Patient Safety and Electronic Health Records

Patient Safety and Electronic Health Records —  What Could Go Wrong?

The Institute of Medicine, in a landmark article published in 1999 titled “To Err is Human,” estimated that nearly 100,000 people each year are killed by preventable medical errors.   Electronic health records (EHRs) have the potential to markedly improve care through the prevention of these medical errors.  The types of errors that are most likely to be prevented by EHRs include prescribing errors, failure to adhere to treatment algorithms, failure to follow-up on orders, and errors related to not having access to the complete medical history available during care. However, the safe use of EHRs requires a detailed understanding of their features, how they influence workflow, and awareness of situations where overreliance on technology could lead to patient injury.

The first goal of an EHR implementation is to make sure that no new patient safety issues are introduced as a byproduct of the EHR’s deployment.  Medical errors caused by EHRs are an order of magnitude less frequent than errors that occur on a daily basis in paper-based practices.  When patient safety issues that can be tied to an EHR occur they are for the most part preventable.  Studies have shown that EHR related patient safety issues are usually tied to inadequate preparation, insufficient training, and issues related to changes in workflow.   Any EHR implementation needs to follow a structured process that includes adequate preparation (e.g., customization of clinical content such as orders sets and templates).   Physicians and other care providers also need to be adequate trained on not only the use of the software but also how changes in workflow could impact patient safety.  For example, what protocol is followed to address critical lab values when the ordering provider is on vacation?  If an order tracking module has been implemented, does it accommodate this scenario?  Other sources of errors, although only rarely reported, are frank software errors (e.g., a wrong medication dose is calculated) and software design errors.

Efforts to gather data on patient safety issues that arise during the implementation and use of EHRs are underway.  Providers can now report patient safety issues to a “safe harbor” consisting of federally certified “Patient Safety Organization” (PSO).  Information submitted to a PSO is not discoverable, a protection put in place to encourage providers to report on actual or potential patient safety issues that arise during the implementation and use of an EHR.  This information will be used to identify trends that can be shared with medical associations, vendors, the federal government and other stakeholders.

What should you do to prevent medical errors that may occur as a byproduct of the implementation or use of you electronic health record?

Preventing errors related to implementation:  If you have not already implemented an EHR or about to replace an existing system, the following steps should be emphasized:

• Training – This is critical and often neglected by busy providers.  Software training can be cumbersome but effective use of an EHR can dramatically impact a provider’s productivity and quality of life.  It is also a fundamental duty, as poorly trained providers are more like to make mistakes that can lead to patient injury.  Make every effort to become a superuser of your EHR before your first use and then on an ongoing basis.
• Workflow – Frequently overlooked but an opportunity to evaluate current practices and implement changes that benefit patient care and provider efficiency. When patient safety errors, including patient mortality, it is most likely tied to workflows that have not been thought through.  In particular the management of critical lab values and how patients are monitored in acute care settings can be impacted by an EHR deployment.  Patient safety should always be the primary consideration when designing and testing any system that includes workflow.
• Data migration and loading – Given that most electronic data is in some sort of proprietary format within each EHR it is often difficult and time consuming to load patient data such as medications, problems, and allergies into a database.  Migration of data from other systems through and automated process typically contains errors that if not identified could lead to patient injury (e.g., a medication allergy that did not carry over from one EHR to another).   For this reason, information in the EHR should be considered suspect until it has been reviewed and validated by a provider.
• Content development – Certain types of content such as order sets and templates require extensive provider input before they can be used effectively, and in some instances safely, by clinicians during patient care.  Order sets that are incomplete or that contain errors can be a direct cause of patient injury.  It is imperative that healthcare organizations take the appropriate amount of time to develop and refine content that has the potential to impact patient safety.
• On-going assessment of workflow and user skills – Very often the initial EHR implementation becomes a point in time where the evolution of a practice becomes frozen.  In most settings providers have only partially learned key clinical functions of their system, and spend inordinate amounts of time on manual processes that can be automated.   EHR superusers need to remain constantly vigilante and observe the use of the system by others.  When errors or inefficiencies are identified, the superuser needs to work with the other providers to determine a resolution.
• Over-reliance on technology – EHRs can fail in subtle ways.  In one instance a free e-prescribing application was being used by a practice in Central Texas and the providers came to rely upon warnings that cautioned them about drug-drug interactions.  If there was not warning, the providers assumed there was no contraindication.  However, the warnings inadvertently became deactivated by the e-prescribing vendorand providers continued to prescribe medications using the tool, assuming they would be warned if there was a contraindication.  After several days it was discovered that the warning function had been deactivated, leaving the practice to wonder how many patients might have been prescribed drugs that were contraindicated.   Reliance on information technology that could fail is a relatively new aspect of healthcare. Clinicians will still need to be fully trained in clinical medicine and remain leery of clinical decision support tools that provide feedback that is not clinically accurate or fail to provide feedback that may impact patient care.

Michael Stearns, MD, CPC, CFPC