Data Integration Strategy
Introduction
There has been an increase in the number of patients requiring medical care. This is due to an aging population and the availability of medical and financial resources necessary to seek medical attention. As most institutions are required to store accurate healthcare information, there has also been an increased reliance on medical records systems. Such information is crucial at the clinical level for improved medical decision-making, at the financial and administrative levels for better management decisions, and quicker reimbursement of claims by insurance agents. Beyond these, quality improvement organizations often use such data to determine the quality of care provided by institutions. Hospitals require accurate Electronic Health Record (EHR) systems to enhance the delivery of such care to patients. Such records include patient demographic records, lab results, physician orders, and treatment routines, as well as account information used to facilitate payments.
Due to several barriers and obstacles, different components of the EHR system may be obtained and purchased from various sources and at different times. This leads to a highly fragmented EHR system where different departments use different methods to process and store their data. The data is highly fragmented due to various coding schemes, presenting barriers to interoperability as well as the flow of data to other partners. For a large institution such as in this case, the amount of data can be overwhelming for all parties, resulting in confusion since the decoding of data takes a long time. Data integration is the only solution that can salvage such a system and result in the seamless flow of data within the institution and between the hospital and other partners involved in value-based care.
This paper looks into such a hospital system to determine the data standardization requirements in the current policies and offer solutions based on the applicable clinical standards theory. The work then looks into how interoperability, decision support, and legacy systems can be implemented in the institution as well as models for clinical data and clinical processes.
Standardization of data dictionaries
Within the institution, the most critical data elements are patient demographic and physical data, laboratory tests and results, clinical orders, and discharge reports. Within the institution, patient data obtained by a physician is coded using the Standard Nomenclature of Medicine Clinical Terms (SNOMED-CT) applications that involve the use of codes for questions that a physician asks as well as phrases presented as answers by the patient. This SNOMED system allows faster storage and ease of retrieval for such data (Lee et al., 2013, 87).
Laboratory data is captured using the LOINC (Logical Observation Identifiers, Names, and Codes) system that has codes representing the various tests and observations as well as codes for the multiple answers and results that can be obtained from the patient. The LOINC system similarly allows such data to be easily stored, transferred, and accessed across different platforms and systems, leading to ease of use for internal and external users.
The Uniform Hospital Discharge Data Set (UHDDS) is used, especially in the provision of care to patients using Medicare and Medicaid, but has been expanded in the institution to be used for all patients. The billing and coding of UHDDS include the hospital identification code, insurance provider code, patient demographics, medical procedures, and the main and additional diagnoses. By using standard ICD-9 CM codes, the UHDDS improves the billing process and simplifies claims for the institution (AHIMA, 2010, 1). The system simplifies the discharge process at the institution and streamlines reimbursement from the government Medicare and Medicaid programs as well as from other insurance providers.
These three systems are crucial to the running of the institution as they affect the most important procedures for clinical, financial, and management decision-making. They directly affect patient care and quality of health services since they offer indicators of excellence that improve the rating of the institution.
To ensure standardization of the institution’s data dictionary, it will require to model all the terminologies used, including the ICD, SNOMED CT, and LOINC. Since LOINC uses the ICD system, the organization can achieve standardization by upgrading its UHDDS system to also use ICD 10 instead of ICD 10, leading to better standardization between LOINC and UHDDS. Maps will need to be established between the systems to ensure that the code groups representing single concepts in the three systems are defined (Wolters Kluwer, 2014, 1). Thus, there should be a group for the SNOMED CT and ICD 10/ICD 9 codes that identify hypertension. This will make the systems user-friendly for the coders as well as users.
Applicable Clinical standards theory and development
Schulz, Stegwee, and Chronaki (2019, 3) identify the most important principles that govern clinical standards theory and development. These will be crucial in ensuring that the systems used in the institution can achieve data integration while following the important elements in defining the needs of the hospital.
The data integration strategy will utilize reference terminologies that are unique and easily understandable by the professionals using the systems or accessing data from the systems. The self-explanatory no-ambiguous labels will ensure that the reference terminologies will have formal definitions in the data sets that are well understood and used by the clinicians and other users of the systems.
The data integration will also include the development of the rules, which will be used in composing new terminologies for the system. This will follow the rules of syntax for all the systems used, whether individually or in collaboration. For example, the rules will define how laterality terms like left and right will be combined with medical terminologies like posterior and anterior with a clear order and definition of the right choice of language.
The development of standards and their use across the systems will also follow the rules of pragmatics, where symbols utilized will include the situational context. While composing expressions, for example, pragmatics calls for identification and embedding of meaning in a given phrase that has been composed. In asthma, for example, suspected asthma, severe asthma, and asthma prevention have different meanings that are important, especially in improving the ability of the systems to provide support in clinical decision making.
Data integration needs
The data integration project within the institution is bound to provide benefits due to the ability of the software systems to link to each other flawlessly. The data integration will enable the systems and users to easily translate information obtained from patients to allow data to be shared among the internal and external users of the systems. The most crucial challenges in data integration will be ensuring interoperability, improving decision support, and building on legacy systems.
Interoperability
System interoperability refers to the ability of the systems to not only be interconnected but also allow seamless flow of data and information between the various users and systems. While interoperability does not necessarily have to be in real-time, there has to be a synchronization of the systems so that the information is constantly being shared among the systems to provide a real-time understanding of the data and shape decision making.
For the institution, interoperability will lead to an improved workflow by reducing the duplication of data in different systems. Data entered in the SNOMED and LOINC systems, for example, can be used to fill the UHDDS forms, saving on time, when compared to the situation where the data has to be filled separately for each system. Interoperability in the systems will ensure that the institution can benefit from innovations that improve patient care quality.
