Elsevier

Injury

Volume 44, Issue 1, January 2013, Pages 132-138
Injury

Comparing ICD-9 and ICD-10: The impact on intentional and unintentional injury mortality statistics in Italy and Norway

https://doi.org/10.1016/j.injury.2012.01.010Get rights and content

Abstract

Background

The international classification of diseases (ICD) provides guidelines for the collection, classification and dissemination of official cause-of-death statistics. New revisions of the ICD can potentially disrupt time trends of cause-of-death statistics and affect between-country comparisons. The aim of this study was to measure how switching from ICD-9 to ICD-10 affected mortality statistics for external causes of death, i.e. intentional and unintentional injuries, in Italy and Norway.

Methods

A sample of death certificates (N = 454,897) were selected in Italy from the first year the ICD-10 was implemented (2003) and reclassified from ICD-10 to ICD-9 by the Italian National Institute of Statistics. A sample of death certificates was also selected in Norway (N = 10,706) from the last year the ICD-9 was used (1995) and reclassified according to ICD-10 by Statistics Norway. The reclassification (double-coding) was performed by special trained personal in governmental offices responsible for official mortality statistics. Although the reclassification covered all causes of death (diseases and injuries) in the sample, our analysis focused on just one ICD chapter XX. This was external causes of mortality (injury deaths), and covered 15 selected categories of injuries.

Results

The switch from ICD-9 to ICD-10 had a significant net impact on 8 of the 15 selected categories. In Italy, accidental falls decreased by 76%; traffic accidents decreased by 9%; suicide by hanging decreased by 3%; events of undetermined intent decreased by 69%; and overall injury deaths decreased by 4%. These net decreases reflect the moving of death records from injury categories in ICD-9 to other injury or disease categories in ICD-10. In Norway, the number of records in three categories decreased significantly: transport accidents, 9%; traffic accidents, 13%; and suicide by self-poisoning, 18%. No statistically significant differences (net changes) were observed in the total number of accidents, suicides and homicides in either country.

Conclusions

Switching to ICD-10 did not change the overall trends for accidents, homicides and suicides in either country. However, the number of records in some injury subcategories e.g. accidental falls and traffic accidents, decreased. Changing classification can thus affect the ranking of causes of injury mortality, with consequences for public health policy.

Introduction

Mortality from injuries (suicide, homicide and accidents) is a public health concern worldwide. In European countries and the United States, 4–12% of all deaths are reported to be caused by intentional and unintentional injuries.1, 2 Differences in the way mortality statistics are produced hinder between-country comparisons and make it difficult to interpret variations in injury mortality statistics in different countries. 3 Official mortality registries are critical for public health planning and for risk identification, as well as for monitoring and epidemiological research on fatal injuries. Other medical data sources, such as inpatient and outpatient registries and survey data, may provide more detailed information about injuries and diseases, but mortality statistics are more broadly attainable at the international level.4, 5

The International Classification of Diseases (ICD) is used to classify and code mortality information worldwide. Changes in new revisions of the ICD, especially changes in the rules for selecting the underlying cause of death (i.e. the cause used in international mortality statistics), affect comparability and time trends for cause-specific mortality statistics.

The tenth revision of the ICD (ICD-10) is more complex than the ninth (ICD-9). Specifically, ICD-10 has more codes as well as re-structured chapters and changes in rules for coding. In particular, the chapter that deals with external causes of injury was changed radically compared to ICD-9. Some of the changes in ICD-10 could affect comparisons of statistics on external causes of mortality. Regarding coding rule changes, the changes impact some deaths classified as being due to a disease (natural cause) in ICD-9. In ICD-10, these deaths should be classified as being due to an external cause of death. So, for example, deaths classified as pneumonia or cardiac arrest in ICD-9, with ‘accident’ recorded as an associated cause of death (in part II of the death certificate), should be classified as having an accident as the underlying cause of death in ICD-10.6, 7

