. It was found that the com- trusted advisors, one piece of advice came through munities already collaborating across departments most saliently: Just get started. before COVID-19 were better positioned to respond to the pandemic, which required a community-wide, organized, multisector approach. For communities not already sharing data across sectors, the realities Methodology required for effective COVID-19 responses helped to This report pulls from research on data sharing projects highlight the advantages in coordinating with partners in communities throughout California and nationwide, in both the homelessness and the health care systems. surveys of health care and homelessness providers In other words, the pandemic further added urgency throughout California, and interviews with county rep- for greater cross-sector collaboration. resentatives and provider organizations. Even with dedicated and committed partnerships Homebase spoke with staff and/or reviewed litera- in place for cross-sector collaboration, data shar- ture of prominent data sharing initiatives across 15 ing efforts have not occurred without challenges. California counties with promising practices and 14 Communities have mentioned a common set of barri- states across the country. These communities were ers they have faced, with four primary ones emerging: identified through findings from two surveys the authors conducted of a wide array of health care and $ Privacy issues Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 3 homelessness providers. The health care and home- The authors' surveys and interviews found that less service organizations chosen to be highlighted homelessness and health care providers have many varied in geographic region, size, and scope of data motivations for sharing data about people experienc- sharing activity. In California, Homebase targeted ing homelessness. Chief among them is coordination communities in Southern, Central, and Northern of care. Communities have recognized that many California regions. For a full list of interviewees, please departments, systems, and organizations have inter- see Appendix C. acted with the same people without coordinating, which has often resulted in duplication of efforts and inefficiencies in delivering services and care. As such, Why Homelessness and they have launched data sharing to support care man- agement, track those receiving care, and facilitate Health Care Providers communications among a disparate set of providers. Share Data Once there is recognition that the same people The homelessness crisis in California is unprec- touched the social services and health care sectors in edented. Never before has the state faced so many various ways, the need to coordinate care between people living without stable housing or supportive health care and homeless systems becomes apparent: services. The main driver for health care and homeless to work together to better address the needs of mutual organizations to exchange data has been to address clients to reduce the high health care costs associated the crisis, especially through efforts such as the Whole with emergency room visits and hospitalizations, espe- Person Care Pilot programs and collaborations aimed cially for the most vulnerable populations, as well as to at improving care for those who frequently touch both help people who do not use health care services get the health care and homeless systems of care - while appropriate preventive or responsive care. Ensuring reducing the costs of the two systems so they can people have a roof over their heads is one important serve more people. way to reduce unnecessary health care visits and has The term "data sharing" encompasses a broad scope of activity. For the purposes of this report, data sharing encom- passes any effort to ensure that data about people served are communicated across organizations or sectors in some way. It can be as simple as getting on a telephone and discussing information about a client multiple people are help- ing. It can be as complicated as creating a central database that pulls in data about people from disparate systems and stores them in a centralized location that many have access to. This report takes a broad view of data sharing in order to learn how different communities in California and across the country exchange information to help people experiencing homelessness who have chronic health conditions. The scope of the inquiry looked at a range of types of data sharing, including: $ In-person meetings where people from different sectors verbally share and discuss those they mutually serve $ Data matching, which includes identifying whether a single person is touching both the health care and homeless systems $ Sharedspreadsheets, which might be exchanged by two organizations to enable data matching for more than one person at a time $ Sharedcare platforms, which are used by multiple staff from multiple agencies to enter data directly into a database, app, or tool about the people they mutually serve (e.g., many communities have created care management portals to centralize and coordinate care) $ Central repositories, which pull data from disparate databases into one shared central system that all providers have access to (or limited access to, depending on privacy concerns) California Health Care Foundation www.chcf.org 4 been a leading incentive to collaborate across the two Benefits to the homeless response system include: sectors. $ Access to clients' public benefits information (e.g., CalFresh, CalWORKS, Medi-Cal) Interviewees also pointed to many additional benefi- cial outcomes of data sharing from both the health $ A trusted partner to call and discuss a client's case and homeless response systems of care. $ Improved ability to keep track of clients $ More robust health information to assist with The homeless system of care (or homeless service prioritization decisions in the Coordinated Entry sector) includes federal, state, and local agencies, System (CES) nonprofit and community-based organizations, $ Maximized resources and increased trust overall service providers, funders, and other groups work- ing to support people experiencing homelessness. within communities A Continuum of Care (CoC) is the umbrella term for $ Greater ability to manage health care for people the group of organizations and agencies that col- lectively coordinates homeless assistance activities with complex issues and resources in a community. At the data sharing $ Stronger systems in place to respond to emer- level, the CoC is the structure that often coordinates cross-sector collaboration. gencies (e.g., communities with preexisting data sharing across the health care and homeless sys- The health care system (or health care sector) spans tems benefitted greatly by being able to quickly various levels and types of organizations. They include hospitals, Federally Qualified Health Cen- respond and house the most vulnerable during ters, health plans, behavioral health providers, and Project Roomkey) local Health Care for the Homeless organizations. Most of the data sharing discussed in this report is shared at a county or local level. Coordinated entry and prioritization. Communi- ties use a process called the Coordinated Entry System (CES) to ensure that people experiencing or at risk of homelessness are prioritized for homeless Benefits to the health care system include: services and resources based on severity of need. Through CES, people are matched to available $ Reduced emergency department admissions resources most suitable to meet their needs. Within $ Reduced inpatient hospital stays each community's CES process, people experienc- ing homelessness are prioritized for housing and $ Advances in screening for people with mental community resources based on factors agreed upon health and substance use disorders by the community, which usually take into account the severity of service needs, considering factors $ Theability to locate patients for follow-up such as risk of illness, death, and/or victimiza- medical treatment tion; history of frequent use of crisis services; and significant physical or mental health challenges, $ Better care coordination for frequent users of substance use disorders, or functional impairments. acute care services Much of the health-related information that feeds into prioritization is self-reported, and people may $ Provision of care to infrequent users who underreport certain conditions or disabilities for would otherwise not be identified in the various reasons. Receiving health data directly from health care system health care providers in addition to self-reported data could provide a fuller understanding of the $ Abilityfor health providers to more effectively severity of clients' needs, thus enabling CES to advocate for housing for people with complex or prioritize people more accurately. severe medical conditions Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 5 What Was Learned: Emerging Stories A Master Person Index (MPI) allows users to match a person to their stored data While the focus of this report is on California initia- records. In cross-sector data sharing, it works tives, before diving into takeaways from California, to identify people in two separate systems by providing them an MPI number that can Homebase conducted a national environmental scan be matched across the two systems while of communities across the country that have under- still maintaining confidentiality. taken cross-sector data sharing to set the context. (See Appendix A for details about cross-sector data shar- ing outside of California). It was found that although each community approaches data sharing differently, Overall, communities across the country have either a number of common attributes across state- and piloted cross-sector data sharing or have adopted countywide programs emerged: programs county- or statewide to enable greater collaboration, to achieve more effective care, and to $ Addressing social determinants of health (SDOH) lower the costs of caring for the most vulnerable peo- is a key driving factor underpinning communities' ple in our system. California is in the forefront of some efforts to pursue cross-sector data sharing. of those efforts. $ Communities tend to start small (e.g., with non- identifiable Homeless Management Information Systems and health information), then add more California Leads the Way data as consent forms are signed and additional While a variety of data sharing models are emerging departments join in. nationally, many California communities have devel- oped some of the most successful approaches to $ The most successful communities embrace central- health care and homeless system cross-sector data izing their cross-sector data sharing, either through sharing in the country. a data warehouse or central repository, in some cases beginning with merging existing Homeless A key reason for this is that California's Medicaid pro- Management Information Systems (e.g., Chicago, gram (Medi-Cal) is implemented on a county level, Connecticut). unlike other Medicaid programs implemented at the $ In the most successful cases, communities uti- state level. The fact that California has 58 counties has lize existing cross-sector partnerships to begin provided communities the opportunity to develop a data sharing efforts. In some communities, while variety of different approaches to collaborate across there may be an initial distrust or data privacy sectors when piloting programs. concerns, existing cross-sector relationships can help overcome early hesitancies. For example, California also has a history of using data sharing to a long-standing relationship between a county's better serve vulnerable populations. Over the past Department of Health and its Department of Social five years in California, several important policy efforts Services would serve as a helpful backdrop for have taken shape that have led to the development coordinating care. of pilot programs for cross-sector data sharing, includ- ing past and upcoming initiatives and programs like $ Some communities use academic and university Health Homes and Whole Person Care (WPC), the partners to help with the data pulls (e.g., King statewide Homeless Data Integration System (HDIS),1 County, Chicago, New York City). and growing adoption of health information exchange $ Establishing a shared Master Person Index helps (HIE) efforts across the state. California continues to track clients across multiple systems. value the importance of cross-sector collaboration California Health Care Foundation www.chcf.org 6 and data sharing as a means to improve outcomes Approximately half of the pilot programs in California for those who need access to both health care and counties identified people experiencing homeless- housing. ness as a specific population they would target. Through Whole Person Care, DHCS set the stage for Additional proposed or upcoming policies seek to the overarching goal of data sharing - to promote further encourage and strengthen communities' cross- community-wide collaboration across sectors. Under sector data sharing efforts, including a new initiative WPC, pilot projects are required to assess each client's for Medi-Cal - California Advancing and Innovating health, housing, and social needs and to coordinate Medi-Cal (CalAIM) - that recognizes housing as care in real time to improve outcomes. Because health care. (See "Policy Opportunities" section on WPC is part of an 1115 waiver, counties could spend page 23 for a deeper discussion of CalAIM and other Medicaid dollars on infrastructure and services typi- upcoming policy that could impact California's data cally not covered under traditional Medicaid, including sharing efforts.) cross-sector IT data systems. The state requires par- ticipants to form new partnerships and to share data. Whole Person Care. WPC was a five-year project Cross-sector data sharing efforts undertaken by WPC (originally 2016–20 and extended through 2021 due Pilot programs have led to many of the successes to the COVID-19 pandemic) initiated by California's highlighted in this report.3 Department of Health Care Services (DHCS) that focused on high-risk and high-utilizing Medi-Cal patients. Through a federal Medicaid 1115 waiver, DHCS provided flexible federal and state funding to Medicaid Section 1115 waiver. A provision under Medicaid that authorizes experimental, pilot programs led by counties to improve health and pilot, or demonstration projects at the state or housing outcomes for targeted populations. With a local level. Projects must promote the purposes $3 billion investment across 25 county pilots (and one of Medicaid but 1115 waivers allow for flexibility city),2 each local WPC pilot has worked to seamlessly and creativity to design projects to better serve coordinate care across different sectors, including Medicaid populations. Successful projects started public health care systems, clinics, behavioral health through a Section 1115 waiver are often adopted as federal policy when evidence shows their value. providers, social service agencies, Medi-Cal managed care plans, sheriff/probation departments, homeless services providers, and food pantries. Each commu- nity effort identified the target population(s) it would focus on, including: $ People experiencing homelessness or "precariously housed" $ People with medically complex situations $ People with alcohol or substance use issues $ People involved in the criminal justice system $ Frequent users of emergency services or crisis health care Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 7 California Communities in Action health information shared, it is not a requirement for HIPAA cov- Many of California's 58 counties have focused on data ered entities. Unlike many communities, the county has included sharing as a central means to enable collaboration clinical mental health data in the information exchange, even between the health care and homeless systems of though it requires consumer consent. Alameda is piloting an care. Three of the communities stand out, as they have opportunity for non-HIPAA covered entities in the next phase of not only taken on cross-sector data sharing, but they the CHR. It will require consumers to provide full consent before have done so by fundamentally shifting their way of data can be shared in the CHR, including all protected personal operating in their communities. All three communities health information. With consumer consent, these providers can - Alameda, San Diego, and Sonoma Counties - have begin on a level playing field (with client consent), giving them created new or different systems of interdepartmen- a more robust picture of the care and barriers their clients face. tal collaboration and coordination, implemented centralized data sharing to better serve high-needs One of the greatest benefits from the SHIE has been for hybrid community members, and adopted a more expansive housing agencies (those covered by HIPAA) that normally are not view of data sharing that goes beyond health care and able to see how their clients move through the health care sys- the homeless system of care. tem. Alameda County anticipates that once the pilot is deemed successful, other noncovered entities will have access to HIPAA- protected data, allowing them to better serve their clients. It ALAMEDA COUNTY SOCIAL INFORMATION EXCHANGE. To has resulted in a fundamental shift in how housing service pro- better serve the thousands of people experiencing homeless- viders work in the county. For example, in the past, they simply ness across the county, Alameda County wanted to develop a would not be able to find a person who was previously living Community Health Record (CHR) that provides a curated and on the street but then disappeared. Now they can go into the real-time view of events and developments in clients' experi- Community Health Record and see if the person was admitted to ences within the health care and homeless system of care. To do the hospital, living with a family member, or returned to a shelter. so, the county created the Social Health Information Exchange (SHIE) to securely collect and integrate people's medical, mental As the project continues, the county is using the rich informa- health, housing, incarceration, crisis response, and social services tion collected through the Social Health Information Exchange to information. The SHIE is a central repository that stores data from undertake data analysis through an equity lens. an ever-growing list of participating organizations and allows the data, under all relevant privacy rules, to be accessible across these sectors. The SHIE helps with overall care management and also has an alert system that notifies participating providers when HIPAA covered entities. Under the HIPAA rule, their client is admitted to the emergency department / inpatient organizations with access to personal health infor- mation (PHI) are considered "covered entities," or booked into jail, as well as when they are discharged from the which means HIPAA privacy rules apply to them. hospital or released from jail, helping ensure providers don't lose As the homeless system of care is structured, track of their clients over time. most housing and shelter providers are not con- sidered covered entities. For providers to view The county collaborated with internal and outside counsel to PHI, clients must explicitly consent to allow develop the data sharing agreements between the partner agen- providers to see their personal information. cies. They also developed policies and procedures, a data security management plan, and the consumer consent sharing releases. In the first phase of development, all participating providers are bound by HIPAA (Health Insurance Portability and Accountability Act), so the Community Health Record lets them share a robust amount of health-based information. To protect sensitive infor- mation, although consumers can consent to have HIV and mental California Health Care Foundation www.chcf.org 8 SAN DIEGO COMMUNITY INFORMATION EXCHANGE. The San a Comprehensive Social Continuum Assessment (CSCA) (PDF), Diego Community Information Exchange (CIE) began as a pilot, which helps providers look at clients as whole people with various with the purpose of centering the patient and of coordinating care factors influencing their experiences. As a multisector collabo- across sectors for people experiencing homelessness. University ration, the CSCA looks at three primary constructs across 14 of California San Diego Health, Father Joe's Villages, City of San domains, which creates a shared language about the risk faced Diego Fire/Rescue and paramedics, and the Regional Task Force by clients in each domain. The data also allow for seeing trends on the Homeless (RTFH) - San Diego's Continuum of Care - and over time for these individuals, in order to better provide patient- other San Diego thought leaders launched the project, with suc- centered care. cess demonstrated through initial return on investment through a reduction in EMS (emergency medical services) transports. CIE was The CIE integrates data from many sources including HMIS then folded into the 2-1-1 San Diego infrastructure in 2016. While and FQHC's electronic health records populating a comprehen- the CIE began as a direct response for immediate needs seen on sive, longitudinal client record. The early partnership with RTFH the ground, it has since turned into an elaborate case coordination addressing homelessness has led to deep integrations of tech- and collaboration system that allows information sharing, referrals, nologies and client consenting processes that have positively and prioritization of care. impacted the region at multiple levels, including data analytics at the community level, changing the way service providers work The CIE platform shares client-level data and facilitates com- together, and offering people a more trauma-informed experi- munity case planning and care team communications. The ence seeking services. CIE began with a cohort of homeless service providers, then expanded to senior services, veterans organizations, and others. On the health care side, FQHCs are active users of the CIE, with It now involves homeless service providers (about 25% to 30% a handful of health plans signed on as well. The extent to which of those who participate); health care organizations including health plans are bought into the CIE framework varies, but the health plans, hospitals, and Federally Qualified Health Centers most invested partners have reengineered workflows, leverag- (FQHCs); and other social service organizations and faith-based ing the CIE to help with complex members. While some health organizations that focus on issues ranging from food insecurity to centers started with read-only access to CIE records, some are transportation needs. contributing patient-level data through system integration. Partners entering data are not sharing "big-ticket" items such as A key component of the success is the strong partnership with the unit utilization and inpatient admission, which could help with the RTFH, which manages the Homeless Management Information cost-benefit analysis. Behavioral health data in the CIE are self- System (HMIS) in San Diego. The partnership established a shared reported but not vetted due to continued legal concerns with 42 release of information, allowing an individual's consent to apply CFR Part 2. (For more information on privacy rules and regula- to HMIS and CIE activities simultaneously. Data integration feeds tions, see "Appendix B. Privacy Primer - Relevant California and from HMIS share valuable data points to support a more compre- Federal Laws".) hensive CIE client profile, creating the opportunity to leverage the data to help highlight housing instability (PDF) inequities and The CIE allows for warm handoffs instead of traditional referral opportunities for systems change. options, and many organizations have adopted CIE frameworks as part of their own shift toward more person-centered care. CIE Creating a shared language for all entities involved in the CIE was staffers have witnessed situations where it's benefitted client care. essential to ensure true community care planning. Participants For example, when a homeless service provider saw that a client have emphasized the importance of having a robust steward- was missing, an alert immediately came from a jail, and the pro- ship infrastructure that includes all partners involved, as well as vider coordinated with discharge to return the person two days having 2-1-1 - a long-term trusted partner - as the backbone later. Several working groups and affinity groups are being con- for building trust for data sharing and local partnerships. Under a vened to better understand the broader population health and social determinants of health framework, the CIE team developed system performance impacts of the CIE. Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 9 The San Diego CIE demonstrates that community collaboration As a result of strong collaboration and partnership, the ACCESS efforts can be incredibly fruitful in coordinating care and changing Sonoma platform was designed to meet the needs of the the culture around traditionally siloed social systems. Communities ACCESS Sonoma team. Recognizing that each input system is around California can learn from San Diego in implementing a different, the platform was designed around the vision of the truly client-centered, community-driven approach to data sharing ACCESS Sonoma initiative and what the other systems could and care coordination. interface with. It uses enabling technology to report, identify, and manage people. While ACCESS Sonoma does not pull a large amount of HMIS data into its central data warehouse, it pulls the ACCESS SONOMA. Sonoma County has become a statewide most important data needed to coordinate care, such as voucher model for data sharing between homelessness and health care and housing-eligibility information from clients. Since ACCESS systems. Its success has, in large part, come from having a big- provides its own care coordinators, case manager data are not picture vision for its work. Sonoma emphasizes culture change specifically pulled from HMIS. in its data sharing efforts, not simply focusing on the technical aspects but moving to fundamentally change the approach to Sonoma's leaders emphasized that communities should sim- how government should work with vulnerable populations. By ply start, regardless of what stage of the process they are at. If ridding itself of the notion that a person "belongs to" a particular communities face barriers or challenges to fully implementing department or set of providers, Sonoma shifted the narrative to Sonoma's model at a large scale, they can always begin on a seeing each person as the community's client. smaller level. For something as intricate as care coordination, it's important to be creative and open to nonideal solutions. Under Accessing Coordinated Care and Empowering Self Sufficiency this philosophy, Sonoma is working with four or five other coun- (ACCESS) Sonoma is a countywide initiative - formed by an ties in California to share data across CoCs - meeting folks interdepartmental multidisciplinary team - that focuses on where they are and helping them get started. the critical needs of residents experiencing physical and men- tal health challenges, economic uncertainty, housing instability, substance use disorders, criminal justice engagement, and social inequity. Born from a Board of Supervisors resolution, ACCESS Sonoma has strong leadership, a pool of funding from all the departments involved, and buy-in from the county's safety-net programs. All this has contributed to the county having a sense of collective responsibility for mutual clients. Individual leaders also played a large role in Sonoma's success. Representatives from Sonoma County shared that it is important to have a leader who can carry the water and advocate internally and with the commu- nity to make the vision a reality. California Health Care Foundation www.chcf.org 10 Common Challenges consent for organizations to share personal data is an underpinning value of all successful collaboratives to Cross-Sector Data examined in this report; consent puts the client at their center of whole-person care. Sharing State and national efforts to engage in cross-sector The health care sector is intimately familiar with the data sharing have uncovered four areas where chal- limitations that HIPAA outlines on the sharing of per- lenges frequently arise: addressing privacy issues, sonal health information. In contrast, HIPAA rules negotiating relationships and collaboration, overcom- do not apply to most providers working within the ing interoperability challenges, and improving data homeless system. Medical information included in the quality. This section examines each of the challenge homeless system tends to be self-reported, mean- areas, explains the types of issues that may arise, and ing that people share their health, mental health, and identifies opportunities to overcome or minimize the substance use information voluntarily to the service challenges. The "Community in Action" sections high- providers who coordinate their care. light how communities have been able to overcome the challenges to advance cross-sector collaborations. Health care partners, governed under HIPAA, cannot share health information about their clients with home- less service providers without client consent. However, Privacy while both systems of care should seriously consider One of the primary issues communities face in data the privacy protections of clients and obtain informed sharing is the challenge to fully address client privacy consent before undertaking data sharing, communi- issues. The most successful and robust data sharing ties should also remember that HIPAA was created to programs have developed data privacy policies that promote data sharing. Rather than shying away from build trust between participating organizations and data sharing efforts, the most successful communities have robust consent policies and systems that build have dug further to better understand what the pre- trust with clients. These organizations typically work cise limitations are and how to address them in order simultaneously with their legal departments to develop to meaningfully share data. For example, ACCESS (1) policies that enable them to share data within the Sonoma moved forward by underpinning participant parameters allowed by federal and