- Email: [email protected]
Safety riding program and motorcycle-related injuries in Thailand
Accident Analysis and Prevention 58 (2013) 115–121
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Safety riding program and motorcycle-related injuries in Thailand Patarawan Woratanarat a , Atiporn Ingsathit b,∗ , Pornthip Chatchaipan b , Paibul Suriyawongpaisal c a b c
Department of Orthopaedics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand Section of Clinical Epidemiology, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand Department of Community Medicine, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand
a r t i c l e
i n f o
Article history: Received 7 February 2013 Received in revised form 30 April 2013 Accepted 2 May 2013 Keywords: Safety riding Motorcycle Injuries Thailand
a b s t r a c t A retrospective cohort study was conducted in Thailand from 2007 to 2009 to evaluate the efficacy of a safety riding program in preventing motorcycle-related injuries. A training group of motorcyclists were certified by the Asia-Pacific Honda Safety Riding Program in either 30-h instruction (teaching skills, riding demonstration) or 15-h license (knowledge, skills, and hazard perception) courses. The control group consisted of untrained motorcyclists matched on an approximately 1:1 ratio with the training group by region and date of licensure. In total, there were 3250 subjects in the training group and 2963 in the control group. Demographic data and factors associated with motorcycle-related injuries were collected. Motorcycle-related injuries were identified using the Road Injuries Victims Protection for injuries claims and inpatient diagnosis-related group datasets from the National Health Security Office. The capture–recapture technique was used to estimate the prevalence of injuries. Multivariate analysis was used to identify factors related to motorcycle-related injuries. The prevalence of motorcycle-related injuries was estimated to be 586 out of 6213 riders (9.4%) with a 95% confidence interval (CI): 460–790. The license course and the instruction course were significantly associated with a 30% and 29% reduction of motorcycle-related injuries, respectively (relative risk 0.70, 95% CI: 0.53–0.92 and 0.71, 95% CI: 0.42–1.18, respectively). Other factors associated with the injuries were male gender and young age. Safety riding training was effective in reducing injuries. These training programs differ from those in other developed countries but display comparable effects. Hazard perception skills might be a key for success. This strategy should be expanded to a national scale. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction Motorcycle-related injuries contribute more than 50% of road traffic injuries in low- and middle-income countries that have a high proportion of motorcycles in road traffic (Hyder et al., 2007). Motorcyclists experienced orthopedic (Amin et al., 2011) and head injuries 20 times as often as four-wheel vehicle drivers (Haworth and Mulvihill, 2005) with a high mortality rate that partially resulted from poor riding practices (Stella et al., 2002). Most motorcyclists start riding at a young age with limited supervision. Immature and risky behavior are most likely the main causes of riding injuries (Wong et al., 2010). The provision of knowledge and riding practice might be important tools to prevent this problem (French et al., 2009).
∗ Corresponding author at: Section of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, 270 Rama VI Road, Rajathewee, Bangkok 10400, Thailand. Tel.: +66 022 011269; fax: +66 022 011284. E-mail address: [email protected] (A. Ingsathit). 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.05.001
Safety riding programs have been established in many countries for decades (Amoran et al., 2005; Baldi et al., 2005; Blanchard and Tabloski, 2006; Braver et al., 2007; Haworth and Mulvihill, 2005; Reeder et al., 1996). However, the effect of riding programs has varied from the successful reduction of road traffic injuries to an increase in such injuries (Kardamanidis et al., 2010). Comprehensive learning steps may start from the learner permit, intermediate, and license courses (Haworth and Mulvihill, 2005; Reeder et al., 1996). After graduating from licensing, riders must be under supervision up to two years depending on regulations, which vary from country to country (Austroads, 1999; Braver et al., 2007). These data were primarily collected in countries that have a small number of motorcycle users, such as Australia, New Zealand, England, and the USA (Kardamanidis et al., 2010). The applicability of the data to developing countries might be limited because of different contextual factors, i.e., a high number of motorcyclists, different road conditions and different degrees of law enforcement. A safety riding program has been implemented in Thailand since 1989. The program has instructed up to 10 million registered motorcyclists. However, there has been no evidence to support the program’s advantages in minimizing injuries or its adverse
116
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Fig. 1. License course practical skills.
consequences. We conducted this study to evaluate the effect of the safety riding program. The results might be useful to inform policy decisions on the prevention of motorcycle crashes in Thailand and developing countries. 2. Materials and methods A retrospective cohort study was conducted in Thailand from December 2007 to June 2009. This analytic study was initially started from two groups of exposed and unexposed risk factors in the past. Then, the outcomes that have already occurred were retrieved (Hennekens and Buring, 1987). We focused only on the Asia-Pacific (AP) Honda Safety Riding Program, which is the best available systematized safety riding program in our setting according to stakeholder perception. Other safety riding programs were provided by the Department of Land Transportation and individual training agents and consisted of 5-h knowledge, a 10-h riding skills course without a riding simulator, a hazard perception test, and teaching skills. The AP Honda Safety Riding Program was established in 1989 and has educated more than 10 million trainees (62.5% of the total number of registered motorcyclists). The multi-facetted courses include 15-min pre-delivery safety advice (P.D.S.A.) for new customers, 2-h instruction and 1-day basic courses for students and general motorcyclists, a 15-h license course for students and general riders, and a 30-h instruction course for trainers and dealer staff. The license course was our particular interest because it permitted riders as young as 15 years old to ride unattended. It was taken into account that the instruction course is perceived as the best available rider training program in Thailand.
Fig. 2. Riding simulators.
After graduation, the riders receive a temporary license. One year later, they automatically receive a full license. 2.1.2. Instruction course This course was only offered to full-license applicants, such as soldiers, policemen, government officers, rider-training instructors and dealer staff. It was aimed to increase teaching abilities, firstaid knowledge, and riding demonstration. All the applicants were trained to detect potential road traffic hazards using riding simulators and instructed in practical riding skills for a total of 30 h (Fig. 2). 2.2. Study population The training group consisted of riders who received certificates from the AP Honda Safety Riding Program in the license or the instruction course between 2005 and 2007 to allow for at least one year’s experience prior to recruitment into the study. Riders aged less than 15 years old at the time of training were excluded because they were not entitled to apply for a rider license. A control group, whose members were at least 15 years old, was matched one-to-one to the training group by region and date of licensure within the same consecutive year. Controls with invalid or unidentified ID numbers from the Department of Land Transport database were excluded. This study was approved by the Institutional Ethics Review Board.
2.1. Safety riding program 2.3. Study factors and outcomes 2.1.1. License course This course provided a 15-h training program for a license permit on a voluntary basis. The applicant should have reached the age of 15 years old. The program consisted of training in standard knowledge and practical skills as determined by the Department of Land Transport. The 5-h knowledge-training segment included traffic law, motorcycle maintenance, riding techniques, safe riding, and hazard perception using riding simulators. The 10-h practical-skills segment (Fig. 1) included preparing, maneuvering and controlling the motorcycle; gear-changing; normal and quickstop braking; negotiating curves, a pylon slalom, and S-, L-, and figure-8 courses; riding on an inclined plane, a bridge, and a narrow road; intersections; balancing; and transporting a passenger.
The study factors were age, gender, region of licensure, date of licensure, history of riding training, date of training, and training course. The outcomes were the number of motorcycle-related injuries, traffic law violations, the date and time of injuries, the injury severity (minimal injury, hospital admission, and intensive care unit admission), and the number of hospital admissions. Additional data, such as riding behavior and self-reported injuries, were collected by mailed questionnaires from the training group before and after training. Motorcycle-related injuries were counted after the participants completed training. That is, the license course started from the date of the license permit, and the instruction course started from the date of training in the training group (Fig. 3).
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
117
2.5. Statistical analysis
License Course Time to RTI
Control group
Analysis Time
Time
Rider license
Training date
Baseline characteristics were described as the mean and standard deviation (SD) or range and percentage for continuous and categorical data, respectively. Student’s t-test or the Wilcoxon rank-sum test for continuous data and the Chi-square test for categorical data were used to compare the groups. The prevalence of motorcycle-related injuries was estimated from the injuries claimed in the RVP and the NHSO data and analyzed by capture–recapture analysis. Based on the assumption that the RVP and NHSO datasets were independent, the number of motorcycle-related injuries retrieved from each dataset (capture1 and capture2) and that were common in both datasets (capture12) were used to calculate the total number of motorcycle-related injuries. The formula is as follows (Chapman, 1951): Total samples =
Instruction Course Time to RTI
Control group
Analysis time
Time
Rider license Fig. 3. Starting date for each type of training.
