*2.1. Variables*

According to capacity and services, hospitals in Taiwan were classified into three levels: medical centers (MC), regional hospitals (RH), and local hospitals (LH). Taiwan's definition for medical centers and regional hospitals includes capacity and services. Empirically, the patients were stratified into nine age groups: 0–14, 15–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, and >84 years of age. The patient area of residency was classified into two groups: urban (including suburban) and rural areas. The average population density (persons/km2) is higher in urban areas (2635 persons/km2) than in rural areas (230 persons/km2). According to the level of monthly income, injured patients were classified into four groups: monthly income less than USD \$660, between USD \$660 and \$1320, more than USD \$1320, and dependents, which are those who were injured but not employed, including those insured (NHI) by a spouse, other family member, or social welfare. During the study period, the average Taiwan monthly living cost was USD \$595/person. Pre-existing conditions were defined as the medical co-morbidities of the injured patients who were diagnosed and recorded with ICD-9-CM codes in the same admission data file [18]. We used Charlson's comorbidity index to quantify the PECs [20]. According to Charlson's study in 1987, 17 diseases are included in the formula for calculating the Charlson comorbidity index [21]. The patients were classified into four levels according to their Charlson comorbidity index: 0, 1, 2, and more than 2.

For assessing injury severity, a computerized mapping method that used the ICD codes to obtain injury severity scores, such as the AIS, was employed [22,23]. A computerized mapping system, ICDMAP, for converting injury-related International Classification of Diseases, ninth revision (ICD-9-CM) rubrics into AIS scores, was proposed by MacKenzie et al. in 1989, and their results have been verified [24]. This ICD mapping system has been refined over the years and has been used in several large or population-based studies to classify severity using ICD diagnostic codes [25,26]. Our study applied ICDMAP to the dataset and derived ICD/AIS scores for each injury diagnosis and an ICD/ISS (injury severity score) for each admission.

As the patient group coded with an AIS head score of 1 or 2, as generated by ICDMAP software, was classified as mild head injury, we defined moderate-to-severe head injury as significant head injury with an AIS head score more than 2, which included moderate (AIS head = 3) and severe head injury (AIS head 4, 5, and 6), as shown in Table 1.

**Table 1.** Abbreviated Injury Score (AIS) head score and head injury severity classification.


ICD E-codes in the claim data were also used to classify the mechanism of injury [27]. A patient is classified as the driver or passenger in a motor vehicle traffic accident if the number after the decimal point is zero or one, respectively (e.g., E811.0, E811.1, E812.0, or E812.1). A patient is classified as a motorcycle rider or passenger injured in a traffic accident if the number after the decimal is two or three, respectively (e.g., E811.2, E811.3, E812.2, or E812.3). A patient is classified as having an injury of an unspecified nature in a motor vehicle traffic accident if the number after the decimal point is nine (e.g., E811.9, E812.9, etc.), as shown in Table 2.


**Table 2.** ICD-E codes and injury mechanisms.

ICD-E code: International Classification of Disease-external cause of injury code.