state law and (2) consent in each of its projects. consent policies and protocols as an important way to express each client's understanding of and commit- ment to be part of the collaboration. The Health Insurance Portability and Accountability Act (HIPAA) is meant Not all data sharing requires explicit client consent. to protect people's individual medical In some circumstances - for example, through an information and applies to data sharing 1115 waiver - special terms and conditions set out that includes personal health information. the ability for agency partners to data match shared HIPAA was created to enable data sharing, clients when it furthers the purpose of the underlying not to prohibit it. It lays out the details program. Federal and state rules for a number of pub- about what data can be shared and how to share them. lic benefit programs - Medicaid, TANF (Temporary Assistance for Needy Families), SNAP (Supplemental Nutrition Assistance Program), and WIC (Special Supplemental Nutrition Program for Women, Infants, and Children) - allow agencies to exchange data about shared clients.4 However, obtaining client Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 11 Communities have had to confront both actual privacy Opportunities to Overcome Privacy risk and risk aversion from county counsel (lawyers that Challenges provide legal services and opinions to the county and There are a variety of ways county departments and county privacy officers) and other lawyers, in addition local providers can share valuable client information to addressing and overcoming potential underlying with each other fairly and legally. Communities have legal challenges to data sharing. In some instances, taken different approaches to overcoming the privacy they are told the risks with privacy are not worth the issues that arise when a community desires to under- effort. In other instances, communities dig deeper and take cross-sector data sharing. Policy approaches that put the time and resources into understanding how alleviate privacy- and security-related concerns about best to protect privacy and undertake data sharing. data sharing can go far in smoothing collaboration Once they addressed the actual risks, they were able going forward. to move forward with addressing some of the per- ceptions around data sharing - determining how to Solicit Participant Consent share data effectively and put strong privacy protec- For many cross-sector collaborations, the most effec- tions in place. tive way to permit data sharing of health and housing information is to develop a specific participant consent form (often referred to as a "release of information," COMMUNITY IN ACTION. In Ventura County, data sharing efforts or ROI) that explicitly states that to be part of the pro- were initiated through the County Executive Office, which over- gram, clients had to consent to data sharing with both sees many county departments. To begin data sharing across housing and health systems. departments, the data sharing team worked with upward of seven lawyers, one for each county department, to overcome each department's interpretation and perception of privacy restrictions. COMMUNITIES IN ACTION. In Ventura County, when this type of consent form was first initiated, people experiencing home- lessness were wary about signing such a broad agreement. To address their concerns, leaders developed a training program for staff who worked directly with clients. They educated staff about privacy rules and the intention of the program, and went through a series of exercises so staff could adequately respond to client concerns. Once the new training program was in place and staff were more knowledgeable about the privacy issues, they were well positioned to communicate clearly with clients about privacy, along with the reasons for sharing and the benefits clients could receive from the data sharing efforts. With better information, they were able to build trust with clients, who were then much more bought into the benefits of data sharing and collaboration between the health care and homeless systems. In Marin County, a "universal release of information (ROI)" was developed that involves 42 entities from a wide variety of pro- vider partners within the community and the county. One of the primary benefits of the universal ROI is that on an ad hoc basis, providers can pick up the telephone and have a client-specific discussion across different parts of the human services system (for clients who have signed the ROI). California Health Care Foundation www.chcf.org 12 Engage Outside Legal Counsel County counsel are usually not privacy experts. Rather Data Sharing Agreements than require county counsel to acquire the type of There are many types of agreements that communi- privacy law knowledge needed to facilitate cross- ties put in place to facilitate data sharing: sector data sharing, communities can engage expert Business associate agreement. A business associ- legal counsel to craft consent forms and data use ate agreement establishes responsibilities around agreements (including data sharing agreements) that safeguarding protected health information between a HIPAA covered entity and a noncovered entity. A address privacy. With experts in privacy crafting the HIPAA covered entity must enter into a BAA when agreements, county counsel tend to be much more it is sharing data with a partner that is a noncovered willing to endorse their departments' participation in entity (such as a social services partner). cross-sector data sharing. Data sharing agreement. A formal contract that establishes what data are being used and how the data will be used. These agreements can be entered COMMUNITY IN ACTION. In Santa Barbara County, it took one into by any organizations or agencies that wish and a half years going back and forth with lawyers to establish to undertake data sharing. Similar to a data use a memorandum of understanding (MOU) between the multitude agreement. of agencies wishing to collaborate and share data. When the Data use agreement. An agreement that governs Continuum of Care shared the final MOU with health care provid- the transfer and use of data between two or more ers, they were impressed at how thorough it was and believed entities. These often explicitly state what data will be shared, the way the data will be used, and the it was sufficient to allow cross-sector data sharing and privacy limitations placed on the use of data. Similar to a protections. Because it was such an intensive effort to get agree- data sharing agreement. ment, they also committed to ongoing on-site workplace audits Memorandum of understanding. An agreement and quarterly privacy and security checklists. Once the health between two or more parties that is not legally providers saw the protocols in place, they were willing to engage binding but that outlines the responsibilities and and share data, which ended up being essential at the start of commitments between the signatories. Can be the COVID-19 crisis. Because data sharing was in place, home- an agreement between all types of organizations, less service providers could view client medical records to ensure agencies, or individuals. that those with the highest health risk were prioritized for Project Organized health care agreement. An agreement Roomkey. entered into by more than one HIPAA covered entity (e.g., hospital or Federally Qualified Health Engage Leadership Team to Address Center) that establishes that they are partnering to work together. Often used when hospitals and phy- Privacy Challenges sician practices agree to partner to care for patients. Leadership from within a community that articulates the value of cross-sector collaboration can also help Release of information. An authorization required by the Department of Housing and Urban Devel- address privacy concerns and lessen the overall resis- opment (HUD) that expresses a client's consent tance to data sharing. Leadership can come in many to allow their personal data to be shared among forms: people, organizations, and policies. providers and others within the homeless system of care. Sometimes leadership can come from a depart- ment or agency lead who has a vision and leads the effort across all partners. Other times, leadership can come through an identified champion, advocate, or staff person who can spearhead the conversation, who understands the delicate balance between the goals of data sharing and protecting people's privacy. Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 13 Someone who knows when to bring in external part- Leverage Policy Frameworks ners to help move the conversation forward can be In addition to staff leadership, communities have been advantageous. In organizational partnerships, lead- able to take advantage of impactful policies already in ership can come from an organization committed to place - either departmental policies, local efforts, or change and willing to direct and lead the collabora- state and federal policies that recognize the value of tion efforts. data sharing. While HIPAA is often cited as a barrier to data sharing, in reality it was intended to protect Some communities have found it particularly helpful health care coverage when people lose or change jobs to have or cultivate a compliance or privacy officer - making it easier for health data to be portable with who can see the value of data sharing. For example, the person while still protecting patient privacy. Other Marin County pointed to their compliance and pri- local and state policies can trigger similar opportu- vacy officer's involvement in Whole Person Care as nities. Some communities in California have been a key reason for success. While some of this success particularly successful at interpreting policies (includ- has been up to individual leaders, communities can ing statutes and/or regulations) to create a rationale take initiative to secure the buy-in of these important or incentive to undertake cross-sector collaboration. leaders. As cross-sector data sharing projects begin, engaging privacy experts early on as goal setters and cocreators may help facilitate and shape their support COMMUNITIES IN ACTION. In Los Angeles County, under the for the project. auspices of AB 210 (Chapter 544 of 2017), the County Executive Office created a partnership to serve some of the highest acuity people in the county to share data, increase care management, COMMUNITIES IN ACTION. In Sonoma County, a privacy and and help obtain housing access for those who need it most. compliance officer who supported the cross-sector data sharing efforts took a leadership role in overcoming the privacy hurdles. Following the passage of AB 210 (Chapter 544 of 2017), Riverside The Department of Health Services did their homework, hired County set up a homeless multidisciplinary team that meets outside privacy experts, and took the issue to the county Board monthly to collaborate on helping those with the most complex of Supervisors. The supervisors adopted a resolution in support cases. Rather than occurring in an electronic system, Riverside of the cross-sector data sharing. The resolution applied across all engages in "data sharing" in person through case conferences. departments, sending a policy message to each that the county Line staff from partner organizations who know the nuances of a as a whole endorsed cross-departmental data sharing. client's situation use the meetings to collaborate on how to best serve the client and commit on the spot to next steps. The result The Marin County compliance and privacy officer was an active has been more efficient care coordination for people with com- proponent of the county's Whole Person Care Pilot program and plex needs who otherwise would have been bounced back and directly worked to enable data sharing and data integration across forth between departments. To address privacy concerns, the programs. As a result of significant effort, the county has a 32-entity Department of Social Services took a state countywide letter and ROI for client consent and is working to develop a care coordina- worked with legal counsel to develop a specific release and new tion platform to serve people more effectively across sectors. confidentiality agreement. AB 210 (Chapter 544 of 2017) allows counties to create a multidisciplinary team to expedite the process by which homeless adults access housing and supportive services, including through sharing confidential information. While the bill did not waive privacy rights conferred by state and federal laws, it paved the way for a structure to support cross-sector collaboration when working with people experiencing homelessness. California Health Care Foundation www.chcf.org 14 Define Collaboration Parameters Relationships and Collaboration Another opportunity to overcome privacy challenges Another primary challenge is ensuring strong relation- is to start small and build trust slowly. Even in com- ships and collaboration across county departments, munities that adopt universal consents that allow for which is necessary to execute data sharing between broad sharing of personal information, the law limits homelessness and health care systems. Due to the what can be shared about mental health and sub- siloed nature of systems of care and abstraction into stance use. Sharing data with additional cross-sector spreadsheets and data warehouses, it is important partners (child welfare, criminal justice, social services) to focus on sharing data for coordinating care across raises additional concerns (e.g., some communities departments. Ensuring that all parties participating in raised the possibility of law enforcement using the data sharing will benefit from the effort is key to mov- data to further criminalize homelessness). ing a partnership forward. In the face of these concerns, communities are encour- Privacy challenges are one reason it is vital to build aged to start small and share only the data they are trust and collaboration between communities. comfortable sharing. Rather than not share any data, Unidirectional data sharing (primarily homeless service communities can begin by outlining clear parameters providers sharing data with health care providers) tends to share limited amounts of personal health informa- to be more common than bidirectional data sharing, tion (with consent). For example, in the case of San where both partners share their data with one another. Diego's Community Information Exchange, providers Patient health information is held to a higher standard limited sharing data relating to mental health and sub- (HIPAA) than housing information, and homeless ser- stance use to include only self-reported data. vices providers often are not equipped to meet HIPAA privacy standards. Unidirectional data sharing can end up frustrating homeless service providers who share COMMUNITY IN ACTION. Substance use and behavioral health their data with health care providers but do not obtain data are often held to a higher standard of privacy than other information about their clients' experiences with the types of health care data. In Riverside County, the challenge of health care system, reducing motivation to share data. integrating behavioral health data governed by 42 CFR Part 2 was It is in the interest of both types of