Article Text
PDF
Abstract
Background The International Collaborative Effort (ICE) on Injury Statistics called for an effort ‘to reach consensus on what are the 10 most important indicators of injury incidence that offer the potential for international comparisons and for regional or global monitoring.’
Objectives To describe the process of developing the ICE indicators and to present the specifications of selected injury mortality indicators, along with comparisons between selected countries for those specified indicators.
Methods Participants on the ICE list had been asked to send to one of the authors what they considered the most important five indicators of injury incidence. These were synthesised and presented under six themes: mortality indicators (general); mortality indicators (motor vehicle); mortality indicators (other); hospital data-based (overall); hospital data-based (traumatic brain injury (TBI)); long-term disability (overall). Following two work group discussions and after drafting and revising indicator specifications, agreement was reached on mortality indicators and specifications. Specifications for those mortality indicators are presented. Morbidity indicators are still to be agreed.
Results The mortality indicators proposed were age-adjusted rates of injury death; motor vehicle traffic crash-related death; self-harm/suicide; assault/homicide; and TBI death. The empirical work highlighted difficulties in identifying TBI in countries where mortality data do not capture multiple injuries, prompting us to drop the mortality indicator related to TBI and moving us instead to introduce an indicator to monitor the use of undetermined intent in the classification of injury mortality.
Conclusion The ICE has reached a consensus on what injury mortality indicators should be used for comparison between countries. Specifications for each of these have been applied successfully to the mortality data of seven countries.
- Injury
- indicators
- mortality
- international comparison
- biostatistics
- methods
- surveillance
- international
- e-code
- terminology
- policy
- severity
- score
- international
- developing nations
Statistics from Altmetric.com
- Injury
- indicators
- mortality
- international comparison
- biostatistics
- methods
- surveillance
- international
- e-code
- terminology
- policy
- severity
- score
- international
- developing nations
Background
Coinciding with the preparation for the 9th World Conference on Injury Prevention and Safety Promotion held in Merida, Mexico in 2008, the International Collaborative Effort (ICE) on Injury Statistics1 called for an effort ‘to reach consensus on what are the 10 most important indicators of injury incidence that offer the potential for international comparisons as well as for regional or global monitoring.’ The ICE on Injury Statistics, established in 1994 and sponsored by the US National Center for Health Statistics, provides a forum for international experts in injury prevention and epidemiology to discuss issues related to data. For more information on projects and participation visit http://www.cdc.gov/nchs/injury/advice.htm.
An injury indicator is defined as a summary measure that denotes or reflects, directly or indirectly, variations and trends in injuries or injury-related or injury control-related phenomena.2
This paper briefly describes the process of developing the ICE indicators and presents the specifications of selected injury mortality indicators, along with comparisons between selected countries for those specified indicators. The purpose of the paper is not to address methodological issues related to the comparison of rates (eg, standardisation) or classification issues, but rather to present consensus on injury mortality indicators. Neither does it expand on the benefits of comparing injury rates for research and intervention prioritisation.
Methods
Consensus work
In 2008 members of the ICE on Injury Statistics were interviewed via email to elicit their suggestions regarding the five most important indicators on injury incidence. A summary of the process followed to distil those suggestions into the indicators presented in this paper is described at the ICE' website. Briefly, 57 ‘candidate’ indicators were discussed at a group meeting (http://www.cdc.gov/nchs/injury/ice/mexico2008/mexico2008_agenda.htm) and clustered into one of the following rubrics: (1) mortality (general); (2) mortality (motor vehicle); (3) mortality (other); (4) hospital (overall); (5) hospital (traumatic brain injury (TBI)); and (6) long-term disability (overall). Subsequent work focused on the mortality indicators which were the focus of discussion of another ICE meeting where indicator specifications were developed (http://www.cdc.gov/nchs/injury/ice/washington2008/wdc_2008_participants.htm). This paper focuses on these mortality indicators.
