Kinkajou : What do patients think of doctors utilising computer data systems?
Erasmus : One of the complaints you here is that patients can become very upset that the doctors never look at them. The doctors spend all their time looking at the screen or the keyboard typing and recording information and ignoring the patient. They spend a lot less time interacting with the person in front of them because they are busy performing another task.
Patients need Medical Records
So the requirements of data entry actually produce extra stresses in the doctor-patient relationship. People no longer look at each other or interact with each other.
Why Computerise Medical records:
Kinkajou : So what is driving computerisation?
Erasmus : The key factor driving computerisation is the removal of paper. A typical doctor’s surgery for one doctor could easily have a filing system stack of single leaf paper weighing 2000 kg, over 10 years.
This would occupy a wall say 8 m wide and 2 ½ m tall, just for a single doctor. Imagine the data file that five doctors could generate over a decade. Accessing and organising this information is very difficult. Files are removed, used and then need to be replaced. This utilises the services of at least a half-time person per single doctor just finding and replacing files.
By computerising the information, all the files are stored on a computer hard drive. This means you don’t need to buy paper, you don’t need to store paper, and you don’t need someone to perform all these tasks for you.
It is more important to save money in processing the files that is to worry about what is recorded in them.
This again is a situation that the design of the medical health systems brings about. This situation does not need to exist.
Kinkajou :
What methods are therefore important to input data into a computerised health system?
Dr Xxxxx :
- Typing
- Mouse operations
- Writing (OCR of handwritten text)
- Voice input
- Touch pen on touchscreens
- Hand driven tablet graphics and Picture coding
- Computerised cut-and-paste commands
- machine generated data such as ECGs, biometric data, Spirometric data, gas readings
- attached files such as scanned letters
Dr Xxxxx :I think a mature system would utilise any or all of these options, as under specific circumstances different solutions may be better suited to different circumstances.
Currently most Data input into computerised medical systems must be text and/or added drawings. (And also in English text only).
Once input, data must exist as text in a particular language, there are also becomes a need to ensure that the data is spell checked for correctness. This can substantially slow down consultations or cause automatic errors due to “AutoCorrect” features misinterpreting data.
Electronic Medical Records Structure
Dr Xxxxx : Some data input directly by humans can bypass the need to be spelt correctly, as for example arrows for showing increasing or decreasing trends in symptom progression. Coding this data element gives you only a right option or a wrong option.
There is essentially no misspelled option. (This type of data is essentially impossible to misspell, as it consists only a single symbol.) This means a simple symbol representing a string of events is easier to represent in human terms than in computer terms. (The key problem here is that computerised keyboards limit the use of logic elements and of complex descriptors that do not fit a standard keyboard.)
Text based data can also be very difficult to interpret by people with visual limitations. It can only be magnified to a specific extent, because the magnification level is set by the amount of data that needs to be represented on screen and the size of the text box limiting that data.
If there are 20 data entry boxes on a screen, each can only be made a little bit bigger. If a vision limited user enlarges all the boxes too much in an effort to make the text larger, they will essentially customise the boxes on the page out of view. They will either disappear of the page and be taken up by scroll menus horizontally and vertically or else begin to overlap each other.
Kinkajou : Two screens can be linked for the visually impaired doctors.
Dr Xxxxx :Remember the medical environment is a high volume, low margin industry and so often work places are cramped. To have two screens, you need space for two screens.
PICTs in Computerised Medical Records
Kinkajou : PICTS are different?
Dr Xxxxx : Where a data entry system such as a PICT is used, these can be displayed at significant magnifications because numbers of unused PICTs do not all need to be displayed on the screen. They are selected individually or in small groups to be displayed.
A PICT based data system also allows far more speed and customisation to a power user. On a computer screen everything is essentially customised by the computer programmer. On a PICT screen, the user himself or herself directly selects the data input grid (versus the programmer customised computer screen “one size fits all” method available in most computer systems)
Data added by people can be text, drawings, abbreviations, logic symbols, and a large range of descriptive modifiers substantially greater than the yes/no allowed by many computer systems. Strangely also much of this data can bypass language coding to directly carry concepts from one human mind to another. Data added by people can be customised easily.
PICTs are readily customisable to allow human users to select disease use/medical system views/or anatomical views to display data. These can be “auto transcribed” into other display portals for other users.
