As described in previous sections, the site visit team observed that while in some instances computers enabled remote access to health information there was no "clinical data interoperation," that is, we did not observe any instance of semantic interoperability that would enable computer-to-computer data exchange and re-use of clinical content between non-affiliated sites. No site reached Level 4 in Table 2 for HIE for any of the four clinical areas that were examined during the course of this study. As previously mentioned, there also were no instances where information was exchanged using both agreed upon messaging and content HIT standards.
To frame our specific observations regarding interoperation (or lack thereof), firm definitions of the relevant concepts in this report are outlined below, because use of these concepts in current health care discussions is so highly variable.
1. Health Information Exchange (HIE)
As described throughout this report, HIE applies to any mode of transmission of HIE--voice, paper, fax, or digital--about a given patient. The broader definition of HIE (rather than considering only electronic (i.e., digital) exchange) for several reasons is used. Not only were instances observed where patients' health information was exchanged using more traditional (voice, paper and fax) methods and in some case standardized (but not semantically interoperable) e-HIE, but it also was observed that HIE was often a hybrid of the more traditional and digital methods. There were instances when the information exchanged resided in a computer on one or both ends of the exchange, but the actual exchange of data was done via phone or fax. For example, site visitors heard a description of a discharge planner viewing her computer screen while talking on the phone to an admitting planner typing things into her computer during an initial conversation regarding placement of a patient. More frequently, it was explained, the person receiving the voice (usually telephonic) information will be taking handwritten notes, and later these notes are transcribed into the local computing system.
Representatives from the selected sites provided the following reasons for not more fully utilizing their HIE capacity.
2. Interoperation and Interoperability
Over the last few years, several efforts have been aimed at developing a universal (beyond health care) definition of interoperation and interoperability. (In this report, the terms will be considered synonymous.) The Presidential Executive Order of August 2006 (E.O. 13410) defines interoperability as "the ability to communicate and exchange data accurately, effectively, securely, and consistently with different information technology systems, software applications, and networks in various settings, and exchange data such that clinical or operational purpose and meaning of the data are preserved and unaltered.19 The ability to exchange data and retain the meaning of such data across different HIT systems and software applications requires the use of agreed up HIT messaging and content standards.
As noted, a variety of methods of HIE were observed during our site visits and in the rare case that HIE was automated such exchange was not standardized. HIE that is automated, but not standards-based, is by definition ad hoc though from a local point of view it is no less useful. For example, the occasional pre-population of records in one system from information in another system, as was observed at Montefiore Medical Center (New York), was always seen as desirable in the few places it was observed, and it was desired in many contexts where it did not yet exist (e.g., as was reported by Erickson). Only rarely did automated (i.e., electronic) data exchange make use of HL7 Version 2 (HL7v2) messaging standards, although intra-enterprise use of such messaging was observed (e.g., at LDS Hospital in Utah where HL7v2 is used extensively). Some sites planned to make use of HL7v2 messages as the basis for inter-enterprise interoperation. No where was it observed that the use of "semantic" standards, wherein information in one computer was transmitted to a computer in a different care enterprise in a way that the meaning, for the sending computer, was retained in the receiving computer. For example, one can imagine an electronic representation of a discharge medication list from a hospital being "understood" by a computer at a PAC site (in the same way that a local medication list would be understood, and, say, checked for drug interactions), but we did not observe such "exchange by meaning." Instead, universally, we saw manual (i.e., human) "medication reconciliation" at receiving sites.
