Secondary Logo

Journal Logo

Information models offer value for nursing

Lytle, Kay S. DNP, RN-BC, NEA-BC, CPHIMS, FHIMS

doi: 10.1097/01.NURSE.0000559937.20007.a8
Department: TECH NOTES
Free

Kay S. Lytle is the chief nursing information officer at Duke University Health System in Durham, N.C.

The author has disclosed no financial relationships related to this article.

CLINICALLY, IT IS easy to understand the need for sharable and comparable data. For example, patients may visit multiple providers within the areas they live, work, and visit. To have the complete picture of their health, clinicians need data from all those clinical encounters.

At the most basic level, clinicians use electronic health records (EHRs) to view their patients' clinical information and document care. EHRs can also be used to share data about clinical effectiveness, benchmarking, value-based care, quality improvement, and research. Clinicians need a well-designed data sharing system to ensure that fall risk or pain scores captured at one institution mean the same thing at another.

Achieving this level of meaningful data sharing requires the use of standardized terminologies and information models, which are agreed-upon concepts for assessments, nursing diagnoses, interventions, and outcomes.1 This article examines standardized terminologies, discusses how information models can be used to improve clinical practice, and describes an ambitious initiative to create nurse-sensitive information models.

Back to Top | Article Outline

Standardized terminologies

The American Nurses Association (ANA) issued a position statement in 2014 supporting standardization and interoperability of nursing data captured in EHRs among all vendor products, throughout the nursing process, and across care settings.2 The ANA issued another position statement in 2015 promoting the use of recognized terminologies for data exchange, including the use of Logical Observation Identifiers Names and Codes (LOINC) and Systematized Nomenclature of Medicine—Clinical Terms (SNOMED CT).3 LOINC is used to describe “questions” such as nursing assessments and outcomes. SNOMED CT is used to describe the “answers” or nursing findings, problems, and interventions.

Other standardized nursing terminologies may be used within a given organization in its EHR. Examples include Nursing Interventions Classification (NIC), Nursing Outcomes Classification (NOC), International Classification for Nursing Practice (ICNP), Omaha System, or Clinical Care Classification System (CCC).

For organizations choosing to use one of these recognized standardized nursing terminologies, data exchange requires mapping the local terms to LOINC and SNOMED CT terms for comparison across systems. This mapping allows for comparison of patient data documented at one organization using NIC and NOC with another using CCC, maintaining meaning locally while facilitating exchange.4

Nurses document much of their care in EHR flowsheets, which contain structured and semistructured data.5 Within organizations, EHR flowsheets are often customized across various care areas such as the ED, ICU, perioperative areas, and acute care units. A lack of internal standardization may result in issues with data sharing and visibility within one organization as well as between health systems. Data standardization is needed within organizations and for cross-organizational use. This standardization can be supported with the use of standardized terminologies and information models.

Back to Top | Article Outline

What are information models?

Information models are organized structures used to represent knowledge about a clinical condition or concept.1 They include the data elements and definitions, their relationships, and the associated data standards. Information models are independent of the EHR design or build.

For example, an information model for hospital-acquired pressure ulcer (HAPU) was developed in a collaboration between Kaiser Permanente Healthcare System and Veterans Affairs.6 The HAPU information model includes various items related to skin assessment such as color, temperature, turgor, and moisture, with each of these having designated allowable values. For instance, values for skin color include normal pertaining to the ethnic group, dusky, icteric/jaundiced, pale, cyanotic, flushed, mottled, and ashen. The question of skin color and the associated values together represent a value set. Typically, this would be presented to the nurse within the EHR as the pick list associated with a particular flowsheet row.

Much of the current work on information models focuses on the medical domain. For example, the PCORnet Common Data Model (CDM) includes many data categories, such as demographics, conditions, diagnoses, medication ordering, pharmacy dispensing, medication administration, lab results, procedures, and vital signs.7 Currently, nursing care is represented in only the medication administration and vitals categories of the PCORnet CDM; most of the work of nursing and the nursing process is not included. This lack of nursing data is common with other data and information models. In order to share nursing data across organizations, we need common data models and information models that include representations of nurse-sensitive data such as pain, falls, or pressure injury prevention.

