If you listen to healthcare industry chatter, you’ll hear that we are in the “post-EHR” world, in which everyone has electronic health records and, with the requirements of the 21st Century Cures Act, systems will soon be required to share data. We may be getting ahead of ourselves because users still cannot find what they are looking for––even in their own systems––and this situation will only get worse when data begins flowing in from others.

This is not a new revelation, as the challenge was raised in a 2017 interview with HealthIT Answers, when we spoke about the coming “data tsunami” (which has since gained buzzword status).

Before we can solve the problem of finding relevant information in the incoming data, we need to address the fact that users of current “legacy” EHRs have trouble finding relevant clinical information in their own systems. Sure, they have a problem list, medication list, lab orders and results, physician and nursing notes and documentation, and other information in a variety of formats. Some is recorded using standard terminologies and code sets such as ICD, CPT, LOINC, RxNorm, SNOMED, etc., but much of it is in free-text notes, and users must usually navigate to a separate section of a chart to see items in a specific domain.

The result is that users spend far too much time searching for items that are clinically relevant for a specific problem. Most current EHRs have no way to provide a diagnostically holistic view for a specific clinical condition. Let’s use renal failure as an example. Clinicians may see a problem list with renal failure listed. But they must navigate to the medication list, then to the laboratory results, then to the encounter notes to determine what is happening with the patient’s renal status––and if the treatment is working and the condition is being managed effectively. The transition to value-based care will only make this problem more acute and will require a new set of tools.

These tools must work with legacy EHRs, since those are not going away anytime soon. The good news is that these systems do have a problem list coded in either ICD-10-CM or SNOMED, laboratory data often coded to CPT or LOINC, and medication lists coded to RxNorm. Unfortunately, all those codes and terminologies are typically located in separate sections of the EHR, have different formats and data schemas, and are not designed to work together to support clinicians at the point of care.

Now, factor in the need to allow clinicians to see the information from free-text encounter notes. Every time they want to review a chart, they must open the dumpster, dive in, and hope they can find what they want––with some in a coded format and some in free-text notes. Of course, the data is probably in there somewhere, but it can take a while to find it.

The good news is that there is a way to avoid repeated dives into the dumpster: enable users to select any item from the problem list, and instantly see the clinically relevant information from all the different places and various formats in the EHR. This requires the integration of a clinical relevancy data engine, with links between diagnoses, medications, lab orders and results, and history and physical examination findings, mapped to the relevant terminologies and code sets in each domain, and made available at the point of care.

Fortunately for clinical users, such solutions are available for integration with today’s EHRs, and can serve as essential building blocks for the next generation of systems to support value-based care.

Dumpster Diving for Data in the 21st Century

So, we’ve discussed the difficulty of finding clinically relevant information in a patient medical record. With the advent of the 21st Century Cures Act, this situation will become even more challenging.

Once the floodgates are opened and systems are required to share information and are sending it back-and-forth, the contents of someone else’s dumpster will be added to your own. I might reach into my own trash can to find something, but I am going to be a lot more hesitant to go digging around in someone else’s garbage. If interoperability is going to improve outcomes and contribute to the success of value-based care, new tools are needed to support integration of relevant clinical data with specific documentation, workflow, and reporting requirements.

This raises a major question: should these tools be based on post-encounter analytics, integrated at the point of care, or both? First, let us look at where we are now and some of the current approaches to solving these problems.

Currently there is a lot of focus on using emerging technologies to capture the encounter and to limit the documentation burden for the clinician. These include ambient artificial intelligence, using microphones to capture sound, turn it into text notes, and applying natural language processing to identify relevant clinical data.

Another much talked about approach is to use scribes to record information. These solutions do not address the challenge of getting clean structured clinical data into the record in the first place. And, with the increasing requirements for data and outcomes reporting, such as with electronic clinical quality measures (eCQMs) and the management of Medicare Advantage patients using hierarchical condition codes (HCCs), today’s legacy EHRs often rely on what they are calling “curation”, using a mix of terminologies and code sets that were designed to meet billing and transaction requirements from the last century.

Imagine if banks captured sound, turned it into text, then analyzed that to determine how much money was in your account, with an error rate of more than ten to twenty percent. Then to find the errors, manual reviewers checked the information and reached out to you to correct any errors.

That, in effect, is what is happening in health care right now with patient medical records. Rather than providing clinicians with the tools they need to find, and record, clinical data at the point of care, systems try to shield providers from the lack of clinical usability of today’s legacy EHRs by using after-the-fact analytics to identify issues in patient care.

There is a better way forward.

Phoenix Children’s Hospital took a proactive approach by asking their clinicians what they need from an EHR. The answer that came back was not surprising: give instant access to the key clinical indicators needed for each specific patient and their specific conditions, organized the way clinicians think and work, summarizing the status of each area of concern, and enabling them to act upon each item at the point of care. Clinicians did not want to worry about coding or reporting and wanted documentation to be done as a result of care, and not as the focus of the encounter.

The hospital had to find a solution that could work with their Allscripts Sunrise Clinical Manager EHR and, fortunately, Allscripts agreed to work with Phoenix Children’s Hospital, if such a solution could be found.

Medicomp’s Quippe solutions were integrated with the existing EHR to provide what the clinicians wanted by using a clinical relevancy engine with workflow customization tools to drive patient-specific dashboards, actions, and documentation. Key clinical indicators were used to drive the capture of structured data needed to manage care and to track, and improve, patient outcomes.

Phoenix Children’s Hospital has found that by working with clinical users to identify what data is needed, integrating it into clinical views and workflows, and linking data to action, providers can be given what they want, the data is available for coding and reporting, users are more productive, and patient outcomes are improved.

And dumpster diving is a thing of the past.