Perspectives on Safety

Updates in the Role of Health IT in Patient Safety

Background

Health information technology (HIT) has the potential to improve patient safety by facilitating evidence-based decision making, minimizing error due to human factors, such as illegible handwriting and imprecise recollection of dosing recommendations, and enhancing the reporting, tracking, and aggregation of patient data.[1],[2],[3]

In 2019, articles published on PSNet on the use of HIT to improve patient safety focused on reducing errors in four areas:

  • Adverse drug events
  • Diagnostic errors
  • Transfusion safety
  • Adherence with evidence-based care

While there are a variety of HIT practices that may address each one of these areas, this essay focuses specifically on HIT practices highlighted in resources published on PSNet.

Reducing Adverse Drug Events

Adverse drug events (ADEs) are a common patient safety event across all settings of care.[4] Specifically, ADEs account for approximately one third of all inpatient hospital adverse events.[5] HIT solutions such as electronic dosing, mobile applications, EHR medication reconciliation tools, and clinical decision support (CDS) alert systems can improve medication safety by increasing the quality of documentation, supporting accurate prescribing practices, and by facilitating information exchange.[6]

One featured approach is the use of barcoding and scanning protocols in the medication delivery process to reduce the risk of ADEs. Several articles recommended barcoding in perioperative care and the operating room settings as risk reduction strategies to prevent medication administration errors. In addition to these examples, barcode scanning nursing protocols may have the potential to reduce the risk of wrong-patient insulin pen errors.  

Another approach highlighted by several articles is the use of computerized provider order entry (CPOE) within CDS to reduce medication errors. The results of a systematic review concluded that the use of CPOE reduces the overall medication error rate as well as specific error rates for wrong dose, wrong drug, administration frequency, administration route, and drug-to-drug interaction errors. However, the implementation of CPOE can introduce new types of errors and may not prevent errors that are not rule-based, such as technical or memory errors. CPOE and continued pharmacist review are both essential for more comprehensive error prevention.

HIT Approaches to Impacting Diagnostic Errors

The true magnitude of diagnostic errors is difficult to quantify. However, the National Academy of Medicine estimates that most people will experience at least one diagnostic error over the course of their life.[7] While diagnostic errors may not always necessarily lead to a harm, they can result in a delayed- or inappropriate- treatment.7 HIT has the potential to help providers improve the diagnostic process and reduce diagnostic error by providing easier access to critical patient data, facilitating information exchange, offering complex analysis, and retrieving workflow and decision making information.7 Specific HIT approaches may include the use of CDS, diagnostic study interpretation, and trigger tools.  

In 2019, PSNet included multiple resources that identified artificial intelligence (AI) systems as a rapidly emerging healthcare technology with the potential to improve diagnostics. Specific examples highlight the use of AI in pediatric diseases. AI-based systems could support physicians in the analysis of large amounts of data for diagnostic evaluations and provide clinical decision support when there is diagnostic uncertainty or complexity. Some deep learning AI diagnostic algorithms achieve similar diagnostic accuracy to physicians, although authors note more testing in clinical settings is needed.

Barcoding to Improve Safety in Transfusions

More than 20 million blood components are transfused annually in the U.S.,[8],[9],[10],[11] with approximately 51,000 adverse reactions related to transfusions.[12] Labeling issues are a major source of transfusion errors, causing patients to receive incorrect blood types.[13] Barcoding can alleviate the risk of labeling errors in transfusions by reducing the potential for human error in the validation processes. One 2019 article demonstrated that barcoding, in conjunction with an electronic auditing system, was safer than a manual verification system for transfusions in the operating room

Clinical Decision Support and Increased Adherence with Evidence-based Care

CDS has the potential to increase provider adherence to best practice, evidence-based care by codifying standards in a way that can be easily accessed by clinicians. This may include, but is not limited to, notifications, alerts and reminders, condition-specific order sets, clinical summaries, and templates.3 One systematic review demonstrated that the use of CDS in oncology care increased clinical practice guideline use and concordance, improved care process measures, and reduced safety events. Further, integration of AI into CDS software has the potential to further improve knowledge management for clinicians and enhance the analytic capabilities and outputs of CDS software.[14] For example, AI systems are being developed that can detect complex, time dependent conditions such as sepsis or warning signs of patient deterioration.  