Furthermore, ICD-10 changed the definition of ‘accidental falls’ and ‘traffic accidents’ in such a way that a smaller number of deaths in these categories are expected. Specifically, hip fractures and fractures at other sites with unspecified causes were included in the ‘accidental falls’ category in ICD-9 but not in ICD-10. In addition, deaths involving traffic accidents in which the death certificate did not explicitly state that the involved vehicle was a motor vehicle were classified as traffic accidents in ICD-9 but not in ICD-10.8 The categories for the late effects of suicide, homicide and injury of undetermined intent were placed in the relevant categories for these causes in ICD-9, but ICD-10 removed late effects and added a new code, code Y87, ‘sequelae of intentional self-harm, assault and events of undetermined intent’. Late effects or sequelae of external causes are defined as conditions that occur at least one year after the original event.9 ‘Sequelae of accidents’ was also moved from the block for accidents in ICD-9 to the new categories for ‘sequelae of external causes of mortality’ in ICD-10 (Y85-Y89, with Y85-Y86 including the categories for accidents).9 The World Health Organization's annual list of Official ICD-10 Updates is also new; prior to ICD-10, updates were not issued between revisions.10 The second edition of ICD-10 has been published and contains all changes up to the 2003 updates.

Double and independent coding of the same death records/cause-of-death information according to two different classification systems is a tool used to measure expected and unexpected effects of moving to a new revision of the ICD. Such ICD-9/ICD-10 comparability (or bridge coding) studies conducted in different countries report major discontinuities for some causes of death. For example, the number of deaths coded as being due to accidental falls was halved in Canada when changing from ICD-9 to ICD-10.8 National bridge coding studies found no impact of ICD-10 on suicide statistics.7, 8, 11, 12 In contrast, Pearson-Nelson et al.13 used pooled (aggregated) data from 71 countries to analyse the impact of ICD changes on suicide rates and concluded that switching from ICD-9 to ICD-10 was associated with an overall decrease in the suicide rate of 0.73 suicides per 100,000 population. If this result could be applied to Italy and Norway, we would expect a reduction of 418 and 32 deaths, respectively, classified as suicide after implementation of ICD-10. The suicide rate was 7.1 per 100,000 in Italy in 2002 (ICD-9) and 2003 (ICD-10) and 12.6 and 11.8 per 100,000 in Norway in 1995 (ICD-9) and 1996 (ICD-10) respectively.

In this study, we used the same group definitions and double coding methods to measure the effects of switching from ICD-9 to ICD-10 on official mortality data in two European countries, Italy and Norway. It is important to compare findings from different countries to assess whether new revisions of the ICD have different effects on mortality data in different countries. The aim of this study was to examine the impact of changing classification from ICD-9 to ICD-10 on official mortality statistics for external causes of death, i.e. intentional (suicide, homicide) and unintentional injuries (accidents). We expected to find fewer accidental falls and traffic accidents using ICD-10 compared to ICD-9, and no impact on suicide and homicide.

Section snippets

Samples

This study used mortality data from Italy and Norway. The material we used were extracted from two samples of death records that included all causes of death and that were coded according to ICD-10 (Italy) and ICD-9 (Norway) and reclassified to ICD-9 (Italy) and ICD-10 (Norway). We used the selected ‘underlying cause of death’ to compare the classification of deaths in a single ICD chapter, external causes of mortality (injury deaths). The underlying cause of death is used for primary

Results

Changing classification from ICD-9 to ICD-10 had no impact on overall trends for accidents, homicides and suicides in either country. There was, however, a significant 69% decrease in the number of undetermined deaths in Italy (Table 3). There was also a decrease in the total number of external causes of mortality records in Italy. The comparability ratio was 0.959 (95% CI 0.949 to 0.970), indicating a 4% decrease with ICD-10.

However, we found statistically significant net changes in the number

Discussion

This study investigated the effects of switching from ICD-9 to ICD-10 on classification of external causes of mortality. To our knowledge, this is the first study that used same group definitions and measures at the aggregate level and at the case level to analyse official mortality data from two European countries. We found that ICD-10 had no significant impact on the official mortality statistics regarding deaths due to suicide, homicide and unintentional injuries (accidents). In the two

Conclusions

Changing the classification system from ICD-9 to ICD-10 had no impact on the overall trends for accidents, homicides and suicides in Italy and Norway. However, the number of records decreased in some injury mortality subgroups, such as accidental falls and traffic accidents. The ranking of causes of injury mortality with public health importance can therefore be substantially affected by new ICD revisions.

Key points

  • Implementation of ICD-10 affected some categories of injury that have public health importance, but the changes in ICD-10 had no overall impact on the statistics for suicides, homicides and accidents in Italy and Norway. Notably, there was a significant decrease in deaths categorised as ‘undetermined deaths’ in Italy.

