CLINICAL NOTES
Compare three types of clinical notes. The Lloyd George record which was the norm in general practice until twenty years ago consists of an A5 opened ended wallet containing a collection of strung together cards with enough space for brief, sometimes indecipherable, notes along with a folded bundled of often trimmed correspondence and results which had to be unfolded and held flat for perusal before being stuffed back into the wallet. The A4 record which succeeded it is a sort of expandable ring folder with a section for extensive clinical notes and another for correspondence, results, prevention, health visitor notes, summary sheet and so on which can be flipped through and scanned or read like a reference book. The computer record typically has plenty of space, requires technical expertise to open, read and add to, is littered with abbreviations and acronyms devised by software writers and their clerical staff rather than by doctors, and files letters and results in such a way that no one page can be viewed in its entirety on the screen, and flipping through is impossible because each document has to be opened and closed individually. In many ways we have taken a step backwards towards Lloyd George days.
The computerised record has two separate advantages over the paper record. Firstly, the record can be available to many people in many sites simultaneously. Secondly, the record can be manipulated and interrogated. These two discrete aspects of computerisation have become confused, with the result that, while some problems have been solved, new ones have unnecessarily been created.
The clinical record is significantly behind other areas of computing. For instance if correspondence used slideshow or photography software I could see many letters on one screen and select the one I want to read; and if clinical notes used spreadsheet software I could hide all the administration entries such as site of encounter and read only the clinical notes.
Some of the problems - and solutions - are elementary. At present, before beginning to write a clinical note, even having set my user options as well as the software allows, I have to perform some combination of: typing in a code or word, selecting from a menu, clicking, and pressing the return key. But the greater part of the clinical record requires no coding and the computer knows the date, time, who I am and where the patient is (from the appointment system) so all I should need to do is to start typing.
When I try to read previous clinical notes I have to pick my way through all sorts of debris - irrelevant, meaningless, nonsensical, fragments: entries with punctuation removed, field names without entries dumped by referral letter software, duplicated words, redundant references to documents that I can see in the document software - in order to find anything of any medical significance. Even were this debris removed our current display defaults to reverse chronology presenting information as a disjointed series of events rather than a continuously unfolding narrative. And for some reason problems are listed in alphabetical order so that we have no sense of major diseases, minor ailments or life events occurring in sequence.
Apart from being inconvenient to use, clinical software systematically leads to inaccuracies in the records. Software for calculating body mass index sometimes inserts height coded as a contemporaneous examination finding when the measurement might be historical and therefore, for a child or young adult, incorrect. Historical drug allergies have to be recorded against a specific Read code even when this is not known, so the user guesses. Past events need a precise date which is usually not known and therefore invented - it looks as if an awful lot of things used to happen on New Year's Day! This reflects poorly on the reliability of the computer record and on our professionalism.
For years I had hoped that things would improve but for every forward step there is at least one backward step. It is only if ungeeky, jobbing, general practitioners - wanting the computer as our servant rather than our master - are at the centre of the design process that the future might be brighter than the present.
PRESCRIBING
Repeat prescribing is the soft underbelly of general practice: we write a prescription of a drug without reviewing the patient and encourage messages to be passed via intermediaries about extra supplies, changes of formulation, additional drugs that were prescribed at the last consultation which the patient would like to have again, and so on. Strangely computer prescribing software, in other respects tediously cluttered with safety features, warnings and rules, has been written with this this unsatisfactory and medicolegally hazardous practice of repeat prescribing at its core.
In our current computer system when I want to prescribe something for the first time I am invited to indicate whether this is to be an acute or a repeat. I cannot possibly know. It makes no clinical sense. I can only decide when I see what effect it has on the patient.
The steps I have to take when I want to re-prescribe a drug that is already on the screen depend on whether the prescription had been set up as an acute or a repeat. This is not intuitive to me nor, as far as I can tell, to any of my general practitioner colleagues. We have to be shown the difference by a computer trainer and remember it in the middle of the consultation. This is an extra, irrelevant, meaningless mental and physical task taking our attention away from patient care.
When I refer a patient to hospital with gastrointestinal bleeding, it is of little importance medically or in communicating with the hospital whether the diclofenac they had been taking was "on acute or repeat" or whether, on discharge, it is continued "on acute or repeat"; what matters is whether or not the patient had been taking it at the time and is to continue taking it.
The consultation is an ideal time to review medication and decide with the patient which prescriptions they may obtain via reception and which ones require a further consultation. The mental process might take a few seconds. Discussion with the patient takes further time. Labeling the drugs as "acute" or "repeat" should take no more than a second per drug. In our system I have to open up each drug in turn and change the number of authorisations, which takes brain, hand and eye away from the patient for more time than I can afford in most consultations. And this part of the process has no medical content: it is simply administrative. When this task is complete and I note that I have done a medication review it remains unspecified whether or not I have asked the patient which drugs they are taking or told them which ones they should be taking: the next doctor seeing the patient is unable to deduce from my record which drugs my medication review refers to unless I enter this information as free text.
There are many other computer foibles that we have to contend with. Amongst these are the windows that pop up requiring us to justify deleting a prescription or prescribing despite a warning. I suspect they result from some sort of conflation of factors relating to patient care, record keeping, clarity of the clinical record, audit, litigation, billing and paperless prescribing. The result is a guddle. A prescription on the computer doesn't have any effect on the patient: it is an epiphenomenon and a record. It is the doctor's signature that is the crucial step leading to cure, care, kill, complaint or court. I can't trust the warnings because there are too many false positives and false negatives. I have just tested our system: it takes the same number of clicks (twelve) to prescribe either paracetamol or warfarin and the number of interaction warnings is the same. As with handwritten prescriptions, I must depend on - and continue to develop - my own skill and knowledge to achieve safe prescribing.
So administrative rather than clinical factors have dictated the design of prescribing software and the emancipation that came with word-processing has been replaced by tyranny.
I have just had a paper published in the BJGP (The primrary care electronic heatlh record: who't righting the software. BJGP, Feb 2011, Pp152-154) and a new organisation has been set up in Scotland called SNUG.