This is the second article in a series about the importance of including technology “usability,” or records-1552781-1920x1440ease-of-use, criteria in the technology procurement process. Thanks to all who commented on or followed up with me personally about the earlier article. It’s exciting to hear that a significant and growing community understands the importance of usability for achieving the promises of healthcare technology.

The previous article highlighted the productivity costs of of poor usability – potentially millions of dollars/year wasted over the life of a product! In this article, we focus on the importance of technology usability – specifically health IT usability – for patient safety.

Poor EHR usability is a known source of patient care errors!

While EHRs and other health IT applications have led to decreases in some types of patient care errors (e.g., misinterpretations of physician handwriting and lost papers), the use of healthcare IT has not eliminated all errors and, in some cases, has proven to be a rich source of new errors. In a report prepared for the ONC, Mardon et al (2014) noted that “health IT-related errors occurred at nearly every stage of the care delivery process.” For instance, medication-related errors occurred at the reconciliation, ordering, administration, and post-administration monitoring stages.

Even before the HITECH Act made EHRs commonplace, academic studies highlighted the impact of health IT usability on patient safety and clinical outcomes. The following list is a sampling of recurring EHR use-related errors appearing in academic reviews of incident report logs:

  • Documenting on the wrong patient’s chart
  • Errors in selecting orders to be cancelled or continued
  • Duplicated or missed care tasks
  • Test results overlooked or misinterpreted
  • Medication dosing errors
  • Wrong blood product given or given to the wrong patient

Seasoned clinicians will note that these same errors occurred in the paper chart world! What makes errors in the EHR even more dangerous, however, is the speed with which a data entry, diagnostic, or treatment error propagates through the care delivery system. Data entered at one point in the care delivery process are immediately visible to anyone with access to that record, potentially impacting diagnostic and care decisions by clinicians in other departments or facilities.

In the report prepared for the ONC using data from two Patient Safety Organizations (PSO), Mardon et al found that 60% – 68% of patient safety events involving health IT reached the level of “incident,” meaning a care delivery error occurred. An additional 14-15% were “near misses,” meaning they persisted up to the point of care but were caught before an actual care error occurred. While events reported as “health IT-related” in that sample were less likely to result in injury or death, they were also more likely to be deemed “preventable.”

EHR usability problems are well-documented.

The list of known usability issues leading to the types of errors listed above is extensive. Drawing on the ONC report and Linda and Craig Harrington’s book,Usability Evaluation Handbook for Electronic Health Records in addition to our own experience, the following list represents a sampling of known usability issues observed in multiple EHRs:

  • Difficult to read displays (gray text; small fonts)
  • Numbers (e.g., weight) presented without labels
  • Icons that are non-intuitive to clinicians
  • Use of non-standard nomenclature
  • Related functions are not logically grouped
  • Inconsistent use of color and bolding to indicate important or urgent information
  • No clear workflow – clinicians uncertain whether they have completed the documentation requirements for a particular step in the care process
  • Default units of measure (e.g., English v. Metric; grams v. kilograms) are not apparent or are inconsistent with common practice
  • Drop-down menus have no discernible order (e.g., alphabetical) so users must scan entire list
  • Multiple tabs or pages with the same label (e.g., “Medications”) but with different content and functionality – users have to learn and remember the functions available on each page

When the interface doesn’t match rushed clinicians’ mental models and vocabulary, important information gets buried in free-text fields where it is likely to be overlooked by the next rushed clinician who relies on the structured information display to make care decisions.

“We cannot continue to give staff more work and poor tools, then admonish them to ‘be more careful.’ These are system problems that call for systemic solutions.”

These are not rare or isolated problems –a single screen “header,” the top section of the screen which contains basic patient information such as name, allergies, weight, and medical record number (Harrington & Harrington, 2014), may contain numerous usability gaffes. Such problems are only further compounded when the interface displays more complex data, such as test results, medication histories, and complex surgical reports.

The good news is that many usability problems are easily corrected!

So why is the usability of many EHRs still so poor?

