ENVIRONMENTAL SCAN. Iowa Health Information Exchange Ecosystem. Qualitative Data Report

Qualitative Data Report ENVIRONMENTAL SCAN Iowa Health Information Exchange Ecosystem Advocate Consulting LLC Lisa Moon, PhD, RN, LNC, CCMC Gregory Clancy, DNP, RN Mindy Hangsleben, Process Engineer Nicole Kapinos, DNP, RN

Executive Summary This report includes background, study-design, data analysis methods, findings, emerging themes from data collected through a convenience sample of 50 key informants, from 32 unique organizations, in 16 cities across Iowa. The purpose was to understand current information exchange capabilities needed to enable care coordination for hospitals, clinics, behavioral health, long term care, home care and federally qualified health care centers. Qualitative data collection using key informant interviews and observations was completed over a six-week period. Two instruments were used: 1) key informant questionnaire including organizational profile, technology infrastructure, information sharing and information use, and 2) observation guide targeting health information technology capabilities and their level of maturity in the health care operation based on requirements for Meaningful Use (MU) and Merit-based Incentive Payment System (MIPS). Data was systematically collected, aggregated and analyzed. Proportions and percentages are presented for quantitative data. Qualitative data were coded, maturation levels scaled and data synthesized using thematic analysis. Findings were reported using charts to represent specific items in the key informant questionnaire. A synthesis of all data was completed and following themes identified. 1. EHR adoption low in some settings. 2. Most data sharing between settings uses paper record transfer. 3. Large information system vendors are used by larger health systems, but electronic exchange outside of those systems is limited. 4. Portals are becoming the norm for view only data sharing. 5. Health information exchange concepts not well understood and in many cases misunderstood. 6. Even when information exchange capabilities exist they are seldom fully implemented, operationalized or generalized creating information silos. 7. Admission, discharge and transfer alerts are desired, however, when available are not well integrated into workflow. 8. Most care coordination services use manual work methods to accomplish broad responsibilities. 9. Significant number of organizations are participating in alternative payment models, but few have technical infrastructure needed to accomplish goals of accountable care. 10. Merit based incentive program system (MIPS) / meaningful use (MU) attestation is a top priority for organizations. Recommendations are categorized using People, Process & Technology Framework. Each category focuses on information exchange capabilities needed to enable care coordination across an expanded care continuum in Iowa. Findings, themes and recommendations create baseline understanding of statewide opportunities for information exchange. 2

Background Iowa Health Information Network (IHIN) aims to improve care, increase security, promote cost savings, streamline health care delivery services and reduce medical errors using health information exchange (IHIN Goals, 2017). IHIN is responsible for the operation and maintenance of a statewide technical infrastructure for information exchange needed to support Iowa heath care delivery system and state agencies. To accomplish the broad goals, an environmental scan of the Iowa s current state as it relates to the information exchange ecosystem was completed. The study data will be used to establish priorities for both short and long-term strategies for operationalizing a state-of-the-art health information exchange by 2020 (IHIN Goals, 2017). Context The total population of Iowa is 3,134,693 1. The median household income $53,183. Eleven percent (11%) persons in poverty and close to eight percent of its population (7.9%) disabled under age 65 1. Six percent of population are veterans (211,066) 1. A population approach for information exchange will need to consider the full population and its unique characteristics. Iowa has 118 acute care settings including 34 hospitals, 82 critical access hospitals and two (2) Veterans Administration facilities. 2 Large health care systems are pervasive with enterprise wide networks: Mercy, UnityPoint, Trinity, Catholic Health Initiative, University of Iowa. An information exchange will want to build on past success and existing community assets. Medicaid beneficiaries include more than 750,000 individuals with total Medicaid program expenditures of almost $5 billion dollars each fiscal year 3. Safety net clinics include 14 federally qualified healthcare centers (FQHC) or look alike clinics caring for more than 180,000 individuals 3. An information exchange focused on requirements specific to the needs of vulnerable populations will be important. Iowa participated in the Affordable Care Act Medicaid expansion through expanding coverage via alternative plans 4. Medicaid Managed Care is accomplished by three managed care organizations (MCO) known as 1) Amerigroup Iowa, 2) AmeriHealth-Caritas and 3) UnitedHealth Plan of the River Valley 5. More than 90% of beneficiaries are covered under the current Medicaid approach 3. The move from fee-based service to value and quality relies on 1 United States Census, 2016. Quick Facts. Retrieved October 25, 2017 from https://www.census.gov/quickfacts/fact/table/ia/pst045216 2 Medicare, Hospital Compare (2017). Retrieved October 30, 2017 from https://www.medicare.gov/hospitalcompare/results.html#dist=25&state=ia&lat=0&lng=0 3 Access Monitoring Review Plan (2016). Iowa, Post Public Comment, Version 2 (10.3.2016). 4 Advisory Board. Where State Stand on Medicaid Expansion. May 19, 2017. Retrieved on October 30, 2017 from https://www.advisory.com/daily-briefing/resources/primers/medicaidmap 5 Medicaid Managed Care (2017). Retrieved October 30, 2017 from https://www.ihaonline.org/managedcare 3