In the integrated system, semantic interoperability will be achieved by building a reference database for all the terminologies used in the database. This will ensure that users of the system have a reference point in case they need assistance on the right terminologies to use. Data exchange between the systems will be structured beforehand to ensure that data from one system can be translated correctly into the other system. The coding of the data and vocabulary of all the systems will enhance the utilization of the information in the different systems. Technical interoperability will be promoted by using Health Level 7 (HL7) standards that offer guidelines on the structuring of messages. This will ensure that the messaging structures between the systems are clear.
Decision Support
The integration of the systems should enhance the decision-making ability of the institution. These decisions range from clinical to financial decisions that are important in the running of the institution. In enhancing the ability of the new EHR system to aid in decision-making, the project will involve a team to define the limits of the decision-making application. This is because the application can offer many notifications that will reduce the productivity of caregivers and lead to burnout (HealthIT Analytics, 2017, 2). The team will decide the most important applications for their EHR system by defining the points in care delivery that need decision support.
The support can range from systems to notify patients about their scheduled appointments, providing timely information to physicians, enhancing nursing care through reminders in important care routines, and improving drug use by advising on adverse effects. After the team determines the most important aspects of decision-making, these can then be implemented by coders into the new EHR system.
Legacy Systems
Legacy systems used in the institution include traditional software that has been involved in the collection and storage of data for a long time. These systems are part of the initial EHR strategies that the hospital used from the beginning. During integration, it is crucial that legacy systems are analyzed to determine their importance in the EHR system. Legacy systems that are obsolete require to be updated so that the new system only consists of modern software that can handle all the responsibilities of the older systems.
During integration, data from the legacy systems will be transferred to new systems so that when the legacy system is discarded, the old data is not lost. After developing the library of terms for the new integrated system, data from the legacy software will be transferred to avoid duplication of roles and loss of information about patients from the previous systems.
The legacy system may contain a lot of information, some of which may be difficult to input and code into the new system. Further, coding all the data in a legacy system can be an expensive routine since it is labor-intensive to code all the information in the system. In order to avoid this expensive and time-consuming affair, a hospital committee will be set up to determine the data to keep from the legacy system and what to discard. This will reduce costs while ensuring that patient safety is not compromised.
Clinical Data and Clinical process modeling
Unified Modeling Language (UML)
The data integration process is first analyzed using the UML approach based on the users (Springer Professional Computing, n.d., 24). The system’s direct users will include nurses, physicians, coders, and administrators who will be involved in inputting data or collecting information from the system. These individuals will interact with r system daily and will access the system based on allocated passwords. The flow of activities within the system will involve data entry by the care provider, storage of the data in the data warehouse, and retrieval from various nodes in the system.
Sequence diagrams for the integration entail the collaboration of the various objects in the systems. Since it is integrated with enhanced operability, the system will allow the continuous flow of information between the systems where the SNOMED and LOINC systems will first receive data about patient demographics and diagnostic and clinical data. The UHDDS system will use information from the other two systems to fill the discharge forms and finalize the process.
Unified Process (UP)
The UP model focuses on the iteration of a system using feedback to improve it and enlarge its scope with time and corrections from the feedback (Springer Professional Computing, n.d., 28). The unified process is also case driven, based on architecture, and acknowledges the risk. For the integration, the implementation of the system will follow a phased approach, beginning with the development of the code libraries. These can then be evaluated on paper before implementation. In order to reduce the level of risk during implementation, the system will first be used on a subset of patients, such as in the oncology department. This will allow errors to be detected early, and the architecture change to enhance the system. Slowly, the integration can then be implemented for the whole organization.
Conclusion
In conclusion, the integration of the systems used in the institution will be a major step towards improving the quality of care that the hospital provides. Standardization of dictionaries will be as important as the execution of the different models involved in implementing the changes. This is because problems in the dictionaries will lead to further problems downstream, causing the system to be ineffective. The integration of the systems should also improve interoperability, at least within the institution. This will ensure that all users of the systems can access data in real-time to reduce duplication. The decision-making ability of the various caregivers will also improve as a result of the system. Lastly, the implementation of the changes will have to follow established approaches. Using the UML and UP approaches will reduce the risks that would face the institution if the new system failed after implementation. With time, the system can be used throughout the organization leading to improvements in the value of care that clients receive.
References
AHIMA (2010). ICD-9-CM Coding Guidance for LTC Facilities. Appendix B: Reporting and Sequencing Diagnoses on the Health Record and UB-04 Claim Form. Journal of AHIMA 81, no.10
HealthIT Analytics (2017). Understanding the basics of clinical decision support systems. Available at https://healthitanalytics.com/features/understanding-the-basics-of-clinical- decision-support-systems
Lee, D., Cornet, R., Lau, F., & de Keizer, N. (2013). A survey of SNOMED CT implementations. Journal of Biomedical Informatics, 46(1), 87–96.
doi:10.1016/j.jbi.2012.09.006
Schulz S., Stegwee R., Chronaki C. (2019). Standards in Healthcare Data. In: Kubben P., a Dumontier M., Dekker A. (eds) Fundamentals of Clinical Data Science. Springer, Cham
Springer Professional Computing. (n.d.). An Introduction to the UML and the Unified Process. Guide to the Unified Process Featuring UML, Java, and Design Patterns, 21–37. doi:10.1007/1-85233-856-3_3
Wolters Kluwer, (2014). What is a data dictionary, and what role does it play in semantic Interoperability. Available at https://blog.healthlanguage.com/what-is-a-data-dictionary- and-what-role-does-it-play-in-semantic-interoperability