2.4. Data collection Several methods were employed to collect data for the different sets of variables described above. A standardized questionnaire was used to obtain the following data for the training group: age, gender, details of riding training, region of licensure, date of licensure, age at licensure, ID numbers, self-reported motorcycle-related injuries, i.e., the history and number of motorcycle-related injuries, traffic law violations, the date and time of the injury, the injury severity, the number of hospital admissions, and the number of crashes avoided using safety riding skills. The data extraction from the AP Honda database was performed for 2005–2007 to obtain the training group profiles, i.e., the date of training, the region of training, and the ID numbers. The licensure data were extracted from the Department of Land Transport dataset for both groups for 1988–2007. The region of licensure and the date of licensure of the training group were used to identify the controls from this database. Then, the ID numbers, age, gender, the region of licensure, and the date of licensure of the controls were collected. For the outcomes of interest, data were retrieved from 2 sources. (1) The Road Accident Victims Protection (RVP) Insurance was used for injury-claim data for both groups for 2005–2009. The RVP databases covered 3.6 million injured motorcyclists during the period (30% of all compensation claims for hospitalized subjects with motorcycle-related injuries). (2) The National Health Security Office (NHSO) dataset was used to obtain the claim data of hospitalized patients for 2005–2009. This dataset included patients classified by diagnosis-related group (DRG) according to the International Statistical Classification of Diseases, tenth revision, ICD-10, as V2044-V2999 (motorcycle-related injuries). The NHSO DRG dataset covered 26 million records annually among its 46.8 million registered individuals (70% of the Thai population).
(capture1 + 1) × (capture2 + 1) (capture12+1)
−1
The 95% confidence interval (CI) of the capture–recapture method was also calculated using a standard formula (Seber, 1970). Factors related to motorcycle-related injuries were analyzed using logistic regression. The coefficient of each risk factor was estimated by Wald’s statistics for level of significance and presented as an odds ratio with a 95% confidence interval. An odds ratio of greater than 1 was considered to be a risk factor, whereas an OR of less than 1 was interpreted to be a protective factor. Factors associated with the outcomes at a p-value of less than 0.2 in the univariate analysis were considered to be potential confounding factors and were included in the multivariate analysis. The final or parsimonious model, which provided the maximum power of the model’s predicted value, was chosen from the largest area under the receiver operating characteristic curve (AUC). Goodness of fit for the final model was performed using the Hosmer–Lemeshow test (Hosmer and Lemeshow, 2000). Survival analysis was used to compare the time-to-injury between groups. The Cox proportional hazard ratio with a 95% CI was presented to determine the magnitude of the association. The univariate analysis was performed first. Factors that had p-values < 0.2 were considered to be potential confounding factors and included in the multivariate analysis. The final model was the model that had the largest AUC and satisfied the Hosmer–Lemeshow goodness of fit test. A p-value ≤ 0.05 was considered statistically significant. All the analyses were performed using STATA 10.0, StataCorp, College Station, TX, USA. 2.6. Sample size calculation The estimated sample size was 1627 riders per group. The sample size was calculated assuming alpha = 0.05, the power of the study = 0.8, the expected proportion of injuries in a control group = 0.023 (Ingsathit et al., 2009), and the expected minimal difference of injuries between the training group and the control group = 0.004. However, we rounded up the sample size to 2000 riders per group. 3. Data The total number of subjects for both groups was 6213 with 3250 in the training group and 2963 in the control group. The training group consisted of 2729 trainees from the license course and 521 trainees from the instruction course. The proportion of males and the average age of the control group were significantly higher than in the training group. The baseline characteristics are shown in Table 1.
118
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121 Table 4 Incidence of motorcycle related injury/1000 persons/year.
Table 1 Baseline characteristics. Factors Gender Male Female Age; mean (SD) Region of licensure (%) Bangkok and vicinity Central North Northeast South East
Training group (N = 3250)
Control group (N = 2963)
2025 (62.31) 1225 (37.69) 24.51 (9.55)
1960 (66.15) 1003 (33.85) 28.64 (10.48)
989 (30.43) 304 (9.35) 1130 (34.77) 665 (20.46) 124 (3.82) 38 (1.17)
992 (33.48) 286 (9.65) 940 (31.72) 613 (20.69) 99 (3.34) 33 (1.11)
p-Value
0.002 <0.001 0.074
Study factors
Gender (n = 6213) 201 Male Female 68 Age (years) 25 <18 ≥18 244 Type of training License 22 Instruction 98 149 Control *
Table 2 Capture re-capture analysis of motorcycle related injuries. NHSO capture
RVP capture
Number of injuries (95% CI)
+ + − −
+ − + −
28 52 177 329 (203, 533)
Total motorcycle related injuries
586 (460, 790)
3.1. Prevalence of motorcycle-related injuries In the NHSO database, 80 out of 6213 riders (1.3%) reported motorcycle-related injuries. The types of injury included 23 subjects with fractures or dislocations (28.8%), 17 with head injuries (21.2%), and 40 with soft-tissue injuries (50.0%). The crash mechanism was reported in only 59 out of the 80 motorcycle-related injuries. The crash mechanism of these injuries consisted of a collision with a car, a pickup truck or a van in 23 cases (40.0%); with a motorcycle or a three-wheeled vehicle in 17 cases (28.8%); with a bus or a large truck in 3 cases (5.1%); with a human, an animal, or a bicycle or other vehicle in 9 cases (15.2%); and a single motorcycle crash in 7 cases (11.9%). The RVP database revealed 705 overall injury claims (11.4%) out of a total of 6213 riders in both groups. However, only 205 events (3.3%) involved riders rather than passengers. The most common motorcycle crash was a single motorcycle crash (50.7%), followed by a collision with another vehicle (38.6%) and a collision with an object other than a vehicle (10.7%). After we combined the 205 RVP claims and the 80 NHSO registered cases, involving riders’ injuries were 28 records common in both datasets, 52 records were only captured by the NHSO data, and 177 records were only captured by the RVP data. The injuries that were not captured by the NHSO and RVP datasets were calculated as 329 cases (95% CI: 203–533). The capture-recapture
Number of motorcycle related injuries
Outcomes
NHSO (N = 6213)
RVP (N = 6213)
AP Honda after training (N = 1209)
Motorcycle related injury (%) Traffic law violation (%) Severity of injury (%) Minimal injury Hospital admission Intensive care unit admission Number of hospital admissions, median (range)
80 (1.30) NA
205 (3.30) NA
124 (10.30) 175 (14.5)
52 (65.00) 28 (35.00) 0
103 (50.24) 102 (49.76) 0
1
1
89 (71.77) 33 (26.61) 0 1 (0–3)
NHSO, National Health Security Office; RVP, Road Accident Victims Protection; AP Honda, Asia-Pacific (AP) Honda Safety Riding Program; NA, not available.
Number of motorcycle related injuries/1000 persons/year
p-Value
3984 2228
20 16
0.001*
565 5635
35 18
0.423
521 2728 2963
13 23 19
0.115
p-Value < 0.05.
technique estimated a total of 586 motorcycle-related injuries (Table 2). The prevalence of motorcycle-related injuries was 586 out of 6213 riders (9.4%). The distribution of the number of motorcycle-related injuries, traffic law violations, and the injury severity obtained from each dataset is shown in Table 3. The number of motorcycle-related injuries was highest in the AP Honda training group, whereas the highest percentage of hospital admissions was in the RVP group. Table 4 shows the incidence of motorcycle-related injuries per 1000 individuals per year according to gender and type of training. There was a significant difference between genders. However, the incidence did not differ between subjects according to training type. 4. Results 4.1. Factors related to injuries Of the 80 NHSO registered cases and the 205 RVP claims, 28 records were shared. After excluding these duplicate data, a total of 257 records of motorcycle-related injuries were retrieved for analysis. The factors associated with motorcycle-related injuries are presented in univariate analysis (Table 5). Age and gender were significantly associated with motorcycle-related injuries. Thus, they were included in the multivariate analysis together with the type of training, which is the primary factor of interest. Multivariate analysis (Table 6) showed that male motorcyclists were significantly associated with an increased risk of motorcycle-related injuries (68%) compared with female motorcyclists. Every 1-year incremental change in the age of riders was associated with an approximately 2% decrease in the risk of a motorcycle crash. Table 5 Univariate analysis of factors associated with motorcycle related injuries. Factors
Table 3 Distribution of outcomes of interest among each data set.