approaches using sophisticated analytics and reporting capabilities that advance new models of care. This work is integral to IHIN mission, vision and values. Before beginning to build an information exchange, a broad understanding of the current state of health providers, their settings, and current methods of health information exchange for care coordination was needed. Subsequently, a qualitative study of the health care environment technical capabilities required to accomplish care coordination between hospitals, clinics, longterm care facilities, home health care, behavioral health and FQHC s was completed. The primary data collection activity used were key informant interviews and onsite observations over a period of six (6) weeks. The report will review qualitative methods, data analysis, findings and recommendations to inform the work of the Iowa Health Information Network. Key Informant (KI) Interviews and Observations The purpose of key informant interviews and observations were to understand first-hand knowledge about information exchange needed to coordinate care. These are qualitative methods. The use of qualitative methods ensures listening to, and consulting with, current and future customers, partners and stakeholders of Iowa Health Information Network. Key informant interviews are a rapid appraisal technique that, when systematically applied, provide phenomenological data needed for thematic analysis of stakeholder perspective, perception and behaviors. Key informant data are used to understand how and why the reasons that supplement and drive findings from sample quantitative data. Key informant profiles for the environmental scan included individuals on staff at health care organizations in the following roles: Health Information Management, Health Information Technology (interoperability focus), Care Coordinators, Case Managers or Individuals in health care operations familiar with health data exchange (internal and external). Key informants from leadership at large organizations were included if their perspective was HIT operations focused. A geographic mapping strategy was developed to organize KI engagement. Sample methods were used to identify, schedule and interview 50 individual key informants from a variety of organizations in the Iowa health care ecosystem based on their specialized knowledge and unique perspectives. Key Stakeholder Target Site Plan The state of Iowa is large - made up of 99 counties. Organizing the work was important to collect essential information needed to answer questions related to information exchange capabilities between health care entities for care coordination. A geographic mapping strategy was developed to guide work (figure 1). An Inventory of health care organizations was cross-walked with geographic mapping strategy to classify key informants by location. The purpose of this step was to ensure an opportunity for multiple 4

perspectives across the diverse ecosystem of Iowa health care stakeholder organizations. Potential key informants were first identified if there was a formal IHIN relationship. The list was then matched with large health care systems and network affiliation to avoid over representation of any one health care enterprise. Figure 1: Geographic Mapping Strategy Then, key stakeholders for groups of interest for home health care, long term care, federally qualified health centers, behavioral health and hospitals (urban and rural, CAH) were identified and added to the geographic mapping strategy. If available, pertinent historic or previously aggregated data for each classified key informant target was added to that health care entity profile on the geographic mapping strategy and contact information added. When background information was not available, a call to the health care setting main number to search out key staff based on key informant profile was completed. This geographic mapping strategy of potential key stakeholders produced an inventory of organizations used to recruit key informant target sites to fit with sample size (50) needed to accomplish data collection. Once KI targets were identified via outreach phone and email were completed to schedule visits. Onsite visits based on geographic mapping strategy and three travel day work week beginning first week of September were attempted. When onsite visit attempts failed, telephone interviews were scheduled and data collected using the environmental scan 5

interview instrument for health information exchange. Instrument testing the first week of September was completed using a pilot site in Centerville, Iowa. The survey instrument was refined and the remainder of key informant interviews were completed in September and October 2017. A total of 242 health care organizations were contacted for interviews. 50 key informant interviews from 32 unique organizations in 16 cities across Iowa, were completed for a final reach rate of twenty-three percent (23%) shown in table 1. Activity Outreach Scheduled / Completed Interviews Pilot Centerville 4 Mercy affiliates 9 5 Independent (HH, BH, Clinic, LTC, FQHC) West, Central & East Swing 64 Health care targets 19 Geo Map Strategy Alternate Sites Geo Map Strategy 62 Health care targets 3 HIMSS conference 11 Health care targets 7 IHIN lists 76 Independent clinic & HH targets 10 Other contacts IHIN board, etc. 2 224 Targets (phone, email, in-person) 50 Table 1: Key Informant Interview Engagement Yield Sampling Methods A convenience sample was used for this study. Evaluators identified groups and organizations for inclusion in a convenience sample based on the key stakeholder target site plan using the geographic mapping strategy and currently available IHIN contact lists. The goal was to intentionally seek at least 50 distinct perspectives on the topics of health care information exchange and computer system interoperability from the KI perspective across multiple roles and settings. When convenience sampling had low yield, snowball sampling was deployed during KI outreach or interview. Snowball sampling was used for two purposes: 1) to solicit other key informants to be interviewed from the target organizations and the surrounding health care community, and 2) as a method of securing onsite observations. This additional sampling method allowed key informants to identify additional sample participants if time permitted, when original key informant was not available in the timeframe dictated by the evaluation and as a means of soliciting needed information rapidly. A secondary benefit was acquiring contact information for key informants used in the 17 onsite observations shown in Table 2. Activity Outreach Observation Pilot Centerville 1 3 F/U after KI Interview 32 14 33 17 Table 2: Key Informant Observation Engagement Yield 6

