The objective of this study was to examine whether the implementation of the helmet law had reduced the likelihood of head injury and the associated medical cost in Taiwan. Data were taken from the 1996 and 1998 population-based data. In total, 888,179 and 921,058 effective samples were used in the study from the two years. Two different types of regression model were adopted to evaluate the impact of the motorcycle helmet use law on incidences of head injury and associated medical cost and hospital length of stay. The results reveal that medical cost is down by 11.5 percent and hospital LOS has fallen by 18.58 percent. Thus, with the introduction of the motorcycle helmet use law having had a demonstrably positive influence on motorcycle head injuries and fatalities, significant savings are clearly being achieved, not only in terms of economic and social costs, but also with regard to medical cost.
The law establishing the mandatory use of motorcycle helmets in Taiwan came into force on June 1, 1997. Riding motorcycle without a helmet will be fined NT$500 (US$15) for each person according to Taiwan Traffic Penalty Rules. It had been previously determined that, in 1996, Taiwan had a total of 9.28 million motorcycles, as compared to 4.99 million automobiles and, according to statistics provided by Ministry of Interior, there were 2,990 traffic-related fatalities, defined as a fatality occurring at the scene of the crash or within 24 hours of the crash, that year [
Numerous studies have indicated a high correlation between fatal head injuries and accidents involving motorcycles, with helmets having been shown to be extremely effective in preventing serious head injury. The enforcement of laws relating to the mandatory use of motorcycle helmets in many countries has also proven to have significantly reduced both head injuries and fatalities [
Thus far, there have been several studies carried out in Taiwan evaluating the law on the mandatory use of motorcycle helmets and the subsequent effects on both head injuries and fatalities. Using data from police accident reports, Tsai and Hemenway (1999) [
Chiu et al. (2000) [
Lee et al. (2010) [
There have been many studies examining the motorcycle helmet law and the tendency has been to focus on reductions in injuries and fatalities using data from specific areas or hospitals. To the best of our knowledge, the association between the implementation of helmet use law and the likelihood of head injury and the associated medical care utilization has drawn precious little attention. Nor has any study addressed differences in such a relationship using population-based database. This study is the first of its kind worldwide to use a selected population-based data in an attempt to estimate the impact of the helmet use law on the likelihood of head injury and the associated medical utilization. If such likelihood and utilization have been effectively reduced by the implementation of the motorcycle helmet law, through a reduction in the severity of head injuries, then valuable medical resources can be used more effectively in other areas of inpatient care.
The data used in this study is obtained from the Taiwan National Health Insurance Research Database (NHIRD). The NHIRD represents one of the most comprehensive population-based databases in the world, given that the National Health Insurance (NHI) system in Taiwan covers approximately 99 percent of the island’s entire population. The data includes information on all inpatient and outpatient medical claims, contracted medical facilities and physicians. The Longitudinal Health Insurance database (LHID) created and released by the Taiwan National Health Research Institute consists of 1,000,000 subjects by systematically selecting a representative database from the 25.68 million enrollees listed in the 2005 Registry of Beneficiaries from NHIRD. There is no significant difference between the patients in the LHID and the NHIRD in gender distribution, age distribution, or a beneficiary’s annual sum of insurance claims. Consequently, the LHID has been widely used in the health researches and policy making.
As the NHI system was implemented in Taiwan in March 1995, we are able to use this database to fully evaluate the impact of the 1997 helmet use law on medical care utilization involving head injuries. We use the LHID dataset released in 2007 as our study sample, which involved the random sampling one million NHI beneficiaries. Our sample comprises of all inpatient and outpatient medical claims one year prior to the implementation of the motorcycle helmet use law (1996) and one year after its implementation (1998).
The dataset provides primary diagnoses from the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) and up to four secondary diagnoses. After deleting some samples with missing information, we were left with a total of 888,179 and 921,058 observations of patients in 1996 and 1998, respectively. ICD-9-CM codes 800-999 and E codes (external injury) 810-819 (motor vehicles) were used to identify the head injury cases, and the final dataset of head injuries for analysis in this study includes 609 and 645 cases. The definitions of the variables selected for this study are provided in Table
Definitions of the dependent and independent variables.