While the site visit team believes that the observed lack of deployment of data, terminology, and messaging standards may result for deep reasons not yet fully understood, the following hypotheses based on the sites visited are offered. First, knowledge of such standards is limited at the sites visited; there is little awareness of messaging standards and no unilateral mention of terminology standards. Further, there seems to be little motivation to acquire this knowledge. The simplest explanation for this first hypothesis is that "anywhere, anytime" access of patient information that follows the patient is a multi-enterprise, multi-disciplinary objective that all stakeholders can understand, and therefore imagine. The idea that computer-empowerment of one computer by another should be part of such access is just not something that the visited sites tended to think of first, and this is in spite of the burden manual re-entry imposes. Second, at sites where inter-enterprise interoperation based on standard messaging is contemplated, the information technology staff are focused on acquiring the resources required for the translation of the patient information, which might be only a summary, into HL7v2 and the translation of the resulting HL7v2 message back into something the receiving computer can (at least) display. Often the business case for paying for the resources necessary to translate patient information in a standardized way is not yet well articulated and the required resources are not always easy to procure. Third, the notion that terms (e.g., the names of diagnoses, laboratory tests, medications, procedures and the like and their meanings (are to be shared across enterprise boundaries seems to cause concerns and sometimes anxiety among some providers. Often these concerns focus on the fact that terms are used differently at different but neighboring enterprises (i.e., the same term may mean something different or different terms are used to name the same meanings). Shared, local use of language may support local care and providers cannot imagine using standardized terms while retaining the way they are used to doing things. A more sophisticated view of the potential use of standard language is that the latter will (locally) reduce the quality of care.
The only way to overcome the challenges implicit in these hypotheses may be to gain experience with terminology standards while simultaneously reducing the technical and financial barriers impeding their deployment and use. It may be for instance that actual use is not nearly as difficult as it is imagined to be.
The lack of agreed upon HIT standards needed by PAC and LTC providers was observed to be a particular challenge. At the time of the site visits, while the NCVHS had endorsed the CHI Patient Assessment and Disability standards,20 the Secretary of HHS had not yet endorsed the use of these standards. PAC/LTC providers are understandably reluctant to demand/request standards-based EHR systems of the PAC/LTC vendors when such standards are not recognized. However, the endorsement by the Secretary in August 2007 of these standards and his approval that these standards be used by federal agencies in implementing new and to the extent possible in modifying existing health information systems should help address providers' uncertainty about which standards should be included in their EHR applications. The commitment by CMS to include CHI standards in the development of the new patient assessment CARE instrument also could help mitigate provider concerns.
Similarly, the recently balloted and approved HL7 CCD standard (i.e., that supports the standardized exchange of human-readable and/or coded documents, such as transfer/discharge documents) and endorsement by the HIT Standards Panel (HITSP) of the CCD as an exchange standard should help reduce provider's uncertainties about the exchange standard that software should support. Further, the expected endorsement of this standard by the American Health Information Community (AHIC) should work to further reduce providers' risks in this area. The CCD could be something that any vendor can understand and any of the vendor's customers can imagine using. This standard specifies a way of exchanging information about a patient so that the receiver can use standard software to display, process, and store that information.
Further, the decision by the CCHIT to specify the certification criteria for NH EHRs by 2009 and develop a roadmap by which future certification criteria will be adopted should assist both NH providers and HIT vendors in anticipating and planning for needed EHR functionality. The CCHIT NH EHR certification criteria likely will include use of the "Patient Assessment and Disability Standards" endorsed through CHI and the HL7 CCD information exchange standard endorsed by HITSP.
While it is unrealistic to assume that the adoption of the CHI Patient Assessment and Disability standards and the HL7 CCD HIE standard will change anything quickly, formalization and use of these standards could promote realistic planning and resource allocation on the part of vendors and providers alike. Providers in communities that were visited could plan to implement these standards in some incremental way as soon as resources could be allocated. For example, Erickson and St. Agnes Hospital (Maryland) could ask their vendors to implement the CCD standard so as to support HIE for patients treated at these two health settings. The fact that implementation of the standard may be a resource-intensive and burdensome undertaking for the vendors could potentially be ameliorated in proportion to the vendor's market share.