Dr Xxxxx : Probably the key issues with medical GUI interfaces are:
- the speed of use
- the ease of use
Any technology that reduces the number of keystrokes involved in data entry will create an imperative for that own use. There will be advantages for any technology that offers a greater range of data entry in terms of either:
- range of data entered: in short a better choice of “nouns”
- Data element descriptors in terms of “adjectives” or “adverbs” or “quantity descriptors” or “direction descriptors” (e.g. increasing or decreasing).
- Availability of logic elements for use
- Better options for answers to questions than simple “yes” or “no” answers (for example: isq, Increasing, decreasing, stable, not tested, sort of like that: symptoms are partially similar or suggest this type of symptoms or could be interpreted in this fashion).
Erasmus : PICTs (graphical data input structures or pictures) creating structured data at data entry points would seem to be the logical long-term solution.
There are also substantial issues to be overcome in accelerating data input through the use of new data input technologies. There are no reasons why a range of existing and new technologies cannot be utilised to input data. Different data entry solutions may be appropriate for different consulting situations. For example, entering data while the patient is present needs to use non-vocal mechanisms, perhaps most obviously touchscreen tablets using graphical input.
If the patient leaves the room, voice input mechanisms could well be used in conjunction with touchscreen tablets. Of interest using a touchscreen tablet may well obviate the use of the mouse, our current cutting-edge modern technology. By definition in using a touchscreen tablet, entering data into particular data entry point selects that data portal as well as entering data into it at the same time. The mouse is no longer needed to find the data, to allow the keyboard to enter data into it.
Dr Xxxxx : Data Input methods into modern medical computer systems.
- Typing
- Mouse operations
- Writing (OCR of handwritten text)
- Voice input
- Touch pen on touchscreens
- Hand driven tablet graphics and Picture coding
- Computerised cut-and-paste commands
- machine generated data such as ECGs, biometric data, Spirometric data, gas readings
- attached files such as scanned letters
Doctor with Tablet Inputing Data
GUI for Health
Kinkajou : The latest trend in the health industry is the progressive growth (intrusion) of computers into all health activities. At a basic level, computerised appointment systems are standard in most medical practices. Computer records are standard too.
Erasmus : There are a few flies in the ointment though. Most people think that anything to do with computers is wonderful. So few people even deign to realise that there are losses as well as gains with medical computerisation. There are a few specific drivers of computerisation in the medical industry. But they are nothing to do with quality of medical records.
Firstly, by not having to file and unfile “charts”, the average business can lose a staff member whose job it is to do the finding of files and the filing of files. This is a huge cost saving.
Secondly, paper files are bulky. The average business would need an extra-large room to hold the files. With time, as people arrive at or leave a practice this could easily become two rooms full of files.
These files need to be held for a number of years after a patient leaves a practice for legal reasons. Computer files are much less bulky and an average hard drive in this day and age can hold a host of computer files. This is easily equivalent to thousands of paper files. There are huge cost savings in terms of space and storage costs.
Thirdly, the information recorded is in plain, very legible typed English. Accessibility and readability is excellent
Medical Files Traditional
Erasmus : That’s where the good features end.
From what I have heard from medical friends of mine:
- Many people record a lot less information than in the days of paper. It “is” just a bit more difficult to type in information than to write it on paper.
- The computerised medical programs generate so much verbage that it can be hard to work out what the hell is going on, without rally concentrating and reading carefully. Most written notes can be read at a glance or scanned much more easily. Automatic verbage tends to disguise a lack of recording significant information with a patina of much less significant information. Doctors look like they are doing more, even as they do less.
- Putting a thought process on a computer record is so difficult that most people do not bother. These become the missing facets of medical records. No one complains because no one cares it’s gone. The trend to most doctors becoming employees rather than owner operators, as cost cutting continues to impact the industry, encourages a lack of concern for work Quality. Just finish it and get on with it. The record is not that important except to the government that underwrites the bills.
- Many things that doctors want to do, can’t be done. So doctors don’t bother. As an individual they have no capacity to change the programming of the medical records, so they don’t. If the fields are inappropriate or set out badly, your only choice is to use them or not, as is.
A critical or serious example is the recording of allergies. If something happens to a patient when they take a pill, often in many medical programs the only option is to record this as an allergy. For example, if an asthma medicine makes you shaky this will often be recorded either as an allergy (or not at all). The reality is that these types of medicines commonly cause this reaction.
Yes, there has been a reaction to a medicine, but it is normal under the circumstances. There is often no capacity to record this “shade” of information. Far easier to call everything an allergy, and then just worry about it when you have to.
- Scanning old information is much slower and more difficult than with paper, so doctors tend to live and work in the present. Old letters from other health providers are available, but are too slowly accessed and often erratically indexed.
