A review of injury epidemiology in the UK and Europe: some methodological considerations in constructing rates

Background Serious injuries have been stated as a public health priority in the UK. However, there appears to be a lack of information on population-based rates of serious injury (as defined by a recognised taxonomy of injury severity) at national level from either official statistics or research papers. We aim to address this through a search and review of literature primarily focused within the UK and Europe. Methods The review summarizes research papers on the subject of population based injury epidemiology published from 1970 to 2008. We examined critically methodological approaches in measuring injury incident rates including data sources, description of the injury pyramid, matching numerator and denominator populations as well as the relationship between injury and socioeconomic status. Results National representative rates come from research papers using official statistics sources, often focusing on mortality data alone. Few studies present data from the perspective of an injury pyramid or using a standardized measure of injury severity, i.e. Injury Severity Score (ISS). The population movement that may result in a possible numerator – denominator mismatch has been acknowledged in five research studies and in official statistics. The epidemiological profile shows over the past decades in UK and Europe a decrease in injury death rates. No major trauma population based rates are available within well defined populations across UK over recent time periods. Both fatal and non-fatal injury rates occurred more frequently in males than females with higher rates in males up to 65 years, then in females over 65 years. Road traffic crashes and falls are predominant injury mechanisms. Whereas a straightforward inverse association between injury death rates and socio-economic status has been observed, the evidence of socioeconomic inequalities in non-fatal injuries rates has not been wholly consistent. Conclusion New methodological approaches should be developed to deal with the study design inconsistencies and the knowledge gaps identified across this review. Trauma registries contain injury data from hospitals within larger regions and code injury by Abbreviated Injury Scale enabling information on severity; these may be reliable data sources to improve understanding of injury epidemiology.