Study population
Gender (%) Male Female Age; mean (SD) Type of training (%) License Instruction Control Region of licensure (%) Bangkok and vicinity Central North Northeast South East *
p-Value < 0.05.
Motorcycle related injuries (N = 257)
No injury (N = 5956)
p-Value
193 (4.84) 64 (2.87) 25.30 (8.92)
3792 (95.16) 2164 (97.13) 26.54 (10.26)
<0.001*
99 (3.63) 18 (3.45) 140 (4.72)
2630 (96.37) 503 (96.55) 2823 (95.28)
0.083
82 (4.14) 20 (3.39) 90 (4.35) 55 (4.30) 8 (3.59) 2 (2.82)
1899 (95.86) 570 (96.61) 1980 (95.65) 1223 (95.70) 215 (96.41) 69 (97.18)
0.941
0.031*
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121 Table 6 Factors associated with motorcycle related injuries.
Table 8 Survival analysis.
Factors
Coefficient
Standard error
Adjusted odds ratio (95% CI)
p-Value
Factors
Male Age (years) Type of training License Instruction
0.52 −0.02
0.15 0.01
1.68 (1.24, 2.26) 0.98 (0.97, 0.99)
<0.001* 0.016*
−0.35 −0.35
0.14 0.26
0.70 (0.53–0.92) 0.71 (0.42, 1.18)
0.012* 0.184
Gender 0.44 Male Female Type of training −0.24 Training Controls
*
p-Value < 0.05.
Trainees in the license course exhibited a lower injury rate by 30% with an odds ratio of 0.70 and a 95% CI: 0.53–0.92. The instruction course was associated with a risk reduction of injuries by 29%. However, the result did not reach statistical significance. 4.2. Effects of training in the training group Of the 3250 trainees, 1209 completed the questionnaire (782 and 427 riders from the license and the instruction course, respectively). The average age was 25.3 ± 10.5 years. Nearly 68% were male, and the average age of licensure was 20.6 ± 7.6 years. The prevalence of self-reported motorcycle-related injuries decreased from 263 crashes (21.8%) before training to 124 crashes (10.3%) after training with a p-value less than 0.001 (Table 7). The median number of motorcycle-related injuries was 2 (range 1–24) and 2 (range 1–15), respectively, without a significant difference. After training, the occurrence of reported traffic law violations was approximately 14.5%. The median number of hospital admissions according to motorcycle-related injuries was 1 (range 0–8) before training and 1 (range 0–3) after training. Most subjects in both training groups reported a perceived improvement of safety riding behavior (93.5%) and felt more confidence in riding (96.9%). Nearly half of the trainees (45.9%) reported successful crash avoidance using saferiding techniques. 4.3. Survival analysis Survival analysis was performed to find factors associated with time to have motorcycle-related injuries. Significant factors were male gender and safety riding training, whereas age did not reach statistical significance in the univariate analysis. The final multivariate model is presented in Table 8. Males incurred injuries 55% faster than females with an adjusted hazard ratio (HR) of 1.55 (1.18–2.05) and p-value = 0.002. The time-to-injury was 22% longer in the training group than the control group with an HR of 0.78 (95% CI: 0.62–0.99) and p-value = 0.049.
Table 7 Results of training in the training group. Factors
Safety riding training (N = 1209) Before
Self reported motorcycle related injuries (%) 263 (21.8) Yes 946 (78.2) No 2 (1–24) No. of motorcycle related injuries, median (range) – Traffic law violation (%) No. of hospital 1 (0–8) admission, median (range)
119
p-Value
After 124 (10.3) 1085 (89.7) 2 (1–15)
18 (14.5) 1 (0–3)
<0.001 0.8440
– 0.8066
Coefficient
Standard error
Adjusted hazard ratio (95% CI)
p-Value
0.14
1.55 (1.18–2.05) 1
0.002
0.12
0.78 (0.62–0.99) 1
0.049
5. Discussion The safety riding program aimed to decrease the number of road crashes that involve motorcyclists. In this study, we found a total of 586 motorcycle-related injuries out of 6213 motorcyclists (9.4%) using the capture-recapture estimation technique. The license course was found to significantly minimize motorcyclerelated injuries up to 30% compared with the control group. Other factors associated with motorcycle crashes were younger age and male gender. The incidence of motorcycle-related injuries among riders in this study might be underestimated because the data from the NHSO and the RVP datasets did not include nonhospitalized claims and non-insurance coverage. However, the estimates based on these datasets correlated well with the incidence of motorcycle-related injury in Thailand according to the National Statistical Office of Thailand (http://service.nso.go.th/ nso/nsopublish/pubs/pubsfiles/safeJourney.pdf). 5.1. Effect of the training program We found that the license course could prevent riding injuries with a magnitude comparable to that reported in other studies (Table 9) (Haworth and Mulvihill, 2005). The instruction course was associated with an injury reduction of 29% but without statistical significance. A possible explanation was inadequate power of the study to detect the difference. The details of the training might affect the results regarding riding accidents. AP Honda Safety Riding Program trainees began as early as age 15. The program provides knowledge and practical skills for both license and instructor courses for 15–30 h. A riding simulator was used with each trainee to improve riding and hazard perception skills. After training, all the trainees could ride unattended under general traffic law restrictions. Worldwide training programs provided approximately the same level of knowledge and off-road riding skills in the various steps of graduate licensure. The age permission for training and license application was lower in New Zealand and Thailand (Table 9). However, the early start of training resulted in promising records of 22–30% in injury reduction. Voluntary-basis training most likely resulted in selection bias toward trainees with extra motivation to ride safely. Therefore, the protective effect of the training in the Thailand and New Zealand studies might be overestimated. In contrast, the injury-reduction effect of mandatory training might be confounded by delayed commencement of motorcycle riding (Kardamanidis et al., 2010). The hazard perception skills gained from practicing with a riding simulator might be another key factor to prevent crashes in our context. Given that in developing countries the roads are crowded with motorcyclists who exhibit risky riding behavior and that the roadside environment is frequently dangerous, the risk of motorcycle-related injuries might be increased. Considering these dangerous circumstances, training to improve safety riding skills and decision-making to avoid injuries has been important. It has been demonstrated that using a simulator equipped with a system that visually depicts various traffic environments in motorcycle
120
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Table 9 Safety riding training programs (Haworth and Mulvihill, 2005). Country
Age at training/license (years)
Course
Duration (h)
Restriction during license application
Injuries
Australia
18/18.3
New Zealand
15/16.5
9, 12 6, 8 NA (voluntary)
Size of motorcycle, passenger, alcohol Size of motorcycle, speed, time period, passenger, alcohol
Not improve attitude, riding skills, and safety Decrease by 22%
England USA
17 14/20
15/15
9–12 40 NA NA 15 (voluntary) 30 (voluntary)
Size of motorcycle Age < 20 years, size of motorcycle, time period, route, passenger, alcohol, guardian Size of motorcycle, alcohol Alcohol
Tend to decrease Decrease by 25%
Thailand
Pre-learner License Pre-learner Provisional Full license Full license Pre-learner Intermediate Full license License Trainer
Decrease by 30% Decrease by 29%
NA, not available.
rider training substantially reduced the post-training mortality rate (Yuhara et al., 1993). These strategies might enhance the awareness of traffic conditions, which could result in effective risk perception, improved attitudes toward unsafe riding, and a decrease in risky riding behavior (Elliott et al., 2007; Wong et al., 2010).
5.2. Age Age has been demonstrated to be associated with motorcyclerelated injuries (Mullin et al., 2000; Woratanarat et al., 2009). Young motorcyclists have been shown to exhibit increased injury and mortality rates because of inexperience and immaturity (Wong et al., 2010). Mullin et al. (2000) reported that riders aged 15–19 years incurred twice as many motorcycle-related injuries as riders aged over 25 years. Our data supported that each increase by 1 year of age significantly reduced the risk of incurring an injury by 2%. The provision of a license permit at a younger age results in an early exposure to injury risks. Gregersen and Bjurulf (1996) found a 24% increased mortality rate after the license permit age was lowered from 18 to 16 years. In Thailand, the license permits are granted to riders as young as 15 years. However, our study could not demonstrate significant differences in the number of motorcyclerelated injuries between motorcyclists younger than 18 years and those who were older (p-value = 0.931). This outcome could have resulted from the inclusion of only 565 young motorcyclists in our study. This figure was most likely too small to detect a level of significance.