AHRQ’s CDS Connect project is facilitating the use of evidence based standards of care via shareable, standards-based CDS resources. In addition to the artifacts already available through the project, the CDS Authoring Tool enables users to create their own CDS artifacts.

What is on the Horizon?

One area of HIT research receiving continued attention is the challenge of HIT implementation in clinical practice. HIT can offer standardization of workflow and creative solutions to countering patient safety events such as those mentioned in this paper. However, with continued implementation and use in diverse clinical settings it becomes apparent that these solutions may introduce new and unanticipated errors. This raises the critical need for research into the safe design, implementation, and surveillance of HIT to reduce the rate of errors directly related to the newly incorporated technology.

In addition to countering new errors, an additional HIT challenge concerns whether these systems are appropriately meeting the cognitive needs of front-line users. While HIT solutions can greatly expand the amount of information available for collection and analysis in clinical use, the right information is not always available or presented to clinicians at the right time. This supports the idea of context dependent design where HIT solutions are designed to meet specific workflow and contexts of the clinical users. For example, when prescribing a medication, the ability to review previous prescriptions, recent diagnostic testing, and/or vital signs are critical to the safe selection of a new medication. Requiring a clinician to navigate across multiple silos of information in the EHR increases the risk of a patient safety event occurring. Close collaboration between HIT vendors, healthcare organizations, and funding agencies will be critical to improving this complex process and supporting the cognitive needs of clinicians.

Another area of continued research and policy discussion is how to best foster interoperability. In 2020, the Office of the National Coordinator for Health Information Technology (ONC) is expected to release rules regarding security, interoperability, usability, and EHR design. The intention is to push for transparency between EHR vendors in an effort to improve EHR usability and safety across the board.[15] However, a report from the Trusted Network Accreditation Program (TNAP) collaborative found that concerns over privacy and data security are a major barrier to achieving interoperability.[16] With clear benefits for increasing provider coordination and continuity of care, and with increasing demand from patients who want access to their information, assuring the security and privacy of data transfer is a critical next step in advancing interoperability goals.    

A final area of note concerns the progression of CDS and the integration of AI. Users call for additional clinical evidence of the utility of AI as well as specific frameworks within the Food and Drug Administration (FDA) to evaluate the effectiveness of these algorithms to ensure safety during real-world use.[17] However, despite these concerns, AI has great potential to alleviate burden on providers and presents a practical way to efficiently and effectively analyze and incorporate vast amounts of patient data. More research is needed to safely integrate AI into the clinical workflow and to proactively address bias that AI may unknowingly introduce into the clinical environment based on what data is used to train the models.

In conclusion, 2019 saw many advancements to our knowledge of the strengths and challenges associated with HIT systems. There are many areas of opportunity left to explore as these vitally important tools are optimized to safely and efficiently support patients and their care team.

 

Dr. Hettinger is an employee of MedStar Health and is a volunteer with the AAMI Health IT Standards Committee, DC Health Information Exchange Advisory Board, and Cerner Strategic Opioid Advisory Board. He has also received grant funding from AHRQ, FDA, ONC, DOD, AMA, and PEW Charitable Trust. 

Aaron Zachary Hettinger, MD, MS, FACEP, FAMIA
Assistant Professor
Medical Director and Director of Cognitive Informatics, National Center for Human Factors in Healthcare
MedStar Health
Washington, DC

Kendall K. Hall, MD, MS
Managing Director, IMPAQ Health
IMPAQ International
Columbia, MD

Eleanor Fitall, MPH
Research Associate, IMPAQ Health
IMPAQ International
Washington, DC

References


[1] Office of the National Coordination for Health Information Technology. Health Information Technology Patient Safety Action & Surveillance Plan. July 2, 2013. https://www.healthit.gov/sites/default/files/safety_plan_master.pdf. Accessed November 19, 2019.