providers to work encountered during a behavioral health screening hotline. The together and build capacity in the homeless services county has a screening access line (as part of a state waiver) that sector to meet data privacy requirements. allows the county to screen people for services. In this process, cli- ents share behavioral health information over the phone. In order Fractured data systems, regulations, political issues, to protect confidential substance use disorder and behavioral and misperceptions all create further barriers to col- health data, the county entered into a business agreement with laboration. In addition, service providers and eligibility providers that outlines the purposes and uses of the data shared workers are not always aware of how the process of on the calls. For the behavioral health treatment components data sharing is essential for coordinating care - espe- included in the data sharing, Riverside utilizes an atypical ROI cially in the context of limited resources, funding, and that comes with both HIPAA and 42 CFR Part 2, along with train- staff capacity. ings for all staff about both. The unique ROI allows the screening entity to discuss behavioral health information shared during the For both health care and homelessness systems, a screening process to accomplish a warm handoff with community reorientation in systems and approach is needed partners. Those partners not signed onto the ROI do not have to successfully collaborate. For example, homeless access to clients' medical records but can still view HMIS data. system databases are used by people representing many organizations and are not accessible by health care system providers. With so many different people entering data about the people they serve, there may be less uniformity in the way data are entered. That is Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 15 not always the case in the health care system, where Communities can work toward overcoming barriers by most data entry is done by staff from the same health clearly articulating the purpose and benefits of their care organization. Another example: Hospital systems data sharing program. Communities may clarify that often have ongoing contact with their patients while data will be used to better target services. In addition they are under their care. As people move through to care coordination, data from adjacent systems of the hospital, staff enter dates when people exit care care afford service providers a larger-picture view of through a discharge plan, etc. In the homeless system client needs. Providers can utilize data to more directly of care, people experiencing homelessness may not tailor services to the clients they serve, even in their touch the system for follow through, and therefore are day-to-day operations. For instance, in Humboldt not "exited" from the database in real time. County, staff found that developing detailed use cases that specified the circumstances and reasons for Each system's rules work appropriately for the kind data sharing helped overcome misconceptions and of care they provide, yet the approach is different allowed them to gain wide support for their data shar- between the two systems. To collaborate effectively ing efforts. in their cross-sector data sharing efforts, both systems need to be flexible and to recognize their different, but complementary, approaches. Integration of homeless- COMMUNITY IN ACTION. Ventura County is working with a vari- ness and health care systems would further diminish ety of county agencies to implement their data sharing program. silos and refocus care on centering the whole person. To ease community concerns about sharing data across sectors, Ventura articulated a clear vision and purpose for why they are Opportunities to Improve Relationships sharing data. They emphasized that the purpose of data sharing and Collaboration was to coordinate care rather than to instate punitive measures Successful data sharing partnerships underscore (such as from law enforcement). The data sharing helped them the importance of culture shift to support successful to quickly identify eligible people for Project Roomkey and to collaboration. Among the key strategies to inspire work with partners to connect them quickly to shelter during the collaboration and secure cross-sector relationships is COVID-19 pandemic. Without broad participation of partners, to communicate a clear vision and purpose for the the county would not have had the data needed to determine desired data sharing and to foster cross-departmental what the community need was for Project Roomkey. relationships. These approaches can ease the process of coordinating care via data sharing, building a strong Center Racial Equity and Systems Change foundation for ongoing partnership between depart- The communities that have developed the most ments and across sectors of care. meaningful engagement and cross-sector collabora- tion are those that see data sharing as a method to Communicate a Clear Vision and Purpose undertake a fundamental shift in how to care for their Having the right people bought into the data shar- most vulnerable residents. Given that homelessness ing process is essential to move efforts forward. Amid disproportionately affects individuals and communi- a variety of important and complex projects, it takes ties of color, using a racial equity framework for data strategic effort to establish the long-term value and sharing has helped communities achieve their goals of potential impact of data sharing. To fully buy in, targeting the most marginalized within their systems. departments and individuals need to understand the potential value of data sharing efforts to their over- By centering goals on racial equity, data sharing arching programmatic goals. Since data sharing can becomes a mechanism to help shift agency systems have various purposes and benefits, it is critical to and cultures in the way they think about people - understand which of those will resonate with which those they serve are not a department or agency's partners, then communicate clearly how data sharing clients, they are "community clients." Successful com- will achieve the benefits each partner cares about. munities have devised systems, tools, and processes California Health Care Foundation www.chcf.org 16 to ensure that they are viewing clients as holistic indi- shared goal of better serving clients through coordi- viduals with complex identities and histories at the nated care. Some communities specifically called out center of their work. They then treat them with the opportunities for expansion in this area, such as home- most seamless and holistic approach possible. lessness partners working with hospital discharge planning, colleges, and LGBTQ+ centers. COMMUNITY IN ACTION. Looking at data through a racial equity lens can help reduce systemwide service disparities in both the COMMUNITIES IN ACTION. In Santa Cruz County, collaboration homeless and health care systems. For instance, in Alameda on the county's COVID-19 response led to long-lasting cross- County, all public dashboards display data broken down by departmental partnerships. The existence of these relationships race and gender. Data showed intersectional demographic has aided in data sharing efforts because they laid the foun- information in ways that helped the county see patterns and dation for working together across departments. Meanwhile, the impacts of their policies (e.g., the Black population is much in Marin County it was the other way around - community older and is homeless at a much higher rate than other groups). building as a result of its data sharing project led to strong In one instance, the county hospital observed an unusual spike trust and shared contacts between departments, which were in COVID-19 cases within a specific indigenous population in then found helpful in coordinating a comprehensive COVID-19 the Alameda Community (Mam/Mayan language speakers). The response. These two communities' experiences illustrate the reci- SHIE was used to create a holistic picture of that population's procity between data sharing efforts and ongoing collaboration demographics (race, ethnicity, residence - zip code and neigh- between departments. borhood) and utilization - type of services accessed, assigned medical homes, assigned heal plans - to support targeted out- Successful data sharing efforts across multiple agencies are reach efforts. Communities can further expand on this work by based on trusted partnerships. In Seattle and King County, looking at racial disparities between clients served, using this Washington, the public health department joined with the public data to inform programmatic changes targeting historically and housing authorities to lead a linkage of health care and housing currently underserved groups. data based on a shared fundamental belief that housing is health. Setting the value proposition and finding ways that the data link- Foster Cross-Departmental Relationships age and results can assist data collaborators were useful tools to Communities that have been successful with data overcoming barriers and keeping the partnership goals aligned sharing built their efforts on cross-departmental rela- and sustained. tionships between homelessness and health care systems. Many of these relationships draw from coop- Going forward, communities can work on expanding eration on previous projects, while others simply have their data sharing to fold in more partners. For instance, leadership that sees the value of cross-sector col- some partners expressed a desire to focus attention laboration. Data sharing projects themselves have on cross-sector data sharing with social services pro- contributed to the development of deeper relation- grams. They expressed interest in knowing if clients ships as well as shared responses to the COVID-19 are on services such as CalFresh or CalWORKS, as well pandemic. as timelines for when clients are up for Medi-Cal rede- terminations - all to more holistically understand the The connection between beginning to share data client's situation and needs. Communities also raised and structurally establishing an ongoing partner- the possibility that a specific role could be designated ship is beneficial when it is a reciprocal one. Building in public assistance systems that helps to input data to cross-departmental relationships is essential to over- share with health care and homeless system partners. coming misperceptions of how shared data might be used, including by emphasizing that colleagues across departments and sectors are similarly invested in the Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 17 Identify Specific Roles and Structures Use Technology to Support Line Staff One potential way to mitigate a lack of - or often Access to care coordination data sharing platforms fragmented - staff capacity is to create a specific role that support staff while in the field would allow them to carry forward data sharing projects within a com- to locate people more readily and identify their imme- munity, such as a dedicated staff member in charge of diate needs. For example, in Santa Barbara, the public the project. The role can be especially dedicated to health nurse who provides services at encampments project management and coordination or can be split reported that she often doesn't know who she will with project managing other community priorities. interact with until she arrives at the encampment, so she can't look up their circumstances before she Setting up longer-term sustainable structures to move leaves her office. If she could access their data while the work forward is another way to overcome staff she's at the encampment, she could look up a client's shortages. The efforts to develop sustainable struc- health history and be better equipped to provide care tures can be led by the aforementioned dedicated in the field. She could also add notes to the system in staff member or by a team but ideally involves rep- the moment rather than needing to remember when resentatives from each partner organization. While she returns to the office. setting up a governance committee may require a great deal of initial work, communities have found that The ability to use a mobile app to geolocate a per- the support of such a committee has been helpful in son's last known interaction with staff would also be seeing through the efforts. It is especially advisable to impactful. Geolocation would help health workers to have a governing committee that represents all par- locate people who need care; for example, if some- ties involved in the project. one living unsheltered has a colostomy bag, a mobile application could support health care staff to more easily locate the client to follow up and provide any COMMUNITIES IN ACTION. Contra Costa County leadership was medical care. convinced by the vision around the county's data sharing efforts, so they hired a program manager to coordinate across programs and enter data more thoroughly. Working with backend HMIS Interoperability teams as well as direct program staff, the manager has played a The health care and homeless systems of care use key role in bringing the county's data sharing efforts to fruition. different technology platforms to move their work forward. Because of the health care system's complex- Orange County had a data sharing relationship with Cal Optima ity, there are many different software systems, tools, (a Medi-Cal health plan) through the Whole Person Care Pilot applications, and other technologies to help provid- for a number of years that allowed them to collaborate around a ers do their work, including different software tools shared set of clients. More recently, the county initiated efforts to to enable data sharing within the health system, such expand its data sharing efforts. To get a broader group of local as electronic health records and health information partners comfortable with data sharing, Orange County estab- exchange processes. In contrast, the homeless sys- lished a governance committee around its data sharing platform, tem of care has one primary data system that it uses the System of Care Data Integration System (SOCDIS). SOCDIS for coordination and collaboration across all service integrates all the traditional HMIS data elements and Whole providers - the Homeless Management Information Person Care data. The governance committee includes all the System (HMIS). agency members involved in data sharing to advise on privacy, security, and compliance. Orange County also had county coun- As a result of the different ways the various data sys- sel and the county privacy officer ensure compliance for data and tems have been used in the health care and homeless information sharing. sectors, technological functionality advances and processes working with the two systems are quite dif- ferent. Enabling the systems to effectively share data California Health Care Foundation www.chcf.org 18 is one of the biggest challenges the two sectors have Opportunities to Address Interoperability had to overcome to collaborate effectively. Differences Technological advances enable disparate systems to in interoperability of the data systems present a set of share data more readily. While not all database