5.3. Gender Male gender is an important factor that affects road traffic injuries (Woratanarat et al., 2009). Because males tend toward sensation-seeking, amiability and impatience, male riders might be associated with a higher risk of injury (Wong et al., 2010). Our study demonstrated that male motorcyclists displayed a 68% higher risk of incurring injury compared with female motorcyclists. Additionally, according to survival analysis, male riders were injured 1.55 times (HR 1.55, 95% CI: 1.18–2.05, p-value = 0.002) more faster than female riders.
5.4. Strength of this study This study was conducted to assess the best available safety riding training in Thailand that was accompanied by a hazard perception test. The outcome assessors were blind to the evaluation of training courses and did not have any conflicts of interest. Moreover, this study reflected the effectiveness of safety riding training in real-life circumstances.
5.5. Limitation of this study Our study employed retrospective data collection not randomization. Therefore, confounding factors may not be adequately balanced among comparable groups. Most of the trainees were willing to participate in the study. Therefore, they might differ from the general motorcyclists who were selected as controls by a higher concern for safety. Moreover, the 30% reduction in accident risk attributed to the safety riding program could have been overestimated. The trainees in the instructor course included soldiers, policemen, government officers, and dealer staff, who might have been more highly educated and socioeconomically advantaged than the controls. Certain factors related to injuries, such as riding behavior, helmet use, alcohol, psychoactive drug use, educational and socioeconomic level, were not available in our datasets. However, in exposed groups, systematic instruction on safety riding behavior and traffic law might reduce the number of traffic law violations, as found in other studies (Elliott et al., 2007). Only hospitalization data were available from the two health-insurance datasets (RVP and NHSO). Thus, minor injuries were missed. Therefore, our findings on the protective effect of the training course might not be valid for minor injuries. Among the training group members, we could not find significant difference in the number of motorcyclerelated injuries and hospital admission before and after training because the number of injured cases was too small. 5.6. Policy implications Given the finding on the positive association between age and injury, it is imperative that the licensing age should be raised to discourage early motorcycle riding. More emphasis should be placed on male applicants completing the license course because it is evident from this study that males are at an increased risk of injury. Given the limitations of the health-insurance datasets, further studies are required based on a prospective cohort study. In addition, future studies should consider a larger number of subjects to ensure adequate test power. 6. Conclusion Safety riding programs for licensure can significantly reduce the road traffic injuries by 30% in developing countries. Each one-year increase in rider age reduces the risk of motorcycle-related injuries by 2%, whereas male gender increases the risk of incurring such an injury by 68%. Hazard perception skills might be a key to success. Our study supports a policy decision to launch this program nationwide. However, a cost-effectiveness analysis of this safety riding program and cooperation among stakeholders are still required before a policy can be implemented.
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Acknowledgements This study was financially supported by the Thai Health Promotion Foundation and the Road Safety Fund. The authors would like to thank Mr. Suthat Churak and Mr. Pracha Somanus of Asia-Pacific Honda for providing safety riding details and facilitating data collection, Dr. Yolsilp Sucholwanich and Mr. Chaiwat Kiri of the National Health Security Office (NHSO) for data retrieval on diagnosis-related group and hospital charges, Road Accident Victims Protection (RVP) for accident claims, Ms. Supaporn Gerdklai and Ms. Thanaporn Rojruengnond of the Department of Land Transport for licensure data processing, the Department of Probation, Samutprakarn, for providing traffic violation reports, and Ms. Maneemon Prommai for data collection and public relations. References Amin, N.H., Jakoi, A., Katsman, A., Harding, S.P., Tom, J.A., Cerynik, D.L., 2011. Incidence of orthopedic surgery intervention in a level I urban trauma center with motorcycle trauma. Journal of Trauma 71 (4), 948–951. Amoran, O.E., Eme, O., Giwa, O.A., Gbolahan, O.B., 2005. Road safety practices among commercial motorcyclists in a rural town in Nigeria: implications for health education. International Quarterly of Community Health Education 24 (1), 55–64. Austroads, 1999. Investigation of learner driver experience under three driver licensing systems in Australia. In: Austroad Report AP-140, Sydney, NSW. Baldi, S., Baer, J.D., Cook, A.L., 2005. Identifying best practices states in motorcycle rider education and licensing. Journal of Safety Research 36 (1), 19–32. Blanchard, H.T., Tabloski, P.A., 2006. Motorcycle safety: educating riders at the teachable moment. Journal of Emergency Nursing 32 (4), 330–332. Braver, E.R., Kufera, J.A., Volpini, K.D., Lawpoolsri, S., Joyce, J.J., Alexander, M.T., Ellison-Potter, P., 2007. Persuasion and licensure: a randomized controlled intervention trial to increase licensure rates among Maryland motorcycle owners. Traffic Injury Prevention 8 (1), 39–46. Chapman, D.G., 1951. Some Proprieties of the Hyper-geometric Distribution with Applications to Zoological Sample Census, vol. 1, issue 7. University of California Publications in Statistics, Berkeley, pp. 131–160.
121
Elliott, M.A., Baughan, C.J., Sexton, B.F., 2007. Errors and violations in relation to motorcyclists’ crash risk. Accident Analysis and Prevention 39 (3), 491–499. French, M.T., Gumus, G., Homer, J.F., 2009. Public policies and motorcycle safety. Journal of Health Economics 28 (4), 831–838. Gregersen, N.P., Bjurulf, P., 1996. Young novice drivers: towards a model of their accident involvement. Accident Analysis and Prevention 28 (2), 229–241. Haworth, N., Mulvihill, C., 2005. Review of Motorcycle Licensing and Training, vol. 240. Monash University Accident Research Centre, pp. 1–83. Hennekens, C.H., Buring, J., 1987. Epidemiology in Medicine. Williams & Wilkins, Lippincott. Hosmer, D.W., Lemeshow, S., 2000. Applied Logistic Regression. Wiley, New York. Hyder, A.A., Waters, H., Phillips, T., Rehwinkel, J., 2007. Exploring the economics of motorcycle helmet laws – implications for low and middle-income countries. Asia-Pacific Journal of Public Health 19 (2), 16–22. Ingsathit, A., Woratanarat, P., Anukarahanonta, T., Rattanasiri, S., Chatchaipun, P., Wattayakorn, K., Lim, S., Suriyawongpaisal, P., 2009. Prevalence of psychoactive drug use among drivers in Thailand: a roadside survey. Accident Analysis and Prevention 41 (3), 474–478. Kardamanidis, K., Martiniuk, A., Ivers, R.Q., Stevenson, M.R., Thistlethwaite, K., 2010. Motorcycle rider training for the prevention of road traffic crashes. Cochrane Database of Systematic Reviews 10, CD005240. Mullin, B., Jackson, R., Langley, J., Norton, R., 2000. Increasing age and experience: are both protective against motorcycle injury? A case–control study. Injury Prevention 6 (1), 32–35. Reeder, A.I., Chalmers, D.J., Langley, J.D., 1996. Rider training, reasons for riding, and the social context of riding among young on-road motorcyclists in New Zealand. Australian and New Zealand Journal of Public Health 20 (4), 369–374. Seber, G.A.F., 1970. The effect of trap response on tag recapture estimates. Biometrics 26 (57), 13–22. Stella, J., Cooke, C., Sprivulis, P., 2002. Most head injury related motorcycle crash deaths are related to poor riding practices. Emergency Medicine (Fremantle) 14 (1), 58–61. Wong, J.T., Chung, Y.S., Huang, S.H., 2010. Determinants behind young motorcyclists’ risky riding behavior. Accident Analysis and Prevention 42 (1), 275–281. Woratanarat, P., Ingsathit, A., Suriyawongpaisal, P., Rattanasiri, S., Chatchaipun, P., Wattayakorn, K., Anukarahanonta, T., 2009. Alcohol, illicit and non-illicit psychoactive drug use and road traffic injury in Thailand: a case–control study. Accident Analysis and Prevention 41 (3), 651–657. Yuhara, N., Oguchi, Y., Ochiai, H., 1993. Riding simulator in safe driving education – a method to evaluate the effectiveness and the simulated fidelity of necessary functions using fuzzy relation. IATSS Research 17 (1), 54–71.
