|
PERSPECTIVERegulatory And Policy Barriers To Effective Clinical Data Exchange: Lessons Learned From MedsInfo-ED
MedsInfo-ED is a proof-of-concept clinical data exchange project that uses prescription claims data to deliver patient medication history to emergency department clinicians at the point of care. This patient safety initiative, while limited in scope and scale, has been crucial in identifying numerous policy and regulatory barriers to successful clinical data exchange. The lessons learned and strategies to overcome the barriers are the focus of this paper. Through commitment and effective collaboration, MA-SHARE was able to address some of these barriers that are embedded in existing government regulations and corporate business practice.
The need to deploy automated clinical information systems in hospitals and ambulatory care settings has recently been identified as a top priority. At the same time, it has become clearer that to optimize the use of electronically stored clinical data, connectivity among health care entities must be established to facilitate clinical data sharing at the point of care.1 Several notable efforts have begun to demonstrate the feasibility and utility of clinical connectivity across sites of clinical care.2 In particular, because of the nature of the patient population served and the frequent lack of up-to-date medical records at the point of care, hospital emergency departments (EDs) and their patients may benefit greatly from access to clinical data stored at other care sites. Among the clinical data of greatest value to clinicians is a patient’s medication history.3 Background: MedsInfo-ED project. Given the ubiquitous and timely availability of dispensed drug history information through prescription claims databases, we initiated the MedsInfo-ED project, a proof-of-concept patient safety initiative to automate communication of medication history to ED clinicians. The goal of the project is to serve as a prototype for building and testing a community clinical data exchange model that uses available technology and operational solutions for improved quality and patient safety outcomes. The MedsInfo-ED project is sponsored by the Alliance for Health Care Improvement, a collaborative of six Massachusetts-based health plans. The project is managed under the auspices of MA-SHARE (Massachusetts Simplifying Healthcare Among Regional Entities) LLC, the regional health information organization for Massachusetts. MA-SHARE was created by the Massachusetts Health Data Consortium (MHDC), a collaborative founded in 1978 by the state’s major public and private health organizations.4 MedsInfo-ED system architecture. Data sources for the initial MedsInfo-ED pilot project include Blue Cross Blue Shield of Massachusetts; Harvard Pilgrim Health Care; Tufts Health Plan; Neighborhood Health Plan; Massachusetts Medicaid (MassHealth); and the Massachusetts Group Insurance Commission (GIC, purchaser of health insurance for Commonwealth of Massachusetts employees and retirees). Overall, these health plans cover nearly 5 million of the 6.3 million people in Massachusetts. The initial pilot hospitals are Beth Israel Deaconess Medical Center, an urban teaching hospital affiliated with Harvard Medical School; Boston Medical Center, an urban, safety-net, teaching hospital affiliated with Boston University Medical School; and Emerson Hospital, a full-service suburban community hospital. Zix Corporation (ZixCorp) developed the MedsInfo-ED application and provides ongoing technical support. Computer Sciences Corporation (CSC) provided technical coordination on behalf of MA-SHARE. RxHub consolidated access to two pharmacy benefit managers (PBMs) representing three of the participating health plans and the GIC through its master patient index (MPI).5 The payer data sources connect to MedsInfo-ED using either a direct connection via a Web service query to the payer’s prescription claims database (Harvard Pilgrim Health Care, MassHealth) or a PBM consolidator model with RxHub (Blue Cross Blue Shield, Neighborhood Health Plan, Tufts Health Plan, GIC). Pilot hospitals are supported by two log-on security models: a Web-based authentication with the users hosted by ZixCorp; or passthrough authentication, with the respective hospital maintaining users within its own ED information system. Searches for patients and dispensed drug history occur simultaneously across each data source. Patients are located using the MPI through a multipass, probabilistic logic algorithm search. MPI logic provides the opportunity to find a patient without having an exact letter or digit match on the search criteria. The search engine uses soundex (White versus Whyte, Jim versus James) and data transposition (02345 versus 02435) to identify close matches, assigning confidence levels to score the match. MedsInfo-ED uses MPI matching both at the ZixCorp application level and from the RxHub consolidator. The MPI works as follows: (1) The required five search criteria fields (first and last names, sex, date of birth, and ZIP code) are entered on the MedsInfo-ED patient search screen; (2) the ZixCorp search engine queries the RxHub MPI for data from the RxHub-affiliated PBMs as well as checking the directly connected data sources; (3) if data are found at multiple sources, they are aggregated for presentation; and (4) MedsInfo-ED displays the prescription history found across the various data sources.
