Enhancing the ICD System in Recording Alcohol's Involvement in Disease and Injury

Abstract

The International Classification of Diseases, 10th revision (ICD-10), published in 1992 (World Health Organization, 1992), is the definitive international system of diagnosis, classification and coding of diseases and related health problems. As such, it is used worldwide to classify and record diagnoses in clinical practice, and to capture disease occurrence for statistical monitoring. Its broader aim is to help in reducing the burden of disease and disability through providing an internationally consistent system for facilitating communication.

The diverse roles and long history of the ICD system (Knibbs, 1929) has resulted in a degree of complexity that makes it difficult to navigate through in certain places. Currently, ICD-10's auspicing organization, the World Health Organization, is undertaking a comprehensive review and redesign with a view to publication of the 11th revision (ICD-11) in 2015. Among the tasks are an examination of the multiple places in which the involvement of alcohol and other psychoactive substances (and their associated disorders) are captured and finding out how this can be optimized for clinical and epidemiological purposes.

In a proposal by Touquet and Harris (2012), a substantial change to the ICD-10 Y91 coding system for alcohol is envisaged. Alcohol use and its related disorders appear in many sections of ICD-10 (Table 1). An important one is within Chapter V, which covers mental health and substance disorders. Here we find, in the second section, harmful alcohol use (coded as F10.1), alcohol dependence (F10.2), alcohol withdrawal syndrome (F10.3) and a number of alcohol-induced mental and neurocognitive syndromes such as alcohol withdrawal delirium (F10.4) alcohol-induced psychotic disorder (F10.5), alcohol-related amnestic syndrome (F10.6) and alcohol-induced residual psychotic and neurocognitive syndromes (F10.7).

Harmful alcohol use represents the repeated consumption of alcohol that has actually caused some form of physical or mental damage. Alcohol dependence is a psychobiological syndrome in which there is a persistent driving force to consume alcohol such that it becomes a central feature of that person's life (Edwards and Gross, 1976; Saunders and Latt, 2011). It corresponds, to an extent to earlier understandings of ‘alcoholism’ but specifically not encompassing the physical, mental and social consequences. In preliminary versions of ICD-10, there was an additional disorder ‘hazardous alcohol use’. Hazardous use denotes a level or pattern of alcohol consumption that confers the risk of harmful consequences (Babor et al., 1994). It was excluded from ICD-10 as published because it was considered not to represent a disorder as such, rather a risk factor.

Alcohol-related emergency presentations including injuries are typically classified using the primary code for the specific disease or injury. Except for alcohol overdose or poisoning, the primary injury codes in ICD-10 (S and T codes for the nature of the injury, and V-Y codes for the external causes) do not include routinely used codes indicating alcohol involvement. Instead, Y90 and Y91 are provided as ‘supplementary codes’, which have not been widely used. The Y90 section classifies alcohol involvement according to the measured blood alcohol concentration. The Y91 system denotes alcohol involvement through clinical measures of intoxication. Physical disorders seen in other areas of practice (e.g. hospital wards) typically employ other sections of the ICD-10 system. Alcoholic cirrhosis, alcoholic cardiomyopathy and alcohol-related peripheral neuropathy are, for example, found in other sections, the respective codes for these examples being K70.3, I42.6 and G62.1, respectively (Table 1).

In their paper in this issue, Touquet and Harris (2012) have suggested a radical revision to the way in which ‘evidence of alcohol involvement’ in diseases and injuries is classified. Presently, the Y91 codes are used as an alternative to the Y90 codes when the blood alcohol concentration at the time of presentation is not determined. There are a number of problems with Y91 as it is presently positioned in ICD-10:

1. The four-level clinical intoxication assessment correlates only moderately (τ = 0.3) with the blood alcohol concentration, among those reporting drinking within 6 h before their injury (Cherpitel et al., 2005)—unsurprisingly given the inherent biological variation in sensitivity to alcohol and the tolerance that develops in response to repeated alcohol consumption. Thus, Y91 is not a good approximation to Y90, as was intended in the ICD-10 formulation.