Once a KI was identified, outreach was made and meetings were scheduled. All KI were contacted by phone and email prior to in-person and telephone interviews (or observations). A standard scripted communication tool was used to describe the purpose of data collection activities and IHIN context to potential KI. Reminder calls and emails were completed one (1) day prior to scheduled KI interviews (KII). If organizational detail was not available prior to visit, a standard set of predetermined questions was used to obtain that information. Sample outreach trials yielded engagement data that were kept in an Excel spreadsheet and used to track and monitor outreach attempts and success. The final KII count by setting was Accountable Care Organization (2), Behavioral Health (5), Clinic (6), Federally Qualified Health Center (3), Hospital (11), Home Health (7), Critical Access Hospital (5), Health System (3), Long Term Care (6) and Specialty Clinic (2). Interview Questionnaire An interview questionnaire was developed. The interview questionnaire included four sections: 1) Organization Profile, 2) Technology Infrastructure, 3) Information Sharing, and 4) Information Use. The questionnaire was available in electronic format and pre-coded with typical answers and a notes section for ease of documenting key informant responses. Probing questions were used to solicit additional detail as needed. The questionnaire was reviewed, refined and then approved by the IHIN staff prior to field testing. Interview debrief methods were used post-interview when applicable. All interview questionnaires were saved in secure filesharing format. Appendix A is the questionnaire used in semi-structured interviews of KI across Iowa health care ecosystem. KI interviews were completed in-person and via telephone. All KII were 30-60 minutes in length. Interviewers used the interview questionnaire to guide the conversation. Notes were taken and typical answers captured for analysis. All data were transferred to an aggregated spreadsheet. Tables were constructed for each question and charts created and analyzed. Observation Guide An onsite observation guide was developed to assist with data collection from end-users in KI organizations. The matrix was used to identify health information technology capabilities and their level of maturity in the health care operation. The tool uses a three-point scale to assess maturity of knowledge, capabilities and implementation. The scale was level 1 no functionality, level 2 some functionality and level 3 high functionality. The observation guide evaluated fifteen different HIT (Health Information Technology) requirements tied to Meaningful Use (MU) and Merit-based Incentive Payment System (MIPS). The HIT requirements included continuity of care document (CCD) components, admission discharge and transfer (ADT) alerts, request / accept care summary, closed loop referral, CCD generation, Direct Secure Messaging (DSM) integration, transfer of information to registries, electronic prescribing, patient portal, patient access to protected health information, patient 7

specific education, patient generated health data, clinical information reconciliation and measure collection and reporting. Observations were performed onsite at multiple locations throughout the state of Iowa. Observations were conducted on 17 health care settings at 12 unique health care organizations. Two health care organizations had more than one health care setting on their campus. The key informant (KI) target plan divided Iowa into three regions: west swing, central swing and east swing. KI observation sites were identified by snowball sampling methods. All KI observations sites were first identified as KI interview (KII) targets. When KI interviews were complete, a request for an onsite observation was made. Of the 50 KII completed, 12 health care organizations agreed to onsite observations. Observations were completed by a process engineer and each observation lasted between 1 2 hours at each KII organization. All data were collected electronically in the observation guide instrument. Post observation setting responses and maturity levels were scaled. The observation data analysis report is available in appendix D. Pilot Test To ensure reliability of instrument, data collection methods, and processes a pilot test was completed. The pilot test was scheduled at Mercy Hospital Centerville and included four settings of care (hospital, clinic, long term care and behavioral health). Semi-structured group interviews were completed over a four-hour period. Onsite observations were completed for hospital, clinic and behavioral health. Long term care did not participate in observations due to the absence of an electronic health record (EHR). Snowball sampling was deployed by asking KII to provide organizations in their local health market with which they typically share patient data for treatment purposes. KII quickly identified another? long-term care setting, Federally Qualified Health Center (FQHC), behavioral health, home health care and a large primary care clinic. Immediately following the KII the evaluation team met and debriefed. The team identified emerging themes and opportunities for instrument refinement. The health care settings identified in the local health care market were placed on the outreach plan for KII, and engagement processes yielded additional KII for the environmental scan. Data collected from these nine (9) settings in Centerville local health market is by far the most complete picture of an information ecosystem in Iowa for this qualitative study. Data Analysis A data aggregation plan was developed and all data was transferred into an Excel spreadsheet that included a data aggregation tool for storage purposes. All personal identifiers of KI were removed, and KI interview instruments were saved in electronic format in secure file share. Organization and setting level identifiers were preserved. The aim was to provide a unique perspective by setting level for each KI. 8