Background
Injury is a major, preventable public health problem in terms of morbidity, premature mortality or disability. Worldwide about 5.8 million people die every year as a result of an injury and the projections for 2020 show that 8.4 million deaths are expected annually [1,2]. Moreover, injuries are an important source of direct medical costs as well as indirect costs resulting from economic production losses; in the Netherlands for example, the direct costs of injury represents 5% of the health care budget whereas in Spain the total costs associated with Road Traffic Crashes (RTC) alone account for 1.35% of the gross national product [3,4].
In the UK, injury is the commonest cause of death in the first four decades of life and amongst the leading causes for ill-health -"for every injury death there are 45 hospital episodes, 630 doctor consultations and 5000-6000 minor injuries" [5]. In 1999 after Government intent to put all non-fatal injury (i.e. injuries defined as those requiring at least a general practitioner consultation) on the political agenda, reducing serious injury became a public health priority [6,7]. At national level official data on injury occurrence comes from health surveys, general household surveys, and morbidity surveys in general practice. For example: The Fireworks Injury Enquiries in Scotland provides information on proportions and trends in firework injuries  by age, place of accident or type of injury and outcome [8]; the ninth and tenth series of the Health Survey for England (2000)(2001) that offer data on nonfatal accidents by socio-demographic characteristics, injury details and consequences [9,10]; the Morbidity Survey in General Practice Fourth National Study 1991-1992 [11] and the Home and Leisure Accident Surveillance Systems reports until 2002 [12] describe reasons for consultation in general practice and injury consultations in A&E departments respectively. In this context it is noteworthy the European Injury Database established in 1999, under the European Prevention Programme, as an enlargement to European Home and Leisure Accident Surveillance System. The purpose of this database is to provide access to national injury data collected from Emergency departments of Member States hospitals for the development of evidence-based injury prevention strategies [13]. English Hospital Episodes Statistics (HES) are a relatively comprehensive national statistics dataset relying on information regarding hospital discharges sent by National Health System (NHS) Trusts [14]. Using HES data Jones et al. [15] collated injury information and underlined the issues related to availability of valid denominators in constructing injury death rates and hospitalization rates; however, some approximate rates per 100,000 population by age group based on finished consultant episodes (i.e. a completed period of care a patient using a NHS hospital bed, under one consultant within one health care provider) from HES data were calculated: 1,100 (0-14 y); 1,100(15-59 y); 1,300 (60-74 y) and 4,313 (75+y). Fatal injury data in England and Wales are reported in the yearly publications of the National Office for Statistics, i.e. Mortality Statistics -Injury and Poisoning DH4 series (between 1974 to 1989 entitled Mortality Statistics -Accidents and Violence) as well as in the Decennial Supplements that deal with trends and geographic patterns [16,17]. However, in order to control injury by prevention strategies incidence rates alone are insufficient. The severity of each injury needs to be characterised as it has been shown to relate to health service resource use and economic loss, as does the causal mechanism of injury [18].
An epidemiological challenge is defining 'serious' injury using existing databases. Indicators for monitoring nonfatal serious injuries have been proposed (i.e. length of hospitalisation four or more days, or admissions for long bone fractures) and further criticised [7,19,20]. The national database (HES) uses the International Classification of Diseases ICD-10 th revision to categorize injuries. However, the ICD contains little indication of injury severity. On the other hand trauma registries (in the UK, Trauma Audit and Research Network (TARN) [21]), in line with international conventions, use the Abbreviated Injury Scale (AIS), which is an anatomically-based injury description system that lists over 1200 serious injuries and scores the immediate severity of each injury from one (minor) to six (maxima). Table 1 offers an example of injury description within ICD and AIS lexicons. The AIS allows computation of several measures of injury severity that have been used in trauma research. All these measures are based on the single AIS severity scores and are constructed in order to measure one's patient global severity when he/she has multiple injuries: Maximum Injury Severity Score (MAIS) which is the highest single AIS score [22,23], Injury Severity Score (ISS) calculated by summing the squares of the highest AIS-scores in three body areas 520208.5 1 superficial, incomplete transection; incomplete circumferential involvement; blood loss ≤ 20% by volume 2 rupture, incomplete transection; segmental loss; complete circumferential involvement; blood loss ≥ 20% by volume 3 in italics the AIS score that allows computation of injury severity score [24], or the New Injury Severity Score (NISS) developed by Osler et al. in 1997 based on the three most severe lesions regardless of the body region [25]. Moreover, in the middle of the 1990s, Osler introduced the ICD-9 based ISS (ICISS) that allows severity to be classified based on the ICD-9 classification of injuries [26]. Injury Severity Score ICD-9 is a product of survival risk ratio from each injury sustained, based on the values of the survival rates of prior patients with similar diagnoses as classified by ICD-9. Hospital databases do contain ICD diagnoses and as a result there is opportunity to derive AIS severity scores ICISS for the estimation of injury severity [27]. Apart from the measures based on AIS, it should be mentioned in this context the Paediatric Trauma Score (PTS) -an injury scoring method specifically designed for children [28,29]. It relies on a six point score calculated from weight, blood pressure, ability to maintain on airway, level of consciousness and presence of fracture/ wound. Several issues have been raised with regard to the use of any of these specific measures of injury severity. Injury Severity Score, MAIS and NISS for example, are 'congenitally attached' to the AIS score for calculation [30], add to this ISS does not take into account age or chronic preconditions and it is limited to one injury per body region [31,32]. Injury Severity Score, MAIS and PTS only apply to physical injury and are not suitable for other events, such as drowning. Moreover, ISS and MAIS are derived from adult norms and consequently are not 'child friendly' [33]. Although developed to correct some of the limitations related to ISS, NISS seems to be a measure 'that tends to overstate severity for lesser injuries' [31]. With regard to ICISS, there is limited evidence on the validation and performance of this scoring method with ICD10 classification of injuries [33][34][35]. Although there is no consensus on the 'best' method for defining injury severity, the ISS as developed by the American Association for the Advancement of Automotive Medicine remains the 'gold standard', most reliable and extensively used measure of injury severity and ' a cornerstone of injury epidemiology' [31,32,36,37]. The ISS characterises the likely threat to life from injury and is widely used in hospital trauma registries to characterise their activity and performance. However, it seems that this data is rarely related to populations to create severity-specific rates of major injury. We have reviewed the injury epidemiology literature with the purpose to present incidence rates and characteristics of injury in UK and Europe, and discuss methodology for constructing rates, including the availability of data sources.