3. World Wide Web
The most powerful example of information technology use today is the World Wide Web. Its use is pervasive and so successful in many arenas such as retail commerce and banking that health care providers (and to a lesser extent patients) are beginning to have expectations for their health care experience that rival these other interactions.21 When care providers are asked what they want from HIT, they often express their wishes in web-based analogies; which for the purposes of this paper has been reduced to personalized "anytime/anywhere" access. However, health care providers (and patients) also want other attributes that come to mind when they think of this immediate access, namely autonomous, maintenance-free, low-cost, system and vendor-independent communication. However, while the web often is used as the model against which other frameworks are compared, the site visitors observed only one new and one planned use of web-based technology. Significantly, they both involved new technology investments. As part of the IHIE, Docs4Docs is deploying remote web-based physician access to local patient encounter information. At Hillside Rehabilitation Center (Utah NH), their ongoing deployment of the BlueStep EHR will take advantage of web-accessible backup for the patient information they accumulate. One important use of this backup will be for "disaster-recovery" (i.e., by having the patients' data off-site, in the event of a disaster, care providers will have intact medical records from which to provide care). Another explanation why only two sites showcased their use of the web may be that the web is pervasive enough to be an assumed tool. Therefore, some sites may be planning use of the web for data exchange but did not make mention of it, and in other circumstances those interviewed may not have been aware that their system makes use of the web.
4. Communication Gaps Across Organizations
One of the biggest challenges observed during this study was the lack of awareness of HIE-related opportunities going on outside the doors of each enterprise. These challenges presented themselves as both technological and organizational issues. As HHS Secretary Leavitt observes, "All health care is local."22 As most organizational energy in most provider enterprises is allocated to existing problems within their own organization, the idea that answers, or, more likely, partial solutions may exist "out there" is only rarely considered. For example, the site visit team observed many instances where organizations unilaterally decided on what information a receiving care provider needed in the event of a patient transfer, without any discussion between the two organizations about the type of information that both organizations actually needed to provide care for transferred patients. Similarly local, unilateral attempts at local HIT solutions often are pursued without any awareness/understanding of the efforts of others in the health care community to address the same or similar problems. Instead, providers tend to look outside their organization for solutions (if they are considered at all) when the solution become overwhelming in size and complexity and only then do enterprises look to partner with nearby providers or otherwise enlarge their search for solutions. Because of this, substantive progress, however partial, in one place is often completely unknown even to nearby providers. Conversely, because trust among the parties exchanging data about a given patient is a major requirement for collective action, and it generally was observed that e-HIE only occurred among regional providers who know one another well or have an agreement with one another to be able to cooperate and share solutions. This happens in spite of the fact that some providers (e.g., Erickson) are committed as a matter of organizational objective to sharing their experiences with other providers. The simple exercise of the site visit team asking the various settings how data are (or are not) transferred illuminated the wide communication gap. Presumably, communication about the fact that HIE is possible needs to precede planning for HIE, as it is hard to implement HIE solutions without thinking what problems one would like to solve.
5. Health Information Exchange Network (HIEN) Involvement
Three of the four site visits were in communities with some type of RHIO/HIEN. One was a fairly established regional network (IHIE), one was established within a state (UHIN), and one was a newly-formed RHIO (Bronx RHIO). The ability for these networks to exchange information was idiosyncratic to the location visited. The UHIN largely exists because of state-regulated requirements regarding claims submission and credentialing of physicians. The IHIE was established over time and spearheaded by a few forward-thinking physicians and informaticists. It was through Drs. McDonald and Overhage's good will and perseverance that the IHIE was able to get initial buy-in with the participating hospitals and physicians. The Bronx RHIO is just getting started, and has been heavily funded by both the state (NY HEAL money) and the Federal Government.
6. Site-Specific Technology
In the remainder of this section, each site will be reviewed against a number of emerging criteria for interoperation readiness, or lack thereof. For more details, see appendices corresponding to each site visit.
a. Erickson Retirement Communities: Hardware and Software. Erickson's main HIT priority has been the Erickson-wide (one or more facilities in each of ten states)23 rollout of GE Centricity functionality for "100% of Erickson physicians."24 While this functionality does not yet include CPOE, it does include web-based "anytime/anywhere" access for physicians. Currently, this deployment does not make use of CHI standards except as are required for reimbursement. However, internal standards are sufficient to create comparable data within the Erickson enterprise and these are being used to predict the incidence of falls, and they will be used to track the efficacy of interventions that attempt to prevent falls. Interestingly, given Erickson's commitment to an internal ethic of "enter once, read many times" (aimed at avoiding the re-entry of internal data already in an Erickson computer), it was surprising that no plans were described to exchange data between Centricity and CareMEDX, the primary record for patients in home health care, Part A SNF, and LTC.25 More predictably, as with most providers, neither system interoperates with external laboratories or pharmacies, although health care providers often are content to have the information they need available even if it is on two computer screens. Still, the fact that Centricity is deployed nationally by Erickson is a singular accomplishment, one from which significant benefits are already accruing. For instance, Erickson's information technology philosophy and systems seemed to contribute significantly to employee satisfaction and a trend toward staff retention.