Data from the environmental scan was analyzed at three levels: 1) semi-structured interview question data, 2) semi-structured interview notes and 3) observation information. Data analysis included moving interview question item responses to tables in the Excel spreadsheet. Charts were created using pivot tables and labeled for each section and question number. Each chart was embedded into a Word document to create Appendix C. An analysis of all interview question notes was completed. Notes were coded by color and theme. The coding was validated by a second evaluator to increase reliability. Following coding procedures, one key point for each setting was captured. A key point would be the significant takeaway arrived at following a synthesis of available note level data. Key points were documented in the data aggregation note tab for each KII. When a key point became a repeating theme, it was broken into discrete types based on issue. Those issues were assigned to each KI data stream in the data aggregation note tab for future reference. Observation data analysis was completed by the process engineer familiar with the onsite context for each KI. Example findings were isolated and documented in the observation data report found in Appendix E. Observation findings were used in this report to support emerging themes and recommendations. Findings Organization Profiles Network affiliation by geographic footprint shows that key informants (KI) report their organizations are local (40%), regional (36%), multi-state (10%), statewide (12%) and national (1%) (Figure 1 SAQ1 6 ). Network affiliation and the structure of enterprise relationships is a key consideration when developing statewide technical infrastructures. For example, one KI in Centerville, IA reported no decisions are made locally related to technical interoperability. Instead, all decisions must be presented to regional system, who then determines if change is beneficial system wide. This is not unusual for large healthcare enterprises, and especially those that span across state lines. Network affiliation may be a factor that slows down implementation, adds to time and materials cost, and slows participation at local level. 6 Denotes the figure number (Figure 1) in the report and the interview section letter and item number (SAQ1) from KI instrument questionnaire for the purpose of mapping findings to Appendix A and Appendix C. 9

25 20 15 20 18 10 5 0 5 6 1 Local Multi-state National Regional Statewide Figure 1: Network Affiliation by Geographic Footprint When asked about the number of sites operated by a KI organization, the sample was almost evenly split with 26 organizations (52%) reporting more than one site. Twenty-four (24) KI (48%) reported only one site of operation for their health care setting (SAQ5). This is important as statewide interoperability design begins to plan for local implementations. Forty percent (40%) of KI report providing services that span beyond their local community. Sixty percent (60%) report providing services only in their local community (SAQ6). When determining local data governance structures, the phenomenon of local care across community partners is critical to consumer trust in health care delivery systems. Most KI report providing services for more than one population (Figure 2 SAQ7). 46 of50 (92%) KI report the most common population served is the aged. 17 of 50 (34%) KI report migrants as the smallest population served. Hospital systems have the most variety in types of populations served. Specialty clinics have the least diverse population types (e.g., pregnant women or pediatrics) per KI report. Knowing the population served by prospective clients will assist in use case development based on local needs and statewide population health issues. 10

Vulnerable Population Pregnant Women Physically Disabled Pediatrics Newborns Migrant Mental Health Indigenious Homebound Elder Care Developmentally Disabled Chemical Dependent Adult Adolescents Academic Student Pop 17 19 21 27 27 28 29 30 33 32 32 31 33 41 46 0 5 10 15 20 25 30 35 40 45 50 Figure 2: Population Served by Key Informant Organizations KI organizations participate in multiple collaboratives or partnerships simultaneously. Understanding health care market relationships is important as new modes of care and payment develop. Fifty-two percent (52%) of KI report participation in Accountable Care Organizations (ACO) and forty-two percent (42%) Managed Care Organizations (Figure 3 SAQ2). Although participation is high in ACO and MCO, KI report their understanding of attribution, risk-stratification and costing models is low. Two ACOs that were interviewed reported sophisticated analytic operations, but admit that their work is structured for corporate consumption and is far removed from directly working with frontline staff. In one example, the ACO provides near real time information to guide the work, but it s up to the individual health care entity to manage workflow, policy and practice. This finding may indicate a need for standardized care management tools and processes available to end-users. Twenty-six percent (26%) of KI report participation in Alternative Payment Models (APM), twenty-six percent (26%) have Memoranda of Understanding (MOU) for shared services, eighteen percent (18%) participate in Integrated Delivery Network (IDN) and sixteen percent (16%) in University Health Systems (UHS). KI note that partnerships improve their access to advanced technical tools needed to manage care delivery. One hospital system KI reported the need to learn how to leverage the information to make changes at the point of care. This may indicate a need for technical assistance and education exists. Thirty-four percent (34%) of KI report they do not currently participate in a partnership or collaborative. A KI from LTC noted that no one seeks us out for participation because we don t have any real value in the market, showing a need may exist for education in post-acute care settings on value-based care topics. 11

30 26 25 21 20 17 15 13 13 10 5 7 5 9 8 0 3C ACO APM FQHC IDN MCO MOU None UHS Figure 3: Types of Collaboratives and Partnerships KI Organizations Participate in Regularly The good news is that the Iowa health care ecosystem is participating in and working on broad health care reform goals through strategic collaborative relationships. The highest level of participation in collaboratives and partnerships by settings shows hospitals (25), health systems (24), ambulatory clinics (24) and specialty care (21). Emerging trends indicate that participants expand across the care continuum with participants reporting mental health (21), university health systems (11), long term care (9), and social services (11). Twenty-two (22) KI reported they do not participate in collaboratives or partnerships (Figure 4 SAQ4) showing available capacity still exists and market saturation is relatively low. KI report innovative new models of care driving them to collaborate with unlikely partners. For example, one behavioral health KI has a formal partnership with the jail system. By proactively targeting clients experiencing mental health, chemical dependency and homelessness they hope to reduce recidivism. University Health System State Agency Specialty Care Social Services Public Health None MH / CD Long-term Care Hospital Health System Ambulatory Care 5 9 9 11 11 21 22 21 25 24 24 0 5 10 15 20 25 30 Figure 4: Collaboratives and Partnership Participation by Health Care Setting 12