Back to Top | Article Outline

Nursing knowledge: Big data science

Nursing informatics experts at the University of Minnesota School of Nursing launched the Nursing Knowledge: Big Data Science Conference in 2013 with an overall goal of developing an action plan for sharable and comparable nurse-sensitive data.8 This annual think-tank conference strives to ensure the standardization and integration of nursing information in EHRs to improve health outcomes. Virtual workgroups were also organized through this initiative to take action between the annual meetings and report their work each year at the conference.

Bonnie Westra, codirector for the Center of Nursing Informatics, and her research team used EHR flowsheet data from one health system with 2.4 million patients to develop 10 information models representing nursing data. Detailed in a September 2017 report, the models included cardiovascular system, falls, gastrointestinal system, genitourinary system, musculoskeletal system, pain, pressure ulcers, respiratory system, venous thromboembolism, and expanded vital signs/anthropometrics.5

To validate the pain information model originally developed by Westra and her team, nurses collaborating on the Validation of Information Models subgroup of the Nursing Knowledge Big Data Science initiative used documented EHR flowsheet data from eight organizations and EHR builds from two additional organizations.9 The consensus-based pain information model represented data from 6.6 million patients and included 30 concepts.

The original pain information model included one pain rating scale. In contrast, the validated model included 12 unique pain scales. These 12 pain scales represent various self-report and observational assessments used by nurses based on patient age, setting, or clinical condition.

The Validation of Information Models subgroup passed the pain information model work to the Encoding & Modeling workgroup, responsible for submitting the concepts to LOINC and SNOMED CT when existing codes do not exist. The subgroup is currently working on validation of the genitourinary model and the fall prevention model, with plans to validate the remainder of the nurse-sensitive and body system models.

Back to Top | Article Outline

Implications for nursing practice

Practicing nurses are needed to participate locally in their organizational EHR optimization and redesign, focusing on use of standardized concepts and definitions. Nursing can collaborate with other disciplines to create opportunities to share data.

Information models organize knowledge about a clinical concept and include the data elements, relationships, and data standards.1 A focus on information models important to nursing practice can identify gaps in nursing concepts in LOINC and SNOMED CT, supporting the work toward sharable and comparable data across care settings and health systems.

Back to Top | Article Outline

REFERENCES

1. Goossen W, Goossen-Baremans A, van der Zel M. Detailed clinical models: a review. Healthc Inform Res. 2010;16(4):201–214.
3. American Nurses Association. Inclusion of recognized terminologies supporting nursing practice within electronic health records and other health information technology solutions. 2018. http://www.nursingworld.org/practice-policy/nursing-excellence/official-position-statements/id/Inclusion-of-Recognized-Terminologies-Supporting-Nursing-Practice-within-Electronic-Health-Records.
4. Matney SA. Semantic interoperability: the good, the bad, and the ugly. Nursing. 2016;46(10):23–24.
5. Westra BL, Christie B, Johnson SG, et al Modeling flowsheet data to support secondary use. Comput Inform Nurs. 2017;35(9):452–458.
6. Chow M, Beene M, O'Brien A, et al A nursing information model process for interoperability. J Am Med Inform Assoc. 2015; 22(3): 608–614.
7. PCORnet. PCORnet Common Data Model (CDM). 2019. http://www.pcornet.org/resource-center/pcornet-common-data-model/.
8. University of Minnesota School of Nursing. 2019 Nursing Knowledge: Big Data Science Conference. 2019. http://z.umn.edu/bigdata.
9. Westra BL, Johnson SG, Ali S, et al Validation and refinement of a pain information model from EHR flowsheet data. Appl Clin Inform. 2018;9(1):185–198.
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.