At the time of the site visit, Erickson discussed its readiness to adopt HL7 messaging standards. They have initiated discussions with St. Agnes hospital (their primary, unaffiliated acute care hospital in Maryland) to deploy a custom data exchange solution with their two EHR systems. The site visit occurred before the CCD standard was an approved and widely available standard.
Summary: Erickson uses a national EHR that supports web-based access. This EHR does not interoperate with non-affiliated sites, although Erickson would like it to interoperate with their preferred local hospital as soon as possible, nor does it interoperate with the Erickson EHR product used by the Erickson PAC and LTC providers. They do not use HIT standards except as are required for reimbursement. Centricity makes use of Oracle to store patient information. Because of organizational priorities and experience and the relatively contemporary, highly scalable implementation of Centricity, Erickson is positioned to productively leverage any relevant national standard (e.g., a standard for patient data exchange). Erickson will probably be able to send and receive data using any emerging national exchange standard before many of the unaffiliated providers in the 16 communities in which Erickson is located are ready to receive or send health data.
b. Montefiore Medical Center: Hardware and Software. Montefiore uses LastWord as the hospital EHR, the exports from which periodically refresh a Sybase data repository. The Sybase supports general querying and aggregation of patient data, functions not available in the LastWord system. Montefiore is the anchor provider in the recently-funded Bronx RHIO. The RHIO will include other hospitals (e.g., the Bronx VA), some ambulatory care sites, and one PAC/rehabilitation/LTC nursing facility (JHHA). Other PAC/LTC sites may join soon, as will local laboratories and pharmacies. Data exchange in the RHIO will depend on a common computer-represented patient summary (which, at the time of the site visit was called a virtual patient object [VPO]). The latter will make use of dbMotion technology to manage secure access, among other things.26 A basic premise of the RHIO is that patient data remain stored at its source and only the patient summaries, in the form of a VPO, are transmitted to requesting providers. Although the Bronx RHIO is committed to the use of relevant standards, the developers have not yet finalized the list of standards they will incorporate. For example, they are not yet considering the use of the CCD. However, the VPO, dbMotion technology, and the Montefiore patient data repository, should allow for incorporation of any emerging national patient data exchange standard.
The VNS of New York exchanges patient data with Montefiore and other Bronx providers using traditional fax and phone methods. However, the VNS of New York has projects underway in which they are exchanging information--computer-to-computer--with physician office EHRs and with Cornell Weil Medical Center (discussed in Appendix C). The VNS of New York's long-standing innovative use of information technology enables it to leverage that technology as part of attempts to exchange data with other providers. The VNS of New York has achieved sufficient intra-enterprise interoperation so that it is now focusing its development efforts on data exchange with non-affiliated providers.
Summary: No non-affiliated interoperability was observed and no use of HIT standards was observed except those used for reimbursement. Montefiore Medical Center, the VNS of New York, and some Bronx RHIO sites are self-reportedly "poised for interoperability." The now-funded Bronx RHIO is well positioned to leverage any emerging national patient data exchange standards. Not coincidentally, the VNS of New York has several HIE pilots underway with non-Bronx providers. (see Appendix C for a discussion of the VNS pilot projects).
c. LDS Hospital: Hardware and Software. LDS Hospital continues to evolve its EHR, which today consists of a heterogeneous mix of internally-developed and commercial-off-the-shelf (COTS) components. The distinguishing features of the IHC system (including LDS Hospital) include a longitudinal data repository that is accumulating both inpatient and outpatient information on the IHC population, and an ongoing effort to make all the components interoperate internally using HL7v2 messaging. An IHC goal is to use its recent partnership with GE Healthcare to move all components, including the data repository, to COTS status. IHC has pioneered the use of HL7 messaging and the use of LOINC, a CHI standard, for use within its own enterprise in these intra-enterprise messages. Other uses of CHI standards are those required for reimbursement. LDS Hospital and almost all other providers in Utah make use of the UHIN for accelerated in-state billing and reimbursement. LDS Hospital will be participating in a pilot project aimed at use of the UHIN to communicate limited clinical information to and from other Utah health settings. No further details, such as whether there will be a UHIN Master Patient Index were provided to the site visit team.