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Safety riding program and motorcycle-related injuries in Thailand Patarawan Woratanarat a , Atiporn Ingsathit b,∗ , Pornthip Chatchaipan b , Paibul Suriyawongpaisal c a b c
Department of Orthopaedics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand Section of Clinical Epidemiology, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand Department of Community Medicine, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok 10400, Thailand
a r t i c l e
i n f o
Article history: Received 7 February 2013 Received in revised form 30 April 2013 Accepted 2 May 2013 Keywords: Safety riding Motorcycle Injuries Thailand
a b s t r a c t A retrospective cohort study was conducted in Thailand from 2007 to 2009 to evaluate the efficacy of a safety riding program in preventing motorcycle-related injuries. A training group of motorcyclists were certified by the Asia-Pacific Honda Safety Riding Program in either 30-h instruction (teaching skills, riding demonstration) or 15-h license (knowledge, skills, and hazard perception) courses. The control group consisted of untrained motorcyclists matched on an approximately 1:1 ratio with the training group by region and date of licensure. In total, there were 3250 subjects in the training group and 2963 in the control group. Demographic data and factors associated with motorcycle-related injuries were collected. Motorcycle-related injuries were identified using the Road Injuries Victims Protection for injuries claims and inpatient diagnosis-related group datasets from the National Health Security Office. The capture–recapture technique was used to estimate the prevalence of injuries. Multivariate analysis was used to identify factors related to motorcycle-related injuries. The prevalence of motorcycle-related injuries was estimated to be 586 out of 6213 riders (9.4%) with a 95% confidence interval (CI): 460–790. The license course and the instruction course were significantly associated with a 30% and 29% reduction of motorcycle-related injuries, respectively (relative risk 0.70, 95% CI: 0.53–0.92 and 0.71, 95% CI: 0.42–1.18, respectively). Other factors associated with the injuries were male gender and young age. Safety riding training was effective in reducing injuries. These training programs differ from those in other developed countries but display comparable effects. Hazard perception skills might be a key for success. This strategy should be expanded to a national scale. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction Motorcycle-related injuries contribute more than 50% of road traffic injuries in low- and middle-income countries that have a high proportion of motorcycles in road traffic (Hyder et al., 2007). Motorcyclists experienced orthopedic (Amin et al., 2011) and head injuries 20 times as often as four-wheel vehicle drivers (Haworth and Mulvihill, 2005) with a high mortality rate that partially resulted from poor riding practices (Stella et al., 2002). Most motorcyclists start riding at a young age with limited supervision. Immature and risky behavior are most likely the main causes of riding injuries (Wong et al., 2010). The provision of knowledge and riding practice might be important tools to prevent this problem (French et al., 2009).
∗ Corresponding author at: Section of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, 270 Rama VI Road, Rajathewee, Bangkok 10400, Thailand. Tel.: +66 022 011269; fax: +66 022 011284. E-mail address: [email protected] (A. Ingsathit). 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.05.001
Safety riding programs have been established in many countries for decades (Amoran et al., 2005; Baldi et al., 2005; Blanchard and Tabloski, 2006; Braver et al., 2007; Haworth and Mulvihill, 2005; Reeder et al., 1996). However, the effect of riding programs has varied from the successful reduction of road traffic injuries to an increase in such injuries (Kardamanidis et al., 2010). Comprehensive learning steps may start from the learner permit, intermediate, and license courses (Haworth and Mulvihill, 2005; Reeder et al., 1996). After graduating from licensing, riders must be under supervision up to two years depending on regulations, which vary from country to country (Austroads, 1999; Braver et al., 2007). These data were primarily collected in countries that have a small number of motorcycle users, such as Australia, New Zealand, England, and the USA (Kardamanidis et al., 2010). The applicability of the data to developing countries might be limited because of different contextual factors, i.e., a high number of motorcyclists, different road conditions and different degrees of law enforcement. A safety riding program has been implemented in Thailand since 1989. The program has instructed up to 10 million registered motorcyclists. However, there has been no evidence to support the program’s advantages in minimizing injuries or its adverse
116
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Fig. 1. License course practical skills.
consequences. We conducted this study to evaluate the effect of the safety riding program. The results might be useful to inform policy decisions on the prevention of motorcycle crashes in Thailand and developing countries. 2. Materials and methods A retrospective cohort study was conducted in Thailand from December 2007 to June 2009. This analytic study was initially started from two groups of exposed and unexposed risk factors in the past. Then, the outcomes that have already occurred were retrieved (Hennekens and Buring, 1987). We focused only on the Asia-Pacific (AP) Honda Safety Riding Program, which is the best available systematized safety riding program in our setting according to stakeholder perception. Other safety riding programs were provided by the Department of Land Transportation and individual training agents and consisted of 5-h knowledge, a 10-h riding skills course without a riding simulator, a hazard perception test, and teaching skills. The AP Honda Safety Riding Program was established in 1989 and has educated more than 10 million trainees (62.5% of the total number of registered motorcyclists). The multi-facetted courses include 15-min pre-delivery safety advice (P.D.S.A.) for new customers, 2-h instruction and 1-day basic courses for students and general motorcyclists, a 15-h license course for students and general riders, and a 30-h instruction course for trainers and dealer staff. The license course was our particular interest because it permitted riders as young as 15 years old to ride unattended. It was taken into account that the instruction course is perceived as the best available rider training program in Thailand.
Fig. 2. Riding simulators.
After graduation, the riders receive a temporary license. One year later, they automatically receive a full license. 2.1.2. Instruction course This course was only offered to full-license applicants, such as soldiers, policemen, government officers, rider-training instructors and dealer staff. It was aimed to increase teaching abilities, firstaid knowledge, and riding demonstration. All the applicants were trained to detect potential road traffic hazards using riding simulators and instructed in practical riding skills for a total of 30 h (Fig. 2). 2.2. Study population The training group consisted of riders who received certificates from the AP Honda Safety Riding Program in the license or the instruction course between 2005 and 2007 to allow for at least one year’s experience prior to recruitment into the study. Riders aged less than 15 years old at the time of training were excluded because they were not entitled to apply for a rider license. A control group, whose members were at least 15 years old, was matched one-to-one to the training group by region and date of licensure within the same consecutive year. Controls with invalid or unidentified ID numbers from the Department of Land Transport database were excluded. This study was approved by the Institutional Ethics Review Board.
2.1. Safety riding program 2.3. Study factors and outcomes 2.1.1. License course This course provided a 15-h training program for a license permit on a voluntary basis. The applicant should have reached the age of 15 years old. The program consisted of training in standard knowledge and practical skills as determined by the Department of Land Transport. The 5-h knowledge-training segment included traffic law, motorcycle maintenance, riding techniques, safe riding, and hazard perception using riding simulators. The 10-h practical-skills segment (Fig. 1) included preparing, maneuvering and controlling the motorcycle; gear-changing; normal and quickstop braking; negotiating curves, a pylon slalom, and S-, L-, and figure-8 courses; riding on an inclined plane, a bridge, and a narrow road; intersections; balancing; and transporting a passenger.
The study factors were age, gender, region of licensure, date of licensure, history of riding training, date of training, and training course. The outcomes were the number of motorcycle-related injuries, traffic law violations, the date and time of injuries, the injury severity (minimal injury, hospital admission, and intensive care unit admission), and the number of hospital admissions. Additional data, such as riding behavior and self-reported injuries, were collected by mailed questionnaires from the training group before and after training. Motorcycle-related injuries were counted after the participants completed training. That is, the license course started from the date of the license permit, and the instruction course started from the date of training in the training group (Fig. 3).
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
117
2.5. Statistical analysis
License Course Time to RTI
Control group
Analysis Time
Time
Rider license
Training date
Baseline characteristics were described as the mean and standard deviation (SD) or range and percentage for continuous and categorical data, respectively. Student’s t-test or the Wilcoxon rank-sum test for continuous data and the Chi-square test for categorical data were used to compare the groups. The prevalence of motorcycle-related injuries was estimated from the injuries claimed in the RVP and the NHSO data and analyzed by capture–recapture analysis. Based on the assumption that the RVP and NHSO datasets were independent, the number of motorcycle-related injuries retrieved from each dataset (capture1 and capture2) and that were common in both datasets (capture12) were used to calculate the total number of motorcycle-related injuries. The formula is as follows (Chapman, 1951): Total samples =
Instruction Course Time to RTI
Control group
Analysis time
Time
Rider license Fig. 3. Starting date for each type of training.