Variable | |
---|---|
Probability | Traffic accident with head injury = 1; other = 0 |
Medical cost | NT dollarsa |
Length of stay in hospital | Inpatient days |
After motorcycle helmet use law = 1; | |
Before motorcycle helmet use law = 0 | |
North | Patient lived in Taipei City, Keelung City, Taipei County, Yilan County, Hsinchu City, Taoyuan County, Hsinchu County, and Miaoli County; yes = 1, other =0 |
Center | Patient lived in: Taichung City, Taichung County, Changhua County, and Nantou County; yes = 1, other = 0 |
South | Patient lived in: Tainan City, Chiayi City, Yunlin County, Chiayi County, Tainan County, Kaohsiung City, Kaohsiung County, Pingtung County, and Penghu County; yes = 1, other = 0 |
East | Patient lived in: Hualien County and Taitung County; yes |
Male = 1; female = 0 | |
Children | Patient’s age from 0 to 14 years old = 1; other = 0 |
Adult 15–24 | Patient’s age from 15 to 24 years old = 1; other = 0 |
Adult 25–64 | Patient’s age from 25 to 64 years old = 1; other = 0 |
Elderly | Patient’s age over 65 years old = 1; other = 0 |
Special municipality | Patient lived in special municipality (regions with populations of not less than 1,250,000 and have special requirements in their political, economic, cultural, and metropolitan developments) = 1; other = 0 |
City | Patient lived in city (regions with populations of not less than 500,000 but not more than 1,250,000 and occupy major political, economic, and cultural roles) = 1; other = 0 |
County administered | Patient lived in county-administered (regions with populations of not less than 150,000 but not more than 500,000, and possess thriving industrial and commercial development, adequate financial resources for self-government, convenient transportation, and comprehensive public facilities) = 1; other = 0 |
Town | Patient lived in town (regions with populations of less than 150,000, and possess thriving industrial and commercial development) = 1; other = 0 |
Village | Patient lived in village (regions with populations of less than 150,000, and less economic development) = 1; other = 0 |
aUS$1 = NT$33.4 in 1998.
bRural/urban stratify according to the Local Government Act.
The analysis in this study is undertaken in two parts. Firstly, we present the sample characteristics of percentage, medical cost, and hospital length of stay attributable to motorcycle head injuries both before and after the introduction of the helmet use law. Secondly, we apply two different types of regression model to explore the influence of helmet use law on probability, medical cost, as well as length of stay. In the first type, we formulated the logistic regression to estimate the impact of helmet use law on the probability of head injury. The first type of model is as follows:
Although the
The second type of estimation considers the amount of medical costs and the hospital length of stay conditional on the occurrence of head injury. Two regression analyses are undertaken in the second type, each with a specific purpose. Since the dependent variable (the amount of expenses) is a continuous variable, we adopted the ordinary least squares regression as the means of investigating whether after the implementation of helmet use law incurred lower medical expenditures than before. Using the “cost-” dependent variable, the
The statistical package, STATA (STATA Corporation, version 11.0), was used for all of the statistical analyses undertaken in this study, with a two-side
The descriptive information on the sample is presented in Tables
Percentage of head injury (1996 and 1998) in Taiwan.
