According to one healthcare intelligence firm, the depth of coding was around four codes on average per patient in June 2009, up from an average of three in 2005. Today’s average is around five codes per patient.

Depth of coding is an arbitrary measurement, however, and something which is not defined by the NHS Clinical Classifications Service. From a coding perspective, the number of ICD 10 codes [assigned in an episode] will always depend on the condition(s) and comorbidities recorded by the responsible consultant. Equally, the source document will play a significant part in dictating the number of codes that can be captured as there will be variance according to whether the full case notes/electronic health record is used as opposed to a proforma/electronic discharge notification.

For a multi-site acute NHS trust, being able to compare the variance in number of codes for the same activity can help to identify issues which may be the result of incomplete clinical documentation at a particular site or by specific clinical team or due to coder (in)experience.

The quoted national average for depth of coding is prone to being artificially inflated if organisations are including irrelevant codes such as ‘personal history of……’ in an attempt to reflect conditions from which the patient once suffered, but which have no bearing on the current health state.

It could be argued that, as long as clinical coders are following the four step coding process, applying the national standards and including all conditions as stated on the mandated comorbidity list, it is not for them to decide what conditions are clinically relevant to include when completing the coding process. This means those responsible for measuring and reporting depth of coding should do so with caution and not show an organisation, considered to be ‘lower than the national average’, in charts using colours that can be interpreted as poor performance.

There is significant emphasis on a coding department to increase its depth of coding. It is worth remembering that this goes against the clinical coding standard described as Coding Uniformity.The standard states:

“Uniformity means that whatever a given condition or reason for a consultant episode is coded, the same code is always used to represent that condition for the encounter. Uniformity is essential if the information is to be useful and comparable.”

General rules for accurate selection of codes apply:

• Code the minimum number of codes which accurately reflect the patient’s condition during the consultant episode
• Code every condition or reason for the encounter which affects the care, or influences health status during the consultant episode, which is available in the classification and supported by the medical record
• Code each problem to the furthest level of specificity i.e. third, fourth, or fifth character, which is available in the classification and supported by the medical record
• Do not code background information or chronic problems which are no longer active and which do not influence the health care being provided in the relevant consultant episode. It is not always intended that symptoms or history be coded. Just because a condition can be coded does not mean it should be coded each time the patient is admitted. Any uncertainty around issues of relevance or inactive problems should be discussed with the responsible consultant. (Ref: National Clinical Coding Standards ICD 10 4th Edition, NHS Clinical Classification Service April 2015).

Denise Blackman ACC
Head of Clinical Coding
East Kent Hospitals University NHS Foundation Trust
deniseblackman@nhs.net