Main findings
We conducted a large-scale cross-sectional study analysing TBI-related deaths in 30 European countries in 2015 among residents and non-residents to the country of occurrence of the death. Overall, 40,087 TBI-related deaths were identified, of which about 3% occurred among non-residents. Countries with the highest proportion of non-resident deaths were Turkey (11% of all TBI-related deaths), Luxembourg (9%) and Cyprus (5%). Recalculated to reflect the intensity of tourism, Bulgaria, Greece and Austria had the highest rates of TBI-related deaths: 0.7, 0.5 and 0.5 per million overnight stays, respectively. The pooled age-standardised TBI-related mortality rate in non-residents in the analysed countries was 0.2 (95% CI 0.1–0.3), while among residents the rate was 10.4 (95% CI 9.4–11.5). The following key differences between residents and non-residents were observed: (1) in non-residents, TBI-related deaths were shifted to younger populations (in residents, 86% of deaths occurred in 35 years or older, while 78% of deaths in non-residents occurred in 15–64 years old); (2) while falls were the predominant cause among residents (47% of deaths), traffic accidents were predominant in non-residents (36%); (3) males were substantially more predominant among non-residents (male to female ratio of 3.9), compared to residents (male to female ratio of 2.1). By extrapolating the observed pooled age-standardised TBI-related mortality rate, we estimate that annually about 891 TBI-related deaths would occur to non-residents in the EU-27, while about 1022 such deaths would occur in the EU-27 + the UK and 1488 in the population of Europe as a continent.
Comparison to other studies and interpretation
To the best of our knowledge, this is the first study to analyse TBI-related deaths in residents versus non-residents on this scale and thus presents a unique set of findings. In general, the published literature on this topic is extremely limited, therefore, the possibility to directly compare our findings to other studies is restricted.
Previous studies have investigated TBI-related mortality in Europe on a broader than the national scale, but none of them was aimed at the resident or non-resident population specifically. For the general population, European cross-sectional studies (using the same case definition and statistical approach) reported a pooled age-standardised TBI mortality rate of 11.7 per 100,000 in 201211 and 11.3 per 100,000 in 201312, with higher mortality rates in men compared to women. Our study yielded a very similar age-standardised rate of 10.6 per 100,000 population—the slight difference may be explained by the true change in the TBI-related mortality in Europe, but also by including a slightly different set of countries for the analyses. Thus, in general, our findings are similar to previous reports, which confirms the robustness and validity of our analyses.
In our study, the age-standardised mortality rate among residents to countries where the deaths occurred was estimated at 10.4 per 100,000. This can be directly compared to the findings from TBI surveillance in the US which for 1997–2007 estimated the TBI-related death rate among US residents at 18.4 per 100,000 (range: 17.8–19.3)21—the higher mortality rate may be caused by general differences in the population or life-style between Europe and the US, with some key factors such as large numbers of firearm-related TBI deaths22 being possible drivers.
The only study to which our findings are directly comparable is the study of Mauritz et al.15 that analysed hospital admissions and deaths due to TBI among residents and non-residents in Austria. Among visitors, the study estimated the TBI-related mortality in males at 0.9 per 100,000 and in females at 0.7 per 100,000. In our study the rates at 0.3 per 100,000 in males and 0.06 per 100,000 in females which is relatively lower, possibly due to the larger scale of the study and the more substantial variation in the analysed population thereof. On the other hand, the mortality rate for non-residents in Austria in our study has been estimated at 0.61 (in 2015), which is reasonably similar to the estimations for 2009–201115 and support the validity of the findings presented in this paper.
In general, when reflecting the intensity of international travelling in the countries, non-residents in Bulgaria, Greece and Austria were at highest risk of TBI-related deaths of all 30 countries (0.5 or more TBI-related deaths per 1 million overnight stays). While this study does not present evidence, we can hypothesize that the relatively high rate in Austria may be associated with the fact that Austria is a major destination for winter sports. This has been pointed out in the study of Mauritz et al. from Austria which concludes that most of the registered cases of TBI in non-residents may have been associated with winter sports15. In our study, in European countries that are traditional destinations for winter sports such as Austria or Switzerland the proportion of non-resident TBI-related deaths was relatively high (around 5%, see Supplementary Table S1 for details), which may support this hypothesis. However, further studies should investigate this in more detail to provide more evidence. Regarding the high risk observed in Bulgaria or Greece, we could hypothesize that these are driven by traffic accidents—based on our findings in both countries the observed mortality rates due to traffic accidents were higher compared to mortality rates associated with falls or other causes (a pattern that is reversed in most of the countries analysed in this paper, consult Supplementary Table S7). However, further studies should analyse this in more detail.