Empirical work
Mortality data from WHO were made available to us for analysis by the Global Burden of Disease Project, particularly the Harvard University Initiative for Global Health at the Harvard School of Public Health (http://www.who.int/whosis/mort/download/en/index.html). These deaths are registered in national vital registration systems, with underlying cause of death as coded by the relevant national authority. These data are official national statistics as they have been transmitted to the WHO by the competent authorities of the countries concerned. Detailed injury death data are country-specific, ICD-10 coded and aggregated by 5-year age groups. Data were aggregated by the Global Burden of Disease into the format of the ICD-10 External Cause of Injury Mortality Matrix.3
In order to have a ‘manageable’ number of countries for this paper, it was agreed to analyse data from seven geographically varied countries spanning continents and hemispheres with complete death registration (USA, Argentina, Venezuela, Spain, UK, Japan and New Zealand). Following WHO guidelines, rates were age-standardised to the 2000–2005 world average population (see Appendix 1 in the online supplement).
Results
The ICE participants agreed that the primary purpose of the ICE injury mortality indicators is international comparison, with a secondary purpose of monitoring the impact of preventive efforts (eg, tracking trends in injury rates).
The mortality indicators proposed were rates of: (1) injury death; (2) motor vehicle traffic crash (MVTC)-related death; (3) self-harm/suicide; (4) assault/homicide; and (5) TBI death. The empirical work highlighted two difficulties:
Identifying TBIs: the underlying cause of death reported to WHO is the external cause (ie, the motor vehicle crash) rather than the injury itself (ie, the TBI). All other diagnoses on death certificates are referred to as the multiple causes of death. Most countries either do not capture multiple injuries or do not make multiple causes of death data available. Changes between ICD-9 and ICD-10 have made the coding of TBI somewhat more arbitrary.
International differences in assigning undetermined intent: this affected the interpretation of indicators 3 and 4.
As a consequence, we dropped the TBI indicator and introduced an indicator to monitor undetermined intent mortality.
The agreed specifications for the mortality indicators are shown in Appendix 2 in the online supplement.
Comparisons across the seven countries of the recent trends in these fatal injury indicators are shown in figure 1A–E. Rates are plotted on a logarithmic scale to allow for fitting of a wide range of rates, emphasising the relative change in trends rather than comparisons of absolute values. Figure 2A–E presents the percentage change from the initial rate ({(rateyear i – rateyear 0)/rateyear 0]}×100) as an effective means of highlighting relative changes within each country.
(A–E) Age-adjusted death rates per 100 000 population for injury mortality indicators country (2001–2006). Note that this figure is presented on a log scale. Suicide rates are restricted to subjects aged ≥10 years. MVT, motor vehicle traffic.
(A–E) Percentage deviation of age-adjusted death rates per 100 000 population from base year for injury mortality indicators. Note that this figure is not presented on a log scale. Suicide rates are restricted to subjects aged ≥10 years. MVT, motor vehicle traffic.
The key results of figures 1 and 2 are as follows:
All injury: There were large differences in the age-standardised death rates across countries, with Venezuela having very high levels compared with the countries with the lowest rates (UK and Spain). Argentina showed a 15% fall in rates over 4 years, New Zealand 10% and Spain 8%. Changes for other countries were more modest. The US rate showed a 5% increase.
MVTC: There was major variation in rates between countries with contrasting trends. Japan and Venezuela showed declines of 30–31%, Spain of 25% and New Zealand of 18%. The falls for Argentina and UK were modest whereas, for the USA, there was little evidence of a decrease.
Suicide: There was as much as a twofold difference in rates across countries. Japan had the highest rates while Venezuela, Spain and the UK had considerably lower rates. Venezuela had a decline of 33%, New Zealand 9% and Argentina 8%. Other countries showed only modest or no apparent trends over time.
Homicide: These rates showed the largest variation between countries. Venezuela had very high rates while Japan and the UK had the lowest. Argentina showed the largest decline (26% over 4 years) while the USA had a 13% decrease over 5 years. Venezuela showed a larger increase followed by a smaller one.
Undetermined intent: There was large variation in the level of death rates coded to undetermined intent, with Venezuela having the highest level and Spain the lowest. Some deaths coded to undetermined intent will be suicide cases and others will be homicide cases, so these findings should be viewed alongside the previous data.
Discussion
Principal findings
This paper reports on the proposed indicators and their specifications related to injury mortality developed consensually by the ICE on Injury Statistics and should be in the statistical armoury of any country. These indicators can be applied wherever there is access to reasonable quality data—either at a national or at a sub-national level—but which can be used to represent the national picture.
The five mortality indicators proposed are: (1) injury mortality rate; (2) MVTC-related mortality rate; (3) self-harm/suicide rate; (4) assault/homicide rate; and (5) undetermined intent mortality rate.
