Study design and setting

A population-wide, cross-sectional epidemiological study was conducted in 16 European countries (Austria, Bulgaria, Croatia, Cyprus, Denmark, Estonia, Hungary, Ireland, Italy, Lithuania, Luxembourg, Romania, Serbia, Slovakia, Slovenia, and United Kingdom) in order to estimate TBI YLLs for the year 2013. The selection of countries was based on the availability of data. The year 2013 was chosen because it was the most recent year for which data were available. The availability of the data in other EU countries for this year was limited because, at the time of this study, not all European countries were submitting data on causes of injury-related deaths in the necessary format (e.g., giving both the external cause and nature of injury) and detail (e.g., giving data in sufficiently small age groups). Thus, the choice was made to use the 16 countries for which data were available, and to extrapolate the findings to the 28 member states of the European Union (EU-28), and this seemed justified under the circumstances.

Data sources

The data used for all analyses in this study were acquired from the statistical office of the European Union (Eurostat). Eurostat routinely collects data from death certificates from the 28 EU member states, the former Yugoslav Republic of Macedonia, Albania, Iceland, Norway, Liechtenstein, and Switzerland and regularly publishes annual overviews of causes of deaths [19]. For our study, a specifically tailored dataset of micro-level data was provided that, in detail of information, went beyond the regularly published reports. This dataset contained a record for each injury-related death that occurred in the included countries in 2013, where the external cause of death (International Classification of Diseases–10th Revision [ICD-10] codes V01–Y98), the specific nature of injury (ICD-10 codes S00–T98, only 1 diagnosis provided for each record), the age at death, and sex were given. All data used in our analyses were collected at the country level and then—following specific and unified guidelines—submitted to Eurostat, which in turn provided them to us. The study used administratively collected secondary data, and no ethics committee approval was required. No ethics approval was required in order to obtain data from Eurostat.

For the purpose of this study, a TBI-related death was defined as a death where the cause of death was a TBI or a TBI sequela, i.e., from the provided database, records in which the nature of injury was coded as ICD-10 S00–S09 (injuries to the head) or T90 (sequelae of injuries to the head).

Variable definitions

The European Union life table published by Eurostat was used to determine the life expectancy at death for each recorded death [20]. The number of YLLs for each death was calculated by subtracting the age at death from the life expectancy at the age of death, and summarized using this formula: (1) where d₁ is the number of fatal cases due to health outcome l in a certain period and e_l is the expected individual life span at the age of death due to health outcome l.

YLLs were summarized into 6 age groups (0–4, 5–14, 15–34, 35–64, 65–84, and ≥85 years) by summing the YLLs in persons in each age group. Crude rates of YLLs were calculated per 100,000 people per year using mid-year populations for each country in total and for the 6 age groups. For comparison purposes (in response to the suggestions of a peer reviewer), number of YLLs and crude TBI YLL rates per 100,000 persons per year are also presented broken down into 5-year age groups (S1 Appendix). In addition, age-standardized rates with 95% CIs were calculated in order to adjust for differences in the age structures of the compared populations (i.e., differences between the analyzed countries). To calculate age-standardized rates, the European standard population was used, which is a theoretical population with its age distribution based on actual age distributions in the populations of the European countries [21]. A pooled estimate was calculated based on age-standardized rates. Further, average YLLs due to TBI per case were calculated with 95% CIs for each country and overall (mean YLLs per case calculated by dividing the sum of YLLs by the number of cases in the respective group or subgroup).

In addition, both numbers of YLLs and YLL rates were stratified by sex, and external cause of injury. Differences between rates (by age group and sex) are presented as rate ratios with 95% CIs.

In order to evaluate the relative importance of TBI in the context of all injuries, proportions of TBI YLLs out of overall injury YLLs were calculated. For this calculation, cases with unspecific codes (e.g., “other and unspecified effects of external causes”); deaths caused by exposure to heat, frost, or intoxication (T15–T65); and cases with other generalized causes (T66–T78; T80–T88) were excluded. Deaths with these causes are either not directly comparable to deaths due to TBI—because the circumstances or mechanisms leading to death are substantially different (e.g., traffic injury versus frostbite)—or the true cause of death is actually unknown and inclusion could bias the calculated proportions by increasing the denominator. Thus, only the following codes were used in the denominator: injuries to the head (S00–S09); injuries involving multiple body regions (T00–T07); injuries to unspecified trunk, limb, or body region (T08–T14); certain early complications of trauma (T79); and sequelae of injuries, of poisoning, and of other consequences of external causes (T90–T98). In order to allow for comparisons, S1–S3 Tables present these analyses with all injury-related deaths (no exclusions) used as the denominator.

The total number of TBI YLLs in the EU was estimated by extrapolating the pooled crude rate of YLLs in the 16 analyzed countries to the EU-28 population count published by Eurostat [22].