Our study identified key differences between the demographic characteristics of non-residents and residents dying as consequence of TBI. First, the age structure of non-residents is shifted towards younger age-groups, compared to residents (see Fig. 3 and Table 1), which corresponds with the finding of the study from Austria where lower mean age was observed among non-residents compared to residents (28 years vs. 28 years)15. While these findings may reflect true differences in the age structure of TBI victims between the two compared groups, they may also be biased by the fact that the population of non-residents in European countries tends to be younger than the population of residents23. Secondly, non-residents in our study were prevailingly men. The generally higher risk of TBI-related death in males compared to females is a well-documented fact10,13,21; however, our findings suggest that in case of non-residents the male to female ratio is nearly two-fold compared to the male to female ratio in residents (2.1 in residents vs. 3.9 in non-residents). This finding is in line with a higher proportion of males among tourist fatalities in Europe, which was estimated at up to 80%24,25. In general, the higher risk of injury incidence and mortality in males can be attributed in part to higher engagement in risk taking behaviour compared to females, as documented in the literature26,27. In summary, the population of non-residents dying as a consequence of TBI in Europe is younger and has a higher proportion of males, compared to residents dying due to TBI. These differences should be observed by public health and travel medicine professionals to better target prevention. Further epidemiological research is needed to elucidate in more detail the causes and drivers of TBI incidence and mortality among non-residents. Targeted action, such as more detailed surveillance implemented by European or national authorities is also warranted and may help to prevent avoidable morbidity and mortality caused by TBI by informing targeted and tailored prevention.
Limitations and generalizability
The data analysed in this study comes from 30 European countries, where some practices in the process of certification of deaths and assigning causes of death may differ, eventually introducing selection bias. However, the data was provided to us by Eurostat which oversees its aggregation on the European level, and the fact that the certification of deaths in the EU is governed by an EU regulation (No 328/2011)18 ensures the highest level of validity and between-country comparability of these data that is possible under the circumstances. The data used for the analyses in this study comes from a single year of 2015 and thus no trends could be analysed and comparisons between multiple years were not possible. We are aware of this limitation and made all efforts to obtain data for further years in the needed detail, but this was not possible. Furthermore, 2015 was the most recent year for which we were able to obtain the data needed for the study and may not reflect the true epidemiological situation at the time of publication of this study. However, previous studies looking at TBI-related mortality in general populations7,10,13 (E.g., not split to residents vs. non-residents) using the same data sources and multiple consecutive years showed little variation in injury characteristics (such as sex, causes, age structure) between the years. This suggests that the findings regarding the structure of residents versus non-residents among the countries would not vary substantially under normal circumstances and could be generalizable to other years as well.
TBI for the purposes of this study was defined using the full range of codes included in the chapter of “Injuries to the head” of the ICD-10 (E.g., S00-S09) and the code for sequelae of injuries to the head (E.g., T90), including a variety of injuries of diverse severities. We have considered all deaths where the primary cause was coded using one of these codes to be TBI-related deaths and included them in the study. We are aware that some of the codes denote “head injuries” and do not directly imply a TBI. However, we have chosen this strategy for the following reasons: (1) previous research has shown that more narrow case definitions may lead to underestimation of the actual incidence or mortality of TBI in population level studies using administrative data sources28,29; (2) previous studies also suggest that for analyses focusing on epidemiology and prevention, a broader definition of TBI is advisable (as opposed to studies looking at clinical characteristics and outcome where the case definition should be more strict)28; and (3) the same case definition has been used by previous European-wide studies of TBI epidemiology and burden11,12, which creates grounds for better comparability of findings across all studies, and allows for bringing the main findings of this study into context with the overall epidemiological patterns of TBI incidence and mortality in Europe. Furthermore, we note that when broken down into specific diagnostic groups, most of the deaths (72%) were coded as intracranial injuries (ICD-10 code S06), about 10% of cases were coded S02 (fracture of skull and facial bones), about 11% were coded S09 (Other and unspecified injuries of head), with the remaining 7% of cases distributed among the other diagnostic groups within the case definition (see Supplementary Table S9 for details). Such distribution suggests the case definition used in the study was robust and valid. On the other hand, we note that the findings of our study should be interpreted with caution and with these characteristics in mind.