Methods
To conduct this review, the authors accessed the extensive bibliography of John Rylands Library, University of Manchester as well as TARN papers and reports. Medline, EMBASE, and Cochrane Library electronic databases have been examined for English language European injury papers published from 1970 to 2008. Search terms used included: 'descriptive epidemiology injury/trauma', 'injury population based study', 'injury morbidity/mortality', 'injury incidence/deaths' and 'injury surveillance'. The review presents epidemiological studies covering injuries all types of causes. Since the primary aim was to identify comprehensively the literature focusing on the epidemiology of all injury (fatal and/or non-fatal) in UK, the papers identified from the search located in England, Wales, Scotland or Northern Ireland were cross-referenced and further reviewed. Studies from Europe were included if they present the epidemiological profile of all causes of injuries. We included papers that present data in the general population as well as within specific age groups (e.g., children, adolescents). We excluded studies covering selected injury categories, for example sport related injuries, work related injuries (by location of injury), traffic accidents, assaults (by injury cause) or head injuries (by body region), as it was out of the scope of this paper to provide the epidemiological profile within a selected group of injuries. The review addresses all three levels of non-specific measures of injury severity: deaths, hospitalisations and Accident & Emergency (A&E) attendances and categorises the papers taking into account study population (e.g. children (0-14 y) and adolescents (15-19 y)/general population -all age groups) and study location (i.e. UK/other countries in Europe). Papers that identified injury data from additional resources (e.g. general practitioners offices) have also been reviewed. The results are presented in terms such as population-based rates, proportions, or temporal trends and evidence of using severity scoring methods such as ISS. The relationship between injury and socioeconomic status (SES), the subject of other published reviews focusing on socioeconomic inequalities [38] or on overall methodological issues [39], has been examined in this review only within studies that were selected as focusing primarily on injury epidemiology. Although an important and recognised areas of injury research, injury consequences such as level of disability, economic and social costs or injury -related health behaviours, are outside the scope of this paper that has been designed to determine European injury rates and levels of severity within morbidity and mortality studies.

Results
Forty-eight articles have been reviewed: 18 focused on national level data and 30 focused on local level data; 29 papers had used mortality data, 30 papers -hospitaliza-tion data and 19 papers -A&E data. Two or more data sources have been used by 21 papers.

UK and Ireland
Twenty-eight studies dealing with fatal and/or nonfatal injuries occurring in the UK and Ireland were identified in the literature.  [40,41] whereas in the overall age group 0-19 y the rates decreased from 16/100,000 (1980) to 7/100,000 (1994) [42]. Travel patterns were responsible for the decreasing in mortality rates from 1985 to 1995 [43,44]. There was no change in the incidence of intentional injury fatalities over 1980-1994 [42]. Overall the largest proportion of injury deaths occurred in boys (e.g. 69% in England and Wales 1980-1994 [42]). Blunt trauma (RTC (Road Traffic Crashes), falls, blunt assault) is the predominant injury type for severe trauma (i.e. almost 95% [45]). Overall more than half of blunt injuries deaths and/or hospitalisation are due to RTC [45][46][47][48]. Social class gradients are present in injury deaths rates (e.g. a correlation coefficient between death due to accidents and social deprivation r = 0.56 and a 17 times higher homicide rate in social class V vs. I in England, 1980England, -1984 and 1980-1995 [42], respectively). Of note, the five basic social classes recognised by Office of Population Censuses and Surveys are described as follows: I Professional occupations; II Managerial and technical occupations; III Skilled occupations; IV Partly -skilled occupations and V Unskilled occupations [49].
In the general population, all injury mortality rates vary from 23.8 per 100,000 (1998/1999, Italy [78]) to 126/ 100,000 (1990-2002, Lithuania [79]). Stable trends over 1990-2002 in mortality rates in Sweden have been shown in the literature reviewed whereas in Baltic communities, over the same study period, the rates registered a tendency to increase until 1994 followed by more stable patterns since 1997 [79]. Incidence rates for all injury treated ranged from 12900/100,000 in Greece 2001 (data from the surveillance system [80]) to 11400/100,000 in Norway 1985/1986 (data based on A&E attendances, hospitalization and general practitioners visits [81]) and 7470/ 100,000 in Spain 1990/1991 (A&E questionnaire data [82]). Incidence rate of severe injuries (ISS>15) is available from one study, i.e. 52,2 per 100,000 1998/1999 in Italy (data including pre-hospital deaths) [78]. Except in the over 65 age group, the injury incident rates in male exceeded those in females [81,83]. Falls and RTC were the prevailing mechanism for both fatal and non-fatal injuries (e.  [83]) and age group (i.e., the proportion of injures ISS>8 vary from 37.3% (children 0-17 y) to 8% (adults >14 y) [72,82]; over half of injuries ISS>15 are in the category 16-25 y [78]). Around 80% of all hospital admissions and general practitioners visits are injury AIS 1 [81].

Discussion
The literature review has identified a significant body of epidemiological research in relation to injury mortality and morbidity.