Statistical methods

Pooled analyses were performed in order to estimate summary age-standardized rates of YLLs. In order to model possible heterogeneity of rates in the different countries, the random effects model was applied [23] by the DerSimonian and Laird method, in line with previous studies [5,24]. To assess the heterogeneity of the pooled estimations, I² values with 95% confidence intervals were calculated. To make our findings as comparable as possible, in S4 Table we provide pooled rates calculated by applying the fixed effects model (in response to the suggestions of a peer reviewer). The original analysis plan for the study is reported in S1 Text.

Main findings

We conducted a large-scale, cross-sectional, population-based analysis of YLLs due to TBI in 16 European countries for the year 2013. We found that in the selected countries a total of 17,049 TBI-related deaths occurred in 2013. These translated into a total of 374,636 YLLs. The pooled age-standardized rates of YLLs per 100,000 were 259.1 (95% CI: 205.8 to 312.3) overall, 427.5 (95% CI: 290.0 to 564.9) in males, and 105.4 (95% CI: 89.1 to 121.6) in females. Males contributed more substantially to the overall numbers of YLLs than females (282,870 YLLs, 76% of all TBI YLLs), which translated into a rate ratio of 3.24 (95% CI: 3.22 to 3.27). Each TBI death was on average associated with 24.3 (95% CI: 22.0 to 26.6) YLLs overall, 25.6 (95% CI: 23.4 to 27.8) in males and 20.9 (17.9 to 24.0) in females. Falls and traffic injuries were the most common external causes of TBI. TBI contributed on average 41% (44% in males and 34% in females) to overall injury YLLs in the 16 countries. Extrapolating our findings, about 1.3 million YLLs were attributable to TBI in the EU-28 in 2013 overall, 1.1 million in males, and 270,000 in females. To our knowledge, this is the largest and most comprehensive analysis of TBI YLLs in Europe to date.

Interpretation and generalizability

For all our analyses, microdata on causes of death obtained from Eurostat were used. Eurostat collects data on causes of death from countries, which extract them from death certificates in accordance with EU Commission Regulation No 328/2011 on community statistics on public health and health and safety at work [25]. This regulation defines the scope, provides definitions of variables and characteristics of the data, and aims to achieve the highest possible degree of harmonization and comparability of the information obtained from various countries [26]. Thus, to the best of our knowledge, for our study we have used the most valid and comparable data that were available—as such, we believe that the information and comparisons presented in this paper are valid. However, the relatively large between-country differences in YLLs suggest that there still may be factors beyond true country variability affecting the size of this variation. In general, countries follow ICD-10 standards, making the data collection procedures on causes of death relatively homogenous; however, factors such as differences in interpretation and use of ICD-10 rules at the national level, nonapplication of WHO updates, and differences in reporting of deaths of residents abroad and deaths of nonresidents in the reporting country may hinder the general comparability of the data on causes of death [26], and thus the generalizability of our findings.

Besides these systematic factors, country characteristics, such as age distribution and general economy level, may influence the numbers of reported TBI deaths. A recent study that evaluated TBI-related mortality in 25 European countries found that countries with higher gross domestic product tended to have higher TBI death rates. Furthermore, this study reports that the substantial between-country differences in TBI death rates could be driven by varying degrees of attributing death to multiple injuries—countries that reported relatively low numbers of TBI-related deaths at the same time reported relatively high numbers of death due to multiple injuries. Thus, a substantial number of TBI deaths may be in some countries “hidden” under multiple injury deaths [5]. These factors may also have influenced the findings of this study, as both studies used data from the same source. In a similar manner, variations in use of “garbage codes” for cause of death (e.g., general causes such as “unknown” or “other”) may have an influence on the between-country variation observed in this study.

Despite all these issues, by using data routinely collected by an official European authority, based on a specific EU regulation, for the same year, and using the same coding system for case ascertainment, our study overcame many limitations of other types of investigations (such as the heterogeneity of time, case ascertainment, and geographical coverage of studies included in systematic reviews of TBI epidemiology [14–16])—which supports the validity and generalizability of our findings.

Comparison to other studies

To our knowledge, only 2 previous studies have specifically analyzed and reported TBI-related YLLs. A study from the Netherlands [18] reported 118,207 TBI-related YLLs annually for the period of 2010–2012. As data for the Netherlands was not made available by Eurostat for our study, we are not able to directly compare these findings to ours. However, by dividing the number of YLLs reported in the Dutch study by the mean total population of the Netherlands for 2010–2012 according to Eurostat (16,653,712) [22], we were able to obtain a crude YLL rate of 710 per 100,000 persons per year. This rate is higher than that of the highest ranking country in our comparison (Estonia, with a crude TBI-related YLL rate of 525.6). However, it is important to note that the number of deaths in the Dutch study was estimated using average case fatality rates, which further limits the comparability with our findings.