  • New revisions of ICD may disrupt trends for specific injury categories; comparability ratios can be used to adjust the statistics.

  • Results from comparability studies are essential

Conflict of interest statement

In this study, none of the authors has declared conflict of interest.

Acknowledgments

We thank the Italian National Institute of Statistics (Istat) and Statistics Norway for the ICD-10/ICD-9 bridge coding used in this study.

References (18)

  • L.A. Johansson et al.

    Unexplained differences between hospital and mortality data indicated mistakes in death certification: an investigation of 1094 deaths in Sweden during 1995

    J Clin Epidemiol

    (2009)
  • Eurostat

    Causes of death – absolute number (annual data)

    (2009)
  • M. Heron et al.

    Deaths: final data for 2006

    (2009)
  • F. Belanger et al.

    Analysis of injury related mortality in Europe. The ANAMORT project final implementation report

    (2008)
  • L.A. Johansson et al.

    Comparing Swedish hospital discharge records with death certificates: implications for mortality statistics

    Int J Epidemiol

    (2000)
  • R.N. Anderson et al.

    Comparability of cause of death between ICD-9 ICD-10: preliminary estimates

    (2001)
  • R.N. Anderson et al.

    Disease classification: measuring the effect of the tenth revision of the international classification of diseases on cause-of-death data in the United States

    Stat Med

    (2003)
  • L. Geran et al.

    Comparability of ICD-10 and ICD-9 for mortality statistics in Canada

    (2005)
  • (1992)
There are more references available in the full text version of this article.

Cited by (18)

  • The Impacts of ICD-10-CM on U.S. Army Injury Surveillance

    2021, American Journal of Preventive Medicine
    Citation Excerpt :

    The most significant contributor to the higher injury rate was CMT injuries, which represented 77% of Army injuries in 2019. A smaller increase in rates of other non-CMT injuries (ACT and nonmechanical injuries) was also observed, which contrasts with results of other studies where ACT injury rates decreased after the implementation of ICD-10-CM.21,22 This is likely because the APHC injury taxonomy includes ACT injuries that were not included in previous studies to capture certain neurologic, acute MSK, acoustic, and ocular injuries.4

  • Looking for an explanation for the excessive male mortality in England and Wales since the end of the 19th century

    2020, SSM - Population Health
    Citation Excerpt :

    However, since its first adoption, the ICD has been changed several times. Every revision can affect time trends in cause-of-death statistics (Gjertsen, Bruzzone, Vollrath, Pace, & Ekeberg, 2010). Perhaps the largest problem in comparing causes of death over time is the fact that deaths attributed to unknown, poorly defined or symptomatic causes reduced dramatically over time: as medical attention during illness increased, medical knowledge improved, and the frequency of autopsy rose, more deaths were assigned to a cause, and these causes were increasingly accurate.

  • Practical use of medical terminology in curriculum mapping

    2015, Computers in Biology and Medicine
    Citation Excerpt :

    Some members of the family of classifications are derived by using a fifth or even sixth character to specify more detail. In others, the categories are condensed to render the broad groups suitable for use, for instance, in primary health care or in general medical practice [43,44]. SNOMED CT as a coding system for clinical problems was created by the merger, expansion and restructuring of two large-scale terminologies: SNOMED RT (Systematized Nomenclature of Medicine Reference Terminology) and CTV3 (Clinical Terms Version 3).

  • Trends in socio-economic inequalities in injury mortality among men in small areas of 26 Spanish cities, 1996-2007

    2013, Accident Analysis and Prevention
    Citation Excerpt :

    Finally, another limitation of this study is the change of coding, from the 9th to the 10th revision of the ICD, in 1999. Studies conducted in Italy, Norway and the United States (Anderson et al., 2001; Gjertsen et al., 2012) have analysed agreement between ICD-9 and ICD-10 for the leading causes of death and found that the introduction of ICD10 did not provoke changes in suicide mortality whereas mortality of traffic injuries decreased. However, two national level studies found that the introduction of the ICD-10 had not provoked changes in either traffic injury mortality, nor suicide mortality (Cano-Serral et al., 2006; Cirera Suarez et al., 2006).

View all citing articles on Scopus
View full text