sys- challenges and opportunities, while data quality and tems have evolved at the same pace, interoperability technology standards present a different set. can be achieved in a number of ways - including through building a central data warehouse, upgrading Policy supporting interoperability for health care sys- systems to include more contemporary interoper- tems has existed since the federal Department of ability functionality, and creating new, external shared Health and Human Services set up the Office of the platforms that allow for collaboration. National Coordinator to focus on interoperability in the health care system. As a result, more movement Create a Centralized Data Warehouse toward interoperability has occurred in the health care Historically, HMIS vendors did not share data out- sector compared to the homeless sector. While health side the homeless system of care. When data sharing information technology systems are still far from fully opportunities arose, many of the HMIS systems were interoperable, the health care industry has taken on not built to be interoperable with other software sys- the issue by, for example, creating systems that allow tems, which limited the ability for health and homeless electronic health records (EHRs) to query each other. systems to intersect in a meaningful data sharing arrangement. This can be overcome by creating a In contrast, most HMIS software was not originally centralized data warehouse to store data from both designed for interoperability, but rather as a single sectors. shared platform - an insular system for all organi- zations in a community's homeless system of care to store and share data among themselves. HMIS's lack COMMUNITY IN ACTION. Contra Costa County's Health Services of interoperability has challenged homeless service Division manages a centralized data warehouse through its inter- providers when they want to use databases for their nal IT team, rather than through a vendor. As part of the Whole internal organizational systems that are incompatible Person Care Pilot program, Contra Costa implemented a new with their HMIS. For example, organizations that serve HMIS that could be fully integrated with the data warehouse. a broader population of people (i.e., not just people Working with the vendor, the county IT team developed a nightly experiencing homelessness) may be tracking and file exchange process to bring the HMIS data into the warehouse administering to their clients through other internal and a patient-matching algorithm to allow for bidirectional data software programs, and different funders often require sharing between the health system's EHR and HMIS. Once the the use of specific databases. Because most HMIS are EHR information is pulled into the central repository, care team not interoperable, program staff have had to enter data - including case manager name, title, and contact infor- data into two different systems, once into HMIS and mation - flows into the HMIS to make it available to homeless again into their own internal software system, resulting service providers. At the bottom of each client's data record in in more time spent on administrative data entry than HMIS is any contact information for case managers in the health time working directly with clients. system, making it easy for homeless service providers to contact their health care counterparts to discuss the client's situation. The same challenge arises with data sharing when HMIS data available in the EHR include housing programs the the HMIS is not designed to be interoperable with patient is actively engaged with and contact information. other cross-sector systems. Without the functionality to share HMIS data with the health care system, com- munities wishing to collaborate have sometimes had to do manual data matching.5 Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 19 Enhance Technological Functionality COMMUNITY IN ACTION. In Humboldt County, North Coast Technological developments in the tools that home- Health Improvement and Information Network (NCHIIN) used a lessness and health care systems use can help ease grant from the Office of the National Coordinator (ONC) on cross- the process of bidirectional data sharing. Creating sector data sharing. They reached out to the county Department new features and functionalities in shared care plat- of Health and Human Services to identify the best programs to forms can lower barriers for providers to input data to undertake a pilot. The county originally believed that their HMIS be shared, easing concerns of needing to enter the could serve as the central database but quickly discovered it same data into multiple systems. lacked the needed functionality to be a coordination platform. While the county's HMIS provides good demographic data and information about the services clients are receiving, it is primar- COMMUNITY IN ACTION. The Santa Cruz County Whole Person ily used for reporting to HUD. When HMIS proved not to be a Care Pilot program developed a cross-sector care coordina- viable option, NCHIIN reached out to a fairly young start-up com- tion tool that both health care and other service providers can pany, Activate Care, which provided the technological platform access called Together We Care. Health care sector stakeholders to undertake care management. While upholding HIPAA compli- expressed concern about signing on to yet another platform. To ance, NCHIIN was able to engage multiple cross-sector partners resolve some of the skepticism, Whole Person Care staff created to use the shared care platform by developing a series of use single sign-on functionality in the data platform so health care cases exemplifying how such a platform would improve their work providers could easily view care coordination information. The and the care offered to clients. technological development eases the process for health care pro- viders to view data, thus enabling bidirectional data sharing with participation from health care. Data Quality Health care partners shared that data quality and Use a Shared Care Platform and Use Cases accessibility issues in HMIS make it challenging to Sometimes, when neither system can provide a suit- share data across the two sectors. HMIS data fields able platform to anchor the data sharing, it can be may be incomplete, data entered may not be accu- helpful to establish a shared third-party platform to rate, and data standards may be different across the hold the data. Housing the data in a third-party plat- two sectors. form can also make all parties feel that they co-own the data, rather than one system subsuming others. It In HMIS, key data elements are frequently text-based can help ease cross-sector concerns about one sector and stored in open-ended "notes" fields. For example, over-compromising to fit the other's standards. HMIS users often put detailed and important informa- tion about the people they serve in the notes field, In addition, outlining specific use cases can help garner rather than in data fields that require specified input partner support for a mutually beneficial data sharing options. In one Bay Area community, telephone and relationship. Homelessness and health care provid- email contact for their clients was stored in "notes" ers can get together and brainstorm situations where rather than in specific data fields, which made it dif- it would be helpful for both systems to share client ficult to collaborate when the county was preparing data (e.g., knowing when a shared client has entered to contact people experiencing homeless who were or exited a hospital and/or shelter, so case managers eligible for priority placement in hotels due to COVID- understand what their health care needs are). 19 vulnerability risk. Health care and IT partners also expressed that many HMIS fields have little to no data validation. If users enter data in the wrong place or in the wrong format, the system is not programmed to reject the entry and California Health Care Foundation www.chcf.org 20 require the user to enter the information correctly. A Opportunities to Improve Data Quality simple example is a field requiring a telephone num- Strengthening data quality in the systems used to ber, but the system allows a user to type in letters. It is share data can facilitate community efforts to col- only when the provider tries to call the client that they laborate across sectors. On the HMIS side, methods discover they cannot because there is no telephone to enhance cross-sector data sharing efforts include number in the system. developing data standards, improving data validation, undertaking more robust education, and imple- menting training for new users to the system. When technology or resources are not available to make Data validation ensures the accuracy, clarity, those changes readily, sharing a limited set of data and details of data before using them. The goal across partners can ensure that data - even a minimal is to have clean, reliable, and accurate data. set - can enable cross-sector collaboration. Technological tools and programming rules can be used to validate data. User training can also assist with accuracy, ensuring that users enter Develop HMIS Data Standards data in the appropriate format. With more efforts to undertake cross-sector data shar- ing, it may be worthwhile for the homeless system of care to develop and adopt HMIS data standards beyond those that HUD currently mandates. Some In some cases in HMIS, there is little uniformity about communities collect and store more data in HMIS how data are entered. For example, in order to match than HUD requires, and many of those data are ripe people across systems, some health care databases for sharing between the health care and homeless required a full Social Security number, when available. sectors. Through local policy or via contracts with the In the partner HMIS, however, users were only entering many organizations that work with the HMIS vendor, the last four digits of clients' Social Security numbers. the community could collectively develop a set of When the two partners wanted to data match to see if broad HMIS data standards for data fields that enable their clients overlapped in any way, they were unable care coordination between homelessness and health to because they could not match people without all care systems that build upon HUD's existing ones. The nine digits. standards could establish requirements for robust data validation, interoperability, and greater transparency. Other data quality issues that make it challenging to effectively data share include: Improve Data Validation Data validation can be accomplished manually, $ Duplicateentries in the database for the through programming, or through a combination of same person both. For example, programs can be designed with $ Undated or untimely data entry rules that prevent a user from entering inaccurate information (e.g., displaying an error message when $ Vendors making constant updates to the letters are entered into a telephone number field). In system without informing users, including some communities, partners from the health care sec- modifying data fields tor contributed dedicated IT staff to help clean up the data elements from HMIS in order to facilitate data Without strong data quality in each of the data sys- matching with their local health care data. Homeless tems, the benefit of data sharing is lessened, as the system providers almost always lack the staff capac- technologists on the receiving end of the data have ity to do this on their own, so health care providers to spend significant time and resources going through with more resources and/or expertise are much bet- the data to make them comparable. ter positioned to undertake data quality efforts for the collective good of both systems. While manually Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 21 validating data can be time-consuming and requires Start Out Small dedicated staff, the investment at the initial data shar- In some instances, the effort needed to improve data ing stage can improve the efficiency and quality of quality may be more time-consuming than it's worth. data sharing over time. While data quality issues can feel insurmountable, cross-sector partners may choose to limit the data they share. For example, starting with identifying COMMUNITY IN ACTION. In Contra Costa County, data teams those who interact with both the homeless system and from the Health Care for the Homeless Program and the Health, the health care system has proven valuable. For health Housing, and Homeless Division undertook efforts to review HMIS care providers who see a person in the emergency data and validate many of the data elements, to match data in the department, a flag on that person's record indicating health system. While time-consuming, the efforts were worth the they are homeless enables that provider to reach out investment. With strong data matching, not only were they able to their local Health Care for the Homeless partners or to collaborate more effectively for their Whole Person Care Pilot be more deliberate about finding the patient a safe program, but when faced with the COVID-19 pandemic, they place to go upon discharge, including by contacting were well positioned to quickly identify the most vulnerable in their partner homeless service provider to make a the homeless community to test or temporarily house. warm handoff. Initiate Education and Training For the small but important population of people Cross-system data sharing requires consistent data experiencing homelessness seen only in emergency entry and quality data in both systems. The simplest departments for all their care, limited data sharing solution that communities use to achieve those goals may result in people falling through the cracks. Local is education and training. Helping on-the-ground staff communities that find themselves limiting data shar- understand why data quality is important for data ing will want to ensure that whatever limitations they matching and cross-system care coordination and how impose, the system still allows for bidirectional data to improve the quality of the data they input results in sharing, ensuring that people are getting all the ser- data quality improvement that enables smoother data vices across both systems they are entitled to receive. sharing. These efforts should be paired with policies and procedures for routine testing and cleaning of data to ensure that the education and training inter- COMMUNITY IN ACTION. For the City of Sacramento, the orga- ventions are successful. nization leading the area's Whole Person Care Pilot program, it was clear early on that they could not undergo bidirectional data sharing through their shared care platform. Rather than give up, COMMUNITY IN ACTION. Orange County recognized that effec- they worked with service providers to determine the bare mini- tive cross-sector data sharing would require a data dictionary to mum of client information someone on a care team would need help ensure the data being shared were useful. The county pro- to know: where the patient is and what their needs are. They then vided information about the data they were sending, told health revised many of the fields in the shared care plan over time to system partners that "we need