2.4. Data collection Several methods were employed to collect data for the different sets of variables described above. A standardized questionnaire was used to obtain the following data for the training group: age, gender, details of riding training, region of licensure, date of licensure, age at licensure, ID numbers, self-reported motorcycle-related injuries, i.e., the history and number of motorcycle-related injuries, traffic law violations, the date and time of the injury, the injury severity, the number of hospital admissions, and the number of crashes avoided using safety riding skills. The data extraction from the AP Honda database was performed for 2005–2007 to obtain the training group profiles, i.e., the date of training, the region of training, and the ID numbers. The licensure data were extracted from the Department of Land Transport dataset for both groups for 1988–2007. The region of licensure and the date of licensure of the training group were used to identify the controls from this database. Then, the ID numbers, age, gender, the region of licensure, and the date of licensure of the controls were collected. For the outcomes of interest, data were retrieved from 2 sources. (1) The Road Accident Victims Protection (RVP) Insurance was used for injury-claim data for both groups for 2005–2009. The RVP databases covered 3.6 million injured motorcyclists during the period (30% of all compensation claims for hospitalized subjects with motorcycle-related injuries). (2) The National Health Security Office (NHSO) dataset was used to obtain the claim data of hospitalized patients for 2005–2009. This dataset included patients classified by diagnosis-related group (DRG) according to the International Statistical Classification of Diseases, tenth revision, ICD-10, as V2044-V2999 (motorcycle-related injuries). The NHSO DRG dataset covered 26 million records annually among its 46.8 million registered individuals (70% of the Thai population).
(capture1 + 1) × (capture2 + 1) (capture12+1)
−1
The 95% confidence interval (CI) of the capture–recapture method was also calculated using a standard formula (Seber, 1970). Factors related to motorcycle-related injuries were analyzed using logistic regression. The coefficient of each risk factor was estimated by Wald’s statistics for level of significance and presented as an odds ratio with a 95% confidence interval. An odds ratio of greater than 1 was considered to be a risk factor, whereas an OR of less than 1 was interpreted to be a protective factor. Factors associated with the outcomes at a p-value of less than 0.2 in the univariate analysis were considered to be potential confounding factors and were included in the multivariate analysis. The final or parsimonious model, which provided the maximum power of the model’s predicted value, was chosen from the largest area under the receiver operating characteristic curve (AUC). Goodness of fit for the final model was performed using the Hosmer–Lemeshow test (Hosmer and Lemeshow, 2000). Survival analysis was used to compare the time-to-injury between groups. The Cox proportional hazard ratio with a 95% CI was presented to determine the magnitude of the association. The univariate analysis was performed first. Factors that had p-values < 0.2 were considered to be potential confounding factors and included in the multivariate analysis. The final model was the model that had the largest AUC and satisfied the Hosmer–Lemeshow goodness of fit test. A p-value ≤ 0.05 was considered statistically significant. All the analyses were performed using STATA 10.0, StataCorp, College Station, TX, USA. 2.6. Sample size calculation The estimated sample size was 1627 riders per group. The sample size was calculated assuming alpha = 0.05, the power of the study = 0.8, the expected proportion of injuries in a control group = 0.023 (Ingsathit et al., 2009), and the expected minimal difference of injuries between the training group and the control group = 0.004. However, we rounded up the sample size to 2000 riders per group. 3. Data The total number of subjects for both groups was 6213 with 3250 in the training group and 2963 in the control group. The training group consisted of 2729 trainees from the license course and 521 trainees from the instruction course. The proportion of males and the average age of the control group were significantly higher than in the training group. The baseline characteristics are shown in Table 1.
118
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121 Table 4 Incidence of motorcycle related injury/1000 persons/year.
Table 1 Baseline characteristics. Factors Gender Male Female Age; mean (SD) Region of licensure (%) Bangkok and vicinity Central North Northeast South East
Training group (N = 3250)
Control group (N = 2963)
2025 (62.31) 1225 (37.69) 24.51 (9.55)
1960 (66.15) 1003 (33.85) 28.64 (10.48)
989 (30.43) 304 (9.35) 1130 (34.77) 665 (20.46) 124 (3.82) 38 (1.17)
992 (33.48) 286 (9.65) 940 (31.72) 613 (20.69) 99 (3.34) 33 (1.11)
p-Value
0.002 <0.001 0.074
Study factors
Gender (n = 6213) 201 Male Female 68 Age (years) 25 <18 ≥18 244 Type of training License 22 Instruction 98 149 Control *
Table 2 Capture re-capture analysis of motorcycle related injuries. NHSO capture
RVP capture
Number of injuries (95% CI)
+ + − −
+ − + −
28 52 177 329 (203, 533)
Total motorcycle related injuries
586 (460, 790)
3.1. Prevalence of motorcycle-related injuries In the NHSO database, 80 out of 6213 riders (1.3%) reported motorcycle-related injuries. The types of injury included 23 subjects with fractures or dislocations (28.8%), 17 with head injuries (21.2%), and 40 with soft-tissue injuries (50.0%). The crash mechanism was reported in only 59 out of the 80 motorcycle-related injuries. The crash mechanism of these injuries consisted of a collision with a car, a pickup truck or a van in 23 cases (40.0%); with a motorcycle or a three-wheeled vehicle in 17 cases (28.8%); with a bus or a large truck in 3 cases (5.1%); with a human, an animal, or a bicycle or other vehicle in 9 cases (15.2%); and a single motorcycle crash in 7 cases (11.9%). The RVP database revealed 705 overall injury claims (11.4%) out of a total of 6213 riders in both groups. However, only 205 events (3.3%) involved riders rather than passengers. The most common motorcycle crash was a single motorcycle crash (50.7%), followed by a collision with another vehicle (38.6%) and a collision with an object other than a vehicle (10.7%). After we combined the 205 RVP claims and the 80 NHSO registered cases, involving riders’ injuries were 28 records common in both datasets, 52 records were only captured by the NHSO data, and 177 records were only captured by the RVP data. The injuries that were not captured by the NHSO and RVP datasets were calculated as 329 cases (95% CI: 203–533). The capture-recapture
Number of motorcycle related injuries
Outcomes
NHSO (N = 6213)
RVP (N = 6213)
AP Honda after training (N = 1209)
Motorcycle related injury (%) Traffic law violation (%) Severity of injury (%) Minimal injury Hospital admission Intensive care unit admission Number of hospital admissions, median (range)
80 (1.30) NA
205 (3.30) NA
124 (10.30) 175 (14.5)
52 (65.00) 28 (35.00) 0
103 (50.24) 102 (49.76) 0
1
1
89 (71.77) 33 (26.61) 0 1 (0–3)
NHSO, National Health Security Office; RVP, Road Accident Victims Protection; AP Honda, Asia-Pacific (AP) Honda Safety Riding Program; NA, not available.
Number of motorcycle related injuries/1000 persons/year
p-Value
3984 2228
20 16
0.001*
565 5635
35 18
0.423
521 2728 2963
13 23 19
0.115
p-Value < 0.05.
technique estimated a total of 586 motorcycle-related injuries (Table 2). The prevalence of motorcycle-related injuries was 586 out of 6213 riders (9.4%). The distribution of the number of motorcycle-related injuries, traffic law violations, and the injury severity obtained from each dataset is shown in Table 3. The number of motorcycle-related injuries was highest in the AP Honda training group, whereas the highest percentage of hospital admissions was in the RVP group. Table 4 shows the incidence of motorcycle-related injuries per 1000 individuals per year according to gender and type of training. There was a significant difference between genders. However, the incidence did not differ between subjects according to training type. 4. Results 4.1. Factors related to injuries Of the 80 NHSO registered cases and the 205 RVP claims, 28 records were shared. After excluding these duplicate data, a total of 257 records of motorcycle-related injuries were retrieved for analysis. The factors associated with motorcycle-related injuries are presented in univariate analysis (Table 5). Age and gender were significantly associated with motorcycle-related injuries. Thus, they were included in the multivariate analysis together with the type of training, which is the primary factor of interest. Multivariate analysis (Table 6) showed that male motorcyclists were significantly associated with an increased risk of motorcycle-related injuries (68%) compared with female motorcyclists. Every 1-year incremental change in the age of riders was associated with an approximately 2% decrease in the risk of a motorcycle crash. Table 5 Univariate analysis of factors associated with motorcycle related injuries. Factors
Table 3 Distribution of outcomes of interest among each data set.
