An HIE is a Health Information Exchange in which various health information for patients can be stored and viewed electronically by various participants depending on the model used. Participants include patients, physicians, insurance companies, and other users authorized to view a patient’s information. There are three basic models of HIE: centralized, decentralized, and hybrid. A centralized, or consolidated, model is one in which participants that can upload and view information. It is a centrally located warehouse that will keep the information stored in a singular electronic storage location.
Authorized participants can access the information from an external location. This model provides the easiest use because information is centrally located. Also, because it is in a central location with many participators, community-wide queries are easier to obtain. One of the drawbacks to this model is that it depends on the participants for accurate information. There is a possibility of data submitted improperly or duplication of data. A decentralized, or distributed, model is one in which data is kept internally, at the point of service.
Those participating in this model are members of the organization in which the model is used and providers control their own data. However, they must agree to share the information they have with other organizations. This is one of the pros of this model in that security of records can be better assured. With this model, a “Record Locator Service” is necessary to find where data is stored and to request information from participants. This can also be a drawback for this model in that there can be any number in points of service.
Because this model is kept at the point of service, there can be variations of the model itself. A hybrid model contains a mixture of elements from both the centralized and decentralized model and is generally the model most used by participants. Generally, data from a centralized model is collected as part of a patient locator service and downloaded to a decentralized EHR system, making analytics easier. Additionally, HIT. gov, cites three other differing types of exchanges: directed, query-based, and consumer mediated.
Directed exchanges allow providers to exchange information strictly and securely between each other in order to effectively coordinate care. If a patient was referred by a primary care doctor to a specialist, that primary care doctor and specialist could freely and securely communicate data related to that patient’s care. Query-based exchanges allows for providers to request information for patients outside of their direct or referred care. For instance if a patient chose to change providers, the new provider could send a request records from the previous provider.
Consumer mediated exchanges put the patient directly in control of their medical record information. A patient portal available through an online EHR system would allow patients direct access to their records and allow them to share, track, and correct their own information as necessary. The integrity of information involves the dependability of its accuracy. The quality of information involves how correct it is. Therefore, data integrity and quality of an HIE involve the dependability of its accuracy. A critical element to Health Information Exchanges is patient identification.
Access to and transfer of correct data will assist in ensuring quality care for the patient. Record duplication poses another problem and hazard for quality patient care. The use of algorithms can assist organizations and exchanges in ensuring data integrity and quality. A basic algorithm will compare data based on matches of certain pieces of information such as name, birthdate, etc.
The “wild-card” is an example of a basic algorithm. For example, when searching for names using the wild card algorithm, the first three letters in a last name can be entered along with a pecified character, such as an asterisk, and the search will return any number of records attached to last names beginning with those three letters. With an intermediate algorithm, other unique identifiers are attached for a more complex search. These searches will yield slightly more specific results such as last names and first names that can be used as either/or (i. e. David, John, etc. ) along with names that have been shortened to nicknames attached to identical birthdates.
The most complex algorithm allowing the most specific results is the advanced algorithm, and generally involves artificially intelligent systems that are capable of learning. These searches yield results that are diversified. One search can return results in with variations in the spelling of names and dates of birth with the same social security numbers or addresses. In order to participate, it would be necessary to have policies and procedures to ensure the information gathered for any Health Information Exchange is reliably correct.
Because the most important information collected is that which identifies the patient, data integrity and quality should begin at the front-end in registration and/or scheduling where initial information is collected. For a decentralized model: Any new patient would need to complete a form with name (first, last, and middle initial), birthdate, social security number (or at minimum the last four digits), current address and phone number along with being assigned an additional unique identifier in the form of a patient number that would correspond with a patient number in the HIE.
Any returning patient would need to also complete a form updating the same information. The registration or scheduling clerk would need to ensure that information was correctly inserted or updated to the Health Information Exchange via data entry so any algorithm used to search for identifiers would reliably return the correct information. This form could also be uploaded to the organizational system so when information is questioned, the form can be relied upon as historical information per the patient.
When searching for patient information, a minimum of two identifiers, such as name and birthdate or name and social security number (or at minimum last four digits), should be used to correctly identify patients. If somehow the results yield more than one patient, one additional identifier should be used until the correct patient is completely identified. For a centralized model: It would be necessary to ensure the information used to identify any one patient was removed before uploading to the centralized system.
Any information other that the state in which the patient lives and year the patient was born and received treatment should be removed for patient privacy. In addition, it would be necessary to provide a way to reduce or closely eradicate the ability to duplicate patient records that would be uploaded to a central HIE or sent between decentralized models. If a central HIE is used, it would be necessary to run occasional manual checks to ensure duplicates have not been uploaded in error.
As a preventative measure, one upload per day at the end of the day would help ensure no more than one copy per patient would be uploaded. For decentralized models that share information between each other, duplicates can be reduced by requesting specific dates or types of service within a request for information. By replying with information with only the specific information requested the responding organization helps ensure the requesting organization only receives what they need.
As a requesting organization, a thorough check within the patient file should be completed to ensure the information needed has not already been received. When receiving information, another check should be completed to make sure what is received is not already saved in the patient file. Training new and occasionally retraining existing staff on data collection and sharing would also help ensure data quality and integrity. Upon completion of training, staff should sign documentation stating they thoroughly understand the policy and procedures of PHI and documentation should be kept in the employee’s HR file.