your data to look something decrease the data entry needed while still providing all users similar to our data," and then provided a data dictionary with access to key information about a client's care. examples to make clear how to translate data between the two systems. The county provided the health system database admin- istrators with a series of bulleted questions, time frames, and other needs. When administrators didn't have the answers, they passed the requests to line staff who were manipulating the data directly. When they finally contracted for data sharing, they were able to formalize the specified outcomes into their latest contract amendment because they had a shared language early on. California Health Care Foundation www.chcf.org 22 $ Help make Medi-Cal more consistent and seamless Policy Opportunities: by reducing complexity and increasing flexibility New Efforts in Place or $ Improve quality outcomes, reduce health dispari- ties, and drive delivery system transformation and on the Horizon innovation through value-based initiatives, mod- There are a number of policy opportunities, either ernization of systems, and payment reform6 currently being implemented or on the horizon in California, that can further efforts to share data across To achieve these goals, CalAIM's key proposals include sectors. The on-the-ground efforts of Whole Person developing managed care plans and patient-centered Care (WPC) Pilot programs and other vital data sharing population health strategies. efforts illustrate the impact policy has already had on data sharing across the homelessness and health care Managed care plans will be required to provide sectors. The pilot programs were extremely successful enhanced care management (building on WPC pilots in helping communities address the needs of some of and other efforts) to people experiencing homeless- their most vulnerable patients. While the WPC Pilot ness, one of seven mandatory high-risk populations programs are phasing out by the end of 2021, they that plans must identify and serve. Enhanced Care have set the groundwork for future policy efforts that Management is intended to provide high-touch, on- build off their progress. the-ground, and face-to-face engagement and should recognize the unique challenges of unsheltered California has recently proposed or initiated policy people as they attempt to navigate and access the changes to embrace at a state level more robust data medical and behavioral health care delivery systems, sharing efforts and cross-sector, cross-community col- as well as social services. laboration through Medi-Cal. The CalAIM initiative presents new opportunities to expand health care CalAIM also includes support for flexible wraparound and homeless system data sharing through improved services, including housing supports, called "In Lieu of systems, broader coalitions, and collaborative efforts. Services," which take the place of more costly medical Furthermore, in 2021, the state has developed a state- services. wide Homeless Data Information System that has the promise for cross-sector data sharing across multiple state systems, in addition to displaying aggregate In Lieu of Services (ILOS). Federal Medicaid law data statewide that can be filtered by community. allows states to substitute nonmedical services Additionally, recent budget and legislative proposals to for traditional medical care. Under California's advance a statewide health information exchange may proposed CalAIM pilot project, ILOS are optional create opportunities for greater data sharing efforts. services that pilot programs can cover under Med- icaid. They are intended to be flexible wraparound CalAIM. California Advancing and Innovating Medi- services. Examples of ILOS include housing transition and sustaining services, recuperative Cal is a multiyear initiative by the DHCS designed care, short-term nonmedical respite, home- and to improve quality of life and health outcomes of community-based wraparound services for benefi- Medi-Cal members through implementation of broad ciaries to transition or safely reside in their home delivery system, program, and payment reforms. or community, and sobering centers. The initiative has three primary goals: $ Identify and manage member risk and need through whole-person care approaches and addressing social determinants of health Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 23 HDIS. Beginning in 2020, California began collect- current systems, including clinical data fragmentation, ing and integrating data from each of the 44 local exclusion of certain sectors in the exchange of infor- HMIS systems into a state Homeless Data Integration mation (including public health, behavioral health, and System. The state's vision for HDIS is to create a tech- social services), and complex and onerous rules and nology platform for centralized aggregation of the regulations.8 data collected by all regional homeless systems of care "to make data-driven policy decision aimed at Policies that support expanded HIE that include robust preventing and ending homelessness in California."7 cross-sector data sharing will better help people with complex health and social needs benefit from HIE The system is administered by the state's Homeless systems. Efforts are underway in California to address Coordinating and Financing Council (HCFC), which is these challenges through policy. housed within the Business, Consumer Services and Housing Agency - a sister agency to the California Health and Human Services Agency. HDIS receives Opportunities for Policy identified client data required by HUD from each The state Homeless Data Information System, health homeless system of care's HMIS and pulls it into a information exchange advancements, and Medi-Cal's statewide deidentified cloud database of homeless CalAIM Initiative present opportunities for California client service activity. Because of the efforts required to improve cross-sector collaboration between the by HDIS to have all 44 HMIS vendors participate, they health care and homeless systems. These efforts have all created some interoperability between their have the potential to be big and bold, fundamentally systems and the newly established HDIS. changing the ways the state supports some of its most vulnerable residents. The upcoming or proposed pol- In future iterations, HDIS will also pull client data from icy changes can be strengthened to ensure that people other state programs (e.g., CalFresh, foster care) to experiencing homelessness who have ongoing health provide a more holistic picture of state and locally conditions are placed front and center as the efforts provided services. The state plans to use the informa- unfold. To that end, the authors recommend looking tion to produce deduplicated estimates of the number at the state HDIS, health information exchange, and of people experiencing homelessness in California, Medi-Cal's CalAIM initiative. enable cross-jurisdictional analysis of homeless sys- tems of care, identify patterns of service use, evaluate Homeless Data Information System the impact of services, and identify gaps in services. As California develops data systems to better under- The hope is to allow for cross-sector data sharing stand homelessness through a statewide lens, the beyond the homeless system of care and to share data opportunities for cross-sector collaboration will grow. with other systems, including Medi-Cal, which would HDIS presents great potential for cross-sector col- help California to better address the needs of individ- laboration. With HDIS, the state can and should lead uals and families experiencing homelessness. the way by doing its own cross-sector data sharing between Medi-Cal and HMIS. Like Sonoma County, Health Information Exchange. There is proposed the state could identify the patients in the community legislation and a new budget allocation in California who interact with both the health care system and to enable a statewide exchange of health-related data HMIS. A simple aggregate report or dashboard could among health care providers and consumers. While begin to tell the story of how people are touching both HIEs already exist in California, there are a number of these important safety-net systems. Through initial of gaps in how the systems function. A recent paper HDIS efforts, the state already has the HMIS data and from the California Health Care Foundation by Manatt the technology system in place to support the con- Health identified a number of shortcomings of the tinued establishment of this cross-sector data sharing. California Health Care Foundation www.chcf.org 24 The state should also create a funding stream to sup- regulates EHR vendors and sets technical interoper- port CoCs to develop or enhance greater capabilities ability standards. No parallel initiatives or guidance in HMIS - particularly efforts that facilitate interop- exists to support CoCs in working with their HMIS ven- erability and improve data validation. The federal dors to achieve interoperability. government allocates only a small percentage of federal homelessness dollars to fund work on HMIS. California's creation of HDIS is a step toward fur- Creating a state funding source that CoCs could access ther government action to create pathways for data to improve the functionality and capabilities of their exchange. The state can build on this progress by HMIS could address some of the major technological establishing financing programs that support local challenges that arise with cross-sector data sharing. cross-sector data exchange, such as what the HITECH Act did for HIE. As shown throughout this paper, local Through state policy, HCFC could lead the community communities often are best positioned to undertake to collectively develop a set of broad data standards such efforts, with deep cross-sector relationships. for all HMIS software used in California. The standards Local pilot programs, with financial support, can help could establish requirements for robust data valida- test the waters for statewide application. The com- tion, interoperability, and greater transparency. munities highlighted throughout this report show that some already have experience with cross-sector data Health Information Exchange exchange and have processes or infrastructures in Because of a variety of events, including the COVID- place that can provide a foundation for wider efforts 19 pandemic and the CalAIM rollout, the need to across California. improve health information exchange in California has become apparent and urgent. Earlier this year, Also noted in this report, one of the primary issues CHCF commissioned a report called Why California that communities face in data sharing is overcom- Needs Better Data Exchange: Challenges, Impacts, ing complex privacy rules, regulations, and consent and Policy Options for a 21st Century Health System,9 requirements. California's broad consumer privacy which looked at the challenges and opportunities policies add to that complexity. The state can support for data exchange across four critical scenarios for communities' ease of data sharing across sectors by California, including the need to provide care for com- participating in efforts to harmonize the rules across plex patients such as those at risk of or experiencing the different systems, while reconciling those policies homelessness. Many key challenges were also uncov- with federal requirements. ered in this report, such as (1) the exclusion of other sectors like the homelessness system of care from Medi-Cal's CalAIM health information technology (HIT) and health infor- The CalAIM proposal formally recognizes that people mation exchange funding and policy conversations experiencing homelessness have not been well served that can leave systems like HMIS behind on develop- by the traditional approach to care and also recog- ing interoperability capabilities, and (2) the complex nizes that housing is health care, an unprecedented and onerous data exchange rules and regulations that position in California that has the potential to uniquely prevent providers from legally sharing data with the leverage resources to strengthen the state's social broader care team. safety net. Setting the groundwork for data exchange is a critical component to ensure that rollout is suc- While data exchange in health care is far from perfect, cessful. Through a series of convenings with safety-net the progress was shaped by major federal policy initia- leaders, the California Health Care Foundation has tives. The HITECH (Health Information Technology for put forward a roadmap10 for implementing HIT and Economic and Clinical Health) Act of 2009 provided HIE needs for two components of the CalAIM pro- funding for the adoption of EHRs and HIE and created gram that focus on housing and case management: the Office of the National Coordinator (ONC), which Enhanced Care Management and In Lieu of Services. Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 25 Among these recommendations are several additional suggestions on privacy rules and consent forms that Conclusion align with the needs featured in this report - with The abundance of efforts in California to undergo potential for far-reaching impacts for people in the cross-sector data sharing illustrates the deep desire health care system also experiencing homelessness. to better serve Californians with the greatest needs. In considering these recommendations, it will be criti- The Whole Person Care Pilot program inspired many cal to build relationships with and involve multiple communities to develop robust data sharing efforts sectors, especially the homeless systems of care, in instrumental in helping them forge long-term collabo- planning and implementation. rations. Yet there is much more to be done, both in improving processes already in place and in expand- The first is the recommendation to develop cross- ing efforts to each and every community in California. sector working groups between housing, health care, and the criminal justice systems as part of the CalAIM With pending policy that would capitalize on Whole initiative. In most instances where such workgroups Person Care and expand cross-sector collaboration have developed in local communities, including part- through Medi-Cal's proposed CalAIM initiative, com- ners representing multiple sectors has enhanced the munities that have yet to fully invest in cross-sector collaboration. Recognizing that people's health care collaboration are poised to do so with support from and experiences with homelessness are also impacted the state. by interactions with the child welfare, social services, and the criminal justice system is essential to ensure There are a number of common elements of success- people's intersecting needs are met. The broader ful cross-sector data sharing efforts. Most communities approach to collaboration exists in the communities recognize that there is a cohort of clients who have highlighted herein for their exemplary cross-sector complex needs and who touch multiple systems of data sharing. care. Through data sharing efforts and other methods of partnership, communities may develop the ability to Second, policy guidance should require that home- see and care for people with complex needs through less service providers