2. It seems to be used only to a limited extent worldwide, with major differences between countries in the extent and accuracy of recording. For instance, an Australian study of injury-related hospitalizations with alcohol involvement indicated in the case notes or ICD-10 codes found that only 38% were identified with an alcohol-specific diagnostic code, and that neither Y90 or Y91 was used at all (most such codes were F10, T51 and/or X45/X65/Y15) (McKenzie et al., 2010). It does not inform clinical decisions about whether an intervention is merited, something that is of considerable importance, given the evidence for the effectiveness of brief alcohol interventions and the contracts requiring such interventions that now exist [e.g. in Scotland; The Scottish Government (2011)].

The essence of Touquet and Harris' proposal is to abandon the clinical intoxication assessment (which is not applied to other psychoactive substances and can be captured in other sections of ICD-10) with a means of classifying longer-term use of alcohol. They argue that, in defining the contribution of a person's long-term alcohol use to the disease or injury, a more accurate assessment of attributable risk (in effect) will be obtained, and the level of alcohol involvement will trigger an appropriate clinical intervention. As they propose it, the person will be classified in a hierarchy from being a lifelong abstainer to current alcohol dependence.

In principle, this could work well, if it were routinely applied to injury cases and if diagnostic codes were recorded to the fourth character. Current evidence indicates that a person with hazardous alcohol use will likely respond to a brief behavioural intervention. Those with alcohol dependence will typically need a greater level of treatment, including detoxification, behavioural interventions and/or medications such as naltrexone (World Health Organization, 2010). One should emphasize that the ICD-10 F10.x diagnoses are not based on the typical level (or pattern) of alcohol consumed, as this proposal might suggest, rather on the existence of specified harm (harmful use, F10.1) or the cognitive, behavioural and/or physiological features of alcohol dependence (F10.2).

Would this merely duplicate the current F10.x series of diagnoses? Not exactly. This section of the ICD, which is used in particular by mental health and addiction services and professionals, does not include an assessment of alcohol involvement between an episode of intoxication (F10.0) and repeated use with harm (F10.1). There is nowhere in the F10.x series where different levels or patterns of alcohol consumption in themselves are captured; as mentioned above, such patterns of use are not considered disorders. Should ‘hazardous alcohol use’ be restored by way of Y91 or a successor code in ICD-11? Should there be a finer grained hierarchy of alcohol use as well as disorder in F10.x?

Possibly, a composite system could be developed for ICD-11 such that the key diagnoses of hazardous use, harmful use and alcohol dependence feature in F10.x, which would be used when an alcohol use disorder is the presenting condition. An expanded hierarchical classification system of alcohol involvement along the lines proposed by Touquet and Harris could then be employed when the presenting disorder is a physical illness or an injury. An alternative would be to use Y91 as a unidimensional supplementary code to record a current pattern of hazardous drinking (Touquet and Harris's Y91.3), which, as they note, is not now captured in the ICD coding system, and which is now seen as a signal for brief health-professional intervention.

Decisions on revisions of codes for ICD-11 in the coming months will be influenced by a variety of inputs and considerations. The suggestions by Touquet and Harris for enhancing the ICD system are valuable input for this process, pointing to the importance of codes that will be used in the emergency-department environment both for capturing alcohol's involvement and to point to the necessary therapeutic response.

A crucial consideration will be how to adjust the current coding system so that the involvement of alcohol in injuries is routinely recorded. As the Network of European NGOs Dedicated to Injury Prevention (2009) observed in a policy statement on ‘Alcohol and Injuries’:

The health sector should systematically collect information pertaining to alcohol use from all injured patients attending emergency units. This requires the inclusion of a special section recording alcohol involvement as a part of the standard surveillance form used in emergency departments (for example, by integration of ICD-10 Y90/Y91 codes into standard forms) supported by training and clinical practice guidelines.

Conflict of interest statement. J.B.S. was a member of the advisory committee for the development of the mental, behavioural and substance-use disorders section of ICD-10 from 1986 to 1994. He is currently a member of the Substance-Related and Addictive Disorders Working Group for ICD-11. As a member of the WHO Expert Advisory Panel on Drug Dependence and Alcohol Problems, R.R. advised WHO informally on ICD-10 and is a technical adviser to WHO for the Substance-Related and Addictive Disorders Working Group for ICD-11. The views expressed in this paper are the authors' own and do not represent those of the World Health Organization.

References