Variable | Before the helmet use law | After the helmet use law | % Change | |
---|---|---|---|---|
0.0812% | 0.0813% | 0.1232% | ||
<0.0001 | ||||
North | 0.0461% | 0.0356% | −22.8046% | |
Center | 0.0703% | 0.0796% | 13.1933% | |
South | 0.1058% | 0.1201% | 13.4863% | |
East | 0.0544% | 0.0705% | 29.6525% | |
<0.0001 | ||||
Male | 0.0949% | 0.1008% | 6.1654% | |
Female | 0.0688% | 0.0637% | −7.4226% | |
<0.001 | ||||
Children | 0.0106% | 0.0100% | −6.0954% | |
Adult 15–24 | 0.1899% | 0.1603% | −15.5572% | |
Adult 25–64 | 0.0797% | 0.0886% | 11.1168% | |
Elderly | 0.0797% | 0.0803% | 0.8056% | |
<0.001 | ||||
Special municipality | 0.0506% | 0.0465% | −8.2351% | |
City | 0.0607% | 0.0474% | −21.9082% | |
County administered | 0.0724% | 0.0710% | −1.8142% | |
Town | 0.1168% | 0.1719% | 47.2089% | |
Village | 0.1134% | 0.0919% | −18.9339% |
Observations = 1,809,237 (before law in 1996 = 888,179; after law in 1998 = 921,058); number of head injury = 1,254 (before law = 609; after law = 645).
Average medical cost per head injury in a motorcycle accident.
Variable | Before the helmet use law | After the helmet use law | % Change | |
---|---|---|---|---|
21318.26 | 18866.76 | −11.50% | ||
0.0289 | ||||
North | 31007.97 | 23769.13 | −23.35% | |
Center | 24298.08 | 15407.69 | −36.59% | |
South | 18047.56 | 17821.99 | −1.25% | |
East | 19979.75 | 21683.67 | 8.53% | |
0.5428 | ||||
Male | 23921.35 | 20804.87 | −13.03% | |
Female | 18071.60 | 16087.79 | −10.98% | |
0.0025 | ||||
Children | 17020.90 | 9428.84 | −44.60% | |
Adult 15–24 | 19818.66 | 16532.17 | −16.58% | |
Adult 25–64 | 22781.54 | 21450.68 | −5.84% | |
Elderly | 21183.06 | 12797.73 | −39.59% | |
0.0173 | ||||
Special municipality | 29674.75 | 21949.89 | −26.03% | |
City | 27034.62 | 19247.94 | −28.80% | |
County administered | 16338.67 | 15193.83 | −7.01% | |
Town | 24339.35 | 19275.41 | −20.81% | |
Village | 23501.87 | 19282.52 | −17.95% |
Observations = 1,254.
Unit: NT $, 1 US$ = 33.4 NT$ in 1998.
Medical costs due to head injuries are only for individuals that sustained a head injury in a motorcycle accident.
Average length of stay per head injury in a motorcycle accident.
Variable | Before the helmet use law | After the helmet use law | % Change | |
---|---|---|---|---|
6.78 | 5.52 | −18.58% | ||
0.0140 | ||||
North | 7.76 | 6.53 | −15.85% | |
Center | 6.66 | 5.36 | −19.52% | |
South | 6.62 | 5.04 | −23.87% | |
East | 5.63 | 7.07 | 25.58% | |
0.0049 | ||||
Male | 6.84 | 5.67 | −17.11% | |
Female | 6.72 | 5.30 | −21.13% | |
0.0008 | ||||
Children | 6.10 | 4.63 | −24.10% | |
Adult 15–24 | 5.94 | 4.71 | −20.71% | |
Adult 25–64 | 7.29 | 6.10 | −16.32% | |
Elderly | 8.60 | 5.07 | −41.05% | |
0.1053 | ||||
Special municipality | 8.46 | 6.46 | −23.64% | |
City | 6.35 | 4.67 | −26.46% | |
County administered | 5.61 | 5.07 | −9.63% | |
Town | 7.24 | 5.44 | −24.86% | |
Village | 7.06 | 5.65 | −19.97% |
Observations = 1,254.
Unit: NT $, 1 US$ = 33.4 NT$ in 1998.
Length of stay due to head injuries are only for individuals that sustained a head injury in a motorcycle accident.
Table
Table
Logistic regression for the likelihood of sustaining a head injury in a motorcycle accident.