MedsInfo-ED introduces a new electronic process into the ED intake workflow. Since one of the objectives of this pilot is to evaluate the system as it might be used in various settings, the actual workflow design was left up to the participating EDs. The degree of technology integration with existing systems and the type of staff interacting with the system vary. Designated ED staff (registrar, nurse, or physician) log on to the MedsInfo-ED system, obtain patients’ permission to query the system, and then query the MedsInfo-ED application. The user can choose to receive data from the past four months, past year, or all available. The user also can limit the initial display to a single line for each drug, regardless of multiple dispensing dates. The displayed drug history can be printed for incorporation into the patient’s chart. A sample screen shot of a typical dispensed drug history returned by the system is available as an online data supplement.6 Although the system’s final version has been operational for only a few months, early experience indicates that MedsInfo-ED is working as intended. During the first few months, pilot hospitals have been adjusting their work flow and assessing the optimal approach to using the system. In all three hospitals, searches are done selectively based on the patient’s clinical situation, assessed potential value of the information, or likelihood that data will be available from participating data sources given a patient’s insurance coverage. The selected work flow differs among the pilot sites. One hospital elected to have the intake coordinators undertake the MedsInfo-ED search as part of the registration process. The search results (or information that no results were available) are printed automatically in the ED record room and appended to the patient’s chart before he or she is seen by the physician. Another hospital, possessing a highly automated ED with electronic records and physicians accustomed to accessing patient information electronically, chose a process in which physicians used the MedsInfo-ED search system directly instead of using a printout. Subsequently, this hospital has expanded access and use to the ED nursing staff as perhaps a more efficient approach. The third hospital designed a process in which the nursing staff were the primary users of the system. Subsequently, because of questions of efficiency, this hospital chose to adjust the process and move the MedsInfo-ED search upstream to the intake coordinators. To date, searches have been initiated for approximately 10 percent of ED patients, but this rate varies widely across pilot sites primarily because of differences in the mix of patients’ insurance coverage. About 4 percent of patients have declined permission. Sixty percent of searches locate the patient in one or more of the data sources, and 77 percent of those located have dispensed drug history information available and returned. Thus, 46 percent of searches returned prescription information relevant to the care of the patient.7
During all phases of the MedsInfo-ED project—from conceptualization, through technology systems architecture design and construction, to operational implementation and stabilization—a number of hurdles related to privacy rules, security regulations, and data availability and standards had to be addressed. In particular, fairly stringent Massachusetts regulations related to the release of data by health plans regarding certain sensitive conditions posed important barriers. However, according to the Healthcare Privacy Project, nearly all states have statutes that impose similar condition-specific privacy requirements, often as stringent as or more stringent than those in Massachusetts.8 Thus, the issues we discuss here are likely applicable in many other states. Privacy and security issues. MA-SHARE convened a Privacy and Security Workgroup in November 2003, leveraging the efforts of the existing MHDC forums for privacy and security officers, which represented payers and providers, government agencies, and the office of the state’s attorney general. Rapid progress in addressing issues occurred because the health care community unanimously supported improved access to electronic prescription medication history in the ED. Even with the help of this committed group of collaborators, however, several barriers posed by privacy and security rules proved problematic. HIPAA barriers to application testing and user acceptance validation. Testing software used across entities can be challenging under the best of circumstances. Add the specter of relatively new Health Insurance Portability and Accountability Act (HIPAA) regulations and HIPAA risk-averse trading partners, and the challenges required considerable extra effort to overcome. For MedsInfo-ED regression and end-to-end testing, as well as validation of user acceptance, MA-SHARE needed the cooperation of the six data sources, participating PBMs, and RxHub to permit the pilot hospitals to validate that MedsInfo-ED worked as designed in the production environment.9 The PBMs and RxHub were concerned that allowing pilot hospitals to validate MedsInfo-ED using test cases before "going live" violated HIPAA. To address their concerns, MA-SHARE engaged its legal counsel, Nutter McClennen and Fish LLP, to clarify that validating clinical applications, conducting quality assessment and improvement activities, and auditing functions constitute health care operations that are allowable by HIPAA’s health care operation exemption. Obtaining sufficient real test cases that would confirm that MedsInfo-ED returned correct eligibility and medication history presented another testing and validation challenge that required substantial effort to overcome. Although the health plans authorized MA-SHARE and ZixCorp staff to use selected test cases for deployment testing, they could not approve disclosure of their members’ protected health information directly to pilot hospitals. Each pilot hospital had to identify its own test cases by asking staff for volunteers who authorized the hospital’s MedsInfo-ED administrator to search for their medication history. The volunteers validated the returned results. To minimize private health information exposure, for MA-SHARE staff validation, ZixCorp deidentified the patients in each of the health plan–provided test cases, which enabled a full user acceptance test without compromising patients’ privacy. Massachusetts Fair Information Practices