June 25, 2007
MPI: The Glue Holding HIE Together
By Laura Gater
For The Record
Vol. 19 No. 13 P. 22
Many insiders view the master patient index as the essential component of any project that involves data sharing.
An enterprise master person index (EMPI), or a master patient index, makes use of sophisticated algorithms to allow for patient identification and link to a patient’s electronic medical record (EMR). Basically, it provides a connection to the correct patient record in organizations that utilize EMRs. An MPI’s accuracy can be attributed to its probabilistic matching techniques that use any available fields for matching despite discrepancies in how various healthcare facilities track information.
“An MPI is the cornerstone of any data-sharing initiative,” explains Beth Just, MBA, RHIA, the CEO and president of Just Associates, Inc. “Very simplistically, it’s the list of all records in that database. Without a list to say that a specific record is in the database, we have no way of knowing what patients are included. An EMR provides the clinical information about a patient, while the MPI is the index for that data. An MPI typically lists data points about a patient, such as a patient’s last name, first name, date of birth, gender, address, phone number, and dates and types of visits to the healthcare organization.”
Keeping It Together
For regional health information organizations (RHIOs) or a national health information network (NHIN) to be successful, they need to have an MPI to locate and link to the correct patient record. They use MPIs to “keep track of information kept in their data warehouse,” explains Victoria Wheatley, vice president of strategic services at Quadramed. The information listed in the MPI provides the institution with the ability to confirm that a particular patient is the correct person. Other information that may be listed in an MPI include Social Security numbers, medical record numbers or patient identification numbers, allergies, physician information, and clinical problems, according to Wheatley.
A community of hospitals—no matter the size—that chooses to share data and utilize an MPI will have to work out a means of sharing and data standards, such as what information will be listed in the MPI.
“The more information a facility stores on its MPI, the more likely it is to have good matches to its records,” says Wheatley. “A common model for an MPI is that of a record locator service. RHIOs position themselves to take that role to track patients, and all RHIOs rely on the concept of an MPI—a repository of necessary information.”
An MPI can link a patient to the correct EMR in a case where there is no common denominator, according to Troy Harrison, vice president of enterprise architecture at Medseek. “All providers will feed the MPI demographic information about patients, and the MPI will determine which patient information matches. Then it can be queried to determine where records are located,” he says. “It determines what patients match the query. An MPI is a very important piece of the solution, and its role is analogous between all three environments—a RHIO, an NHIN, and a community of hospitals. The MPI is the glue that ties all the information together.”
MPIs are also valuable in the case of a patient who visits different hospital emergency departments (EDs) in an attempt to obtain narcotics prescriptions. If their MPI includes data such as previous visits and clinical problems, as many do, ED providers will be notified of previous attempts when they enter the patient’s name into the computer upon check-in.
A particularly critical role of an MPI is to help medical providers determine if they have the correct patient—in the case of a patient who may have checked into her physician’s office as Jane Smith, into the laboratory as Jane E. Smith, and into the ED as J. E. Smith, all on various occasions over a period of time. Some healthcare organizations, such as Physicians Medical Group in Santa Cruz, Calif., have specially trained personnel responsible for verifying and merging the data of duplicate patients to eliminate future confusion.
MPIs and Patient Safety
“You have to have a strong baseline of patient information in the MPI in order to have an accurate EMR,” says Cynthia S. Hyde, RHIA, chief information officer (CIO) and chief security officer at Providence Hospital in Mobile, Ala. “Our MPI is a strong back-end reporting tool [that monitors] everything that happens.”
Providence Hospital shares information with its sister hospital in Pensacola, Fla., and the two often collaborate on sharing their databases for their senior health program and patient care efficiencies. An MPI makes this joint venture possible because it provides the link between the databases and EMRs.
Providence had several goals for utilizing an MPI. One was to ensure patient registration took less than two minutes. The software it uses has also helped eliminate duplicate patient information. By 2008, the hospital hopes to have no unexpected mortalities, and the MPI is playing a major role in this effort by maximizing patient safety. If patient records are accurate, up-to-date, and matched with the correct patient, Hyde says medication errors can be reduced and other patient safety factors can be minimized, all through maximizing technology.
Established in the 1980s, EMPIs were one of the earliest patient safety initiatives, according to Lorraine Fernandes, senior vice president of healthcare practice at Initiate Systems. Their usage increased as healthcare organizations merged and grew larger and, thus, needed an efficacious method to match patient names with the correct data. The MPI, at source level, has been around for decades.
“EMPI software was used by other types of industries first, and I think healthcare drove its use because of patient safety, the need for correct patient identification, and the importance of recognizing patients when they come in,” explains Fernandes. “In the commercial world, the driver for the use of an EMPI was customer satisfaction. For example, customer call centers use an EMPI to retrieve and manage data on each call and thus know who’s calling.”
