Session 12 Client Registries Identifying and Linking Patients
Client Registries: Identifying and Linking Patients PERTEMUAN Ke - 10 Taufik Rendi Anggara., MT Manajemen Informasi Kesehatan Fakultas Ilmu Ilmu Kesehatan
LEARNING OBJECTIVES
- Defne a client registry and describe why such registries are needed in health information exchange.
- Detail common strategies for implementing a client registry.
• Discuss common challenges encountered when implementing
a client registry.- Highlight the critical role a unique identifer plays in implementing a client registry.
- Distinguish between the common methods of patient matching.
PATIENT IDENTIFIERS
EHR systems manage records. Much of the information maintained in EHR systems pertains to patients—or clients—who receive health care services from a clinic or hospital. In order to efectively manage information about patients, EHR systems must assign and use patient identifers, such as a medical record number (MRN), when performing tasks such as creating a new record, updating an existing record, or deleting a record. Identifers (or IDs) ensure that the correct record is added, updated, or deleted. Patient identifers can be thought of as belonging to two general classes: (1) a unique code or set of codes specifcally designed to uniquely identify a patient in a given system and (2) an aggregate set of demographic and related attributes used to describe a patient uniquely, such as name, sex, date of birth, address, etc. Whether utilizing existing or creating new patient identifers, it is prudent to understand the context in which the identifers will be deployed.
UNIQUE PATIENT IDENTIFIERS
Strategy for Assigning UPIs
A UPI requires a sequence allocation sufciently large to cover an entire population over time, theoretically for as long as the number will be in use. Sequencing schemes available for development of a UPI generally fall into three numbering systems: serial, derived, and composite.
- In serial numbering systems, each individual is assigned a number from a central location. These numbers are automated and do not assimilate any nonunique characteristics of the individual. England’s National Health Service number is an example of serial numbering system, with some added functionality.
- As the name suggests, derived numbering systems create a number based on, or derived from, a personal trait of the individual. In contrast to serial numbering, assignment of a derived number can take place anywhere but runs the risk of failing to be unique when derived from a personal trait which is shared by other individuals.
- Composite numbering systems assign part of the number from a central location and the other part is derived from personal traits, thus represents a combination
Cont.
Attributes of Ideal Identifers
Much of the concern around utilizing a universal identifer for health care, particularly in the United States, is fear of malicious intent in the event of a data breach. The only true way to prevent the possibility of confdentiality breaches is to exclude personal information; such as name, social security number (SSN), date of birth, etc., and to eliminate the possibility of linking the health care identifer to databases which contain personal information. At the time of writing, and for the foreseeable future, no identifer exists which will perfectly ascertain all individuals uniquely across all scopes of health care; at least not one that is neither physically invasive (eg, implanted device) nor invasive of a citizen’s right to privacy. Health care identifers are often assigned within individual information systems (eg, laboratory, radiology) as well as organization-wide (eg, MRN). However, as discussed, these identifers are meaningless outside of the domain in which they operate. One solution to this identity challenge is to establish a unique identifer that crosses organizational boundaries. The American Society for Testing and Materials (ASTM), which is a standards development organization accredited by the American National Standards Institute (ANSI), describes 30 criteria which should be used to evaluate the efcacy of a candidate identifer.
Cont.
Attributes of Ideal Identifers (cont)
Meeting all of the proposed criteria would lead to a UPI that achieves
the following:- positively identifes patients.
•automatically links and collates patient records from disparate
electronic sources, creating a longitudinal care record,- protects patient’s personal health information and privacy, •efectively minimizes the cost of patient record management.
No identifer exists which could meet all of the criteria proposed by the
standard because some ideal attributes confict with each other. For
example, a ubiquitous and easily accessible identifer may not
adequately preserve privacy. The following is a description of selected
ideal UPI attributes; for succinct defnitions and examples, refer to
Cont.
Attributes of Ideal Identifers (cont)
Meeting all of the proposed criteria would lead to a UPI that achieves
the following:- positively identifes patients.
•automatically links and collates patient records from disparate
electronic sources, creating a longitudinal care record,- protects patient’s personal health information and privacy, •efectively minimizes the cost of patient record management.
No identifer exists which could meet all of the criteria proposed by the
standard because some ideal attributes confict with each other. For
example, a ubiquitous and easily accessible identifer may not
adequately preserve privacy. The following is a description of selected
ideal UPI attributes; for succinct defnitions and examples, refer to
Cont.
Attributes of Ideal Identifers (cont)
Cont.
Attributes of Ideal Identifers (cont)
- A unique identifer, by defnition, can never be associated with more than one individual. That is, once assigned, the possibility of another person being assigned the same number must be eliminated, or infnitely minuscule.
