Article Text


Error classification
Classifying and identifying errors
  1. A J Avery
  1. School of Community Health Sciences, The Medical School, Queen’s Medical Centre, Nottingham NG7 2UH, UK;

    Statistics from

    There are no easy answers to developing an all encompassing medical error classification system in primary care

    Classifications can help us to make sense of the world and, in the field of medical errors, they can help us to assess what are the most important problems. There have been numerous attempts to develop classifications of medical error, although relatively few of these have come from primary care. This commentary highlights some issues in the classification of medical errors and considers the case for more systematic attempts to determine the incidence of these errors.

    One of the problems with classifying medical errors is that there are so many ways of doing it. For example, one can focus on processes such diagnosis1 or describe underlying system failures.2 Also, one can classify errors in terms of the types of disease, drug, or procedure most commonly associated with error or in terms of the severity of outcomes.

    The paper by Rubin and colleagues3 in this issue of QSHC provides a classification containing six categories of error. The classification was based on broad themes coming out of the analysis of 65 events and it may prove to be useful in primary care along with other classifications.4 In common with other classification systems, there is room for overlap between categories. For example, a “clinical error” might also be a “communication error”. Some investigators have tried to get around this problem by “cross cutting” between different types of classification. For example, Bhasale and colleagues1 have classified errors in primary care in terms of types of incident such as those relating to prescribing or diagnosis. They have then described contributing factors such as communication problems or errors of judgement. Nevertheless, Bhasale et al note that there is still overlap between some of the categories that they developed.

    Most attempts to classify errors in primary care have been based upon the reporting of incidents by health professionals.1,3–5 This type of approach provides valuable information but it is problematic because of under reporting and reporting bias. This means that classifications will be based on incomplete data and estimates of the incidence of error are likely to be inaccurate. It is therefore worth asking if there are better alternatives.

    There is little doubt that the reporting of incidents by health professionals can provide useful feedback on errors. This is the approach that is being adopted by the National Patient Safety Agency in the UK with the aim of helping the NHS to learn more from mistakes. This type of approach has been used successfully in other industries and there are good reasons to expect it to have a positive impact in health care. Nevertheless, a more systematic approach to identifying errors could provide better evidence on which to identify priorities for action. This has occurred in the field of prescribing in primary care5 where, for example, a recent prospective cohort study has examined the frequency of preventable adverse events resulting from prescriptions,6 and another study has identified the errors in medicines management associated with preventable medication related admissions to hospital.7

    In secondary care there have been a number of descriptive studies documenting the nature and scale of errors. Rather than using incident reports from healthcare professionals, many of these studies have worked on the basis of detailed review of patients’ records, sometimes with further information being collected from healthcare staff. Review of patients’ records could be used in primary care, although there might be problems due to incomplete documentation. A combination of record review and interviews with staff and patients might therefore be more successful in providing the information needed to judge whether errors have occurred and their underlying causes. This approach would lend itself particularly well to analysis of serious adverse events such as hospital admissions, but it might also work if a study were to be done of all healthcare encounters in a primary care setting. Even so, this approach would not pick up errors such as the equipment failures identified by Rubin and colleagues.3

    It seems, therefore, that there are no easy answers when it comes to the development of all encompassing classification systems for medical error in primary care. Approaches based on self-reporting clearly provide useful information and more robust descriptive studies would still not give the full picture. Nevertheless, I think it would be worthwhile undertaking more systematic analyses of primary care encounters to determine the most important errors and how frequently they occur.

    There are no easy answers to developing an all encompassing medical error classification system in primary care

    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.