Data sources
Official statistics on injury mortality have been used in some research papers that offer national (e.g. England and Wales) representative data [e.g., [40,43,44,46]; however, these are only the tip of the iceberg on injury data or the first level of injury pyramid (i.e., deaths/hospitalisations/ A&E attendances/general practitioners data/self reported). Hospital discharge data alone does not contain information on injury severity and has also the disadvantage of multiple counting for transfers from one unit to another or for readmissions for the same injury. Hospital as well as mortality data from the Central Statistics Office in Ireland have allowed description of injury epidemiology [67,68]. Although the authors managed to control for double counting of transfers, multiple admissions for the same injury were not reliably detected.
The issue of multiple counting due to several treatments for the same injury applies also to A&E departments' data. It is noteworthy in this context that in a study in West Glamorgan County, Wales, data from emergency departments are being used that are based on a 'diagnostic filter', so that only first attendances due to a new injury are included [57]. Use of trauma registry data in providing epidemiological pictures of injury occurrence is highly desirable, although some limitations of these data sources have been acknowledged. These include lack of information on death or injury before reaching the hospital and the selection criteria for entry into a trauma register. Trauma registers tend to include more severe trauma patients as determined by the length of stay of 2, 3 or 4 and more days [89]. On the other hand trauma registries, with the costs of data collection shared by participating hospitals, allow more appropriate codification of injury (i.e. Abbreviated Injury Scale), categorization by severity (i.e. ISS), exclude multiple admissions for the same injury, permit tracing of transfers, and capture information on post discharge deaths [e.g. [21]]. Moreover, there are trauma registries that include deaths on the scene, for example the French Rhone Road trauma registry covering injuries from road crashes in the Rhone county [90] or The PaEdiatric Network around TraumA registry that provides population-based data of injury in children and youngsters for the Flemish region [72]. North American literature shows that combining trauma registry data with pre-hospital deaths (coroner's department or medical examiner's data [91][92][93]) allows computation of major trauma death and morbidity rates. Some others authors use questionnaires to collect information on socio-demographic characteristics of the patient and injury details in studies of children under treatment or admitted to a hospital [76], patients admitted to a hospital and a sample of patients treated by general practitioners [81], or children treated at an emergency department only [58,75]. Questionnaires of self-reported illness as a primary data source have the advantage that the items to be collected can be tailored to specific research questions, but are time consuming to administer and usually involve smaller sample size compared to secondary data (data collected by people other than the researcher in question, for example, public vital statistics records). A further disadvantage is recall bias amongst respondents.

The epidemiological profile
The identified studies from UK [40,[42][43][44][45][46]48] and Europe [69,71] over the past decades indicate that child death rates from unintentional injury have decreased. The changes are mainly attributed to the decrease in RTC (trauma care improvements and traffic safety programmes [48,69,71] and changes in travel patterns (declines in walking and cycling activities -based on national data, England andWales, 1985-1995 [43,44]). There are no significant trends in children for change in incidence of intentional deaths rates [42,48,69]. In this context two studies from North America are noteworthy that show a similar pattern of decreasing rates of child injury deaths based on RTC decline [94,95]. In USA homicide account for 36% of child injury deaths overall [94] whereas the same figure for UK is 3% [48]. Trends for child non-fatal injury rates are available from Scandinavian studies that shows little changes over time [70,71] Overall in the general population there has been a decrease in fatal injury rates in studies from UK [65,66] and Europe [84,87] as a result of declining fatal RTC. The changes in fatal injuries have been attributed to trauma care improvements and safety measures [65,66] as well as to economic and 'autonomous' factors (i.e. demographic, socio-cultural and technological trends [84]). Socio-economic, historical and political factors have been used to explain the differences in mortality rates and trends between communities in Sweden and Baltic States [79]. Also in USA has been reported a decline in fatal injuries rates; however, it is noteworthy that intentional injuries here account for up to 42% of all injury deaths in USA, violence being a major cause [91]. The pyramids also shows higher number of (hospitalisation or A&E) non-fatal injuries corresponding to one fatality in children [96][97][98] vs. general population [5,59,67,102,103] or adult population (i.e., 1/6/133 [82]).
All the studies in which data have been analysed from a gender perspective alone show that both fatal and nonfatal injuries occurred more frequently in males than females. Whereas no consistent results for children age groups at higher risk, a well-recognised pattern of higher fatal and serious injury rates in males up to 65 y old then in females ≥ 65 years, has been shown in the papers reviewed [59,61,81,83]. Peak values of all injuries in males aged 15-24 y and 65+y and in females aged 65+y have been shown in the literature in UK [59] and elsewhere [83,102]. In the general population as well as in children RTC and falls are the predominant mechanisms for both fatal and non-fatal injuries. Moreover, higher RTC and assault rates in males compared to females have been reported [59,83,102]. Within a study in UK [61] sport and leisure activities were the third (male) and fourth (female) cause of injury after home, work/school and street/public locations whereas in Belgium 53.8% of A&E child injuries were sport related [72]. Disparities by race/ethnicity and (especially for fatal injuries) by SES persist over time. Fractures, dislocations and sprains, stains are cited as the most common causes of hospital admissions and A&E attendances [59,61,81,88].