- A ubiquitous identifer is available and accepted across the health care spectrum. For example, a nonubiquitous identifer would identify a patient for a hospitalization but not the subsequent primary care visit. Ubiquity also requires the identifer to be durable and made readily available at the time of service.
- Unchanging. In order for an individual identifer to be efective, every individual should have an identifer that applies only to that individual and does not change over time. This requires foresight on the part of the issuing agency because enough numbers must be generated to support the population throughout the life span of the identifer.
Cont.
Attributes of Ideal Identifers (cont)
•Uncontroversial. The identifer should help minimize the opportunities
for crime and abuse and should not contain substantive information
about the individual. Similarly, the various stakeholders must perceive
the identifer to be minimally invasive. The subjectivity of what is and is
not invasive makes universal acceptance difcult, if not impossible.•Uncomplicated. An identifer or identifer system that is not practical to
implement or that does not meet the requirements of administrative
simplifcation must be deemed unacceptable.•Inexpensive. The costs of implementation and use of the identifer
must be within an acceptable range. Analysis of costs across all health
care settings should be considered including, patients, providers,
payers, and government agencies. For example, it has been estimated
that a national health identifer implemented in the United States would
cost between $4.9 and $12.2 billion to deploy and $1.5 billion per year,
which many view as unsustainable
Cont.
Existing Unique Patient Identifers
The Health Insurance Portability Accountability Act of 1996 (HIPAA)
recognized the need to uniquely identify patients for managing care and
administrative purposes, thus proposing a unique health identifer for all
individuals. UPIs have been implemented in several international health
care systems but have yet to come to fruition in the United States for a
number of reasons: a funding embargo imposed by Congress; costs to
implement such a system; privacy concerns; difculties in conforming
existing information systems to utilize the identifer; and a lack of
consensus on selection of an identifer. The interested reader is referred to
a 1998 Health and Human Services white paper, developed prior to a
congressional ban on government funded UPI discussion, which discusses
relevant legislation, privacy and confdentiality concerns, as well as
strengths and weaknesses of several proposed implementations of a
national UPI. The following is a discussion centered on commonly
broached UPIs—SSN, biometric identifers, and a voluntary universal
Cont.
Social Security Number
The SSN has been advanced as a candidate for a UPI in the United States because it is theoretically unique, ubiquitous, unchanging, most adults can recite it from memory, and would not require additional infrastructure to implement (ie, inexpensive and uncomplicated). However, in reality, the SSN is sometimes shared by multiple individuals, not all people are eligible for an SSN, in rare circumstances an individual may possess more than one SSN, SSNs are not universally available at birth, and there is no legal protection for maintaining SSN confdentiality in nongovernment organizations.
Using the SSN also comes with controversy, largely as a result of its universality. At its inception, the SSN was for use only within the context of the Social Security program and intended to identify an account, not a person. Today, the SSN has become tightly linked to an individual’s credit score, among other things, and is now largely
Cont.
Biometric Identifers
Biometric identifers such as fngerprints, iris scans, voice recognition, and facial shapes ofer unique, ubiquitous, and relatively unchanging identifers. In developed nations, implementation of biometrics has been scarce, mainly because of concerns with privacy and the potential for law enforcement to use these data. However, underdeveloped nations often lack universal coverage by civil registration systems leading to an identifcation problem that reaches far beyond health care. Recent technological advances are allowing cheaper, more accurate identifcation using biometric indicators such as fngerprints and iris scans. In a report for the Center for Global Development, Gelb and Clark found that 160 countries have deployed biometrics to address the identify gap covering over 1 billion people. While highly discriminatory, biometric identifers do not eliminate the need for sophisticated matching algorithms to uniquely identify an individual. Measurements of a person’s fngerprint or iris will possess variability within persons, due to changes in age, environment, disease, stress, and occupational factors, as well as variability due to hardware (eg, sensor calibration and age of the device). In the case of fngerprints, any single individual can produce several distinct images depending on the angle of depression, pressure, presence of dirt, moisture, and sensor characteristics.
Cont.
Voluntary Universal Healthcare Identifer (VUHID)
An example of VUHID was deployed by Global Patient Identifers, Inc. (GPII)
and attempts to completely eliminate patient matching errors by augmenting
existing probabilistic matching algorithms with a VUHID. The patient may
request one open voluntary identifer (OVID), which as the name suggests
shares all information with all providers, and multiple private voluntary
identifers (PVID) that will only disclose information to patient-selected
providers. A key design element for the VUHID proposed by GPII is that once a
patient requests to participate through their provider, they are assigned a
unique identifer that, unlike the SSN, is used only to identify them in the
health care setting. The VUHID then routes the unique identifer back through
the HIE CR to be associated with the patient’s records. The unique identifer
(implemented by barcode, magnetic strip, smart card, etc.) is given to the
patient and presented when visiting a provider. Drawbacks of the proposed
VUHID include the complexity introduced by the PVID, which assumes that
patients will understand when it is clinically important to share information
from one provider with another, and concerns that the failure to share all
records may impede hospitals from producing complete billing records.