Act (FIPA). Massachusetts state law is stricter than HIPAA with respect to consent. HIPAA allows covered entities participating in the MedsInfo-ED pilot to disclose medication history, in most circumstances, for treatment, payment, or health care operations without the need for consent. The Massachusetts attorney general’s office took the position that under the state’s FIPA, which applies to the GIC and Massachusetts Medicaid, providers are obligated to obtain and document oral consent to use MedsInfo-ED. Further, if a patient is mentally incompetent and a MedsInfo-ED–dispensed drug history inquiry is performed (with a clinician override to patient notification documented), when the patient’s condition improves, the hospital must inform the patient that MedsInfo-ED was accessed. In taking the FIPA requirements into account, the Privacy and Security Workgroup considered and determined the value of informing all patients when private health information is to be accessed. The MedsInfo-ED user interface screen captures and records the patient’s verbal agreement to allow the search, even though the medication history is being sought for treatment. Further, if the patient’s condition (such as an emergency) precludes agreement, the physician may record a clinical override and proceed to query MedsInfo-ED for medication history; the system incorporates a FIPA reminder on the results screen. Regulations regarding sensitive medications. Massachusetts insurance regulations require health plans to obtain advance written consent for the release of any AIDS-related information, and a specific detailed consent form is mandated. This requirement covers medications specific to combating HIV and prescriptions for antibiotics or other medications in dosages that are used only to treat AIDs-related conditions.10 Two additional Massachusetts statutes specifically prevent Blue Cross Blue Shield (BCBS) of Massachusetts and commercial indemnity plans from releasing mental health information without advance written consent of the subscriber.11 BCBS of Massachusetts applies these statutes to all of its products, even though there is no such requirement for its Massachusetts health maintenance organization (HMO) products. Thus, these statutes prohibit the disclosure of all mental health medications, such as antidepressants, anti-anxiety agents, and antipsychotic medications, without additional patient consent. The consent requirement itself is not problematic; these laws fulfill an important purpose by protecting patients’ privacy. It is how the consent must be obtained that is problematic to a clinical connectivity initiative such as MedsInfo-ED: It is on a carrier-specific consent form and must occur prior to release of information. Given the "real-time" release of information by MedsInfo-ED, complying with these laws in the ED was seen as almost impossible. Similar to these state regulations, the use and disclosure of medication information collected by federally funded substance abuse treatment programs, and regulated by 42 CFR, Section 2, also had to be blocked from a patient’s MedsInfo-ED–dispensed medication history display, because these medications indicate substance abuse treatment and can be found in the data source/payer prescription claims databases. To address these regulatory barriers, MA-SHARE worked with the data sources to establish a sensitive drug list of more than 150 medications that could indicate the treatment of HIV/AIDS, mental health disorders, or substance abuse. The resulting drug filter, added to and applied by MedsInfo-ED to prevent the display of sensitive drug information, varies by health plan (as described above), since some of the statutes apply to specific insurers or products, and the statutes were interpreted differently by the health plans. A single filtered drug list had been a goal but would have resulted in defining a large "lowest common denominator" list, preventing the release of medication history that would be clinically useful. ED clinicians expressed concern that the sensitive drug filter would limit the usefulness of MedsInfo-ED. Specifically, they noted that information regarding mental health medications is both critical to medical decision making and often difficult to ascertain accurately from patients. Furthermore, MedsInfo-ED does not alert the clinician that a sensitive drug was filtered, because this would violate the spirit, if not the letter, of the regulations. To quantify the impact of the filtered drugs, a preliminary analysis of one of the health plan’s prescription claims data for a commercial population was requested; results indicate that approximately 5 percent of its members seen in an ED in 2004 had a previous prescription for one or more of the filtered drugs. The percentage is likely higher for the Medicaid and Medicare populations. Patient identification. Impact of HIPAA’s "minimum necessary" provision. MedsInfo-ED patient eligibility and dispensed drug history searches occur simultaneously across each data source. Searching across multiple data sources required a more stringent interpretation of HIPAA’s "minimum necessary" provision. The Privacy and Security Workgroup determined that MedsInfo-ED should not return multiple patient records from which the user would select the correct patient. Instead, MedsInfo-ED should alert the user to enter additional search criteria (such as a health plan ID number) to narrow the search to one patient. In contrast, hospital information systems might allow users to enter nonspecific search criteria (for example, name equals J. Smith) and typically will return multiple similar matching patient records, such as John Smith, Joe Smith, and Johnny Smithwick. Although ED staff understood the MedsInfo-ED privacy rationale, they expressed concern that limiting the inquiry results to one probable patient potentially reduces the likelihood of finding patients and their medication history. Impact of HIPAA electronic data standards. MedsInfo-ED encountered a difference in how RxHub implemented the HIPAA eligibility and eligibility transaction. Referred to as the ASC X12N 270 standard eligibility inquiry, this transaction poses the question: "Is the patient eligible for benefits on a specific date of service?" In reply, the ASC X12N 271 standard eligibility response transaction answers in effect "yes" or "no" and may provide additional benefit information. RxHub’s interpretation was to return the patient’s name, birth date, sex, and