The stumbling block, or bottleneck of electronic medical records is usually the interface between physician and machine. The goal is to capture the data the physician thinks is important, efficiently and accurately and affordably. Here are the major present-day answers to the problem:
Dictation to a medical transcriptionist is the gold standard to compare anything to. A dictated note can be typed into a word processor or any sort of database program. That is the fastest way for the doctor to say just what they want in the record. It is expensive to transcribe, however, and isn't available for a variably long while after the encounter is done.
Dictation into computerized a voice recognition program is a developing technology. In the comparative demonstration (see previous page) the two EMR's that used voice recognition were about 90-95% accurate, leaving a host of incorrect words, some of which made humorous sentences. They were awkward to correct. The documentation of the visit was painfully slow, with corrections being made a few seconds after the words were spoken, and other corrections missed completely.
This does, however, avoid the cost of medical transcription. Also, the note is available immediately, for the patient to read or take to a referring doctor, for example. My friend who uses it in his internal medicine practice likes it.
My prediction is that a future generation of voice recognition software will use the SnoMed or similar medical terminology database along with one of several attempts to organize medical words by meaning and context, to "guess" what words would fit in the context of a sentence. Then the dictation will be 98+ % accurate, more like a medical transcriptionist. That may require the next generation of computer hardware to do, too.
Boilerplate templates consist of pre-written sentences or phrases, with blanks that can be filled in by clicking on a list of choices or typed in. This is faster than typing the whole text, but limits the structure of the text to what was planned ahead of time.
Cascading glossary templates allow choosing a succession of words to generate a phrase. For example clicking on "Heart" opens up a group of words including "Rhythm" which opens up "Regular" and "Rapid." The sentence will then say "Heart rhythm is regular and rapid." This is the format our AutoChart uses. It is more flexible in what phrases can be made, but requires every possible combination to be programmed in to begin with. Thus, some people simply program in long, wordy phrases. For instance, "Heart" and "Normal" could generate the phrase, "The heart rhythm is regular with a normal rate. There are no murmers, gallups or rubs. The paplpation of the point of maximal impact is in the normal position." The user can then highlight and delete the text that doesn't apply.
Combinations of the above techniques are used by most vendors. They like the templates, because they record discrete data points that can theoretically be used to research quality concerns, etc. For example, you could query, " How many patients with asthma had rapid heart beats?" However, I have yet to see this put to any practical use, and would not advise weighing that as a benefit in any decision about software for the forseeable future (about 3 years.) You could do almost as well querying free text notes for the words "Asthma" and "rapid" and "heart*" at this point, since the templates aren't really standardized or used that consistently.
Templates can be faster than dictation in a few specfic situations, including documenting a physical examination (especially a normal exam, as in pediatrics). However, I can say from experience that the overall effect is to slow down data collection. For example, history-taking often doesn't lend itself to pre-defined phrases and has to be augmented by typing or dictation to be inserted into the note. Or, for example, generating a screening sigmoidoscopy note may only take 30 seconds, but it takes much longer than that to get to a computer, pull up the patient's record, open a new note importing the correct template, document the procedure, then save the note. Therefore, we usually just dictate them. Dictating also gives the luxury of saying just what we think is important, with emphasis in the right places, while templates can sound stilted and artificial.
We hoped our EMR would speed us up, make us more efficient and save us money. It did none of that, but I have a different perspective now. I feel like the quality of my care is enhanced by keeping an up-to-date problem list so that I don't forget to address something in a complicated visit, and the readable notes are available immediately for sending with a patient to a referral doctor. Medication errors should be lessened by legible prescriptions and cross-referencing programs, though that benefit could be better achieved by a prescription writer program better than ours.