Study population
Gender (%) Male Female Age; mean (SD) Type of training (%) License Instruction Control Region of licensure (%) Bangkok and vicinity Central North Northeast South East *
p-Value < 0.05.
Motorcycle related injuries (N = 257)
No injury (N = 5956)
p-Value
193 (4.84) 64 (2.87) 25.30 (8.92)
3792 (95.16) 2164 (97.13) 26.54 (10.26)
<0.001*
99 (3.63) 18 (3.45) 140 (4.72)
2630 (96.37) 503 (96.55) 2823 (95.28)
0.083
82 (4.14) 20 (3.39) 90 (4.35) 55 (4.30) 8 (3.59) 2 (2.82)
1899 (95.86) 570 (96.61) 1980 (95.65) 1223 (95.70) 215 (96.41) 69 (97.18)
0.941
0.031*
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121 Table 6 Factors associated with motorcycle related injuries.
Table 8 Survival analysis.
Factors
Coefficient
Standard error
Adjusted odds ratio (95% CI)
p-Value
Factors
Male Age (years) Type of training License Instruction
0.52 −0.02
0.15 0.01
1.68 (1.24, 2.26) 0.98 (0.97, 0.99)
<0.001* 0.016*
−0.35 −0.35
0.14 0.26
0.70 (0.53–0.92) 0.71 (0.42, 1.18)
0.012* 0.184
Gender 0.44 Male Female Type of training −0.24 Training Controls
*
p-Value < 0.05.
Trainees in the license course exhibited a lower injury rate by 30% with an odds ratio of 0.70 and a 95% CI: 0.53–0.92. The instruction course was associated with a risk reduction of injuries by 29%. However, the result did not reach statistical significance. 4.2. Effects of training in the training group Of the 3250 trainees, 1209 completed the questionnaire (782 and 427 riders from the license and the instruction course, respectively). The average age was 25.3 ± 10.5 years. Nearly 68% were male, and the average age of licensure was 20.6 ± 7.6 years. The prevalence of self-reported motorcycle-related injuries decreased from 263 crashes (21.8%) before training to 124 crashes (10.3%) after training with a p-value less than 0.001 (Table 7). The median number of motorcycle-related injuries was 2 (range 1–24) and 2 (range 1–15), respectively, without a significant difference. After training, the occurrence of reported traffic law violations was approximately 14.5%. The median number of hospital admissions according to motorcycle-related injuries was 1 (range 0–8) before training and 1 (range 0–3) after training. Most subjects in both training groups reported a perceived improvement of safety riding behavior (93.5%) and felt more confidence in riding (96.9%). Nearly half of the trainees (45.9%) reported successful crash avoidance using saferiding techniques. 4.3. Survival analysis Survival analysis was performed to find factors associated with time to have motorcycle-related injuries. Significant factors were male gender and safety riding training, whereas age did not reach statistical significance in the univariate analysis. The final multivariate model is presented in Table 8. Males incurred injuries 55% faster than females with an adjusted hazard ratio (HR) of 1.55 (1.18–2.05) and p-value = 0.002. The time-to-injury was 22% longer in the training group than the control group with an HR of 0.78 (95% CI: 0.62–0.99) and p-value = 0.049.
Table 7 Results of training in the training group. Factors
Safety riding training (N = 1209) Before
Self reported motorcycle related injuries (%) 263 (21.8) Yes 946 (78.2) No 2 (1–24) No. of motorcycle related injuries, median (range) – Traffic law violation (%) No. of hospital 1 (0–8) admission, median (range)
119
p-Value
After 124 (10.3) 1085 (89.7) 2 (1–15)
18 (14.5) 1 (0–3)
<0.001 0.8440
– 0.8066
Coefficient
Standard error
Adjusted hazard ratio (95% CI)
p-Value
0.14
1.55 (1.18–2.05) 1
0.002
0.12
0.78 (0.62–0.99) 1
0.049
5. Discussion The safety riding program aimed to decrease the number of road crashes that involve motorcyclists. In this study, we found a total of 586 motorcycle-related injuries out of 6213 motorcyclists (9.4%) using the capture-recapture estimation technique. The license course was found to significantly minimize motorcyclerelated injuries up to 30% compared with the control group. Other factors associated with motorcycle crashes were younger age and male gender. The incidence of motorcycle-related injuries among riders in this study might be underestimated because the data from the NHSO and the RVP datasets did not include nonhospitalized claims and non-insurance coverage. However, the estimates based on these datasets correlated well with the incidence of motorcycle-related injury in Thailand according to the National Statistical Office of Thailand (http://service.nso.go.th/ nso/nsopublish/pubs/pubsfiles/safeJourney.pdf). 5.1. Effect of the training program We found that the license course could prevent riding injuries with a magnitude comparable to that reported in other studies (Table 9) (Haworth and Mulvihill, 2005). The instruction course was associated with an injury reduction of 29% but without statistical significance. A possible explanation was inadequate power of the study to detect the difference. The details of the training might affect the results regarding riding accidents. AP Honda Safety Riding Program trainees began as early as age 15. The program provides knowledge and practical skills for both license and instructor courses for 15–30 h. A riding simulator was used with each trainee to improve riding and hazard perception skills. After training, all the trainees could ride unattended under general traffic law restrictions. Worldwide training programs provided approximately the same level of knowledge and off-road riding skills in the various steps of graduate licensure. The age permission for training and license application was lower in New Zealand and Thailand (Table 9). However, the early start of training resulted in promising records of 22–30% in injury reduction. Voluntary-basis training most likely resulted in selection bias toward trainees with extra motivation to ride safely. Therefore, the protective effect of the training in the Thailand and New Zealand studies might be overestimated. In contrast, the injury-reduction effect of mandatory training might be confounded by delayed commencement of motorcycle riding (Kardamanidis et al., 2010). The hazard perception skills gained from practicing with a riding simulator might be another key factor to prevent crashes in our context. Given that in developing countries the roads are crowded with motorcyclists who exhibit risky riding behavior and that the roadside environment is frequently dangerous, the risk of motorcycle-related injuries might be increased. Considering these dangerous circumstances, training to improve safety riding skills and decision-making to avoid injuries has been important. It has been demonstrated that using a simulator equipped with a system that visually depicts various traffic environments in motorcycle
120
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Table 9 Safety riding training programs (Haworth and Mulvihill, 2005). Country
Age at training/license (years)
Course
Duration (h)
Restriction during license application
Injuries
Australia
18/18.3
New Zealand
15/16.5
9, 12 6, 8 NA (voluntary)
Size of motorcycle, passenger, alcohol Size of motorcycle, speed, time period, passenger, alcohol
Not improve attitude, riding skills, and safety Decrease by 22%
England USA
17 14/20
15/15
9–12 40 NA NA 15 (voluntary) 30 (voluntary)
Size of motorcycle Age < 20 years, size of motorcycle, time period, route, passenger, alcohol, guardian Size of motorcycle, alcohol Alcohol
Tend to decrease Decrease by 25%
Thailand
Pre-learner License Pre-learner Provisional Full license Full license Pre-learner Intermediate Full license License Trainer
Decrease by 30% Decrease by 29%
NA, not available.
rider training substantially reduced the post-training mortality rate (Yuhara et al., 1993). These strategies might enhance the awareness of traffic conditions, which could result in effective risk perception, improved attitudes toward unsafe riding, and a decrease in risky riding behavior (Elliott et al., 2007; Wong et al., 2010).
5.2. Age Age has been demonstrated to be associated with motorcyclerelated injuries (Mullin et al., 2000; Woratanarat et al., 2009). Young motorcyclists have been shown to exhibit increased injury and mortality rates because of inexperience and immaturity (Wong et al., 2010). Mullin et al. (2000) reported that riders aged 15–19 years incurred twice as many motorcycle-related injuries as riders aged over 25 years. Our data supported that each increase by 1 year of age significantly reduced the risk of incurring an injury by 2%. The provision of a license permit at a younger age results in an early exposure to injury risks. Gregersen and Bjurulf (1996) found a 24% increased mortality rate after the license permit age was lowered from 18 to 16 years. In Thailand, the license permits are granted to riders as young as 15 years. However, our study could not demonstrate significant differences in the number of motorcyclerelated injuries between motorcyclists younger than 18 years and those who were older (p-value = 0.931). This outcome could have resulted from the inclusion of only 565 young motorcyclists in our study. This figure was most likely too small to detect a level of significance.