Injury pyramid
Many studies rely solely on mortality rates from injury for describing epidemiology [e.g. [42][43][44]46,69,87]]. Some investigators have used data on hospitalisations or A&E admissions only [e.g. [41,57,58,85]] but there are only a few studies giving data on deaths, hospitalisations and A&E attendance [50,64,84]. Ideally the epidemiological picture should include all levels of the injury pyramid. On the other hand, most minor injuries are difficult to estimate, unless through a specific survey, since most people with minor injuries will not present to health care services. To describe the total burden of injury within a partic-ular community would require specific studies that are time-consuming and expensive to mount. As a minimum, we suggest that researchers should describe injury burden in terms of fatal and non-fatal serious injury.
Key issues in computing incidence rates are representativeness and generalisability. There are only a few studies that have used large national databases to estimate national injury rates (e.g. Patients Episodes Database for Wales [60]; Road Traffic Accident Registry, Occupational Registry, National Medical Registry and Dutch Central Bureau of Statistics in The Netherlands [84]; National Hospital Discharge Register in Finland [69,71,87,88]). Many of the articles we reviewed focused on particular communities usually defined by geographic areas, with analysis extended from one to several years. Some were based on proportions (socio-demographic characteristics and injury details) rather than rates [e.g. [77,85]]. When seeking to extrapolate results from certain population sub-groups to the general population researchers consider carefully the limitations e.g. the socio-demographic structure of the population studied compared with the general population. To discern time trends analyses should include more than two year's worth of data where available. Finally proportions should complement rates in any descriptive epidemiological paper dealing with injury occurrence, rather than be a substitute for them.

Matching numerator and denominator
The definition of appropriate numerators and denominators is crucial for calculating population-based rates. This issue applies not only to injury research but also to the larger area of population based epidemiology. A number of approaches to determine the denominator or 'population at risk' have been described within the literature presenting data from the general practice [104,105] or from sentinel practice networks [106,107]. Whereas use of the number of consultations (i.e., yearly or weekly contact group) has proved to be the step forward to determine the population practice, the ideal denominator seems to be the total population within a clearly defined geographic area, i.e., the practice catchment area [107]. Within this review, most of the articles simply use as the numerator the number of discharges and/or A&E visits from hospitals located in a study area. The denominator is the regional census population. For example one study provides rates for the town of Trondheim based on hospital discharges from one hospital that is judged to be the only provider for the region [81]. However, a number of researchers acknowledge that population movement occurs resulting in a possible mismatch between numerator and denominator. For example in one study from UK [61] the authors tried to account for the admissions of area residents outside their area (i.e. postcode EH45 -Livingstone town), although the exact figures were unavailable. In a remote area in France [83] researchers took into account area cases diagnosed outside region by extrapolation based on mortality rates. The authors themselves recognised the possibility of an inaccurate estimate. However, in this study they attempted to include commuter populations in their rate estimates. Some other studies exclude out-ofarea residents treated at area hospitals as well as out-ofarea hospitalisation of area residents when defining numerator [78]. The ideal solution to these methodological inconsistencies would be a small area analysis in the form of aggregation based on home address of the patients and of the census population in a corresponding way (e.g. aggregation by electoral wards over Wales [60]). However, this is not a viable option when the database is restricted to information from a limited number of hospitals. This is the case of data provided by national registries [e.g., [21]] that relies on a selection of hospital reporting on a voluntary basis; regional registries using patient inclusion criteria based on place of residence [93] or place of injury occurrence [78,90,91] or hospital discharge databases. We suggest that, when possible, researchers should consider all cases from area residents in constructing a numerator. Further research should focus on development of an appropriate methodological approach to deal with this issue.
Another issue in the context of defining numerators and denominators for further calculation of population based rates is related to data quality. Although completeness and accuracy of the records fields within trauma registries have rarely been reported in the literature, there is evidence that validated trauma registries are a reliable data source for population based epidemiology [108,109]. Apart from the validity, an other component of registry data quality is the completeness of case ascertainment defined as the proportion of all cases in the target population that appear in the data base [110]. Case ascertainment rates of 90% of all major trauma (injuries ISS>15) for Victoria State trauma registry [111] or 87% of serious injuries (NISS9+) for French Rhone road trauma registry [90] have been shown in the literature. Use of methods such as capture-recapture through record linkage between several data sources (for example, within road injury research, police records-trauma registry [112], police records-hospitalisation records [113] and hospitalisation records-police records, deaths certificates, trauma registry [114]) can overcome incomplete registration of injuries and provide extensive scientific data for population based epidemiologic research. The capture-recapture method ca be applied when several data sources are available and under certain circumstances (i.e., for the traditional two sample method -independent samples, closed population, homogeneity of capture and perfect matching) [115,116]. It has been used in injury research [e.g., [112][113][114]117]] as well as within larger area of epidemiology [116]. Within injury research, the variety of data sources (e.g., death certificates, hospitalisation discharges, trauma registry data, ambulance records, physician's records, survey data, police records) can only be in favour of the widespread application of this technique in assessing data quality and estimating injury population-based rates.