Given that (a) usability issues are not impossible to fix; (b) the consequences of an adverse patient event are significant for all involved; and (c) the relationship between IT usability and patient safety is well known, how did confusing, difficult-to-use EHRs make it onto the market?

Good question. Unfortunately, there is no simple answer.

  • The FDA has long required documentation of “user-centered design” practices and usability testing before approving medical devices for clinical use, but EHRs and related health IT products do not fall under the purview of the FDA.
  • The ONC has included a few usability measures in the EHR Certification criteria for such high-risk functions as medication allergies and drug-drug interaction alerts, but an independent study of EHR vendors showed that only one-third of the sampled vendors had fully complied.
  • The FDA also monitors and tracks device-related errors, including those resulting from poor usability, but no equivalent performance monitoring exists for health IT. The Common Format Version 1.2(2012) for voluntary reporting of patient safety issues does include fields for indicating health IT involvement in a safety incident, but use of the Common Format and incident reporting are both voluntary.

In addition, it has been our own experience that EHR vendors often “reinvent the wheel” for each client rather than continuously improving their product. For instance, in one implementation of an enterprise EHR in a multi-site clinic organization, one of the physicians familiar with an earlier version of the product complained, “They’ve been in business for twenty years but still don’t have a decent Urgent Care form!”

Furthermore, healthcare facilities purchasing the EHRs often lack the internal resources to assess the usability of these “new” modules as they are presented. For example, in another facility, the implementation team was presented with a “hot-off-the-press” template for documenting in-office biopsy collection, a template every other multi-specialty clinic customer must have required. The template was riddled with usability problems, some severe enough to have resulted in procedures being performed on the wrong body part! The team leaders did not recognize the problem until a lower level team member with previous usability testing experience brought it to their attention.

“They’ve been in business for twenty years and still don’t have a decent urgent care form!”  – Physician in EHR training

To be fair to the EHR vendors, it’s also important to appreciate that creating an easy-to-use EHR interface is a huge challenge. Medical data are vast (terabytes) and the information collected and used varies greatly across specialties – numerical inputs, free text, x-ray images, EKG and EEG tracings, audio files, pressure readings…You get the picture. Working on tight schedules and with limited budgets, product development teams have to make trade-off decisions, and if the client isn’t demanding a good user experience (UX), then UX won’t be high on the vendor’s list of product requirements.

So what’s a CIO or CNO to do?

RAISE THE BAR! We can’t turn back time, but neither do we have to accept the status quo. Begin by educating yourself a bit about usability – check Wikipedia or skim the Executive Summary of the HIMSS and AHRQ reports (see “Additional Resources”). You don’t have to become an expert. You only need to learn enough to ask tough questions and know that better usability is possible.

Before buying a new EHR or related health IT application (e.g., CDS, patient portal):

1. Budget time and money for comparative usability testing beforechoosing a vendor.

Ad hoc evaluations by staff are helpful but do not typically result in solid performance data for a consistent set of tasks. If you don’t have a usability professional on staff, hire professionals to do the testing for you. It will be money well spent.

2. Do a Risk Analysis of any problems identified in the testing:

How could the user errors observed in testing impact patient care? This may be included in the analysis completed by the usability consultants, but your own clinical staff leaders should review the issues from the perspective of your unique care routines.

3. DO check references:

Ask the vendor for the names of customer facilities similar to your own and speak with both clinical and technical people there about the types of errors they are experiencing.

4. Work with your legal team to add usability requirements to the vendor contract:

(a) Require the vendor to improve any interface issues identified in testing before implementation. This should NOT be treated as a customization – your organization has paid to help them identify usability problems that will help them to improve their product.

(b) Ask for indemnification from the vendor for usability-related errors. This will require your organization to closely track errors related to EHR use.

(c) Ensure the contract does NOT contain any type of “gag” clause preventing your organization from disclosing your incident tracking data to the ONC, KLAS, or other current or potential customers.

If your organization is dissatisfied with your current EHR/CDS/patient portal:

The first step is to know that better is possible! Your contract terms, financial situation, and network dependencies will determine whether your best option is to rip-and-replace or to improve the current product, but don’t think you have to be stuck with the status quo!