il-fede - Fotolia
The ability to identify patients, match them to their medical records and provide care based on accurate information is a management issue that could determine whether the healthcare industry can successfully implement value-based care -- or if it will continue to struggle with a costly system that compromises the safety and quality of patient care.
Healthcare providers have always sought to develop a record of truth concerning their patients. During the last decade, the Obama administration's electronic health record (EHR) incentive program shifted healthcare industry record-keeping to digital versions of paper charts. Vendors and providers likewise worked to design an electronic system that accurately documents patient data, avoids duplicate medical records, facilitates data exchange with speed and allows data to be quickly analyzed.
While it was expected that transferring patient data to an electronic format would help curb patient ID errors, duplicate medical records as well as human error when entering information into EHRs still continue to plague the healthcare system. Many organizations recognize an urgent need to address the problem.
Expect 10% of records to be duplicates
In the case of the University of California Irvine Healthcare (UCI Health), which operates the Chao Family Comprehensive Cancer Center in Orange, Calif., the organization took the opportunity in the quarter of 2016 to reconcile its duplicate medical records when it switched EHR systems from Allscripts to Epic Systems.
As part of the project, UCI Health used Epic's native tools and patient matching algorithm to convert patient data that included names, addresses, dates of birth, telephone numbers and financial information. Mismatched patient data was further matched with medical records in the Allscripts EHR by consultants with expertise in patient matching technologies.
"We put our patient records through this patient matching algorithm, and the system spit out the duplicates," said Sriram Bharadwaj, director of information services at UCI. Once duplicates were identified, he said, hospital staff members cross-checked the information, including the spelling of names and addresses. This information was matched with other health-related data, such as which doctor saw the patient, what diagnosis the patient received and what medications the clinician administered.
Making sure accurate patient records went into the Epic EHR was a major part of the project, according to Bharadwaj. The entire EHR conversion took 18 months to complete. "In any system, out of 1.8 million records, approximately 10% -- or 180,000 -- would be duplicates if continuous cleanup is not carried out using matching algorithms," he explained. "It's never good to add data without understanding the ramifications of that data to the system. For one thing, getting rid of duplicate records has helped our workflow be more efficient. But more importantly, it's helping our hospital manage our patient records and, by extension, patient care in a more precise way."
Inaccurate patient info is costly
To appreciate how dangerous duplicate medical records can be to the business operations of a healthcare organization, Black Book Research interviewed 1,392 health technology managers and found that the average expense of repeated medical care because of duplicate records was $1,950 per patient per inpatient stay and more than $800 per emergency department visit.
Additionally, respondents estimated that 33% of all denied insurance claims resulted from inaccurate patient identification or information, costing the average hospital $1.5 million in 2017 and the U.S. healthcare system more than $6 billion annually.
"If a physician has the wrong patient record," Bharadwaj said, "that physician might order a mammogram on the wrong patient, prescribe medications to the wrong patient or perform surgery on the wrong patient. Patient safety and quality suffers a lot because of duplicate records."
As the healthcare industry moves toward value-based care initiatives -- in which providers receive Medicare or insurance reimbursements when they can prove that they raised patient-care quality metrics in a cost-effective manner -- the momentum has grown to improve patient matching and identification, said John Halamka, M.D., CIO at Beth Israel Deaconess Medical Center (BIDMC) in Boston. "If a medical center is paid based on outcomes, it's imperative to gather as much data about a patient as possible from all their sites of care," Halamka said. "Without patient matching, it's impossible to improve quality and reduce total medical expense."
Blockchain stands ready to take on duplicate records
Health IT experts are eyeing blockchain as another technological weapon in the fight against duplicate medical records.
"The application of blockchain technology to patient identity management, clinical workflow optimization and health information exchanges is probably the next wave of exploration," said Mutaz Shegewi, research director of provider IT transformation strategies at IDC Health Insights.
Blockchain can strengthen different facets of data exchange at a foundational, structural and semantic level, Shegewi said. "What blockchain can do that technologies currently can't do -- or don't do very well -- is to add decentralized, distributive and immutable qualities to data," he said.
For certain cases in healthcare, blockchain can essentially reinforce data integrity through its architecture and the way it's programmed, which may be able to further help organizations reduce duplicate patient records and data errors.
Widespread patient misidentification errors exist across the healthcare system, according to the "2016 National Patient Misidentification Report" from the Ponemon Institute and sponsored by Imprivata, a health IT security vendor. Of the 503 clinicians and health IT professionals polled, 86% said they witnessed or know of a medical error that was the result of patient misidentification.
To understand where, how and why problems of patient data capture at healthcare facilities arise, the poll further indicated that 63% of patient identification errors occur because of inaccurate information given during patient registration, 60% because clinicians face time pressures when treating patients and 35% because of insufficient employee training. Meanwhile, 34% of respondents said too many duplicate medical records exist, and 32% said human errors occur when inputting patient data.
Progressive steps to battle duplicate patient records
For many health IT executives, the way forward is to implement technology that can offer new techniques to improve patient identification, stem duplicate medical records and assist related administrative workflows.
According to Halamka, the pathway to avoiding duplicate patient records and preventing patient identification errors includes the following steps:
- Start with an exact match of name, gender and date of birth, which will work 60% of the time to avoid duplication later on.
- Move on to probabilistic matching algorithms to generate a percentage of how likely a record matches a patient, which will work 80% of the time.
- Add in non-healthcare-related referential matching data, such as property and car owned as well as voting place, which will work 90% of the time.
- Add in biometrics, such as palm vein geometry or a fingerprint scan, which will work 99% of the time.
- Establish a national patient identifier, which is an idea that has political minefields because it would require legislation to enact it. A less complicated option would be to start with a voluntary opt-in for patients. Halamka predicted a national identifier would work 99.9% of the time.
Halamka said BIDMC is currently at the probabilistic matching algorithms stage and starting to embark on referential matching to sort through its patient matching efforts. He doubted any academic medical center is beyond that as yet.
Another way to tackle the problem is to allow health information exchanges, which connect to healthcare organizations in a city or region, to play a role in facilitating patient data reconciliation, Bharadwaj said. Additionally, health IT executives should evaluate technologies that make it easier to match patients with their records, a process that still relies heavily on manual tasks.
"Biometrics, which can strengthen patient identification processes, and machine learning algorithms, which can over time learn the mistakes people make to avoid future errors," Bharadwaj explained, "are technologies that can drive efficiency and patient identification matching capabilities into healthcare workflow processes."