Posted by: Jenny Laurello
Speech recognition, Voice recognition
Guest post by: Dr. Andres Jimenez, CEO, ImplementHIT
Recently I wrote about the importance of training and computer hardware as it pertains to the implementation of medical speech recognition (SR) technology. In this post, we will take a closer look at the importance of clinical workflow design.
Let me start by saying that there is no one-size fits all approach when rolling out a speech-enabled clinical workflow to your clinician base, which is why it’s best to begin the process by designing a clinical workflow plan. Clinical workflow describes the process of managing patients through a clinical setting. When you add SR to the mix you need to take into account the various documentation settings you find yourself in. For example, consider the middle aged, obese, diabetic patient with hypertension coming in for an annual visit. Several dictation scenarios exist; the provider speaks with, examines and takes care of the patient. The provider could either a) dictate at the point of care (least common), b) immediately following the patient encounter or c) in batches; [seeing two or more patients, sometimes an entire day’s worth before dictating]. In the worst case scenario, patient notes aren’t finished until the next day, or in some rare cases, days to weeks later.
Clinical workflow design can lean providers toward one scenario or the other with the goal of facilitating dictation and shouldn’t be arbitrary. For instance, if your SR software is on your office work station, and not your laptop, then dictating in front of your patient in the exam room is less likely. If your office workstation is on the other side of the clinic then that might further lean your providers toward dictating in batches. If you move from room-to-room or around the hospital with a laptop loaded with your SR technology (my favorite approach), then all scenarios remain possible.
Today, several EHRs even offer the ability to go ultra-mobile by speech enabling iPad-like devices to not only produce text but hands-free navigation through the patient’s electronic chart (discussed further in our next post on EHR integration). Some might consider this is just hardware selection, or even office interior design, but both should be guided by clinical workflow from the perspective of specialty, practice setting and SR use. This is because factors, such as time between seeing your patient and documenting your note, and seeing additional patients during that interval, greatly impact note quality. What’s more, if you have a busy practice, neither of these factors are in your favor.
Over 1.2 billion patient encounters are documented by physicians each year, and a single percentage point decrease in quality can have a serious nationwide sequel. Critical documentation errors are defined as errors that could impact patient safety or continuity of care, the most common of which are wrong patient, wrong drug name or dosage and left/right discrepancy. Take for example a busy orthopedic practice where the importance of laterality is easily appreciated, time is scarce and not facilitating the use of SR quickly and easily in between patients can significantly impact critical error rates. In contrast to a critical care specialist receiving a new patient at 2 a.m., where laterality can be just as important, but time and seeing additional patients before you create your note are more controllable. In this scenario clinical workflow should lean toward SR in a comfortable setting on a workstation (next to a coffee machine as well). By accounting for clinical workflow for providers in these different specialties and practice settings, you could lean them toward the voice recognition scenarios best suited for efficiency and quality.
There is one more aspect regarding clinical workflow that should be considered. Most practitioners would agree that dictating in front of the patient can be uncomfortable (usually more so for the provider). For this reason, their clinical workflows always account for SR outside of the room, creating an additional step before moving on to the next patient. For encounter types such as post-operative visits, which are less variable, common and typically uncompensated within the global period, a simple clickable template within the EHR system might allow you to complete your note before leaving the room.
In a primary care setting, consider a monogamous female patient with a urinary tract infection. Although you now have to also document to support coding, these tend to be quick lower level visits easily captured with a clickable template completed while in the room with the patient. In both scenarios the note can be completed before leaving the room, eliminating a step before moving on to the next patient, which greatly enhances overall efficiency. Of course, we must never force the patient with the uncommon complaint or complex usually straight forward complaint down this simplified documentation approach, as that would decrease note quality as well. Furthermore, we shouldn’t ignore provider documentation preferences either, which is why these types of “quick” clickable templates should always contain placeholders for custom text generated by SR to fill in the gaps when appropriate.
The key to speech recognition success is training focused initially to leap past the accuracy hurdle, followed by hardware selection (discussed in our previous post) and carefully taking into account unique specialty and practice setting workflow characteristics as part of your speech-enabled clinical workflow design. The process should ultimately facilitate dictation to ensure quality patient records, and decreased error rates. Do not forget to check back later this month for the third and final post on the importance of EHR integration toward finding success with medical speech recognition.
Dr. Andres Jimenez is CEO of ImplementHIT, a leading Health IT training firm and creators of the OptimizeHIT training platform, which is rapidly becoming the “new standard in health IT training”. Dr. Jimenez is a Nuance Healthcare physician advocate, former Allscripts Clinical Director of Content and Online Training, and is still clinically active using voice recognition for all his clinical documentation.