Speech vs. Pen vs. Keyboard - A question of Interface

I have created a new discussion area called 'Data Input' because this is the critical barrier that faces many physicians who use EMR systems. There is a constant debate raging about the best methods to input data into an EMR system.

Should one use a Tablet and pen because this more closely matches the way that physicians have always worked? Is speech recognition input critical if physicians are going to adopt EMR systems in significant numbers? Will it still be necessary to have access to a keyboard so that information can be keyed into the computer? How important is the data input interface in the process of the clinical encounter. My answer to these questions is Yes, Yes, Yes and Extremely.

As EMR systems mature, they will naturally evolve to match the style of practice of the physician user. If systems do not do this or are not capable of evolution becuase of initial design, they will become relics and will be absorbed by competitors or disappear entirely. In the same physician office, I can see a need for all three methods of data input. The pen and tablet work very fast in a setting that requires numerous quick decision e.g. prescribing. It is much quicker to move a pen over a touch interface than a mouse. Speech works well in situations that require extensive data input - e.g. dictation of a specialty consult letter or a medico-legal decritpion of a motor vehicle accident. The keyboard is the old standalone input device that has become indispensable and is good for entering the short encounter or a complex encounter that requires input of data into fields on a form or template - but does not need to be present on every computer. Obviously there is much cross-over between interfaces, but in my opinion, the average physician will require all three although at differing times.

You might make comment that you have survived without speech input, however if speech input were a viable option, how could you improve efficiency if it were available to you. Many developments are also taking place at present because of developments with the Tablet PC platform and handwriting recognition.

My final comment about interface is that of standards. We hear talk about standards for data transfer e.g. HL7, standards for coding e.g. ICD10, Snomed etc. How about standards for interface design in EMR systems? Collectively could a national body not review the most popular EMR systems currently available and put together a list of standard elements that should be present in each part of the encounter and make this information available to the vendor community. The concept behind this idea is that physicians should not have to completely relearn a new system if systems are changed or if locum physicians come into a practice for temporary relief. If there were some standards in the user interface for EMR, any physician should intuitively be able to use another EMR system without a day of training. Make sense?

This is a very interesting discussion area an one in which I have seen little written. If you would like to make a comment, please click on the 'Comments' link below this posting.

Comments

[mark roper]

As a GP involved in regional organisation of healthcare services I am very keen on maximising adoption of an EMR by all physicians in the territory. For this reason I think the Tablet and its Pen input is the way to go. Virtual paper charts with the advantage of IT. If we can say to a physician that it will be exactly as he works now but with the advantage of having medication lists and history automatically updated we will get the greatest buy in.

Let me clarify that by pen input I mean probing, icons, handwriting recognition and free text. You can even call up a keyboard format.

A big issue is the need to digitize all information in a satndard format. My research colleagues are very quite rightly keen on this. However in primary care especially, there is a need for complex notes that are just not amenable to the restricted options most emr's offer. If we allow free text, as paper charts are, but insist on digitized areas such as diagnosis, and medication we will make good progress.

Another big issue is patient interaction and acceptance.I recently purchased a video which awas a faceoff between the ten leading EMR's in the US. The goal was to do the complete chart in 10 minutes. All of them had their eyes glued to the keyboard for the whole time looking very much like the trained rats in a psychology experiment.
Pity the poor patient who is watching his physician perform like that! As for dictation software, who has time to dictate for ten minutes for every patient at the end of a busy clinic.
Free text is the fastest and most natural way of recording the encounter. If we aim at getting a minimal amount of key elements digitized in standard format then we are winning the game.

[Karim Keshavjee]

Hmmm...thanks for starting the discussion on input devices. My take on input devices: "Humans are too fast for computers, so they invented 'input devices' to slow us down enough so that computers can keep up."

Humans think at 700-900 'words' per minute.
Humans speak at 200-400 words per minute.

Keyboards require humans to type which they can only do at 30-50 words per minute.
Voice recognition requires humans to speak at 150-200 words per minute --slower than they actually do.
Hand-writing recognition requires the human to write at 25-30 words per minute --again, slower than they actually do.

The current set of input devices are inefficient. You can still get a lot more done with dictation, so don't undermine its usefulness. Most EMRs should be able to accomodate dictated notes.

Dictating a 2 minute note can potentially capture 500 words --more than enough to capture most encounters. Its getting all the details that takes time.

It is very difficult to dictate or do voice recognition in the presence of the patient. I have seen it done --it just appears too impersonal.

If you want to chart in real-time, i.e., in the presence of the patient, then typing and handwriting are unobtrusive enough to potentially work. I have seen this successfully done in the COMPETE project by many, many physicians. They typed and wrote in real-time, during the patient encounter.

I have tried handwriting recognition in the encounter --the recognition has to be very, very good to make this work. Otherwise, you are making corrections while the patient has merrily carried on talking!

Making the input work with patient interactions is an important point. Let me say that physicians need to be trained on how to intermix data entry and empathy. It is possible. I know that Kaiser Permanente in Ohio had developed some videos on how to interact with patients and how not to interact with patients while using an EMR. I have not been able to get my hands on them, but I know they were developed.

