In today’s enterprise healthcare environment, patient information is incredibly disconnected. Each clinical, accounting, and registration system for every stop a patient makes within a healthcare ecosystem may have a different way of identifying that patient. One system may use the SSN, one may use a chart number, and another may identify the patient by their first name, last name, and date of birth (DOB). In addition to these inconsistencies, lack of data integrity caused by typos, misspellings, changes in identifying information (a marriage, a divorce, a move), or a patient not providing consistent identification may create multiple records for one patient within a system. Due to similar identifying information between patients, records could even on occasion be assigned to the wrong patient in a system.
This lack of a unique patient identifier throughout the enterprise results in inaccurate patient data, reordering of lab work, delayed treatments, denied insurance claims, overworked Health Information Management teams, and decreased registration and scheduling efficiency. A patient’s life is in danger if clinical data is separated into multiple charts or if the wrong patient’s chart is pulled at the point of care. In a patient-centric, value-based healthcare environment, there is even more pressure for patients to be identified across disparate systems to ensure accurate reporting percentages to support care quality and decrease costs.
Organizations adopting new technology and implementing new applications to enhance their enterprise workflows must address the inefficient patient identification across various systems and resolve duplicate records within these systems. It is imperative that patient records be matched and assigned to a master patient index across the enterprise.
Enterprise patient master index (EPMI) solutions are built to identify disconnected patient data in multiple systems. These data registries maintain consistent and accurate patient demographic data, assigning a unique identifier that links the correct patient’s disparate records at an organizational level. Matching records across all systems—EHRs, laboratory ordering systems, radiology, billing, outpatient clinics, rehabilitation facilities, registration, scheduling, etc.—EMPI solutions are designed to provide a complete, longitudinal view of a patient’s health.
EMPI solutions use multiple data attributes such as first name, last name, DOB, Social Security number (SSN), maiden name, alias, prefix, suffix, gender, driver’s license number, phone number, etc., and use both deterministic and probabilistic matching algorithms to best identify matches.
Deterministic matching identifies exact matches based on rules using specific attributes. For example, two records are determined an exact match when they share the same first name, last name, DOB, and SSN. Probabilistic matching evaluates the probability that two records belong to the same patient by considering key elements, phonetic variations, nicknames, abbreviations, and other various criteria, assigning weights to these attributes, and grouping patient records based on matching scores. EMPI solutions will also identify false positive errors, or records that are matched but should not have been: and false negative errors, records that should have been matched but have not been: and helps to resolve these inaccuracies.
It is important to evaluate the quality of the algorithms used in an EMPI solution, as the weight assigned to each data attribute will have a significant effect on the accuracy of the match score.
The Office of the National Coordinator for Health Information Technology (ONC) notes that patient matching is one of the key barriers to achieving interoperability, causing issues for over 50% of health information managers. They believe that the problem will increase as we increase the volume of health data sharing, and that data quality issues make matching more complicated.
To bring about transparency and gather data on the performance of patient matching algorithms, encourage the use of performance metrics to evaluate these algorithms, and positively impact other aspects of patient matching, the ONC hosted a challenge in 2017 to find the best patient matching algorithms. The challenge attracted over 140 teams and thousands of submissions. Participants were provided a data set, and their answers were evaluated and scored against a master key. Winners were scored on measures of accuracy that factored in both precision and recall, and $75,000 in cash prizes were awarded.
ELLKAY feels it is important for providers to have the ability to view all of a patient’s data as one record to deliver the best patient care. We partnered with one of the winners of the ONC Patient Matching Algorithm Challenge, combining their proven algorithms with our software to offer a platform that empowers clients with accurate, reliable patient matching and patient identification across their enterprise.