Variable | Coefficient | Odds ratio | Standard error | |
---|---|---|---|---|
−8.1149*** | 0.0983 | <0.0001 | ||
−0.2693** | 0.764 | 0.1057 | 0.0108 | |
Policy*Center | 0.3664** | 1.443 | 0.1643 | 0.0258 |
Policy*South | 0.3881*** | 1.474 | 0.1334 | 0.0036 |
Policy*East | 0.5242 | 1.689 | 0.4110 | 0.2021 |
Center | 0.3881*** | 1.474 | 0.1334 | 0.0036 |
South | 0.5242 | 1.689 | 0.4110 | 0.2021 |
East | 0.0561 | 1.058 | 0.3108 | 0.8568 |
(ref.: north) | ||||
0.3301*** | 1.391 | 0.0572 | <0.0001 | |
Children | −2.0731*** | 0.126 | 0.1647 | <0.0001 |
Adult 15–24 | 0.6227*** | 1.864 | 0.0606 | <0.0001 |
Elderly | −0.0431 | 0.958 | 0.1192 | 0.7178 |
(ref.: adult 25–64) | ||||
City | −0.2614** | 0.770 | 0.1260 | 0.0380 |
County administered | 0.3246*** | 1.383 | 0.0947 | 0.0006 |
Town | 0.9101*** | 2.485 | 0.0927 | <0.0001 |
Village | 0.4130*** | 1.511 | 0.0891 | <0.0001 |
(ref.: special municipality) |
a***Denotes significant at 0.01 level; **denotes significant at 0.05 level; *denotes significant at 0.1 level.
bReference groups: north, female, adult 25–64, special municipality, Policy*North.
Observations = 1809237.
The relationship between medical cost attributable to head injuries and the implementation of the law on motorcycle helmet usage is shown in Table
OLS regression for medical cost per head injury in a motorcycle accident.
Variable | Coefficienta | Standard error | |
---|---|---|---|
9.49096*** | 0.06281 | <0.0001 | |
Policy*Center | −0.26555* | 0.16026 | 0.0978 |
Policy*South | −0.10464 | 0.12879 | 0.4166 |
Policy*East | 0.26161 | 0.40985 | 0.5234 |
Center | −0.00096 | 0.12829 | 0.9940 |
South | −0.06548 | 0.09799 | 0.5042 |
East | −0.30702 | 0.33084 | 0.3536 |
(ref.: north) | |||
0.13557** | 0.05293 | 0.0106 | |
Children | −0.32607** | 0.14953 | 0.0294 |
Adult 15–24 | −0.21468*** | 0.05513 | 0.0001 |
Elderly | −0.10688 | 0.10941 | 0.3288 |
(ref.: adult 25–64) | |||
City | −0.02904 | 0.12142 | 0.8110 |
County administered | −0.11522 | 0.08924 | 0.1969 |
Town | −0.01480 | 0.08723 | 0.8653 |
Village | 0.00583 | 0.08519 | 0.9454 |
(ref.: special municipality) |
a***Denotes significant at 0.01 level; **denotes significant at 0.05 level; *denotes significant at 0.1 level.
bReference groups: north, Female, adult 25–64, special municipality, Policy*North.
AdjR2 = 0.0211.
Observations = 1,254.
As can be seen in Table
Poisson regression for length of stay per head injury in a motorcycle accident.
Variable | Coefficienta | Standard error | |
---|---|---|---|
2.0449*** | 0.0265 | <0.0001 | |
−0.0114 | 0.0407 | 0.7797 | |
Policy*Center | −0.2175*** | 0.0708 | 0.0021 |
Policy*South | −0.2780*** | 0.0567 | <0.0001 |
Policy*East | 0.3236* | 0.1840 | 0.0785 |
Center | 0.0383 | 0.0555 | 0.4897 |
South | −0.0161 | 0.0424 | 0.7036 |
East | −0.2333 | 0.1538 | 0.1293 |
(ref.: north) | |||
0.0123 | 0.0238 | 0.6056 | |
Children | −0.2245*** | 0.0709 | 0.0015 |
Adult 15–24 | −0.2330*** | 0.0253 | <0.0001 |
Elderly | −0.0042 | 0.0471 | 0.9292 |
(ref.: adult 25–64) | |||
City | −0.1608*** | 0.0567 | 0.0045 |
County administered | −0.2153*** | 0.0411 | <0.0001 |
Town | −0.0525 | 0.0388 | 0.1756 |
Village | −0.0141 | 0.0375 | 0.7079 |
(ref.: special municipality) |
a***Denotes significant at 0.01 level; **denotes significant at 0.05 level; *denotes significant at 0.1 level.
bReference groups: north, female, adult 25–64, special municipality, Policy*North.