Overall, I agree with Alan's point --we will need all the different forms of input for different situations. Finding the right mix for the right situations will take a bit of trial and error, but once you have a 'flow' of charting, watch out...you'll never go back to paper.

[mark roper]

Very interesting numbers on speed of processing. Thank god we do not have to write down everything we think.

The written chart is a synthesis of that thought process highlighting those areas we find relevant.Handwriting (without recognition) has served that purpose in the doctors office for centuries. Its advantage is that it may be done at the same time as talking to the patient without too much distraction to the physician or patient. (I wonder if there are any studies on the opinions of patients on input?)

Handwriting is unfortunately illegible in many cases.(even mine sometimes)I have tried the tablet PC and it has the ability to take free text very naturally, and then the functionality to circle the key words for the recognition process. This at least is an improvement.

I agree that having as many input options as possible is best. The Tablet offers the most in my opinion.

[Ray Simkus]

User interfaces need to be intuitive. The designers of the system need to think of what information the user wants at any particuluar time and what are the next steps that the user is likely to want to do. The interface is the application as far as the users are concerned. They do not care about data structures and what type of database the application runs on. All that they see is the interface. The interface should be designed that clicks and screen changes are minimized. Meaningless clicks should be eliminated. Care should be taken to help the user avoid mistakes and to allow the user to gracefully recover when mistakes are made. As physician users we are trying to pay attention to the patient rather than doing complex data entry.

An earlier comment mentioned the speed of thinking and various data entry. This is an important point. During the patient encounter we are working at thinkspeed. It is very distruptive to be in effect stalled when you have to wait 2-3 seconds for the computer to do something.

The primary care office is a high pressure environment. The speed at which physicians work is staggering. Yet they make it look easy. This is a tough environment for an EMR to be up to the job. There are of course many advantages of having an EMR but physicians in general will just not tolerate systems that get in the way.

There is a rich literature on user interfaces and information visualization. There are a number of rules of thumb in the infovis community and EMR applications are commonly guilty of violating those rules of thumb.

[Dr. Mark Roper]

I can't believe it has been almost a year since I last commented on this topic. However I can explain.I purchased a tablet pc last october and have been enmeshed in trying different types of data entry from Windows Journal to Transnote.
I tried a few beta versions of One note emr and others. Their main failure was a lack of defined text boxes that could synchronize with databases. I flirted with a private company to make my own forms but balked at the expense.

Then I discovered InfoPath with the ink upgrade. Oh my, what a programme. I can make my own forms, benefit from all types of data entry and link easily with xml databases and excel spreadsheets.It saves my forms as read only documents in HTML. When linked to a regional or hospital database is an EMR really necesary?

This programme is included in Windows Office 2003 professional corporate edition so it is relatively ubiquitous and expensive and complicating IP right are avoided.

Worth a look, especially for us Tabletpc fans.

[Ron Joe]

I have the enterprise edition of MS Office 2003 and it includes Infopath as well. I admit the pen feature is very nice. As far as I understand, many of the British ancillary health care providers use it in conjunction with Sharepoint portal.

[Yves Raymond]

Great topic. I have tried Speech, keyboard and handwriting recognition and have come to this conclusion

Intrerface with your EMR is one thing but data entry is another.

Interfacing with your EMR is much faster using the keyboard (if the program is setup that way) then with a pen/tablet system (unless the program doesn't have any shortcuts). Every time you take your hands off the keyboard to touch the mouse, you use up a lot of time. Every time you use a drop down menu, you use up a lot of time.

The best interface for me is the keyboard when it comes to "surfing" the EMR since my EMR has tons of keyboard shortcuts that permit rapid screen switching. I daresay that my ability to quickly view patient information directly impacts on my ability to see patients faster and to more thoroughly review the chart.

My EMR does not accept voice "surfing" but this would be the only way to increase keyboarding speed. (Although this is speculation on my part)

As it pertains to charting...hands down, voice recognition wins. No matter how you slice it. Faster...faster...faster.

Once trained, you can have somebody review your dictation to ensure that what you said is what was typed and then they can correct your text.

Be careful since line speeds and connection methods can slow down any interface.

[jonathan Marcus MD]

Thanks for posting this great topic.

Hallelujah to the physician who wrote that we should have the option of having free text as an input option. This method of input allows the greatest direct interaction with the patient while still recording information. The obvious downside is that since it is not digitized, it can't be searched electronically. But as long as the Assessment and Plan portion of the SOAP model is digitized, the free text note could be easily found.

Free text is important for some of the more complex things we see in family practice such as marital problems. Unless a physician is an extremely proficient typist, free text is the way for most of us to go in this context.

When I discussed this issue with EMR vendors they looked at me like I was from Mars and tried to explain how I could fit my work patterns into their systems. It was amazing to me that so many vendors told me that they felt the one of the reasons that doctors are slow to adopt EMR's is that they are afraid of technology. But I say that if the technology doesn't fit, don't wear it.

In my opinion, we as health care professionals should demand comfortable and flexible EMRs.