Clinical coding is the translation of terminology used in healthcare into a coded form. It involves standardisation of the terms, and the placing of the codes in a structured hierarchy. Because healthcare is immensely complex, a system of codes where similar problems or treatments are grouped is required.
Clinical terminologies are comprehensive lists of terms used in the care, treatment and management of patients, which enable computer systems initially to store, and then to retrieve, patient information in a natural clinical language.
There are different types of clinical terminology, for different purposes. The three main types of terminology and their respective purposes are shown in Table 1 (below).
|Table 1: Functions of clinical coding systems|
|To create an electronic patient record||Read Codes, SNOMED|
|To summarise the incidence or prevalence of diseases on a national or worldwide basis||ICD-9, ICD-10, OPCS-4|
To manage a payment system in the private sector
For the majority of GPs the most important use of clinical coding is in the creation of an electronic patient record (EPR).
Clinical terminologies used for this purpose have a multitude of functions. These include:
- Allowing the development of a computer system that uses clinical data
- Creating a standard language for use in healthcare computer systems
- Enabling decision support to be performed by computer systems, e.g. PRODIGY,1 which allow an EPR to be searched to see whether a particular drug is suitable for a specific patient, given his/her medical history
- Enabling clinical management based on collated patient data, i.e. clinical audit
- Enabling research.
Clinical terminologies support structured clinical information, which in turn can be searched or reported on by a computer.
An example of a consultation using free text, clinical terminology and Read Codes is shown in Table 2 (below).
Table 2: Record of a consultation using free text, clinical terminology and Read Codes
|Free text||Clinical terminology||Read Code Version 2|
|The patient presented with a history of recurrent central chest pain. |
There was an associated swelling of the ankles and shortness of breath on minimal exertion
Examination of the cardiovascular system was normal.
|The patient was prescribed a GTN spray|| |
|The patient was referred for an ECG with a suspected diagnosis of angina pectoris|| |
|The patient was referred to a cardiologist for an opinion|| |
A clinical record that is completely free text offers little advantage over a conventional written record. A clinical record with only coded entries, with no descriptive text, provides an incomplete clinical record. Clinicians need to strike a balance between the two.
What to do
- Use a code wherever possible
- Use the appropriate code; these can usually be divided into the following groups:
- Clinical measurements
- Laboratory tests
- For major areas, such as coronary heart disease, agree a standard set of codes that will be used in the practice or across the local health organisations
- Code any information that may need to be searched for in future, if possible
- Use free text appropriately. A description is very important when recording a history.
What to avoid
- Avoid adding a qualifying statement. For example, adding a code for angina pectoris and then qualifying it with 'excluded on exercise ECG' means that any future search will detect that patient as having a diagnosis of angina pectoris. One way to achieve a meaningful entry would be to add the code for chest pain and then free text '?angina'. If the subsequent tests prove positive, the code for angina pectoris should then be added.
- Avoid mixing categories. For example, do not add the descriptive code for inguinal hernia and then add free text 'repair' once the operation has been performed.
- Avoid spending so much time coding that the entries are either meaningless or so brief as to be unsatisfactory from a medico-legal standpoint.
The different types of coding systems are listed in Table 3 (below).
|Table 3: Clinical coding systems|
|Read Version 2||Clinical record||UK standard, 4 digits, still used by many GP System suppliers|
|Read Version 3||Clinical record||UK standard, 5 digits, full implementation halted because of introduction of Snomed CT in 2003|
|Snomed RT||Clinical record||Under development for the US|
|Snomed CT||Clinical record||Under development – a merger of Read Codes Version 3 and (Clinical terms) Snomed RT, due to be introduced in the UK in 2003|
|ICD-9||Summary||Replaced by ICD-10, still in use in many countries; this contains diseases only|
|ICD-10||Summary||Used for epidemiological purposes|
|Summary||Used for operations, procedures and interventions|
Read Codes are a comprehensive list of clinical terms intended for use by healthcare professionals to describe the care and treatment given to patients. They include signs, symptoms, treatments, investigations, occupations, diagnoses and drugs and appliances.