[2] Feldman, SS, Buchalter S, Hayes LW. Health information technology in healthcare quality and patient safety: Literature review. JMIR Med Inform. 2018;6(2):e10264. doi: 10.2196/10264. [PubMed]

[3] Alotaibi YK, Federico F. The impact of health information technology on patient safety. Saudi Med J. 2017;38(12):1173-1180. doi: 10.15537/smj.2017.12.20631. [PubMed]

[4] Wittich CM, Burkle CM, Lanier WL. Medication errors: an overview for clinicians. Mayo Clin Proc. 2014:89(8):1116-25. doi: 10.1016/j.mayocp.2014.05.007. [PSNet]   

[5] Adverse Drug Events. Health.gov Website. https://health.gov/hcq/ade.asp. Accessed on November 19, 2019.

[6] Seidling HM, Bates DW. Evaluating the impact of health IT on medication safety. Stud Health Technol Inform. 2016;222:195-205. [PubMed]

[7] National Academies of Sciences, Engineering, and Medicine. 2015. Improving Diagnosis in Health Care. Washington, DC: The National Academies Press. https://doi.org/10.17226/21794. [PSNet]

[8] Rogers MAM, Blumberg N, Heal JM, Langa KM. Utilization of blood transfusion among older adults in the United States. Transfusion. 2011. doi:10.1111/j.1537-2995.2010.02937.x. [PubMed]

[9] Qian F, Osler TM, Eaton MP, et al. Variation of blood transfusion in patients undergoing major noncardiac surgery. Ann Surg. 2013. doi:10.1097/SLA.0b013e31825ffc37. [PubMed]

[10] Frank SM, Savage WJ, Rothschild JA, et al. Variability in blood and blood component utilization as assessed by an anesthesia information management system. Anesthesiology. 2012. doi:10.1097/ALN.0b013e318255e550. [PubMed]

[11] Shander A, Fink A, Javidroozi M, et al. Appropriateness of allogeneic red blood cell transfusion: The international consensus conference on transfusion outcomes. Transfus Med Rev. 2011. doi:10.1016/j.tmrv.2011.02.001. [PubMed]

[12] Whitaker BI, Hinkins S. The 2011 National Blood Collection and Utilization Survey Report. U.S. Department of Health and Human Services (DHHS), Office of the Assistant Secretary for Health; 2011. [website]

[13] Dubin CH. Technology, vigilance, and blood transfusions: How U.S. hospitals and the federal government are working to reduce adverse events. P T. 2010;35(7):374-6. [PubMed]

[14] Daniel D, Silcox C, Sharma I, Wright MB. Current state and Near-term priorities for AI-enabled diagnostic support software in healthcare. Duke Margolis Center for Health Policy website.  https://healthpolicy.duke.edu/sites/default/files/atoms/files/dukemargolisaienableddxss.pdf. Accessed November 19, 2019.

[15] Improving EHR Usability, Interoperability to Aid Patient Safety. ehrintelligence.com website. https://ehrintelligence.com/news/improving-ehr-usability-interoperability-to-aid-patient-safety. Accessed January 23, 2020.

[16] The Trusted Network Accreditation Program (TNAP) Collaborative Survey Finds Concern Over Privacy and Security Key Barrier to Interoperability. ehnac.org website. https://www.ehnac.org/?press-release=the-trusted-network-accreditation-program-tnap-collaborative-survey-finds-concern-over-privacy-and-security-key-barrier-to-interoperability. Published June 26, 2019. Accessed January 23, 2020.

[17] Waddell K. Medical AI has a Big Data Problem. Axios website. https://www.axios.com/medical-ai-data-problems-041773b4-5dc8-4558-8173-46f623054627.html. Published May 30, 2019. Accessed January 23, 2020.