CareSpark’s MPI is a “kind of backup safety net for physicians,” says Liesa Jenkins, executive director of CareSpark, a nonprofit organization that is developing a secure health information network in the central Appalachian region. For example, the automation features enable physicians to focus on discriminating factors between symptoms and diagnoses. When physicians review a patient’s test results, they are presented with reminders of best practice recommendations, another safety feature that benefits physicians and patients.
Data Sharing Cornerstones
The MPI is the “key piece” for RHIOs and other data-sharing networks, according to Sheri Stoltenberg, CEO of Stoltenberg Consulting, Inc., because the technology promotes data integrity by allowing healthcare providers to access the correct patient information. In the case of data sharing across a server or network, if the MPI data is not clean and complete, then it isn’t worthwhile because there may be doubts about its accuracy, she says.
“It should be a prerequisite that inconsistencies and inaccuracies in an MPI should be deleted before any data is shared, whether in a RHIO or on a local or national level,” says Stoltenberg.
Salim Kizaraly, product manager at California-based InterComponentWare, Inc., acknowledges that MPI content may vary widely from vendor to vendor. He points out that an MPI’s role is to aggregate data and help medical providers uniquely identify the data and link it with the correct patient.
“Multiply that role by the number of providers that care for a single patient over the years, such as clinicians, specialists, laboratories, radiology and imaging centers, emergency rooms, and pharmacies, and you can see why an MPI is so important for reconciling all of this scattered information,” he says.
Kizaraly believes an MPI is particularly useful in an NHIN for large-scale health studies, disease registries, national health-related situations such as disasters, and cases where an MPI can help providers locate people or compare demographic data.
“An MPI provides for the optimization of healthcare by using technology to help us identify the at-risk patients and help physicians collaborate electronically on the care of patients across a network,” explains Dick Thompson, executive director of Colorado’s Quality Health Network. “Having one method of identifying a patient, regardless of where that patient might present, is the advantage of linking clinical information. There is no single identifier to use, so patient ID is the key to improving care.”
The MPI is the foundation of Physicians Medical Group, which provides care for approximately 280,000 patients across two hospitals, four labs, a radiology facility, a county healthcare organization, and 100 provider offices—a total of roughly 700 users. Each entity in the group enters data into the MPI in real time or on a nightly basis. The MPI is the key for health information exchange (HIE) in order for physicians to make matches to the correct patient.
“We find that the MPI works very well,” says Bill Beighe, the group’s CIO. “Our offices have a variety of paper and electronic records. Some are paperless, some aren’t, but they all use the MPI efficiently. Our MPI has the ability to identify many patient identifiers using a minimum number of matching fields, including a patient identifier number, allowing for a more accurate match.”
RHIOs have different data-sharing models, according to Just, and not many are involved in what she calls “full-blown” clinical data sharing, such as with Michiana Health Information Network. Some RHIO providers or participants have a central database with patient-level data, while others have what is known as a record locator service, which contains MPI data that links the patient data across the RHIO participants (such as hospitals, clinics, physician providers, etc). Just, a member of the AHIMA’s work group for HIM principles in HIE, says one of the group’s tasks is to develop more definitive descriptions of different data-sharing models because the idea of sharing data among competing providers is new to healthcare.
CareSpark wanted as little burden as possible placed on patients and physicians while keeping software add-ons to a minimum, yet it needed a viable solution to match patients with the correct medical record to maintain patient privacy and safety. According to Jenkins, an MPI was the perfect solution, one that would impact users’ functionality. Surveys of the organization’s physicians revealed they wanted information pulled up at a maximum of three mouse clicks. The MPI helps them achieve this goal by checking multiple determinants to ensure that the correct patient information is retrieved.
“We are confident that our MPI has a very accurate level of discriminating among patients,” says Jenkins. “We wanted a machine doing the matching and discriminating rather than a person doing it by hand because it protects patient privacy under HIPAA rules.”
On a National Level
Wheatley says if the United States eventually has a single national healthcare network, an MPI will be needed to ensure accurate patient identification. In this instance, the NHIN will be more of a series of RHIOs and community networks held together by an MPI. “The national concept works best in a place like Canada, which has a national health plan. The concept of creating a national identifier is pretty daunting,” she says.
As the idea of an NHIN evolves, Jenkins believes there will be a need to look at healthcare data on a global scale—for biosurveillance, for example. “It’s obviously going to be beneficial to work with organizations already involved in information sharing,” she says. “It’s important to be in alignment with technologies that are emerging in healthcare.”
According to Stoltenberg, MPIs need to be accurate—or “cleaned up”—before they will be of value in an NHIN data-sharing initiative. “People are so busy trying to get to the end step, the EMR, of a national initiative, that they are skipping other steps, like ensuring the accuracy and completeness of information contained in MPIs,” she says. “The information in MPIs won’t be useful to clinicians anywhere at all if it’s not accurate.”
— Laura Gater’s medical and business trade articles have been published in Healthcare Traveler, Radiology Today, Corrections Forum, Credit Union BUSINESS, and other national and online publications.