This coding system was invented and developed in 1982 by a GP, Dr James Read, following the introduction of a GP-based computer system called Abies. Initially Read Codes were only used in general practice, but they have been significantly developed and improved over the years and are now used throughout the NHS.
In 1990, the NHS purchased the Read Codes from Dr James Read and made them the UK standard coding system for the NHS. The NHS then decided to expand the codes to cover all areas of clinical practice to include physiotherapy, nursing and health visiting, and commissioned the Clinical Terms Project. This led to the production of Version 3 Read Codes.
Read Codes are currently the best coding system available to capture a complete medical record, which can be used for audit, analysis and research. They allow coding of all diagnoses, procedures, drugs and administration.
Version 2 has approximately 105 000 codes, and version 3 approximately 270 000 codes.
The entry NOS after a clinical entry using Read Codes means 'not otherwise specified'.
Many Read Codes have a format such as 'H/O diabetes', which means history of diabetes. These codes were introduced in Version 2 Read Codes to enable storage of summary information, but are best avoided as the latest versions of Read Codes allow a complete clinical record to be created.
SNOMED stands for Systemised Nomenclature of Medicine.
SNOMED CT (Clinical Terms) is a collaboration between the College of American Pathologists (CAP) and the NHS, combining SNOMED RT and Clinical Terms Version 3 (Read Codes).
SNOMED CT is an attempt to produce a worldwide system of clinical coding to enable the development of a complete electronic health record. It covers not only medicine but also veterinary practice.
Ùt is a joint venture between the UK and the US to combine the two leading coding systems available worldwide, namely Read Version 3 and SNOMED RT. SNOMED RT is produced by CAP. The first version of Snomed CT will be available in 2002 and become more widely available in 2003.
SNOMED CT will look very much like Clinical Terms Version 3 (Read Codes), the major difference being in the realms of pathology.
International Coding of Diseases
International Coding of Diseases (ICD) is produced by the World Health Organization (WHO) and is used throughout the world for coding diseases. ICD is intended to be used for epidemiological purposes and not as a coding system for a complete medical record. There is no internationally agreed system for coding procedures.
OPCS-4 stands for the Office of Population, Censuses and Surveys – Classification of Surgical Operations and Procedures – 4th revision.
OPCS was devised for the classification of all operations and surgical procedures carried out on a patient during an episode of healthcare.
Hospitals in England are required to use this coding system. Anyone involved with the contracting process during fundholding will remember OPCS, as this was the system by which healthcare was 'bought and sold'.
The MIQUEST approach, incorporating Health Query Language, is used to collect and analyse health data. It is currently used by many projects throughout the NHS. It was developed in the early 1990s for the NHS to help with the electronic collection of health data from the major GP computer systems.
MIQUEST is now specified as part of the Requirements for Accreditation (RFA) for GP computer systems.
PRIMIS (Primary Care Information Services; www.primis.nottingham.ac.uk)2 is a national project run by a team at the Nottingham University Department of General Practice.
The project provides training and support in data management as well as data input and extraction (using MIQUEST). Many primary care organisations throughout the country have successfully implemented it.
In the continuing struggle to improve clinical care, the recording and retrieval of important health data are important. Many clinicians can see the importance of clinical coding when health data are required for population studies, but are less convinced about the benefits to individual patients.
Consider the following situations and think about which would benefit the patient more – an EPR with good clinical coded entries or a written or free text computer record:
- A drug is withdrawn after potentially fatal side-effects are identified (e.g. cerivastatin). How do you identify patients taking these drugs?
- A disease register needs to be created for all patients with coronary heart disease. Although this can be done manually, isn't it easier to do it electronically?
- You want to identify all patients aged >65 years of age who had influenza vaccination last year but have not attended for it this year, and send them a letter inviting them to attend for vaccination. Can you do this manually? If so, how long will it take?
All clinicians need to understand the importance of clinical coding and clinical terminologies, but should be reassured that they do not need to learn the individual codes or be frightened of future development.
- Watson N. Improved PRODIGY will include chronic disease management. Guidelines in Practice 2001; 4(9): 96-101.
- Watson N. PRIMIS will ensure the effective use of practice computers. Guidelines in Practice 2001; 4(7): 76-8.