KI organizations employ a variety of health care professionals. KI report that nursing staff are employed by 47out of 50 of their organizations (Figure 5 SAQ9). In some cases, nurses are the only health care professionals on staff based on KI responses. Case managers are employed in 23 of 50 organizations and care coordinators in 22 of 50 organizations. KI report that in several organizations nurses make up many of these roles. To increase adoption the information exchange may need to structure education and technical assistance specifically for front line staff in the nursing profession. Licensed providers are employed by 30 of 50 interviewed organizations, but KI reported during interviews that in some cases providers are contracted, work remotely (telehealth), or are shared resources with other settings. Support staff are employed in 29 of 50 organizations. KI report that some support staff have low technical literacy, making it difficult to implement new technology. Mid-level care providers are employed in 32 of 50 organizations. Rehab providers (physical therapy, occupational therapy and speech therapy) are employed in 27 of 50 organizations and mental health providers in 26 out of 50 organization, showing a wide range in roles, responsibilities and data management issues exist across the broad health care ecosystem. Support Staff (MA, Front Desk, Rehab Staff (OT, PT, ST) Nursing Staff Mid-Level Care Providers Mental Health Providers Licensed Providers Dental / Eye Case Managers (TCM, DM, CCM) Care Coordinator 15 23 22 29 27 31 26 30 0 10 20 30 40 50 47 Figure 5: Health Care Professional Types Knowing who end-users are in practice settings will ensure that education is delivered based on profession, setting and workflow. The majority of KI report that all health care staff have access to the electronic health record (47/50) (SAQ11). Only 3 out of 50 KI report that not all providers have access to the EHR. KI interview responses include examples of primary care, specialty provider and long-term care settings that do not use EHRs. Multiple KI reported physician providers that refuse to document in an EHR for a variety of reasons including advanced age of the provider, administrative burden and knowledge gaps. To increase physician participation, several KI reported that provider education related to documentation, quality metrics and the use of tools and information exchange mechanisms available in current information systems is needed. 13

Technical Infrastructure The 50 unique KI report organizations using 17 different electronic health record vendor information systems (Figure 6 - SBQ1). Three (3) KI report their organizations do not have an electronic health record and document patient records on paper. The most prevalent EHR systems are Cerner (13 of 50) and Epic (12 of 50) making up 50% saturation of the sample EHR market in organizations interviewed. The other 50% of the sample use a variety of EHR systems, with no more than any three organizations using the same system in the remainder of KI organizations. Vision successehs Paper NextGen Netsmart ModernizingMedicine McKessonParagon Kinnser HomecareHomebase GECentricity ExtendedCarePreferral(ECP) Evident-CTSI Epic EDCS Credible Cerner AxisCare Allscripts 1 1 1 1 1 1 1 1 1 1 2 2 2 3 3 3 0 2 4 6 8 10 12 14 12 13 Figure 6: Types of Electronic Health Records Used by KI Organizations Information technology staffing models vary among KI organizations (Figure 7- SBQ2). Some KI report using more than one staffing model to manage Health IT assets and services. The most common IT staffing model is one shared between more than one organization (30 out of 50), with some or all IT staff onsite in 24 of 50 key information organizations. Only 12 of 50 KI reported using virtual / hosted IT services. KI report that onsite IT staff in many cases are a critical asset and at risk of becoming non-existent because of budget demands. KI note a disconnect exists between the C-Suite leadership understanding the day to day needs for onsite IT staff and budget constraints. When onsite IT staff are not available health care settings are left to manage their own onsite health IT issues through telephone triage systems and remotein access with virtual IT staff. This practice limits onsite, in-setting problem solving often required for health care transformation. 14

#N/A 1 Virtual / Hosted 12 Shared 30 Other 1 Onsite 24 0 5 10 15 20 25 30 35 Figure 7: Health IT Staffing Models Used by KI Organizations Knowing how staffing resources are structured is valuable information when determining approaches for design, planning and implementation of new information technology capabilities. Many third-party services are used to managed complexity of IT needs in KI organizations. Some KI reported one or more types of third-party services (SBQ3). 28 out of 50 KI report using some type of report writing application to meet quality reporting requirements and 22 of 50 KI reported analytics or business intelligence software, citing many of these services as being contracted. 26 out of 50 KI report Direct Secure Messaging (DSM) capabilities through a health information service provider (HISP), though most organizations don t use the HIE capability. Only 4 out of 50 KI reported health information exchange (HIE) capabilities beyond DSM. When probed on what HIE capabilities were available, most cited the ability to send and receive a CCD. Figure 8 shows meaningful use attestation (SBQ2). Forty-eight percent (48%) of KI report successful attestation of MU stage 2 or updated MU stage 2 requirements. Eighteen percent (18%) report successful attestation of MU stage 3 requirements. Four percent (4%) report successful attestation of MU stage 1 requirements. Thirty percent (30%) report that MU is not applicable to their setting. Observations found technical capabilities like sending and receiving CCDs, DSM integration and alert notifications most often fall into the some-functionality category. KI report low practice integration and no workflow redesign as the two biggest issues. Observations show that even when technical capabilities exist, staff don t know if the information is received, reviewed, or used by the receiver. Other KI report that information received in document architectures like the CCD don t provide value in the clinical setting. A perception of low value may decrease adoption and use. 15