Observations = 1,254.
Studies undertaken in many Western countries have found that the introduction of laws mandating the use of motorcycle helmets can effectively reduce both injuries and fatalities [
Statistics provided by the Ministry of the Transportation and Communications in Taiwan also reveal a significant fall in motorcycle traffic fatalities, from 1,318 in 1996, to 1,042 in 1998 (a 20.9 percent reduction), following the introduction of the motorcycle helmet use law. Furthermore, this significant reduction in the total number of motorcycle fatalities has been achieved despite the steady increase in the total number of motorcycles, from 9.28 million in 1996, to 10.52 million in 1998 [
Although a number of studies have emerged suggesting decreased risks of head injury and fatalities associated with the passage of helmet use law, no nationwide population-based study has ever been undertaken investigating the impact of the motorcycle helmet usage law on the likelihood of head injury and the associated medical care utilization relating to incidences of head injury. This study is the first of its kind worldwide to use a selected population-based data to explore the association between the motorcycle helmet use law and the associated medical utilization.
Our empirical results revealed that following the 1997 implementation of the motorcycle helmet use law, significant reductions in the northern region have been achieved, not only in the probability of head injury, but also in the medical cost and hospital length of stay attributable to such injuries. The probability of motorcycle head injury has been reduced by 22.8 percent, whilst medical cost and hospital length of stay are down by around 23.5 percent and 15.8 percent, respectively. However, regional differences do exist. The results showed that non-northern motorcyclists increased their probability of head injury after the law. Nevertheless, except eastern region where most of the mountains were located and were less developed and populated than the other regions, central and southern regions had lower medical cost and hospital length of stay. As compared to the most populated area of northern Taiwan, other regions in Taiwan seem to have much higher incidence of motorcycle head injuries in the logistic regression. We surmise that this is because, unlike the northern area, the limited number of police in other areas tends to hamper the enforcement of motorcycle helmet use law, which thereby contributes to higher numbers of head injuries/fatalities attributable to motorcycle accidents. Nevertheless, medical cost and hospital length of stay of head injury were reduced significantly in the central and southern regions compared to the northern region after the implementation of motorcycle helmet use law in the OLS and Poisson regression results. In other words, although the likelihood of head injury incidence was not reduced after the law, the severity of head injury incidences was reduced significantly after the law.
This finding is consistent with the previous study [
Two limitations to this study need to be addressed. First, being an administrative claims dataset, it lacks information on some demographic characteristics such as education and income. Second, other factors, such as weather conditions and wearing helmet or not which might contribute to incidence of head injury, are also not available in this database. Accounting for these factors would enable more refined conclusions and,therefore, targeted policy initiatives.
Nevertheless, incidences of motorcycle head injury and the resultant mortality and morbidity rates impose serious socioeconomic costs, in terms of lost working time and income, on both the victims and their families. Our results indicate that medical cost and hospital length of stay have been reduced in Taiwan due to the reduction in the severity of motorcycle head injuries. Thus, with the introduction of the motorcycle helmet use law having had a demonstrably positive influence on motorcycle head injuries and fatalities, significant savings are clearly being achieved, not only in terms of economic and social costs, but also with regard to medical cost and medical resources.
This should be reassuring news for the authorities in Taiwan, proving that their efforts to promote motorcycle helmet usage are delivering positive results. Such an accomplishment may also provide reassurance to other developing countries in which motorcycles tend to be the primary means of transport, such that they may also consider the introduction of laws mandating the use of motorcycle helmets.
This study was supported by the National Science Council of Taiwan (NSC 96-2416-H-305-002-MY3). This study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health, Taiwan and managed by the National Health Research Institutes.