**4. Discussion**

Head injury continues to be a major health problem around the world in both developed and developing countries [1–4]. The aim of this study was to examine the incidence and in-hospital head injury-related mortality in Taiwan. Compared with other regions or countries, the incidence rate in our study, 220.6/100,000, was higher than in the U.S. (103/100,000), Finland (101/100,000), and India (160/100,000) but lower than in Europe (235/100,000) [2,9,16,28]. However, the definitions and inclusion criteria are different, which complicates comparisons between studies and regions.

Among those aged 15 to 74 years, the most common mechanism of injury is motorcycle-related traffic accident because motorcycles are the most popular transportation vehicles in Taiwan. Motorcycle riders and passengers should be the target population for injury prevention, even in the age group between 65 to 74 years. For those older than 65 years, falls caused more than two-thirds of the injuries and hospitalization rates in this age group. This deserves more attention because trauma does not only affect the younger population; therefore, a focus on injury prevention programs should be aggressively advocated for the elderly. For the people younger than 15 years old in Taiwan in the study by Tsai et al. in 2004, the incidence rate was higher in the age groups of 4 to 9 and 10 to 14 years. The main cause of pediatric head injury was traffic injury followed by falls. Of all pediatric traffic injuries, motorcycle-related injury had the highest incidence, followed by pedestrian and bicycle-related injury [29].

In the United States, head injuries cause about 2.5 million emergency department (ED) visits, 280,000 hospitalizations, and 50,000 deaths annually [4]. In Taiwan, our study showed a hospitalization death rate of 11.8/100,000. Compared with the head injury-related death rates in the U.S. (18.1/100,000) [18], Finland (18.3/100,000) [28], India (20.0/100,000), and Europe (15.4/100,000) [22], the standardized death rate in Taiwan, at 11.8/100,000, was lower than in these other studies. However, our dataset only included patients who had been admitted to hospital, and head injury-related accidents caused 4935 in-hospital deaths among the total 7640 hospitalization mortality cases in 2007–2008 in Taiwan. The dataset did not include deaths at the scene of the accident, during transportation, or in the emergency department. In Taiwan, there were 14,207 accidental deaths in 2007–2008 and our in-hospital dataset only included half of the deaths. It is reasonable to assume that the annual head injury-related mortality rate is underestimated in this study. With the aging population, the devastating effects of head injury will become increasingly serious.

Our regression model showed that only pedestrians hit by a motor vehicle and those who had a fall from a height were associated with a higher probability of mortality; other causes of head injury were not. This result might be caused by the lack of helmet protection and pedestrians hit by a motor vehicle is a well-known dangerous mechanism of injury [30]. In Taiwan, NHI provides a high accessibility to hospital resources to all citizens, but our results show that rural residents and the subgroup with lower monthly income were associated with lower survival probability after moderate-to-severe head trauma. Some researchers have found that the medical institutions providing higher levels of trauma care are often located in urban areas [31]. Their results showed that the odds ratio for death was higher in the hospitals located in rural areas than in hospitals located in more urbanized areas. One study found huge rural-urban disparities in mortality from unintentional injuries [32]. Lower monthly incoming and living in a rural area are associated with many factors leading to higher mortality rate, such as low education and health literacy, risky environments, and low economic support for medical care. We need to focus on these vulnerable populations, reinforce cause-specific prevention programs to reduce the mortality rate, relocate more resources to fairly provide optimal care, and periodically monitor the effectiveness of these programs.

This study had some limitations. First, this is a retrospective study including all the NHI claim data about injury-related hospitalizations during the study period. However, the accuracy of NHI claim data is assured given the severe penalty for fraud and erroneous claiming. Second, this dataset does not include the Glasgow Coma Scale (GCS), which may be misleading for assessing the severity of head injury. The NHI covers brain CT for head injury in an emergency setting, and the severity score ICD/ISS in this study was basically determined by brain CT. The third limitation is that nearly 10% of all admissions were recorded as traffic accidents without further information about the mechanisms, and one-quarter of admissions were recorded without any information about the mechanism of injury. This is reasonable because there was a higher proportion of trauma patients who were found injured without any witnesses or other information source compared to in the United States [14]. Because this study is based on the NHI research database, some detailed information that was not included may lead to ignoring some clinical important factors such as alcohol or drug consumption, related to head injury mortality.