25 23 20 15 10 15 10 5 0 2 N/A Stage 1 Stage 2 Stage 3 Figure 8: Meaningful Use Attestation Levels The 35 KI organizations that participated in Meaningful Use / MIPS requirements for information exchange reported varying levels of technical capabilities. E-prescribing capabilities are reported by 34 out of 35 KI organization. Patient access (e.g., portal) is reported by 32 of 35 KI organizations. On average sixty-eight percent (68%) of KI interviewed report some level of information exchange capabilities based on MU / MIPS requirements in 2017 (SBQ5). Though KI report successful attestation for MU / MIPS requirements for Information Exchange, KI organizations are in many stages of implementation. KI were asked to report MU / MIPS HIE capabilities they were planning to implement, and data collected shows the same response pattern (SBQ6). KI indicate that some MU requirements were enabled for a short time and then turned off. Others note that practice level adoption is poor, so capabilities may be available, but not used. Observations indicate that settings attested to MU criteria for data exchange, but implementation may not be complete and in many cases HIE concepts are misunderstood. For instance, interviews and observation data show that KI repeatedly confused encrypted email with direct secure messaging. This is a knowledge gap and may cause data security issues, indicating that technical assistance and education are necessary. Likewise, many KI reported view only access through portals as a MU compliant work around to data exchange. Though portal technology is part of MU for patient access, this capability is confused with provider portals used in health care settings to view patient data in another health system s EHR. The common practice when using provider portals is for the viewer to locate pertinent data and then manually enter it into the local information system, creating source of truth issues and potential for human error. Other KI report they transfer CCD via pdf to providers who are capable of exchanging. KI widely note (interviews and observations) that CCD information set was inadequate in most settings. To supplement, health IT in many cases have added data fields to the CCD or created other work around documents (e.g., summary of care or progress notes). 16

The 35 KI working at organizations that attested to Meaningful Use report several types of barriers, including: Competing Priorities (57%), Funding (31%), Skills / Competency Level (17%), Technology (e.g., EHR) (20%) and Workforce Capacity (17%). Only 4 of 35 that attested to MU report no barriers (SBQ7). Workforce capacity 6 Technology [EHR, other] 7 Skills / competency level 6 NONE 4 N/A 18 Funding 11 Competing priorities 20 0 5 10 15 20 25 Figure 9: Barriers related to MU Attestation Information Sharing The goal of information exchange is to make health care data available at the point of care, when and where it is needed to improve care delivery. Figure 10 shows the types of internal providers KI organization staff need to communicate with to ensure continuity of care (Figure 9 SCQ1). Nursing (41/50), licensed providers (35/50) and support staff (35/50) being the types of internal providers most communicated with in the practice setting. Interestingly, nurses (staff, case management and care coordinators) make up the largest group of internal providers needing communication on key patient information. Support staff Rehab staff [OT, PT, ST] Nursing staff [LPN, RN, APRN] Mid-Level care providers [APRN, PA] Mental health [PSYCH, PSYCHIATRIC] Licensed providers [MD, DO] Case Manager [TCM, DM, CCM, ACT] Care coordinator 27 29 28 29 31 35 35 41 0 5 10 15 20 25 30 35 40 45 Figure 10: Types of Internal Providers KI Org Staff Communicate with for Continuity of Care KI report different types of external providers they need to communicate with to ensure continuity of care (Figure 11 SCQ2). 47 of 50 KI report licensed providers (MD, DO) as the most frequent recipients of external communication. This may indicate that communication outside a health care organization relies on key treatment level information. Additionally, KI report needing to communicate with external providers that include mid-level care providers 17

(35out of 50), nursing staff (34 of 50), mental health providers (34 of 50), care coordinators (32 out of 50), support staff (31 of 50), rehabilitation staff (29 of 50), case manager (25 of 50), dental (24 of 50) and hospitals (1out of 50). KI noted that external information is critical to care and safety. For example, one long-term care has clients transported to and from health care appointments by family members and care cabs. On more than one occasion patients have returned to the facility, but staff are not being notified of the client return. Having notifications would provide insight that staff don t have today and would support planning for change in client disposition and increased safety. Another KI described trying to provide in-home care to clients, but not having any communication from primary care outside of the typical physician order set. This KI noted that lack of communication close to real time, in some cases impedes care and delays treatment. Support staff Rehab staff [OT, PT, ST] Nursing staff [LPN, RN, APRN] Mid-Level care providers [APRN, PA] Mental health [PSYCH, PSYCHIATRIC] Licensed providers [MD, DO] Hospital Dental Case Manager Care coordinator 1 31 29 34 35 34 24 25 32 47 0 10 20 30 40 50 Figure 11: Types of External Providers KI Orgs Communicate with for Continuity of Care To coordinate care, KI report primary care (46 of 50) and specialty care (39 of 50) as the most frequent setting they need to communicate with (Figure 12 SCQ3). More than half of KI report Home care (29 out of 50), long term care (25 of 50), mental / behavioral health (25 out of 50), hospice (24 of 50) were identified as important for care coordination communication. Chemical Dependency (19 of 50), women s health (19 of 50) and inpatient rehab (12 of 50) were identified as important settings for communication in 33% of KI interviewed. However, the ability to communicate using electronic exchange varies. One KI on Epic determined that external electronic exchange isn t an interoperability problem. Instead, the KI reported it is an issue created because receiving organizations have not configured their information systems to meet Epic standards for data integration. Observations show that hospitals (acute care and critical access) were farther along the information exchange capabilities maturity scale than long-term care, home care and some behavioral health facilities. But hospital KI report that sharing data outside their electronic records is still difficult. 18