ZIP code exactly as these had been entered by the user, even if the PBM claims database contained different data. After identification of this issue, ED staff raised questions and concerns about the potential to return false positives. Briefly, during testing, users discovered that a patient’s name intentionally misspelled could be found with the MPI logic and return corresponding medication history. However, the patient identification information originally sent in the ASC X12N 270 standard eligibility inquiry was returned in the ASC X12N 271 standard eligibility response and displayed "as is," even when that information differed from the information in the PBM database and, importantly, without a validation alert to the ED. The users, viewing the "wrong" (misspelled) name but the "correct" medication history, were concerned about the veracity of the information presented by MedsInfo-ED. This difference in standards interpretation affected providers’ acceptance of MedsInfo-ED because of patient care and safety concerns. The pilot hospitals needed assurance that they had accurately identified the patient before relying on the medication history information for treatment. While to the best of MA-SHARE’s knowledge, MedsInfo-ED has never returned a false positive (that is, has never incorrectly identified a patient as "found"), the eligibility standards interpretation differences had to be managed. MA-SHARE initiated two activities: submitted an enhancement request to RxHub to return demographics different from the PBM in its ASC X12N 271 implementation; and required MedsInfo-ED to display some type of flag that alerted ED staff to further validate the patient to the information provided. MA-SHARE, working with RxHub, initiated a workaround that uses existing data content from the returned ASC X12N 271 optional data field to display the cardholder’s name in addition to the name displayed "as entered." With the two names presented on the results screen, ED staff are able to compare the information and validate that it is for the correct patient. Additionally, RxHub’s PBM participants have accepted the MA-SHARE enhancement request to its ASC X12N 270/271 implementation interpretation and are in the planning phase of developing a product. Data limitations. As noted earlier, approximately five million lives are covered collectively by MedsInfo-ED’s six data sources. However, medication history is not available to MedsInfo-ED for several critical populations. Fee-for-service Medicare patients. Lack of medication history for the fee-for-service (FFS) Medicare population was very disappointing for the MedsInfo-ED pilot hospitals. On average, patients covered by Medicare are prescribed and take more medications than younger, healthier populations. ED clinicians noted that a reliable prescription history is most difficult to obtain from elderly patients and is also most valuable in this population. PBM "carve-outs." Some employers have chosen to contract with a different PBM than that used by their health plans. For health plans participating in MedsInfo-ED, this carve-out population represents about 12 percent of the collective health plan membership. In this situation, the medication history is not available from an employer-contracted PBM. In some cases, access to medication history could be achieved through negotiated agreement between MA-SHARE and an employer, followed by agreement between the employer and its PBM. In other cases, in addition to the contracting issues, the MedsInfo-ED technical connection to a new PBM data source must be established. Each scenario begins to push beyond the scope of the initial pilot. Uninsured patients and self-pay medications. The source for MedsInfo-ED medication history is prescription drug claims. If patients pay out of pocket for a prescription, over-the-counter medication, or herbal product, no drug claims will exist to be accessed by MedsInfo-ED. For one of the urban pilot hospitals, uninsured patients represent more than one-third of its ED patients, and thus lack of information for this population poses a major limitation. MA-SHARE is exploring accessing prescription data directly from the pharmacies, which would also address the lack of Medicare prescription data. This solution pushes beyond the scope of the initial pilot. Terminated members. Citing privacy concerns, five of the six data sources do not return information for terminated members, even though the Privacy and Security Workgroup was comfortable with MedsInfo-ED releasing a former member’s medication history for treatment purposes. Further, any disclosure to a former health plan via MedsInfo-ED audit reports for a terminated member was viewed as an incidental disclosure, and within the acceptable range for use and disclosure. However, the understandable PBM business practice of establishing an effective date (to and from) before processing a prescription created a barrier to obtaining medication information for MedsInfo-ED inquiries on terminated members. This data gap is large, given that on average nearly one-third of employers switched health plans in 2004.12 The inability of health plans and PBMs to modify their business practices for clinical connectivity information purposes, in the near term, prevents access to many patients’ complete medication history. Variable data currency and timeliness. Each health plan’s data source refreshes its drug claims file on a different schedule. Prescription history updates can occur as fast as the same day or as slowly as two weeks to three months from the date the prescription was filled. The length of time that data are retained also varies, ranging from twelve to twenty-four months. Changes to either data timeliness or data retention periods will require policy changes by the data sources or standards set at the federal or state level. Missing data elements. The various transaction standards (for example, eligibility request and response, dispensed drug history) allow for optional data fields. Consequently, data sources may or may not store and transmit certain optional data elements consistently. For example, the "prescriber" data element was not returned in 40 percent of cases. Although this did not present a major barrier for MedsInfo-ED, as long as there are optional data elements in electronic data interchange (EDI) standards, the completeness of data will always be inconsistent, which could be problematic for other clinical data exchange implementations.