be part of the development a whole-person lens, rather than by segregating their of shared care plans that involve people experienc- care by department. Collaboration leads to a common ing homelessness or at risk of homelessness. There understanding that there is value in cross-sector data is a wealth of experienced providers in the homeless sharing for both health care sector staff and homeless system of care who have worked intimately with the system providers. Many of the most successful data target populations to be served under CalAIM. The sharing efforts include organizations willing to try new CalAIM program should seek out and incorporate things, take risks, and not accept the current way of their insights about opportunities and challenges serving clients through highly siloed systems. working with people experiencing homelessness. California is in a unique position and moment in time to address some of the fundamental challenges that communities face when undertaking cross-sector collaboration. California Health Care Foundation www.chcf.org 26 As CalAIM rolls out, communities around the state can take advantage of the policy frameworks and fund- Advice from Communities in Action ing from the state to initiate or advance existing data Communities that have had the opportunity to sharing efforts. Keeping in mind the ultimate goal undertake cross-sector data sharing between health care and homeless systems have a great deal to of improving service delivery and care coordination, share. Collectively, they shared a number of key below are a few steps communities can take to assess takeaways: data sharing possibilities: Just do it. Get started. Don't wait to have every- $ Reach out to partners in the health care and home- thing perfect and in place. Even small steps can lessness sectors to scope out what data sharing have meaningful impact in improving health care efforts have historically and currently occurred and housing outcomes for people and can set the foundation for larger data sharing information within the county. exchange systems. Start out small if you need to. $ Take a step back and brainstorm what the goals But do something. and potential gains of data sharing would be for the Be patient. Don't lose patience before you have county, and how efforts to share data could work in a chance to see successes. It might be three steps conjunction with other efforts. forward and two steps back. Alameda County went through years of engaging partners and negotiating $ Follow the examples of the many "communities before it was able to put its systems in place. Some in action" highlighted in this report, based on partners dropped out and never came back. Others which stories and situations resonate most with the were wary, but in the end became part of the cross- sector data sharing efforts. respective county. Don't be afraid to make people uncomfortable. For a long time, the perception has been that it is not okay to share data. As communities have learned over the years, that is not true. Provide details, develop use cases, help explain the inten- tions and values of data sharing, and share the strong protections you can put in place. Engage with organizations who support the efforts, and ultimately others will come along. Remember how complex the systems are. It might not be easy to do cross-sector data sharing. Be willing to adjust or even pull back goals and expectations rather than stop the project because of seemingly insurmountable barriers. Downsize expectations or take on a small piece. One staff person said that it's all about "relentless incremen- talism." But if you have the will and capacity to go big, go for it. There are a wide range of models for cross-sector data sharing. There is not one cookie cutter solu- tion to cross-sector data sharing. There are a range of models - county-led, nonprofit-led, funded by Whole Person Care, funded by large local grants, etc. Develop the model that works best for your community, your location, and the people willing to be at the table and engage. Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 27 Appendix A. National Examples - Communities in Action National models offer innovative approaches to data sharing between health care and homeless systems. They span project type and geographic region, and offer lessons learned and potential paths forward for California communities seeking to implement similar data sharing programs. Beyond that, these examples may serve as models for what a cultural shift to focus on data sharing may look like for a health care / homeless response system partnership. SEATTLE, WASHINGTON: DATA ACROSS SECTORS FOR HEALTH STATE OF MICHIGAN. The Michigan Department of Health and AND HOUSING. King County's Data Across Sectors for Housing Human Services (MDHHS), along with the Coalition Against and Health (DASHH) program, led by Public Health - Seattle Homelessness, sought to change the way the state delivered and King County (PHSKC), integrates housing and Medicaid services in order to help people rather than simply administer pro- and Medicare data to improve the health and well-being of grams. As such, they advocated that the state adopt integrated low-income public housing residents, the majority of whom are service delivery to achieve a more person-centric care model. The insured through Medicaid or Medicare. The data sharing effort goal was accomplished in part by the 2017 merger between the involves multiple agencies and is based on trusted partnerships Michigan Department of Health and the Department of Human between the Health Care Authority, PHSKC, and the Seattle and Services, which became the MDHHS. Core principles of the new King County housing authorities, the two largest public housing department included "people not programs," "root causes not authorities in the county. symptoms," and "engage the community." Initially funded by the Robert Wood Johnson Foundation's Data Targeting "high utilizers," or very frequent users of medical ser- Across Sectors for Health program, DASHH developed data use vices who are experiencing homelessness, MDHHS decided agreements, data sharing agreements, and new releases of infor- that cross-sector program collaboration would be the lifeblood mation to undertake data integration. The comprehensive data of their new system. The state was well positioned to undertake sharing has been invaluable in helping all parties understand data cross-sector work between homelessness and health care systems from the disparate systems. because all counties used the same HMIS vendor to support their local Continuums of Care (CoCs). In effect, they already had an In addition to interdepartmental and interagency data sharing, HMIS data warehouse. The state brought together funding from the DASHH data set is available in aggregated form through a across all programs to fund one statewide system. dynamic, web-based dashboard featuring filters for condition, housing subpopulation, and time period.11 Using the dashboard, MDHHS also engaged its Office of Privacy and Security, which public health and housing agencies can identify prevalent chronic became very active in the project. The office spearheaded the conditions and analyze health care utilization trends among pub- data use and data sharing agreements. It developed a robust lic housing residents. ROI for people covered through HMIS, which is renewed annu- ally. The ROI broadly covered all kinds of data sharing and did The DASHH program has provided valuable insights into future not have opt-out provisions for highly sensitive data (e.g., HIV/ policy planning, program evaluation, and case management. To AIDS, substance use).12 Either the individual opts out of the con- date, the integration between public housing and public health sent or agrees to have all their data shared - even sensitive has led to the development of prevention programs to address information.13 specific housing needs (e.g., through the deployment of com- munity health workers). It was instrumental in the coordination of Rather than creating a data sharing warehouse, the state focused a swift response for people at risk of COVID-19. on data matching. The HMIS vendor sends to MDHHS a monthly list of first names, last names, and Social Security numbers of everyone in their system. The HMIS data indicate if the person is active in HMIS, housed, etc. The vendor then runs a data match of that list against its master patient index to see if the person is on California Health Care Foundation www.chcf.org 28 Medicaid. For those in the Medicaid data warehouse, the vendor The program developed into the Community Partners project, runs the people in HMIS against their Medicaid utilization and with 1,200 patients and is still ongoing. Data activities include expenditure data to identify who is or is not enrolled in Medicaid, electronic notification systems and the use of the warehouse to actively participating, etc. The vendor places an indicator into the aggregate information such as claims data. The program tracks Medicaid data warehouse for enrollees in the homeless system housing outcomes in real time and updates housing informa- of care. tion daily. In addition, BHCHP is now looking at breakdowns by race/ethnicity and disability with the claims data coming in from Through data sharing efforts, MDHHS learned that not all high Medicaid. In line with its mission, it aims to look with a racial equity utilizers were enrolled in Medicaid, but it also found more chil- focus at how the program affects marginalized populations. dren than expected who were high Medicaid users and were also homeless. The data match allowed the state to quickly prioritize housing those children. The pilot project has been eye opening for YELLOWSTONE, MONTANA: UNITED WAY. In Montana, United the state, and CoCs have modified priorities as they have learned Way Yellowstone began a data sharing program motivated more through the data match. by upstream factors such as compounding needs for people who both experienced homelessness and had complex medi- cal issues. With a $60,000 grant over 12 months (2018–19) in BOSTON, MASSACHUSETTS: BOSTON HEALTH CARE FOR THE partnership with the Montana Healthcare Foundation, United HOMELESS. Building on a history of collaboration among the Way Yellowstone identified high utilizers of community services Social Determinants of Health Consortium, the Boston Health via data sharing between health care, homeless services, hous- Care for the Homeless Program (BHCHP) received a $750,000 ing, criminal justice, child welfare, and emergency systems.17 It grant from December 2016 to 2018 to coordinate care across also worked in partnership with the Corporation for Supportive diverse agencies to better serve people experiencing homeless- Housing (CSH) to analyze common data sources using CSH's ness, improve access to services that address SDOH, and reduce Frequent Users Systems Engagement model.18 avoidable emergency department and hospital utilization by 20%.14 They hired a law firm that drafted an organized health care With the shared data, United Way Yellowstone studied patterns agreement, the release of information, and the business associate of systems utilization for existing affordable housing locations agreement. Nearly all patients consented to participate.15 and resources, especially among the most vulnerable popula- tions (in this case encompassing seniors, persons with disabilities, The program used HMIS to house the data warehouse, creating and families with children). It used the information to identify a separate section of the database controlled by BHCHP (the possible funding sources and to further coordination between majority of HMIS is controlled by the CoC). To better understand new supportive housing and services, in addition to establish- clients' needs and service usage, data from all participating agen- ing permanent supportive housing services for people in Billings, cies were drawn into the SDOH Coordinated Care Hub, including Montana. United Way Yellowstone is now working with commu- Medicaid claims, electronic health records, local emergency nities across Montana to potentially replicate the data sharing department data, and data from the City of Boston. BHCHP model, spanning the Missoula, Great Falls, Butte, Helena, and reviewed the data every month so that nurse navigators and case Bozeman areas. managers could more easily reach patients. Evaluators reviewed the pilot and identified a number of suc- cesses. The program resulted in a 23% reduction in the average number of emergency department visits, a 4% reduction in the average number of inpatient admissions, and a 72% increase in time elapsed between inpatient admissions. In terms of housing, 21 of the 50 active participants (42%) were housed, including eight (16%) who started housed and remained housed and 13 (26%) who were unhoused and became housed.16 Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 29 ALLEGHENY COUNTY, PENNSYLVANIA: DEPARTMENT OF vulnerable homeless that seek health care services. The ulti- HUMAN SERVICES. In 2001, the Allegheny County Department of mate goal is for improved health outcomes and housing stability Human Services (DHS) began an effort to create a central reposi- among participants. tory of human services data, eventually growing to incorporate data from overlapping systems such as the Allegheny County Jail, Adult and Juvenile Probation, the county medical examiner, city CONNECTICUT COALITION TO END HOMELESSNESS AND THE and county housing authorities, and the Pennsylvania Department CONNECTICUT DEPARTMENT OF HOUSING. The Connecticut of Public Welfare. The driving goal was improving care coordi- Coalition to End Homelessness and the Connecticut Department nation within human services and between departments. It also of Housing established an HMIS/Medicaid data sharing program incorporated historical data for a fully multidimensional picture of that targets Medicaid enrollees who use housing services. The people's experiences across the many systems. goal is to track which populations are using housing services and have the greatest unmet need, leading to the improvement of It can display the data in a "client view" portal and generate health through housing. In partnership with New York University, reports that are client-specific, provider-specific, and/or program- Coordinated Access Networks, the attorney general's office, and specific. Case workers and clients can access the data online. the Connecticut Hospital Association, the program matched Allegheny's data warehouse has frequently been pointed to as Medicaid data with HMIS data. a national best practice in part because of the numerous com- munity-level benefits it has led to. On the client side, individuals The first match occurred 10 years ago with support from the and families can access aggregate provider and service data Corporation for Supportive Housing and focused on a criminal to make decisions. For planners and program staff, the ware- justice reentry program. Data sharing with the corrections depart- house enables better decisionmaking. Educational and research ment led to a huge revelation regarding the overlap between institutes, including the RAND Corporation, the US Centers for homeless and incarcerated people. The data sharing led to a Disease