Careful with your coding

Originally published in GP magazine .

Written for GP magazine around 2003. This is the version I submitted - the published article may have been edited.

When I first started general practice as a registrar I hadn’t heard of Read codes, and the Windows-based computer system that my training practice at the time was using didn’t have, to my mind, the easiest of browsers to help you find the correct Read code. I was aware that I didn’t want to miscode someone as having something they didn’t have. I distinctly remember an assessment session after a couple of months during which I stated I was fed up trying to find meaningful codes and unless it was an important diagnosis I was going to put everything in under “chat to patient”. In retrospect I don’t think this was a bad policy.

A friend who is a GP at a nearby practice told me the story of one of her practice nurses who, after several training sessions on what Read codes were and how to enter them for each consultation, dutifully entered hypertension as the problem for each blood pressure reading she took. Of course some of these patients were attending for blood pressure series for raised blood pressure and didn’t turn out to have hypertension. It took several hours of admin time to sort this out.

Another related example was when we first changed to our current system: we weren’t used to using the templates. Our local MAAG installed its preferred IHD templates, which had boxes for things like MI, angina etc. These were mainly present so that the existing value would be on screen so you could see when it was. However, for quite some time people started pressing yes to the prompts and gave people second MIs. Again this took quite some time to sort out.

Does any of this matter? Yes. Apart from the fact that we are responsible for maintaining accurate notes and that giving patients diagnoses they haven’t had may have medico-legal implications, there is the more important consideration that by mislabelling people we may be causing more work for ourselves and may lose out in quality payments, assuming the new proposed system goes ahead. Conceivably, labelling a lot of normotensive patients hypertensive will make your percentage of patients with a blood pressure under target easier to hit, at least initially. However, you will still have to recall them for annual reviews. Not only is this an additional workload burden, they may be reluctant to turn up, giving you a poor attendance record at your clinic! Even if the new contract doesn’t go through in its current form, practices like mine which are PMS already have quality markers in our contract, and it would seem sensible to try to have disease registers as accurate as possible to minimise the amount of work that has to be done.

While attending a medicines management workshop recently I took the opportunity to ask several people around the table what they did with similar problems. It was interesting that although there is a push by some enthusiasts to problem code all consultations, several people didn’t put in a code, used a vague code like my registrar example above or entered a problem in free text.

Very few people had heard of or used the [D] codes. These are basically symptom codes that can be used as working diagnoses. They are in a separate Read code hierarchy, so running a search on angina patients won’t bring up anyone with these codes.

If you aren’t familiar with them it is well worth having a browse through them as some are invaluable. It might even be worth putting some up on a list on your wall. They can be found under the R0 (abnormal symptoms) and R1 (abnormal findings) chapters. I find I use them as a presenting complaint type problem and once the diagnosis is clear I then code the problem as the diagnosis. In the blood pressure example above, I would code someone with “[D] raised blood pressure reading”, which our nurses would use when seeing them, and when they come back to me I would then code them as hypertensive if appropriate.

Another classic example is chest pain. Entering a diagnosis of angina on a first consultation is rash even if it turns out to be, and most cases of chest pain I see aren’t. There are numerous [D] codes available to use including [D] Chest pain, [D] Anterior chest wall pain, [D] Pleuritic pain and [D] Chest tightness.

The symptom codes are subdivided into general, nervous and musculoskeletal, affecting skin, nutritional, cardiovascular, respiratory, digestive etc. There are several codes appropriate to dizziness, which always come in handy. Other codes I find useful are [D] Palpitations, [D] Lymph node enlargement, [D] Change in bowel habit, [D] Nausea and vomiting, Multiple symptomatology [D] and [D] Sleep disturbances.

I am not suggesting that these codes are used instead of accurate diagnostic codes. However, for a lot of my consultations they are very appropriate, because that is what the person has come in with, and until a firm diagnosis is made either by me or by secondary care it would be wrong to choose a diagnosis that turns out to be false, as this can take ages to sort out. And by problem coding consultations they are much easier to follow - indeed our computer system, and I’m sure others, allows you to filter by just one problem to allow all related consultations to be read.