All PAC/LTC sites visited in Utah had some HIT in place. One site, Hillside Rehabilitation Center, is installing a relatively advanced EHR system developed by BlueStep, which is entirely web-based and hosted remotely. Features include a focus on workflow management and web-accessible disaster-recovery backup. For further background information, see Appendix D.
Summary: Interoperability with unaffiliated sites was not observed, however all sites made use of the UHIN for claims submission, which is significant in and of itself. No use of CHI standards was observed, except for LDS Hospital's extensive use of HL7 messaging and LOINC for laboratory test result reporting. However, all sites could be poised to exchange clinical information with one another by building on their current use of the UHIN for financial transactions. Predictably, the first types of electronic clinical data exchange through UHIN will likely be laboratory results and medication ordering, which have standardized codes already used for payment.
d. Indiana Health Information Exchange/Indiana University: Hardware and Software. The Indiana University's EHR (Indiana Network for Patient Care) is a locally developed, deployed, and maintained system that is a pioneer system in the development and evaluation of physician order entry and decision support and the creation and use of longitudinal patient records (Overhage, Suico, & McDonald, 2001). Today, it makes use of commodity hardware to run legacy (MUMPS and Windows) software. More significantly, this system was one factor that enables the retrieval of previous encounter information from different hospitals by local EDs. This retrieval does not pre-populate the local system today, but pre-population is one planned feature of the Indianapolis-wide Docs4Docs system. This feature supports web-based physician access to available patient encounter information stored in the EHRs of Indianapolis hospitals by collecting this information in a regional data repository. An example provided during the site visit described a patient seen in one ED for chest pain who proved to have had a recent negative cardio-vascular work-up at another hospital. The patient was later discovered to have a pulmonary embolism and a second, costly cardio-vascular work-up was avoided because the ED physicians could see the (negative) results of the cardio-vascular work-up. Currently, this repository does not feature a uniform patient summary. The IHIE is an expanding umbrella that leverages these and other planned data exchange projects including the data to be collected from state-mandated e-prescribing. Thus, the goal of being able to retrieve all available information about a patient is being approached incrementally. The IHIE is open to all settings that would like to participate, and generally project leaders assist with the inclusion of each site. At the time of this report, the only members of the IHIE are local hospitals and physicians.
Summary: Non-affiliated, human-readable HIE exists on a systematic basis between Indianapolis EDs and selected local hospitals, including Wishard Hospital, which was visited by the site visitors. Indianapolis-wide physician access to available (i.e., data that are collected and stored for other reasons) patient encounter information is supported by the web-based Docs4Docs system, and plans are in place for historical information on the patient to also be entered into the participating providers' EHRs. CHI standards were not being used (other than those required for reimbursement) with the notable exception of the widespread use of LOINC to normalize laboratory report data and for widespread use of HL7v2 messages. The Beverly Healthcare Nursing Home at Brookview made use of relatively advanced national (Beverly) proprietary HIT that does not use HIT (e.g., CHI) standards. At present there are no plans for Beverly to join the emerging IHIE. The VNS of Indiana supports its home care nurses and other clinicians with a laptop-accessible EHR, which also does not use HIT (e.g., CHI) standards. The full site visit report can be found in Appendix E.
"HIEcase.pdf" (pdf, 1.68Mb)
"HIEcase-A.pdf" (pdf, 236.81Kb)
"HIEcase-B.pdf" (pdf, 115.31Kb)
"HIEcase-C.pdf" (pdf, 169.18Kb)
"HIEcase-D.pdf" (pdf, 134.56Kb)