Measures of injury severity
We reviewed an extensive number of articles especially from UK; however, only few of them presented data using a specific measure of injury severity. Some papers refer to categorisation of injuries by severity based on non-specific general measures: mortality, hospitalisations and A&E admissions [e.g., [80]]. However, death is an injury outcome rather than a severity measure, whereas hospitalisations and A&E attendances are subjective measures (for a detailed discussion related to non-specific measures of injury severity see the review by Beattie and al [29]). Some other papers used different measures. For example in one study severity is arbitrarily defined by four categories based on nature of injury and treatment [53] whereas in another study 'sever trauma' is defined by length of stay over 48 hours including all non-survivors [72]. Other authors used case-fatality rates as a function of injury severity [84] or fractures as an indicator of severity [57]. This is probably explained by data limitations -the researchers analysing routinely collected data or primary data lacking information on injury severity. Injury Severity Score remains the most widely accepted measure of severity in the field of injury, although some limitations of this measure have previously been acknowledged. Without a standard severity coding it is difficult to define clearly the category of major trauma (i.e. ISS ≥ 16) for those requiring more attention in terms of both prevention and care, to interpret injuries and to make comparisons across studies. Walsh and Jarvis [55] have pointed out that any epidemiological study using ISS to select patients with more severe injury may overcome issues related to 'selection bias' (i.e. factors apart from the injury itself influencing health care use such are hospital policies, bed supply, SES of the patient) that applies both to hospital admissions and A&E attendances.

Relationship of injury to socio-economic status (SES)
Only one study outside UK has analysed the relationship of rates of injury to SES [74]. Whereas a straightforward inverse association between injury death rates and SES has been observed from the literature review, the evidence for socioeconomic inequalities and injury morbidity has not been wholly consistent. For fatal injuries socioeconomic gradients or inverse association with SES [40,42,46,48] have been shown in descriptive studies of childhood injuries. For non-fatal injuries results vary from presenting no relationship [57,62] to associations [50,55,58,61,74] of injury requiring hospitalisation and/or A&E visits with SES. Evidence comes mostly from ecological studies (children alone as well as from the general population) that are limited by the 'ecological fallacy'. It is noteworthy researchers have been used complex measures of socioeconomic status (e.g. Townsend deprivation score, Jarman 8 Index, Carstairs Depcat Score) or one indicator only (e.g. parent education, insurance status, social class). Either measured at an individual level or area based, SES remains a multidimensional concept, subject to confounding variables such as race/ethnicity (a detailed, complex description on the issue of injury -SES can be found in the literature review by Cubbin & Smith [38]). Finally, we would recommend the researchers to use adequate, complex measures for SES and when possible to apply an analytical design to confirm their findings.

Conclusion
In this paper we reviewed articles dealing with all injury epidemiology expressed as mortality and/or morbidity rates. Some methodological considerations have been discussed as well as possible solutions to the difficulties in counting injuries and constructing population-based rates. New methodological approaches should be developed to deal with the study design inconsistencies and the knowledge gaps identified across this review. Probably, in future, population-based trauma registries collecting data from all trauma hospitals within a larger region will be used and will allow information on severity as defined by the ISS.