To bridge the gap, data exchange use cases may need to include value for the full complement of care settings as the ecosystem matures. 50 45 40 35 30 25 20 15 10 5 0 4 19 9 29 24 4 12 25 25 3 17 46 4 39 19 Figure 12: Setting Types the KI Orgs Need to Communicate with to Coordinate Care Interview and observation data show that communicating with external providers (those outside their organization, network and EHR) using electronic information exchange is a problem (Figure 13 SCQ4). KI report limited or no electronic exchange with primary care (45of 50), home care (40 of 50), specialty care (39 out of 50), long term care (39 of 50), mental health (37 out of 50), inpatient rehab (36 of 50), public health (36 of 50), palliative care (35 of 50), dental (34 out of 50), chemical dependency 33 of 50), alternative therapies (32 of 50) and Indian health services (28 of 50). Observations show that complex work arounds are used to complete otherwise simple tasks because information exchange capabilities are seldom fully operationalized. KI report the need to decrease administrative burden and increase communication needed for care coordination. How they manage the two divergent goals differs. For instance, some KI report using webbased mailboxes where files are transferred in pdf format. It is unclear if this description of information sharing uses Direct Secure Messaging or encrypted email, because observation data shows both concepts are misunderstood. KI report many barriers to information exchange. Fifty-five percent of KI report feasibility (29 of 50) as a barrier. Feasibility may include technical, operational, organizational capacity, sustainability, etc. Observation data supports the notion that technical feasibility is a problem. For example, KI report that information exchange capabilities are implemented, but observations show a complex system of work-arounds persist to accomplish information sharing. Likewise, cost (23 out of 50), competing priorities (23 of 50) and workforce capacity (22 of 50) were reported as barriers to information exchange by 44% of KI interviewed. 19

Observations showed staff knowledge gaps are also a barrier to information exchange needed to coordinate care. For example, one hospital has staff trained to use exchange mechanisms only during the typical work week. Weekend staff don t have training, so patient information transfer waits until Monday. Women s health Specialty care [TYPE] Public health Primary care Palliative care Other Mental / Behavioral Long-term care Inpatient rehabilitation Indian health services Hospice Home care Dental Chemical dependency Alternative therapies 3 28 34 39 36 35 37 39 36 36 40 34 33 32 45 0 10 20 30 40 50 Figure 13: Setting Types that KI Orgs Cannot Communicate with Using Electronic Data Exchange Generally, KI report protected health information (PHI) moves between organizations and settings that are on the same electronic health record. KI responses show that outside an organization s electronic health record, PHI is shared mostly through portal access (view-only) and paper mechanisms. Observations confirm these findings. KI report phone (49 of 50) and fax (48 of 50) are the methods most used by their organizations to receive referrals from internal and external providers (Figure 14 - SCQ7). Only 10% of KI (5 out of 50) report Direct Secure Messaging is used to receive referrals from internal and external providers. Observations show those that report having Direct Secure Messaging often confuse it with encrypted email, putting secure in the subject line of the message. Additionally, observations show that few organizations close the loop on referrals. KI report that even if they have Direct Secure Messaging, they don t have a local directory of Direct addresses, so it is difficult to use the electronic mechanism in practice. KI report that providers have multiple Direct addresses depending on which organization they work in and many providers have privileges at multiple health care entities. Subsequently, efax has taken a primary role in the transfer of health information. 20

Phone Other Fax Encrypted email Email Direct messaging 2 5 8 11 49 48 0 10 20 30 40 50 60 Figure 14: Methods Used by KI Orgs to Receive Internal and External Referrals The majority of KI report that information is not received electronically from external providers. KI that receive electronic health information from external providers report that information as medical records (14 of 50), lab results (10 of 50), diagnostic imaging (7 out of 50), consults (7 out of 50) and advanced directives (pdf version) (1 of 50). Observations show very little electronic data enters an organization from external sources. Even CCD samples reviewed showed incomplete information available on the aggregated record, and when received parsing capabilities are limited. Low value is a deterrent to adoption of information technology. Report Encrypted email 10 Phone call 40 Fax 43 CCDA from hospital 9 ADT alert 5 0 10 20 30 40 50 Figure 15: Methods used to Receive Information for Admission, Discharge and Transfer Events Interview data shows that ten percent of KI (5 of 10) report that Admission, Discharge and Transfer (ADT) alert notifications are used in their organizations (Figure 16 - SCQ7). Phone calls (40 of 50) and fax (43 of 50) are reported by more than 83% of KI interviewed to receive patient disposition information. Additionally, 20% of KI (10 of 50) report encrypted email is used for sending information. When KI were asked about Admission, Discharge and Transfer (ADT) notifications, a few KI reported that ADTs from IHIN are in their data stream and used by health care entities. 21