Health information is recorded in all the right boxes, assuming doctors bother being responsible for records to which they only contribute a little, but accessing this information is difficult. It creates a situation where the record is all there and looks good, but practically due to time constraints in medical consultations, it cannot be used.
- Computers tend to force people to use English. This is a very verbose and inefficient way of transmitting information and hard to input visually into a human brain. It requires a number of transcription events. I.e. Words to meaning> meaning to clinical picture. This is not an efficient way of biologicals (doctors or humans) processing information. Much is said of the value of plain English, but there a cost in terms of information Quality and extent that appears as a trade-off, if plain English is used as a recording medium.
Nested menus are another thing that computers do well but humans do not. A human brain is better able to handle or linear or low branch process than a complex branching tree of choices. I have not seen any computer systems that don’t follow the nested menu scheme, no matter how foreign it may be to human logic processes.
Dr Xxxxx : Many of these issues are not so much about computers as about the medical computing interface. The interfaces currently available are “awful”, for all the above reasons. They do not have to be.
Erasmus : Doctor friends of mine have demonstrated using abbreviated paper based graphical interfaces for recording medical records. With their system, their can record, two to three times as much information as current computer based records systems and in half the time as well.
The information presents well at a glance and lends itself to heavy customisation, which a computer can reinterpret from doctor to doctor. If doctor A likes display A they can have doctor B’s information translated into this format by the computer with no loss of information carrying capacity. These default information modules or "Picts" are Quite easily transferable by doctors from system to system, using for example a USB stick with their preferred settings (defaults and default Picts) recorded.
Using this system also allows bypassing the design restrictions on current programs as the Picts are dictionary linked and would enable the definition and intertranslation of a much wider range of fields. The format allows an enhanced recording of information about each significant event in a patient’s experience.
It applies itself to a new method of diagnostic labelling which is more natural. A doctor friend of mine describes seeing a diagnosis of “non-venomous, non-toxic, nonsystemic skin rash for a skin abrasion a patient of his experienced. The description is close to useless too anyone except a bean counter.
DataBase Icons
Finally, the most powerful consequence is that information is stored in structured data fields not as text in data fields. This makes a huge difference in the long term when extra meaning or consequences are extracted from coalesced data. Currently, most medical records by comparison would not achieve anything of the sort, predominantly focussing on searching diagnosis or treatment data layers.
Kinkajou : There needs to be some big changes in the industry to let it grow to these types of record systems. But, from what you’ve told me, the technology is pretty old or mature now. It just demands a new way of thinking and a substantial redesign of medical information recording systems.
By using “Picts” as prompts it could even extend a doctor’s skill at dealing with diseases as the Picts prompt new Questions and avenues of exploration in working towards a diagnosis. It literally makes a doctor a doctor and a half. Computer systems can do what they are good at, (generating lists and never forgetting any).
Human beings can do what they are good at. (Interpretation of ambiguous responses or inappropriate responses, reasoning and deduction, and pattern recognition from raw or partially corrupted input data: people’s clinical histories)
Kinkajou : I think the progress will sort of mirror the phone industry. Lots of phone manufacturers produced phone models with no thought as to how people might like to use them. Thy made them and sold them and cared little about how these appliances interfaced with people. Then Apple appeared with its iPhone and literally blew most of them out of the water. Innovation is risky, but sometimes not to innovate is the most risky or costly of all.
Kinkajou : So what do you think Goo?
Goo : It is obvious that computer systems for recording medical information have a long way to go. The hardware needs to be reconsidered.
The software needs to be reconsidered.
The data recording method needs to be reconsidered.
The data structures are need to evolved to greater recording of structured data. I think one day we will look back on current systems and consider them to be archaic and primitive. Future systems will record more, faster, in a more structured format and provide thinking prompts for medical practitioners.
Bedside Medical Translation
Dr Xxxxx : There is a considerable upside to medical records development and one day our record system may look like a cluster of Chinese like glyphs. However, medicine is a hostile environment, always pushing for more to be done faster and cheaper. There no point having a Rolls Royce system if you just want to pop next door to the shop for a quick sausage roll.
Many medical consultations are quite mundane. There is no “extra” time to do anything. The only things that will drive the introduction of a new system are time savings, maybe some intelligent prompting for critical information and ease of extracting information from the system.
Current systems do all these things very poorly due to the focus on an inefficient record language: namely English and a reliance on verbage to make it look like you are doing a good job.