Health plans/payers and their business partners (such as employers and PBMs) are new players in clinical data exchange. A new strategic alignment is needed among payers, employers, and providers regarding policies concerning data access, currency, and retention. The known data limitations have inhibited clinicians’ adoption of MedsInfo-ED. Although clinicians confirm the value of having more information about their patients’ medication history at the point of care, many are disappointed with a low match rate and the need to filter sensitive medications. Two of the pilot projects are reconsidering having registrars, instead of clinicians, conduct the MedsInfo-ED inquiry to increase utilization compliance and reduce clinicians’ frustration. MA-SHARE is also actively exploring additional data sources. Regulatory barriers can be harder to overcome. Privacy and security issues can be resolved with regionally empowered privacy and security workgroups. Reaching consensus among regional stakeholders regarding reasonable and appropriate privacy and security practices is critical. In certain situations, regulatory amendments may be required. In Massachusetts, after consultation with the MHDC member organizations, a bill that obviates the need for payers to obtain a special consent to disclose medication history for treatment purposes for HIV/AIDS and mental health conditions was drafted and filed on behalf of MA-SHARE with the Massachusetts legislature. MedsInfo-ED is a clinical data exchange pilot with relatively limited scope. Nonetheless, we encountered numerous barriers to maximizing value to clinicians and patients. Through commitment and effective collaboration we were able to overcome some of the barriers, which are embedded in existing government regulations and corporate business practices. As larger clinical data exchange projects—such as the Centers for Medicare and Medicaid Services (CMS) e-prescribing initiative—are undertaken in Massachusetts and elsewhere, it will be critical to communicate effectively among public- and private-sector stakeholders to identify and effectively address these barriers. Status reports like this as well as more broadly based collaboratives such as the eHealth Initiative and Connecting for Health will be important mechanisms for such communication.
Lawrence Gottlieb (l.gottlieb{at}comcast.net) is executive director of the Alliance for Health Care Improvement, a collaboration of Massachusetts health plans. Until his death in April 2005, Elliot Stone was executive director and chief executive officer of the Massachusetts Health Data Consortium in Waltham. Diane Stone is project manager of the MA-SHARE MedsInfo-ED Pilot Project in Waltham. Lynne Dunbrack is product manager, Community Utility Services, at MA-SHARE. John Calladine is principal consultant at CSC Consulting, also in Waltham. The MedsInfo-ED project was made possible, in part, by Grant no. 1D1BTM00095-01 from the Office for the Advancement of Telehealth, Health Resources and Services Administration, U.S. Department of Health and Human Services, to the Foundation for eHealth Initiative. This work was supported in part by a grant to MA-SHARE from the Robert Wood Johnson Foundation for MedsInfo-ED project evaluation. The authors acknowledge the following people for their leadership of critical aspects of the MedsInfo-ED system design and implementation: Meg Aranow (Boston Medical Center, or BMC), Gary Austin (Blue Cross Blue Shield of Massachusetts), Joseph Bergen (Emerson Hospital), Jim Feldman (BMC), Mark Gingrich (RxHub), Leon Goldman (Beth Israel Deaconess Medical Center), Mick Kowitz (ZixCorp), Pamela Muccilli (Emerson Hospital), Larry Nathanson (BIDMC), Deborah Stevens (Tufts Health Plan), Jim Stowe (Massachusetts Health Data Consortium), David Szabo (Nutter McClennen and Fish), and Bruce Wilkinson (ZixCorp).
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
System Use In Pilot...
Lessons Learned