However, KI also reported that when health care entities are on the same EHR, ADT notifications are available but not always operationalized at the practice level. Interview and observation data show that most organizations obtain Admission, Discharge and Transfer data through manual processes (portal access, electronic flat file or faxed report). KI report that these manual processes take an exorbitant amount of time, human resources and information lags, rendering it unusable in some cases. Several KI reported that IT labor hours preclude any value Care Coordinators may derive from the information set available in an ADT, so implementation of the electronic notifications have not been completed. Both return on investment and knowledge deficits must be considered to ensure success and decrease low value perceptions. All KI report receiving some form of paper-based health information related to medical records (Figure 16 SCQ9). The majority of KI report that paper-based health information includes consults (46 of 50), lab results (45 of 50), advanced directives (41 of 50) and diagnostic imaging (39 of 50). Observations confirmed this finding. KI have limited technical capabilities, training, and insufficient implementation in many cases. None 35 Medical records 14 Laboratory results 10 Diagnostic imaging 7 Consults 8 Advanced directives 1 0 5 10 15 20 25 30 35 40 Figure 16: Types of Electronic Health Information Received from External Entities More than fifty percent (50%) of KI (29 of 50) report no health care settings using paper health care records in their region (Figure 17 SCQ10). Those KI that report paper health care records being used in their region include the following settings: Long term care (11/50), mental / behavioral (8/10), specialty care (6/10), home care (2/50). Public health, primary care, and palliative care organizations were reported by at least one KI as being still on paper records. Observation data shows that long term care settings lag in adoption of EHRs and health IT. KI informants provided information showing regional adoption patterns in northwestern Iowa for long term care may lag the rest of the state. Lags in adoption deter community exchange and impact care quality and safety. 22

Women s health Specialty care [TYPE] Public health Primary care Palliative care Other NONE Mental / Behavioral Long-term care Home care Alternative therapies 1 1 1 1 2 1 3 6 8 11 0 5 10 15 20 25 30 35 29 Figure 17: Health Care Setting in your Region Still Using Paper Health Records Information Use Understanding how information is used is important. Use cases drive the values of health care information exchange ecosystems. KI reported very few use information exchange use cases enabled in their organizations. Transitions of care is an information exchange use case. KI report examples of information used for transitions of care in case management (31 of 50), hospital readmission prevention (30 of 50), communication between settings (24 of 50), and the identification of high risk patients (19 of 50). A low number of KI report that ADT alerts with a C-CDA (4 of 50) and ADT alerts only (2 of 50) are an information type used for transition of care. Twenty-four percent (24%) of KI (12 of 50) report no ADTs information is available for care transitions (Figure 18 SDQ1). KI settings that would like to participate in a transitions of care use case include long term care, home care, case management and patient care attendant services. Transitional case management NO ADTs Identification of high risk patients Decrease hospital readmission Communicate between settings ADT Alerts only ADT Alerts + CCDA 2 4 12 19 24 31 30 0 10 20 30 40 Figure 18: Examples of Information Used to Inform Care Transitions Care coordination has several information exchange use cases. Seventy percent (70%) of KI (35 of 50) report that patient information is used to inform care coordination (Figure 19 SDQ2). 23

Examples of how that information is used are related to improvement of post-acute outcomes (35 of 50), enabling social services (25 of 50), managing the care between the care (24 of 50), disease management (19 of 50), case management (19 of 50) and care integration (19 of 50). Observations show that care coordination is an integral service provided by many health care settings. KI report that care coordination is accomplished by a wide variety of staff (professional and non-licensed). That means that when implemented information exchange must span a continuum of technical literacy levels. Referral management 21 None 4 Manage care-between-care 24 Improve post-acute care outcomes 35 Enable social services 25 Disease Management Program Case Management Program Care integration 19 19 19 ADT Alerts + CCDA 9 0 5 10 15 20 25 30 35 40 Figure 19: Examples of Information Use to Inform Care Coordination Care coordination is supported by multiple use cases including; closed loop referral, CCD and ADT alert. KI reported care coordinators use mostly manual processes and receive very little real-time or close-to real-time information on clients they manage. This makes care coordination difficult. KI provided examples of delayed information patterns and the workarounds they deploy to keep up with increasingly difficult workloads. These patterns of delayed information were described as reactive and not proactive. Observation data confirmed this finding. KI report a desire to use sophisticated tools, but worry that staff don t possess the technical literacy skills necessary to successfully operationalize new tools. Having user-defined tools accompanied by education and technical assistance will be important. Quality metrics are widely adopted and quickly becoming information exchange use cases. KI report a variety of quality metric examples. KI participate in payer incentive programs (32 of 50), experience of care (30 of 50), benchmark performance (29 of 50), federal programs (27 of 50), state or national registries (25 of 50) and national registries based on topic (18 of 50). Only twenty-four percent of KI (12 of 50) reported not participating in quality metric reporting (Figure 20 SDQ3). Observations showed that quality metric reporting is accomplished through mostly manual processes in smaller organizations and a hybrid approach (manual / automated) in larger organizations or those in more heavily populated regions of Iowa. For those 24