Control and Prevention, and Carnegie Mellon University, coordinated effort to prevent a revolving door between home- have used the county's warehouse to conduct studies. In addi- lessness and the criminal justice system. tion, representatives from Allegheny DHS spoke to the power of data sharing in telling an integrated story about people and their needs when making a case to increase funding. More funding 2004 HMIS DATA AND TECHNICAL STANDARDS. The Privacy and would allow for added flexibility in programs and a greater ability Security Standards section of the 2004 HMIS Data and Technical to serve the community. Standards describes how data are to be collected and safe- guarded in HMIS. The standards apply to a "covered homeless organization" (CHO), which is any organization that records, uses, CHICAGO, ILLINOIS: COMMUNITY HEALTH PEER LEARNING or processes protected personal information (PPI) for an HMIS. PROGRAM. In Chicago, a desire for better integration of housing, Any CHO that also is covered under HIPAA is not required to health, and human services delivery systems at a national level led comply with the privacy or security standards in the HMIS notice to a variety of data sharing pilots. One program, funded by the if it determines that a substantial portion of PPI about people Community Health Peer Learning Program, was a partnership with experiencing homelessness already is protected health informa- the University of Illinois Hospital and Health Sciences System (UI tion (PHI) under HIPAA rules. Health) and All Chicago Making Homelessness History. The project sought to improve care coordination by communicating homeless The standards mandate the collection of PPI by lawful and fair status from an HMIS and UI Health's electronic health record (EHR). means with the knowledge and consent of the individual, where appropriate, and further require that a notice be posted for con- The partnership was able to leverage the deep technical exper- sumers that describes the general purposes for which PPI will tise at the University of Illinois at Chicago. Although it was a be used. While not strictly required by the standards, local CoC planning grant, All Chicago and UI Health were able to develop policies typically mandate that consumer information can be col- and test a prototype system that could communicate housing lected in an HMIS and shared with other partner agencies only if status between the HMIS and an EHR while adhering to patient the consumer authorizes that in a release of information. Further, health information protections. The program envisions that data the 2020 HMIS standards instruct that consumer consent should integration can lead to better coordination and care for the most be procured if information is shared with other agencies. California Health Care Foundation www.chcf.org 30 Appendix B. Privacy Primer - Relevant California and Federal Laws Insurance Portability and Accountability Act records relating to the identity, diagnosis, prognosis, The Health Insurance Portability and Accountability or treatment of any patient. In general, Part 2 pro- Act (PDF) is the primary federal law that addresses health grams are prohibited from disclosing any information information privacy. It applies to "covered entities," that would identify a person as having or having had which include health care providers, health insurers, an SUD unless that person provides written consent, health care clearinghouses, and business associates. as specified under Part 2. The HIPAA Privacy Rule establishes when and how PHI held by covered entities can be accessed and Homeless service providers that provide referrals disclosed. It establishes standards for privacy, secu- to SUD treatment are not typically deemed Part 2– rity, and standardization of electronic transactions that covered entities unless (1) substance use disorder restrict the use or disclosure of people's PHI. Covered diagnosis, treatment, or referral is their primary func- entities can share PHI with third parties so long as they tion and (2) the service provider promotes itself to the have direct consumer authorization to do so. HIPAA community as providing those services. Thus, service provides detailed rules about what constitutes con- providers that refer consumers to SUD treatment as an sumer authorization. incidental service or as one primary function of many may not be covered entities under Part 2. Health Information Technology for Economic and Clinical Health Act Confidentiality of Medical Information Act The CMIA (PDF) is a California law that protects the pri- The HITECH Act strengthened HIPAA to provide addi- vacy of a person's medical information (in electronic tional protections and privacy restrictions on PHI. It or paper format) from unauthorized disclosure by extended HIPAA's coverage to include "business asso- limiting disclosures by providers, health plans, and ciates," people or other entities that perform certain contractors. CMIA extends privacy protections to PHI. functions or activities that involve the use or disclosure Covered entities include health care providers, health of PHI on behalf of, or provide services to, a covered service plans, and individuals and businesses that con- entity. Because of HITECH, HIPAA coverage now tract with those entities for work that involves access extends to any entity that "creates, receives, or trans- to medical information. The CMIA's basic prohibition mits" PHI on behalf of a covered entity or on behalf of against disclosure provides that "no provider of health a business associate, including contractors of business care, health care service plan, or contractor shall dis- associates. HITECH further expanded HIPAA require- close medical information regarding a patient of the ments regarding notification of affected people when provider of health care or an enrollee or subscriber of health information is compromised. a health care service plan without first obtaining an authorization unless an exception applies." The CMIA Confidentiality of Substance Use Disorder also mandates specific consent requirements for cov- Patient Records: 42 CFR Part 2 ered entities. Entities covered under the CMIA are Confidentiality of Substance Use Disorder Patient typically also covered under HIPAA. Records, 42 CFR Part 2, protects the confidentiality of substance use disorder (SUD) patient records by Lanterman-Petris-Short Act restricting the circumstances under which Part 2–cov- The LPS Act is a California law with the stated purpose ered programs or other lawful holders can disclose of ending the inappropriate, indefinite, and involuntary such records. Covered entities under Part 2 are fed- commitment of people with mental health disorders. erally assisted programs that "hold themselves out" It also establishes a right to prompt psychiatric evalu- as providing diagnosis, treatment, or referral for treat- ation and treatment and sets out strict due process ment for an SUD. Covered information includes all protections for mental health clients. In addition, the Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 31 LPS Act contains patient consent requirements for the disclosure of mental health information. Information covered under the LPS Act includes records obtained in providing psychiatric and mental health treatment to voluntary or involuntary recipients of services. The LPS Act mandates that in communications between qualified professionals providing services or refer- rals (or in conservatorship proceedings), the consent of the patient or their guardian/conservator shall be obtained before information or records may be dis- closed by a person employed by a facility to a person not employed by the facility who does not have the medical or psychological responsibility for the patient's care. Entities covered under the LPS Act are typically also covered under HIPAA. California Health and Safety Code § 1280.15 California Health and Safety Code § 1280.15 man- dates that covered entities - clinics, health facilities, home health agencies, and hospices shall - prevent unlawful and unauthorized access to medical informa- tion. Covered information includes patient medical information protected under the CMIA, as described above. Entities covered under § 1280.15 typically are also covered under HIPAA and the CMIA. California Health Care Foundation www.chcf.org 32 Appendix C. Table of Interviewees GEOGRAPHIC AREA NAME ORGANIZATION Alameda, CA Cristi Iannuzzi Alameda County Health Care Services Agency Contra Costa, CA Alison Stribling and Linae Young Contra Costa County Public Health Fresno, CA Doreen Eley Fresno Housing Authority Humboldt, CA Kelly Escudero, Martin Love, and Humboldt Independent Practice Association, North Jessica Osborne Coast Health Improvement and Information Network Los Angeles, CA Daniel Reti Los Angeles Homeless Services Authority Marin, CA Charis Baz Marin Department of Health and Human Services Monterey Park, CA Carmen Katsolov and Jocelyn Monterey Park Blue Shield Smart-Sanchez Monterey / San Benito, CA Roxanne Wilson Coalition of Homeless Service Providers Orange, CA Nicole LeMaire, Natalie Dempster, Orange County Health Care Agency, Office of Care Zulima Lundy, and Melanie McQueen Coordination Riverside, CA Marcus Cannon and Rhyan Miler Riverside University Health System Sacramento, CA Lisa Chan-Sawin and Alexis Sabor Transform Health - Sacramento Whole Person Care San Diego, CA Luke Mellis, Meili Hau, and Megan Partch Father Joe's Villages Kris Kuntz and Karis Grounds San Diego Regional Task Force on the Homeless; 2-1-1 San Diego Santa Barbara, CA Lucille Boss, Kimberlee Albers, and Santa Barbara CoC Jett Black-Maertz Santa Cruz, CA Lynn Lauridsen Santa Cruz Health Services Agency Sonoma, CA Barbie Robinson, Tina Rivera, and Sonoma Department of Health Services Carolyn Staatts Ventura, CA Tara Carruth Ventura County CoC Connecticut Linda Casey Connecticut Coalition to End Homelessness Chicago, IL Beth Horwitz All Chicago Stephen B. Brown Center for Health Information Technology, Illinois Public Health Boston, MA Keely Benson and Elizabeth Reardon Massachusetts Technology Collaborative Mary Takach Boston Health Care for the Homeless Michigan Paula Kaiser VanDam and Lynn Hendges Michigan Department of Health and Human Services Allegheny, PA Kathryn Collins Allegheny County Department of Health King County, WA Annie Pennucci King County Housing Authority Amy Laurent Public Health - Seattle and King County Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 33 Appendix D. Advisory Group Members NAME ORGANIZATION Dana Bailey Stanislaus County Community Service Agency Jackie Bender California Association of Public Hospitals and Health Systems Ashley Brand CommonSpirit Health Amanda Clarke California Health Care Safety Net Institute Lynnell Fuller Stanislaus County Community Service Agency Cristi Iannuzzi Alameda County Care Connect, a Whole Person Care Pilot and Alameda County Health Care Services Agency Margot Kushel University of California, San Francisco Daniel Reti Los Angeles Homeless Services Authority Ané Watts Anthem Blue Cross California Health Care Foundation www.chcf.org 34 Endnotes 1."Homeless Data Integration System," State of California. 11.King County Data Across Sectors for Housing and Health, 2018 (PDF), King County, April 2018. 2.Counties participating were Alameda, Contra Costa, Kern, Kings, Los Angeles, Marin, Mendocino, Monterey, Napa, 12.The state includes only physical health data, not mental Orange, Placer, Riverside, San Bernadino, San Diego, San health and/or substance use data, from its Medicaid Joaquin, San Francisco, San Mateo, Santa Clara, Santa utilization and expenditures into the shared data system. It Cruz, Shasta, Solano, Sonoma, Ventura, the Small County doesn't pull behavioral health data into the data warehouse, Consortium (Mariposa, Plumas [later withdrew], San Benito, and HMIS does not share that data, which meant that the and perhaps others) and the city of Sacramento. issues around privacy regarding substance use disorder data have not been implicated in the data sharing efforts. 3.For additional information on data sharing in the Whole Person Care Pilot programs, see Keira Armstrong, Mark Elson, 13.As an aside, they shared with the authors that virtually all and John Weir, Catalyzing Coordination: Technology's Role in clients agreed to the broad language of the ROIs, which California's Whole Person Care Pilots, California Health Care meant they could share substance use disorder information Foundation (CHCF), April 2019. that often is protected by 42 CFR Part 2. 4.Khaliyl Lane, Maximizing Linkages: A Policymaker's Guide to 14.Boston Health Care for the Homeless Program, presentation, Data-Sharing, Alluma, April 2019. July 2019. 5.The lack of interoperability also can be a barrier to data 15.Boston Health Care for the Homeless Program, presentation, sharing for policy initiatives in the health care sector that April 2018. require collaboration for vulnerable patients (WPC and the 16.Boston Health Care for the Homeless Program, presentation, proposed CalAIM being the most significant such initiatives). July 2019. Many health systems are considering independently setting up their own data sharing hubs outside of HMIS or electronic 17."Helena Housing and Regional Health Care Initiative," health systems because many of the data governance Montana Healthcare Foundation. structures and standards needed for broad centralized data 18.The Coalition for Supportive Housing initiative Frequent sharing efforts are not addressed directly when new policies Users Systems Engagement (FUSE) "helps communities or programs are introduced. Homeless system staff working break the cycle of homelessness and crisis among individuals on behalf of a Continuum of Care have shared their concern with complex medical and behavioral health challenges who that if separate systems are established, service providers are the highest users of emergency rooms, jails, shelters, would need to enter data into more than one data sharing clinics and other costly crisis services." hub in their community (for example, ones hosted by the local hospital and a managed care organization), adding more work and creating disparate systems. Homeless system providers also raised concerns that health plans may be developing systems in silos without the opportunity for input and coordination with potential partners and experts in the local homeless system of care. Homeless system staff urged health care systems to coordinate with the larger communitywide efforts rather than create their own separate solutions. 6."California Advancing and Innovating Medi-Cal," California Dept. of Health Care Services, accessed on February 16, 2021. 7."Homeless Data Integration System (HDIS)," State of California. 8.Jonah Frohlich, Eric Bartholet, and Jonathan DiBello, Why California Needs Better Data Exchange: Challenges, Impacts, and Policy Options for a 21st Century Health System, CHCF, February 2021. 9.Frolich, Bartholet, and DiBello, Why California. 10.Jonah Frohlich, Kevin McAvey, and Jonathan DiBello, CalAIM and Health Data Sharing: A Road Map for Effective Implementation of Enhanced Care Management and In Lieu of Services, CHCF, May 2021. Breaking Down Silos: How to Share Data to Improve the Health of People Experiencing Homelessness www.chcf.org 35