5.3. Gender Male gender is an important factor that affects road traffic injuries (Woratanarat et al., 2009). Because males tend toward sensation-seeking, amiability and impatience, male riders might be associated with a higher risk of injury (Wong et al., 2010). Our study demonstrated that male motorcyclists displayed a 68% higher risk of incurring injury compared with female motorcyclists. Additionally, according to survival analysis, male riders were injured 1.55 times (HR 1.55, 95% CI: 1.18–2.05, p-value = 0.002) more faster than female riders.
5.4. Strength of this study This study was conducted to assess the best available safety riding training in Thailand that was accompanied by a hazard perception test. The outcome assessors were blind to the evaluation of training courses and did not have any conflicts of interest. Moreover, this study reflected the effectiveness of safety riding training in real-life circumstances.
5.5. Limitation of this study Our study employed retrospective data collection not randomization. Therefore, confounding factors may not be adequately balanced among comparable groups. Most of the trainees were willing to participate in the study. Therefore, they might differ from the general motorcyclists who were selected as controls by a higher concern for safety. Moreover, the 30% reduction in accident risk attributed to the safety riding program could have been overestimated. The trainees in the instructor course included soldiers, policemen, government officers, and dealer staff, who might have been more highly educated and socioeconomically advantaged than the controls. Certain factors related to injuries, such as riding behavior, helmet use, alcohol, psychoactive drug use, educational and socioeconomic level, were not available in our datasets. However, in exposed groups, systematic instruction on safety riding behavior and traffic law might reduce the number of traffic law violations, as found in other studies (Elliott et al., 2007). Only hospitalization data were available from the two health-insurance datasets (RVP and NHSO). Thus, minor injuries were missed. Therefore, our findings on the protective effect of the training course might not be valid for minor injuries. Among the training group members, we could not find significant difference in the number of motorcyclerelated injuries and hospital admission before and after training because the number of injured cases was too small. 5.6. Policy implications Given the finding on the positive association between age and injury, it is imperative that the licensing age should be raised to discourage early motorcycle riding. More emphasis should be placed on male applicants completing the license course because it is evident from this study that males are at an increased risk of injury. Given the limitations of the health-insurance datasets, further studies are required based on a prospective cohort study. In addition, future studies should consider a larger number of subjects to ensure adequate test power. 6. Conclusion Safety riding programs for licensure can significantly reduce the road traffic injuries by 30% in developing countries. Each one-year increase in rider age reduces the risk of motorcycle-related injuries by 2%, whereas male gender increases the risk of incurring such an injury by 68%. Hazard perception skills might be a key to success. Our study supports a policy decision to launch this program nationwide. However, a cost-effectiveness analysis of this safety riding program and cooperation among stakeholders are still required before a policy can be implemented.
P. Woratanarat et al. / Accident Analysis and Prevention 58 (2013) 115–121
Acknowledgements This study was financially supported by the Thai Health Promotion Foundation and the Road Safety Fund. The authors would like to thank Mr. Suthat Churak and Mr. Pracha Somanus of Asia-Pacific Honda for providing safety riding details and facilitating data collection, Dr. Yolsilp Sucholwanich and Mr. Chaiwat Kiri of the National Health Security Office (NHSO) for data retrieval on diagnosis-related group and hospital charges, Road Accident Victims Protection (RVP) for accident claims, Ms. Supaporn Gerdklai and Ms. Thanaporn Rojruengnond of the Department of Land Transport for licensure data processing, the Department of Probation, Samutprakarn, for providing traffic violation reports, and Ms. Maneemon Prommai for data collection and public relations. References Amin, N.H., Jakoi, A., Katsman, A., Harding, S.P., Tom, J.A., Cerynik, D.L., 2011. Incidence of orthopedic surgery intervention in a level I urban trauma center with motorcycle trauma. Journal of Trauma 71 (4), 948–951. Amoran, O.E., Eme, O., Giwa, O.A., Gbolahan, O.B., 2005. Road safety practices among commercial motorcyclists in a rural town in Nigeria: implications for health education. International Quarterly of Community Health Education 24 (1), 55–64. Austroads, 1999. Investigation of learner driver experience under three driver licensing systems in Australia. In: Austroad Report AP-140, Sydney, NSW. Baldi, S., Baer, J.D., Cook, A.L., 2005. Identifying best practices states in motorcycle rider education and licensing. Journal of Safety Research 36 (1), 19–32. Blanchard, H.T., Tabloski, P.A., 2006. Motorcycle safety: educating riders at the teachable moment. Journal of Emergency Nursing 32 (4), 330–332. Braver, E.R., Kufera, J.A., Volpini, K.D., Lawpoolsri, S., Joyce, J.J., Alexander, M.T., Ellison-Potter, P., 2007. Persuasion and licensure: a randomized controlled intervention trial to increase licensure rates among Maryland motorcycle owners. Traffic Injury Prevention 8 (1), 39–46. Chapman, D.G., 1951. Some Proprieties of the Hyper-geometric Distribution with Applications to Zoological Sample Census, vol. 1, issue 7. University of California Publications in Statistics, Berkeley, pp. 131–160.
121
Elliott, M.A., Baughan, C.J., Sexton, B.F., 2007. Errors and violations in relation to motorcyclists’ crash risk. Accident Analysis and Prevention 39 (3), 491–499. French, M.T., Gumus, G., Homer, J.F., 2009. Public policies and motorcycle safety. Journal of Health Economics 28 (4), 831–838. Gregersen, N.P., Bjurulf, P., 1996. Young novice drivers: towards a model of their accident involvement. Accident Analysis and Prevention 28 (2), 229–241. Haworth, N., Mulvihill, C., 2005. Review of Motorcycle Licensing and Training, vol. 240. Monash University Accident Research Centre, pp. 1–83. Hennekens, C.H., Buring, J., 1987. Epidemiology in Medicine. Williams & Wilkins, Lippincott. Hosmer, D.W., Lemeshow, S., 2000. Applied Logistic Regression. Wiley, New York. Hyder, A.A., Waters, H., Phillips, T., Rehwinkel, J., 2007. Exploring the economics of motorcycle helmet laws – implications for low and middle-income countries. Asia-Pacific Journal of Public Health 19 (2), 16–22. Ingsathit, A., Woratanarat, P., Anukarahanonta, T., Rattanasiri, S., Chatchaipun, P., Wattayakorn, K., Lim, S., Suriyawongpaisal, P., 2009. Prevalence of psychoactive drug use among drivers in Thailand: a roadside survey. Accident Analysis and Prevention 41 (3), 474–478. Kardamanidis, K., Martiniuk, A., Ivers, R.Q., Stevenson, M.R., Thistlethwaite, K., 2010. Motorcycle rider training for the prevention of road traffic crashes. Cochrane Database of Systematic Reviews 10, CD005240. Mullin, B., Jackson, R., Langley, J., Norton, R., 2000. Increasing age and experience: are both protective against motorcycle injury? A case–control study. Injury Prevention 6 (1), 32–35. Reeder, A.I., Chalmers, D.J., Langley, J.D., 1996. Rider training, reasons for riding, and the social context of riding among young on-road motorcyclists in New Zealand. Australian and New Zealand Journal of Public Health 20 (4), 369–374. Seber, G.A.F., 1970. The effect of trap response on tag recapture estimates. Biometrics 26 (57), 13–22. Stella, J., Cooke, C., Sprivulis, P., 2002. Most head injury related motorcycle crash deaths are related to poor riding practices. Emergency Medicine (Fremantle) 14 (1), 58–61. Wong, J.T., Chung, Y.S., Huang, S.H., 2010. Determinants behind young motorcyclists’ risky riding behavior. Accident Analysis and Prevention 42 (1), 275–281. Woratanarat, P., Ingsathit, A., Suriyawongpaisal, P., Rattanasiri, S., Chatchaipun, P., Wattayakorn, K., Anukarahanonta, T., 2009. Alcohol, illicit and non-illicit psychoactive drug use and road traffic injury in Thailand: a case–control study. Accident Analysis and Prevention 41 (3), 651–657. Yuhara, N., Oguchi, Y., Ochiai, H., 1993. Riding simulator in safe driving education – a method to evaluate the effectiveness and the simulated fidelity of necessary functions using fuzzy relation. IATSS Research 17 (1), 54–71.