As you probably know GPs use Read codes to code their clinical notes. They aren’t perfect but they are well known. Any day now they are about to be replaced with Snomed CT codes. I’m not sure if this has been widely understood in general practice – few GPs I have spoken to have any idea about it – though luckily most Practice Managers have!
I guess the idea is to harmonise the codes between secondary care and primary care. I suspect some big data people think it will make doing big data easier. However, I really worry that we aren’t going to see any real benefits/ improvements we are just swapping one code for another.
No one appears to be doing anything about making choosing codes easier or creating a formulary of codes to choose from that narrows down the choices. No one appears to have cleared up some of the uncertainties of working diagnosis and definitive codes.
Patients come in breathless, not in heart failure. Its only heart failure once proved. However, do we all agree on the diagnosis? Perhaps you might say Snomed match ICD 10 diagnoses better but I can show you numerous examples of people coded with one diagnosis that subsequently changes as more evidence comes to light or thinking changes or indeed as the definition of diagnosis changes.
We used to call this associated features but do Snomed Codes tell me who made the diagnosis? where? when? how certain they were? what criteria did they use?
Do I accept a code automatically into my system from a junior doctor on a discharge slip who makes a point of mentioning they’ve never seen the patient but (s)he had urosepsis despite any evidence they did?
There is an old adage Garbage in Garbage out – there are lots of people who promote shared records yet never seem to talk about the quality of those records.
A few years ago we had the IMT DES in primary care – we were incentivised and given some tools to improve our coding but that was a one year scheme and who knows what the quality of coding is now?
I know PRIMIS still have a toolkit – I know people have lots of tools around disease register validation. No one is making anyone use them before we do the switch. I think we should be massively investing in data quality – No one is pushing coding training to clinicians and admin staff. I think this is a mistake. For big data to work it needs to be good data!