In this work we compared trends in these fully specified five mortality indicators between seven selected countries for illustrative purposes rather than for comparisons between these specific countries. It was clear, however, from these example countries that there was much variation in rates between countries. The wide variation in undetermined intent cannot be ignored. Despite WHO coding guidelines for assigning undetermined intent, there is country-specific use of the codes, sometimes for cultural reasons and sometimes for death certification reasons. This needs to be carefully distinguished from true variation in the rates. This paper acknowledges the variation but does not attempt to disaggregate its usage.
When presenting these indicators, we favour presenting deviations from baseline (figure 2A–E) rather than absolute rates, since comparing fatal injury rates can produce results that are unreliable or difficult to interpret. The way an injury-related death is defined, the way it is recorded on a death certificate and the sources of information used to collate the statistics may all differ between countries and thus affect the validity of the comparisons made.4 These threats to validity are less for comparisons of trends, provided consistent methods are used within each country across the time period considered.
We are only aware of one other project that measured injury mortality data across countries—ANAMORT—which was partially funded by the EU DG SANCO.5 However, the aim of this project (completed in 2008) was to evaluate differences in the specific ICD-9 or ICD-10 codes that were being used in eight European countries in order to improve comparability of vital death statistics reported, for example, to EuroStat. In ANAMORT the emphasis was on the selection of indicators that are based on those ICD codes, but not the actual variability or validity of those codes—albeit the issue of validity was a concern expressed in the ANAMORT discussion section.
Limitations of the mortality indicators
Despite the use of common specifications for these indicators, these comparisons could be misleading for a number of reasons. Valid international comparisons depend on good quality data and consistent definitions. Potential limitations identified by the ICE work group on mortality indicators were: variability in the quality and completeness of deaths recording; variability in specifying the manner of death on the death certificates which could influence assignment of the underlying cause of death code; and completeness, specificity and accuracy of coding underlying cause of death across countries. For example, in 2006, 18% of injury deaths in the UK were coded to unspecified mechanisms of injury while in Venezuela this proportion was only 5%. Other potential limitations were: possible use of non-comparable sources of data and, when classifying by year of death, the occurrence of undercounts where deaths registered after the year of death are not used to update the counts. If the international comparisons presented in this paper were being carried out for more than illustrative purposes, each of these potential limitations would need to be investigated for each country involved.
A further problem is that, when comparing injury death rates, we have to accept that the likelihood of death given an injury is influenced not only by severity of the injury but also by access to and the quality of health services (eg, speed of emergency services). The effect of such external factors can be reduced through international comparison of serious injury with intelligent case selection.2 We aim to develop valid international comparisons of serious (fatal and non-fatal) injury in the medium term.
Implications for prevention
Providing a set of agreed injury mortality indicators should facilitate setting priorities for research and intervention around the world. Although injury prevention experts know the benefits of investigating differences in rates and trends (whether because of coding differences or because of effective interventions), this paper presents the only effort to date to create internationally agreed indicators to use in the foreseeable future as developed by an expert group of injury researchers from almost 20 countries.
Conclusions
The ICE has reached a consensus on what injury mortality indicators should be used for comparison between countries. Specifications for each of these have been produced and applied successfully to the mortality data of seven countries.
What is already known on the subject
This paper is the first to provide an international consensus on injury mortality indicators.
What this study adds
Consensus on the most important mortality indicators for use for international comparison.
The operational definitions for five injury-related mortality indicators for international comparison of rates and trends: (1) injury mortality rate; (2) MVTC mortality rate; (3) self-harm/suicide mortality rate; (4) assault/homicide rate; and (5) undetermined intent mortality rate.
Acknowledgments
We thank Kavi Bhalla, Harvard University, for the WHO Mortality Database aggregated by 5-year age groups into the format of the ICD-10 External Cause of Injury Mortality Matrix and by detailed ICD-10 codes. We also thank Gordon Smith, Yvonne Robitaille and J Lee Annest for their comments on an earlier draft. We thank all the participants in the ongoing meetings of the ICE on Injury Statistics. Those who participated in the two meetings highlighted in this paper are listed in http://www.cdc.gov/nchs/injury/ice/washington2008/wdc_2008_participants.htm. Montserrat Ruiz-Pérez assisted in the manuscript preparation and submission process.
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.