Cont.
International Unique Patient Identifers
UPIs have been developed and implemented in low-, middle-, and high-income countries.
Details on the purpose, formats, validation methods, technical architectures, and lessons
learned from UPI implementations in England, Newfoundland and Labrador, Australia, New
Zealand, and Germany may be found in a review from 2010 by the Health Information and
Quality Authority. Specifc use cases for implementing a UPI in Denmark, Botswana, Kenya,
Brazil, Malawi, Ukraine, Thailand, and Zambia are discussed in a UNAIDS white paper
addressing developing and using individual identifers for health services including HIV .
Although no single identifer currently exists with each of the ideal attributes, Figure 11.1
depicts the degree to which the examples discussed meet the expectations of an ideal
identifer. As will be discussed, a key strength of the CR is the ability to utilize multiple
identifers through matching algorithms to positively identify a patient. That is, the CR does
not rely on the necessity to produce an ideal UPI; however, when a UPI is available,
matching efciency is improved. UPIs hold promise for simplifying patient identifcation, but
even in the case a UPI is in use, there is a need to match patients using other attributes,
such as demographics, when a UPI is unavailable at the time of service (eg, forgot card or
emergency situations). Similarly, if a temporary identifcation must be issued, the central
authority will need to utilize other identifers, such as demographics, to link a patient’s UPI
to the temporary ID to prevent record fragmentation. In general, the implementation of a
UPI will improve the accuracy of matching procedures but will not replace the need for
demographic attributes.FIGURE 11.1 The ideal attributes of unique personal identifers, highlighting where commonly used
identifers exist within this framework. UPI, unique
patient identifer; SSN, social security number.THE ENTERPRISE MASTER PATIENT
INDEX
Each health care organization’s eMPI associates its own
unique identifer with patient data; thus, a single patient may
have several “unique” eMPI identifers, one within each
organization where care occurred. The lack of a shared
identifer across organizations makes integration of patient
data difcult. So if Bob visits an ED outside his usual
integrated health care system, data about the visit would not
be integrated into the EHR system record utilized by his
primary care physician and, unless Bob discloses this
information, his primary care physician may never be aware of
the incident. The unique identifer problem has an intuitive
solution that creates a unique identifer to span all health care
organizations.THE CLIENT REGISTRY
The eMPI uniquely identifed and linked each of Bob’s specialty care visits because
they were within the same network or enterprise, but data from an ED visit outside
Bob’s usual integrated health care system would remain in a silo. A natural, logical
solution might be the use of an UPI that could span all of Bob’s health care services
and allow for positive identifcation and linkage of Bob’s ED visit to his other health
care data. However, no single UPI is currently available that would facilitate linking
all of Bob’s records. Thus, in order to link data across the health system, we must
rely upon an assemblage of existing identifers. A CR operates by adjudicating the
assemblage of demographic attributes and other personal identifers from each data
supplier (eg, hospitals, clinics, etc.) to create a single record for each patient within
the HIE. Figure 11.2 demonstrates how the CR fts into the scheme of the HIE model.
Organizational EHR systems and point-of-care applications interact with the CR
through the interoperability layer allowing users to retrieve, add, and edit patient
records. Once positive identifcation is made, the CR associates each patient’s
records with a unique number called the master patient index (MPI). Although similar
in concept to the unique number assigned by an eMPI, the MPI—in the context of the
CR—acts as the controller for all other local identifers that may be associated with
the patient, including one or more MRNs assigned by individual facilities andFIGURE 11.2 The client registry architecture within the OpenHIE model. EMR, electronicmedical record; EHR, electronic health record.
SUMMARY
Health care data are generated from numerous clinical sites,
often by diferent EHR systems. Each organization has its own
method of uniquely identifying patients, but these identifers
are incapable of identifying patient records from outside
organizations. Currently, no single unique patient identifer is
capable of spanning the entire health care environment, and
record linkage must rely upon multiple patient attributes and
identifers. The CR handles inconsistent completeness and
quality of identifying data by standardizing incoming data and
employing statistical matching algorithms to link data from
separate organizations for the same patient. Once identifed, a
single, unique identifer (the MPI) is associated with the
patient, allowing HIE networks to create and maintain patient-
centric records and services.