Another study analyzed TBI-related YLLs in New Zealand [12] and reported a total of 14,386 TBI-related YLLs in 2010. Using this value and New Zealand’s 2010 population estimate of 4,353,000 [27] yields a crude rate of TBI-related YLLs of 330.5 per 100,000 persons per year. Such a rate is within the range of crude rates reported for the analyzed countries in our study and is similar to rates we found for Slovakia (319.4), Serbia (319.7), and Croatia (293.3). However, we note that this study used different methods for YLL estimation (e.g., different life table) and a different definition for TBI death, which hinders the general comparability of its findings to ours.

An earlier study assessed the overall burden of injuries in 6 European countries [28]. In both this study and ours, rates of overall injury-related YLLs are reported for Austria, Denmark, and Ireland, and thus can be compared. For Austria the previous study reported 1,710 YLLs per 100,000 persons per year; in our study, the crude rate was 1,078. For Denmark the rates were 1,550 versus 766, and for Ireland the rates were 1,530 versus 1,035. Thus, the rates found in our study are consistently lower, which may be caused by the relatively large time lag between the 2 studies (e.g., 1999 versus 2013).

The pooled age-standardized all injury YLL rate in our study is lower than the global age-standardized YLL rate reported by the Global Burden of Disease Study 2013 [6]—1,355.5 (95% CI: 1,083.8 to 1,627.1) in our study and 2,945 (95% uncertainty interval: 2,796 to 3,129) in the Global Burden of Disease Study 2013 injury study. This might be explained by the fact that the latter study reports a global estimate, and the observed difference may well reflect the patterns of injury deaths globally. On the other hand, the reported global rate falls within the range of age-standardized all injury YLL rates reported in the 16 analyzed countries in our study and is comparable to the rates found in Lithuania (3,554.2 [95% CI: 3,533.0 to 3,575.5]) and Estonia (2,339.1 [95% CI: 2,313.2 to 2,365.1]).

Implications for policy-making and research

In this study we performed to our knowledge the largest and most comprehensive analysis of TBI-related YLLs in Europe to date. Previous studies relied on analyzing and presenting death rates or case fatality rates in order to describe the magnitude of the burden of fatal TBI for the populations of countries [14–16]. Although important, such analyses do not put the problem of TBI deaths into the broader context of social and economic affairs of the respective countries. Our findings emphasize that with each death, large numbers of years of life are lost in the age groups of economically active people, which underlines the significant burden TBI imposes on the economy of countries and the serious impact on the life of families. We quantified the average number of YLLs due to TBI deaths per 100,000 persons and the average number of YLLs per TBI death in 16 countries, and provided these findings stratified by sex, age, and external cause of injury. We believe that this information could facilitate policy-makers in tailoring preventive action so that actions are targeted to the high-risk populations. Communicating the implications of TBI deaths using YLLs as a measure (rather than numbers of deaths) may help the general public to better grasp the magnitude of the problem, and could help to raise awareness about TBI as a major public health problem in general.

Although YLLs provide a more comprehensive measure of the burden of TBI deaths on a population, they do not capture the burden imposed by nonfatal TBI. Our findings can serve as a basis for analysis of the overall burden of TBI using metrics such as DALYs.

Limitations

There are some limitations to this study that we would like to acknowledge. The differences in the calculated YLLs and all related variables inherit all possible bias and errors that were present in the raw data provided for us by Eurostat. We were not able to control these, or mitigate them in any way. All interpretations of our findings should be made with this in mind. The extrapolations to the population of the EU-28 are based on 16 countries, and it is possible that they are biased—they should be considered an estimation only. The reason for analyzing data from 16 countries and extrapolating the results to the whole EU-28 instead of using data from all countries was that the data for the rest of the EU countries were not available in the necessary format or with sufficient detail (e.g., countries did not provide the ICD-10 codes for nature of injury along with the ICD-10 codes for external causes, or they provided data grouped into larger age groups). Although this limits the validity of our extrapolations, the approach seemed justified under the circumstances. In order to analyze the full population burden of TBI, nonfatal cases must also be taken into consideration using metrics such as YLDs or DALYs. In this paper we were not able to estimate these, due to the nonavailability of data. Future research should be oriented towards these analyses.

Conclusions

Our study showed that TBI-related deaths have a substantial impact at the individual and population level in Europe and present an important societal and economic burden that must not be overlooked. We provide information valuable for policy-makers, enabling them to evaluate and plan preventive activities and resource allocation, and to formulate and implement strategic decisions. In addition, our results can serve as a basis for analyzing the overall burden of TBI in the population.