When I trained in general practice in the early 1990s, computers were gradually becoming integrated into surgery life. We still hand-wrote our notes, but prescribing was done on the PCs that sat on every consulting room desk, and important diagnoses would be entered as both paper records and electronic notes.
This was not mere duplication. Diagnostic information on the computer was entered as Read codes – a heart attack, for example, was G301; cancer of the bowel was B130. Data entered in this way could be systematically evaluated to improve care. It took just minutes to compile a list of every patient with H330 asthma, for example, and to identify any that weren’t on the right inhalers. Done by hand that would have taken days.
Read codes were named after a Loughborough GP, Dr James Read, a pioneer of medical computing. In the 1980s several rival organisations were racing to devise systems for encoding medical information. Read invested thousands of hours in authoring his. In 1988, the British Medical Association and the Royal College of General Practitioners recommended that Read codes should be adopted as standard throughout the NHS.
The original code library comprised 25,000 terms, and the attempt to describe every conceivable clinical scenario led to some wonderfully esoteric examples. In (presumably) rare instances a GP could code E845, “accident involving spacecraft injuring occupant”.
The success of Read’s system arose from the elegant logical hierarchies embedded within his design. It was like a tree, with each main branch dividing into more and more specific diagnoses. Branch B was the branch for all cancers, and this divided by organ and tissue and cell to give unique four-digit codes for each of the hundreds of types of the disease.
Before long, however, staggering advances in medical science – with dramatic increases in the numbers of investigations, treatments and procedures – outstripped Read’s original design. By the early 1990s, a five-byte version had been launched, creating much greater capacity to encode new knowledge. By the end of that decade, a further revision abandoned the hierarchical structure, which had become too constraining. By 2010, the library that started with 25,000 terms had expanded to just under 300,000.
One of Read’s most important insights was that health is affected by much more than disease. From the outset, he had codes to capture every major life event, from the death of various family members or friends, through to marriage, house moves or unemployment. And a whole class of Read codes attempted to describe the different occupations people may follow, reflecting the important impact of work on health.
If anything, it has proved more difficult to keep these latter codes updated than it has to keep pace with medical advances. While I occasionally come across a novel diagnostic test or surgical procedure that has yet to be coded, around half my patients’ job titles now have no matches in the code library. Roles such as “operations supervisor” or “business analyst” have no equivalent. James Read’s codes reflect a far simpler occupational era.
If you want to operate health care as a market, you need your chargeable activity coded. Every test and procedure my patients undergo in hospital is itemised for accounting and billing purposes. The world leader in computerised medical billing technology is the US. In 2001 the Read code system was merged with its American counterpart Snomed to form Snomed CT, which will become the NHS standard system in 2020, when the Read code is retired.
Future generations of doctors will doubtless become used to referring to “Snomed codes” and the term “Read code” will become as quaint and anachronistic as snuff. But James Read’s pioneering work in medical informatics, particularly his emphasis on codes that captured the holistic nature of health and illness, has left a legacy that will endure.
This article appears in the 27 Jun 2018 issue of the New Statesman, Germany, alone